blob: 3056ea09f62a4fea29a554816ecdb9cf06b98a12 [file] [log] [blame]
Willy Tarreau6a06a402007-07-15 20:15:28 +02001 ----------------------
Willy Tarreau8317b282014-04-23 01:49:41 +02002 HAProxy
Willy Tarreau6a06a402007-07-15 20:15:28 +02003 Configuration Manual
4 ----------------------
Willy Tarreau1db55792020-11-05 17:20:35 +01005 version 2.4
Christopher Faulet9d532c42022-04-29 16:04:08 +02006 2022/04/29
Willy Tarreau6a06a402007-07-15 20:15:28 +02007
8
9This document covers the configuration language as implemented in the version
Davor Ocelice9ed2812017-12-25 17:49:28 +010010specified above. It does not provide any hints, examples, or advice. For such
Willy Tarreau0ba27502007-12-24 16:55:16 +010011documentation, please refer to the Reference Manual or the Architecture Manual.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012The summary below is meant to help you find sections by name and navigate
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013through the document.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015Note to documentation contributors :
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016 This document is formatted with 80 columns per line, with even number of
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017 spaces for indentation and without tabs. Please follow these rules strictly
18 so that it remains easily printable everywhere. If a line needs to be
19 printed verbatim and does not fit, please end each line with a backslash
Willy Tarreau62a36c42010-08-17 15:53:10 +020020 ('\') and continue on next line, indented by two characters. It is also
Davor Ocelice9ed2812017-12-25 17:49:28 +010021 sometimes useful to prefix all output lines (logs, console outputs) with 3
22 closing angle brackets ('>>>') in order to emphasize the difference between
23 inputs and outputs when they may be ambiguous. If you add sections,
Willy Tarreau62a36c42010-08-17 15:53:10 +020024 please update the summary below for easier searching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020025
26
27Summary
28-------
29
301. Quick reminder about HTTP
311.1. The HTTP transaction model
321.2. HTTP request
Davor Ocelice9ed2812017-12-25 17:49:28 +0100331.2.1. The request line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200341.2.2. The request headers
351.3. HTTP response
Davor Ocelice9ed2812017-12-25 17:49:28 +0100361.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200371.3.2. The response headers
38
392. Configuring HAProxy
402.1. Configuration file format
William Lallemandf9873ba2015-05-05 17:37:14 +0200412.2. Quoting and escaping
William Lallemandb2f07452015-05-12 14:27:13 +0200422.3. Environment variables
Willy Tarreau4b103022021-02-12 17:59:10 +0100432.4. Conditional blocks
442.5. Time format
452.6. Examples
Willy Tarreauc57f0e22009-05-10 13:12:33 +020046
473. Global parameters
483.1. Process management and security
493.2. Performance tuning
503.3. Debugging
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +0100513.4. Userlists
Cyril Bontédc4d9032012-04-08 21:57:39 +0200523.5. Peers
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200533.6. Mailers
William Lallemandc9515522019-06-12 16:32:11 +0200543.7. Programs
Christopher Faulet76edc0f2020-01-13 15:52:01 +0100553.8. HTTP-errors
Emeric Brun99c453d2020-05-25 15:01:04 +0200563.9. Rings
William Lallemand0217b7b2020-11-18 10:41:24 +0100573.10. Log forwarding
Willy Tarreauc57f0e22009-05-10 13:12:33 +020058
594. Proxies
604.1. Proxy keywords matrix
614.2. Alphabetically sorted keywords reference
62
Davor Ocelice9ed2812017-12-25 17:49:28 +0100635. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200645.1. Bind options
655.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200665.3. Server DNS resolution
675.3.1. Global overview
685.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020069
Julien Pivotto6ccee412019-11-27 15:49:54 +0100706. Cache
716.1. Limitation
726.2. Setup
736.2.1. Cache section
746.2.2. Proxy section
75
Willy Tarreau74ca5042013-06-11 23:12:07 +0200767. Using ACLs and fetching samples
777.1. ACL basics
787.1.1. Matching booleans
797.1.2. Matching integers
807.1.3. Matching strings
817.1.4. Matching regular expressions (regexes)
827.1.5. Matching arbitrary data blocks
837.1.6. Matching IPv4 and IPv6 addresses
847.2. Using ACLs to form conditions
857.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200867.3.1. Converters
877.3.2. Fetching samples from internal states
887.3.3. Fetching samples at Layer 4
897.3.4. Fetching samples at Layer 5
907.3.5. Fetching samples from buffer contents (Layer 6)
917.3.6. Fetching HTTP samples (Layer 7)
Christopher Faulete596d182020-05-05 17:46:34 +0200927.3.7. Fetching samples for developers
Willy Tarreau74ca5042013-06-11 23:12:07 +0200937.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020094
958. Logging
968.1. Log levels
978.2. Log formats
988.2.1. Default log format
998.2.2. TCP log format
1008.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +01001018.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +01001028.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001038.3. Advanced logging options
1048.3.1. Disabling logging of external tests
1058.3.2. Logging before waiting for the session to terminate
1068.3.3. Raising log level upon errors
1078.3.4. Disabling logging of successful connections
1088.4. Timing events
1098.5. Session state at disconnection
1108.6. Non-printable characters
1118.7. Capturing HTTP cookies
1128.8. Capturing HTTP headers
1138.9. Examples of logs
114
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001159. Supported filters
1169.1. Trace
1179.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001189.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001199.4. Cache
Christopher Fauletb30b3102019-09-12 23:03:09 +02001209.5. fcgi-app
Miroslav Zagoracdc32cd92020-12-13 18:32:57 +01001219.6. OpenTracing
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200122
Christopher Fauletb30b3102019-09-12 23:03:09 +020012310. FastCGI applications
12410.1. Setup
12510.1.1. Fcgi-app section
12610.1.2. Proxy section
12710.1.3. Example
12810.2. Default parameters
12910.3. Limitations
130
Emeric Brunce325c42021-04-02 17:05:09 +020013111. Address formats
13211.1. Address family prefixes
13311.2. Socket type prefixes
13411.3. Protocol prefixes
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200135
1361. Quick reminder about HTTP
137----------------------------
138
Davor Ocelice9ed2812017-12-25 17:49:28 +0100139When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200140fully analyzed and indexed, thus it becomes possible to build matching criteria
141on almost anything found in the contents.
142
143However, it is important to understand how HTTP requests and responses are
144formed, and how HAProxy decomposes them. It will then become easier to write
145correct rules and to debug existing configurations.
146
147
1481.1. The HTTP transaction model
149-------------------------------
150
151The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100152to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100153from the client to the server, a request is sent by the client through the
154connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200155will involve a new connection :
156
157 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
158
159In this mode, called the "HTTP close" mode, there are as many connection
160establishments as there are HTTP transactions. Since the connection is closed
161by the server after the response, the client does not need to know the content
162length.
163
164Due to the transactional nature of the protocol, it was possible to improve it
165to avoid closing a connection between two subsequent transactions. In this mode
166however, it is mandatory that the server indicates the content length for each
167response so that the client does not wait indefinitely. For this, a special
168header is used: "Content-length". This mode is called the "keep-alive" mode :
169
170 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
171
172Its advantages are a reduced latency between transactions, and less processing
173power required on the server side. It is generally better than the close mode,
174but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200175a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176
Willy Tarreau95c4e142017-11-26 12:18:55 +0100177Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200178keep-alive, but the client does not wait for the first response to send the
179second request. This is useful for fetching large number of images composing a
180page :
181
182 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
183
184This can obviously have a tremendous benefit on performance because the network
185latency is eliminated between subsequent requests. Many HTTP agents do not
186correctly support pipelining since there is no way to associate a response with
187the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100188server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200189
Willy Tarreau95c4e142017-11-26 12:18:55 +0100190The next improvement is the multiplexed mode, as implemented in HTTP/2. This
191time, each transaction is assigned a single stream identifier, and all streams
192are multiplexed over an existing connection. Many requests can be sent in
193parallel by the client, and responses can arrive in any order since they also
194carry the stream identifier.
195
Willy Tarreau70dffda2014-01-30 03:07:23 +0100196By default HAProxy operates in keep-alive mode with regards to persistent
197connections: for each connection it processes each request and response, and
198leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100199start of a new request. When it receives HTTP/2 connections from a client, it
200processes all the requests in parallel and leaves the connection idling,
201waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200202
Christopher Faulet315b39c2018-09-21 16:26:19 +0200203HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100204 - keep alive : all requests and responses are processed (default)
205 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100206 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100207 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200208 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100209
Willy Tarreau95c4e142017-11-26 12:18:55 +0100210
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200211
2121.2. HTTP request
213-----------------
214
215First, let's consider this HTTP request :
216
217 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100218 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200219 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
220 2 Host: www.mydomain.com
221 3 User-agent: my small browser
222 4 Accept: image/jpeg, image/gif
223 5 Accept: image/png
224
225
2261.2.1. The Request line
227-----------------------
228
229Line 1 is the "request line". It is always composed of 3 fields :
230
231 - a METHOD : GET
232 - a URI : /serv/login.php?lang=en&profile=2
233 - a version tag : HTTP/1.1
234
235All of them are delimited by what the standard calls LWS (linear white spaces),
236which are commonly spaces, but can also be tabs or line feeds/carriage returns
237followed by spaces/tabs. The method itself cannot contain any colon (':') and
238is limited to alphabetic letters. All those various combinations make it
239desirable that HAProxy performs the splitting itself rather than leaving it to
240the user to write a complex or inaccurate regular expression.
241
242The URI itself can have several forms :
243
244 - A "relative URI" :
245
246 /serv/login.php?lang=en&profile=2
247
248 It is a complete URL without the host part. This is generally what is
249 received by servers, reverse proxies and transparent proxies.
250
251 - An "absolute URI", also called a "URL" :
252
253 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
254
255 It is composed of a "scheme" (the protocol name followed by '://'), a host
256 name or address, optionally a colon (':') followed by a port number, then
257 a relative URI beginning at the first slash ('/') after the address part.
258 This is generally what proxies receive, but a server supporting HTTP/1.1
259 must accept this form too.
260
261 - a star ('*') : this form is only accepted in association with the OPTIONS
262 method and is not relayable. It is used to inquiry a next hop's
263 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100264
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200265 - an address:port combination : 192.168.0.12:80
266 This is used with the CONNECT method, which is used to establish TCP
267 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
268 other protocols too.
269
270In a relative URI, two sub-parts are identified. The part before the question
271mark is called the "path". It is typically the relative path to static objects
272on the server. The part after the question mark is called the "query string".
273It is mostly used with GET requests sent to dynamic scripts and is very
274specific to the language, framework or application in use.
275
Willy Tarreau95c4e142017-11-26 12:18:55 +0100276HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100277assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100278
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200279
2801.2.2. The request headers
281--------------------------
282
283The headers start at the second line. They are composed of a name at the
284beginning of the line, immediately followed by a colon (':'). Traditionally,
285an LWS is added after the colon but that's not required. Then come the values.
286Multiple identical headers may be folded into one single line, delimiting the
287values with commas, provided that their order is respected. This is commonly
288encountered in the "Cookie:" field. A header may span over multiple lines if
289the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
290define a total of 3 values for the "Accept:" header.
291
Davor Ocelice9ed2812017-12-25 17:49:28 +0100292Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200293their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100294"Connection:" header). In HTTP/2, header names are always sent in lower case,
Willy Tarreau253c2512020-07-07 15:55:23 +0200295as can be seen when running in debug mode. Internally, all header names are
296normalized to lower case so that HTTP/1.x and HTTP/2 use the exact same
297representation, and they are sent as-is on the other side. This explains why an
298HTTP/1.x request typed with camel case is delivered in lower case.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200299
300The end of the headers is indicated by the first empty line. People often say
301that it's a double line feed, which is not exact, even if a double line feed
302is one valid form of empty line.
303
304Fortunately, HAProxy takes care of all these complex combinations when indexing
305headers, checking values and counting them, so there is no reason to worry
306about the way they could be written, but it is important not to accuse an
307application of being buggy if it does unusual, valid things.
308
309Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000310 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200311 in the middle of headers by LWS in order to join multi-line headers. This
312 is necessary for proper analysis and helps less capable HTTP parsers to work
313 correctly and not to be fooled by such complex constructs.
314
315
3161.3. HTTP response
317------------------
318
319An HTTP response looks very much like an HTTP request. Both are called HTTP
320messages. Let's consider this HTTP response :
321
322 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100323 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200324 1 HTTP/1.1 200 OK
325 2 Content-length: 350
326 3 Content-Type: text/html
327
Willy Tarreau816b9792009-09-15 21:25:21 +0200328As a special case, HTTP supports so called "Informational responses" as status
329codes 1xx. These messages are special in that they don't convey any part of the
330response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100331continue to post its request for instance. In the case of a status 100 response
332the requested information will be carried by the next non-100 response message
333following the informational one. This implies that multiple responses may be
334sent to a single request, and that this only works when keep-alive is enabled
335(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
336correctly forward and skip them, and only process the next non-100 response. As
337such, these messages are neither logged nor transformed, unless explicitly
338state otherwise. Status 101 messages indicate that the protocol is changing
Daniel Corbett9f0843f2021-05-08 10:50:37 -0400339over the same connection and that HAProxy must switch to tunnel mode, just as
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100340if a CONNECT had occurred. Then the Upgrade header would contain additional
341information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200342
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200343
Davor Ocelice9ed2812017-12-25 17:49:28 +01003441.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200345------------------------
346
347Line 1 is the "response line". It is always composed of 3 fields :
348
349 - a version tag : HTTP/1.1
350 - a status code : 200
351 - a reason : OK
352
353The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100354 - 1xx = informational message to be skipped (e.g. 100, 101)
355 - 2xx = OK, content is following (e.g. 200, 206)
356 - 3xx = OK, no content following (e.g. 302, 304)
357 - 4xx = error caused by the client (e.g. 401, 403, 404)
358 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200359
Lukas Tribus23953682017-04-28 13:24:30 +0000360Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100361"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200362found there, but it's a common practice to respect the well-established
363messages. It can be composed of one or multiple words, such as "OK", "Found",
364or "Authentication Required".
365
Davor Ocelice9ed2812017-12-25 17:49:28 +0100366HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200367
368 Code When / reason
369 200 access to stats page, and when replying to monitoring requests
370 301 when performing a redirection, depending on the configured code
371 302 when performing a redirection, depending on the configured code
372 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100373 307 when performing a redirection, depending on the configured code
374 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200375 400 for an invalid or too large request
376 401 when an authentication is required to perform the action (when
377 accessing the stats page)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200378 403 when a request is forbidden by a "http-request deny" rule
Florian Tham9205fea2020-01-08 13:35:30 +0100379 404 when the requested resource could not be found
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200380 408 when the request timeout strikes before the request is complete
Florian Tham272e29b2020-01-08 10:19:05 +0100381 410 when the requested resource is no longer available and will not
382 be available again
Daniel Corbett9f0843f2021-05-08 10:50:37 -0400383 500 when HAProxy encounters an unrecoverable internal error, such as a
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200384 memory allocation failure, which should never happen
Daniel Corbett9f0843f2021-05-08 10:50:37 -0400385 501 when HAProxy is unable to satisfy a client request because of an
Christopher Faulete095f312020-12-07 11:22:24 +0100386 unsupported feature
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200387 502 when the server returns an empty, invalid or incomplete response, or
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200388 when an "http-response deny" rule blocks the response.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200389 503 when no server was available to handle the request, or in response to
390 monitoring requests which match the "monitor fail" condition
391 504 when the response timeout strikes before the server responds
392
393The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3944.2).
395
396
3971.3.2. The response headers
398---------------------------
399
400Response headers work exactly like request headers, and as such, HAProxy uses
401the same parsing function for both. Please refer to paragraph 1.2.2 for more
402details.
403
404
4052. Configuring HAProxy
406----------------------
407
4082.1. Configuration file format
409------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200410
411HAProxy's configuration process involves 3 major sources of parameters :
412
413 - the arguments from the command-line, which always take precedence
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100414 - the configuration file(s), whose format is described here
Thayne McCombscdbcca92021-01-07 21:24:41 -0700415 - the running process's environment, in case some environment variables are
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100416 explicitly referenced
Willy Tarreau6a06a402007-07-15 20:15:28 +0200417
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100418The configuration file follows a fairly simple hierarchical format which obey
419a few basic rules:
Willy Tarreau0ba27502007-12-24 16:55:16 +0100420
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100421 1. a configuration file is an ordered sequence of statements
422
423 2. a statement is a single non-empty line before any unprotected "#" (hash)
424
425 3. a line is a series of tokens or "words" delimited by unprotected spaces or
426 tab characters
427
428 4. the first word or sequence of words of a line is one of the keywords or
429 keyword sequences listed in this document
430
431 5. all other words are all arguments of the first one, some being well-known
432 keywords listed in this document, others being values, references to other
433 parts of the configuration, or expressions
434
435 6. certain keywords delimit a section inside which only a subset of keywords
436 are supported
437
438 7. a section ends at the end of a file or on a special keyword starting a new
439 section
440
441This is all that is needed to know to write a simple but reliable configuration
442generator, but this is not enough to reliably parse any configuration nor to
443figure how to deal with certain corner cases.
444
445First, there are a few consequences of the rules above. Rule 6 and 7 imply that
446the keywords used to define a new section are valid everywhere and cannot have
447a different meaning in a specific section. These keywords are always a single
448word (as opposed to a sequence of words), and traditionally the section that
449follows them is designated using the same name. For example when speaking about
450the "global section", it designates the section of configuration that follows
451the "global" keyword. This usage is used a lot in error messages to help locate
452the parts that need to be addressed.
453
454A number of sections create an internal object or configuration space, which
455requires to be distinguished from other ones. In this case they will take an
456extra word which will set the name of this particular section. For some of them
457the section name is mandatory. For example "frontend foo" will create a new
458section of type "frontend" named "foo". Usually a name is specific to its
459section and two sections of different types may use the same name, but this is
460not recommended as it tends to complexify configuration management.
461
462A direct consequence of rule 7 is that when multiple files are read at once,
463each of them must start with a new section, and the end of each file will end
464a section. A file cannot contain sub-sections nor end an existing section and
465start a new one.
466
467Rule 1 mentioned that ordering matters. Indeed, some keywords create directives
468that can be repeated multiple times to create ordered sequences of rules to be
469applied in a certain order. For example "tcp-request" can be used to alternate
470"accept" and "reject" rules on varying criteria. As such, a configuration file
471processor must always preserve a section's ordering when editing a file. The
472ordering of sections usually does not matter except for the global section
473which must be placed before other sections, but it may be repeated if needed.
474In addition, some automatic identifiers may automatically be assigned to some
475of the created objects (e.g. proxies), and by reordering sections, their
476identifiers will change. These ones appear in the statistics for example. As
477such, the configuration below will assign "foo" ID number 1 and "bar" ID number
4782, which will be swapped if the two sections are reversed:
479
480 listen foo
481 bind :80
482
483 listen bar
484 bind :81
485
486Another important point is that according to rules 2 and 3 above, empty lines,
487spaces, tabs, and comments following and unprotected "#" character are not part
488of the configuration as they are just used as delimiters. This implies that the
489following configurations are strictly equivalent:
490
491 global#this is the global section
492 daemon#daemonize
493 frontend foo
494 mode http # or tcp
495
496and:
497
498 global
499 daemon
500
501 # this is the public web frontend
502 frontend foo
503 mode http
504
505The common practice is to align to the left only the keyword that initiates a
506new section, and indent (i.e. prepend a tab character or a few spaces) all
507other keywords so that it's instantly visible that they belong to the same
508section (as done in the second example above). Placing comments before a new
509section helps the reader decide if it's the desired one. Leaving a blank line
510at the end of a section also visually helps spotting the end when editing it.
511
512Tabs are very convenient for indent but they do not copy-paste well. If spaces
513are used instead, it is recommended to avoid placing too many (2 to 4) so that
514editing in field doesn't become a burden with limited editors that do not
515support automatic indent.
516
517In the early days it used to be common to see arguments split at fixed tab
518positions because most keywords would not take more than two arguments. With
519modern versions featuring complex expressions this practice does not stand
520anymore, and is not recommended.
521
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200522
William Lallemandf9873ba2015-05-05 17:37:14 +02005232.2. Quoting and escaping
524-------------------------
525
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100526In modern configurations, some arguments require the use of some characters
527that were previously considered as pure delimiters. In order to make this
528possible, HAProxy supports character escaping by prepending a backslash ('\')
529in front of the character to be escaped, weak quoting within double quotes
530('"') and strong quoting within single quotes ("'").
William Lallemandf9873ba2015-05-05 17:37:14 +0200531
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100532This is pretty similar to what is done in a number of programming languages and
533very close to what is commonly encountered in Bourne shell. The principle is
534the following: while the configuration parser cuts the lines into words, it
535also takes care of quotes and backslashes to decide whether a character is a
536delimiter or is the raw representation of this character within the current
537word. The escape character is then removed, the quotes are removed, and the
538remaining word is used as-is as a keyword or argument for example.
William Lallemandf9873ba2015-05-05 17:37:14 +0200539
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100540If a backslash is needed in a word, it must either be escaped using itself
541(i.e. double backslash) or be strongly quoted.
542
543Escaping outside quotes is achieved by preceding a special character by a
544backslash ('\'):
William Lallemandf9873ba2015-05-05 17:37:14 +0200545
546 \ to mark a space and differentiate it from a delimiter
547 \# to mark a hash and differentiate it from a comment
548 \\ to use a backslash
549 \' to use a single quote and differentiate it from strong quoting
550 \" to use a double quote and differentiate it from weak quoting
551
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100552In addition, a few non-printable characters may be emitted using their usual
553C-language representation:
554
555 \n to insert a line feed (LF, character \x0a or ASCII 10 decimal)
556 \r to insert a carriage return (CR, character \x0d or ASCII 13 decimal)
557 \t to insert a tab (character \x09 or ASCII 9 decimal)
558 \xNN to insert character having ASCII code hex NN (e.g \x0a for LF).
559
560Weak quoting is achieved by surrounding double quotes ("") around the character
561or sequence of characters to protect. Weak quoting prevents the interpretation
562of:
William Lallemandf9873ba2015-05-05 17:37:14 +0200563
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100564 space or tab as a word separator
William Lallemandf9873ba2015-05-05 17:37:14 +0200565 ' single quote as a strong quoting delimiter
566 # hash as a comment start
567
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100568Weak quoting permits the interpretation of environment variables (which are not
569evaluated outside of quotes) by preceding them with a dollar sign ('$'). If a
570dollar character is needed inside double quotes, it must be escaped using a
571backslash.
William Lallemandb2f07452015-05-12 14:27:13 +0200572
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100573Strong quoting is achieved by surrounding single quotes ('') around the
574character or sequence of characters to protect. Inside single quotes, nothing
575is interpreted, it's the efficient way to quote regular expressions.
William Lallemandf9873ba2015-05-05 17:37:14 +0200576
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100577As a result, here is the matrix indicating how special characters can be
578entered in different contexts (unprintable characters are replaced with their
579name within angle brackets). Note that some characters that may only be
580represented escaped have no possible representation inside single quotes,
581hence the '-' there:
William Lallemandf9873ba2015-05-05 17:37:14 +0200582
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100583 Character | Unquoted | Weakly quoted | Strongly quoted
584 -----------+---------------+-----------------------------+-----------------
585 <TAB> | \<TAB>, \x09 | "<TAB>", "\<TAB>", "\x09" | '<TAB>'
586 <LF> | \n, \x0a | "\n", "\x0a" | -
587 <CR> | \r, \x0d | "\r", "\x0d" | -
588 <SPC> | \<SPC>, \x20 | "<SPC>", "\<SPC>", "\x20" | '<SPC>'
589 " | \", \x22 | "\"", "\x22" | '"'
590 # | \#, \x23 | "#", "\#", "\x23" | '#'
591 $ | $, \$, \x24 | "\$", "\x24" | '$'
592 ' | \', \x27 | "'", "\'", "\x27" | -
593 \ | \\, \x5c | "\\", "\x5c" | '\'
William Lallemandf9873ba2015-05-05 17:37:14 +0200594
595 Example:
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100596 # those are all strictly equivalent:
William Lallemandf9873ba2015-05-05 17:37:14 +0200597 log-format %{+Q}o\ %t\ %s\ %{-Q}r
598 log-format "%{+Q}o %t %s %{-Q}r"
599 log-format '%{+Q}o %t %s %{-Q}r'
600 log-format "%{+Q}o %t"' %s %{-Q}r'
601 log-format "%{+Q}o %t"' %s'\ %{-Q}r
602
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100603There is one particular case where a second level of quoting or escaping may be
604necessary. Some keywords take arguments within parenthesis, sometimes delimited
605by commas. These arguments are commonly integers or predefined words, but when
606they are arbitrary strings, it may be required to perform a separate level of
607escaping to disambiguate the characters that belong to the argument from the
608characters that are used to delimit the arguments themselves. A pretty common
609case is the "regsub" converter. It takes a regular expression in argument, and
610if a closing parenthesis is needed inside, this one will require to have its
611own quotes.
612
613The keyword argument parser is exactly the same as the top-level one regarding
Thayne McCombsf87e4f72021-10-04 01:02:58 -0600614quotes, except that the \#, \$, and \xNN escapes are not processed. But what is
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500615not always obvious is that the delimiters used inside must first be escaped or
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100616quoted so that they are not resolved at the top level.
617
618Let's take this example making use of the "regsub" converter which takes 3
619arguments, one regular expression, one replacement string and one set of flags:
620
621 # replace all occurrences of "foo" with "blah" in the path:
622 http-request set-path %[path,regsub(foo,blah,g)]
623
624Here no special quoting was necessary. But if now we want to replace either
625"foo" or "bar" with "blah", we'll need the regular expression "(foo|bar)". We
626cannot write:
627
628 http-request set-path %[path,regsub((foo|bar),blah,g)]
629
630because we would like the string to cut like this:
631
632 http-request set-path %[path,regsub((foo|bar),blah,g)]
633 |---------|----|-|
634 arg1 _/ / /
635 arg2 __________/ /
636 arg3 ______________/
637
638but actually what is passed is a string between the opening and closing
639parenthesis then garbage:
640
641 http-request set-path %[path,regsub((foo|bar),blah,g)]
642 |--------|--------|
643 arg1=(foo|bar _/ /
644 trailing garbage _________/
645
646The obvious solution here seems to be that the closing parenthesis needs to be
647quoted, but alone this will not work, because as mentioned above, quotes are
648processed by the top-level parser which will resolve them before processing
649this word:
650
651 http-request set-path %[path,regsub("(foo|bar)",blah,g)]
652 ------------ -------- ----------------------------------
653 word1 word2 word3=%[path,regsub((foo|bar),blah,g)]
654
655So we didn't change anything for the argument parser at the second level which
656still sees a truncated regular expression as the only argument, and garbage at
657the end of the string. By escaping the quotes they will be passed unmodified to
658the second level:
659
660 http-request set-path %[path,regsub(\"(foo|bar)\",blah,g)]
661 ------------ -------- ------------------------------------
662 word1 word2 word3=%[path,regsub("(foo|bar)",blah,g)]
663 |---------||----|-|
664 arg1=(foo|bar) _/ / /
665 arg2=blah ___________/ /
666 arg3=g _______________/
667
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500668Another approach consists in using single quotes outside the whole string and
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100669double quotes inside (so that the double quotes are not stripped again):
670
671 http-request set-path '%[path,regsub("(foo|bar)",blah,g)]'
672 ------------ -------- ----------------------------------
673 word1 word2 word3=%[path,regsub("(foo|bar)",blah,g)]
674 |---------||----|-|
675 arg1=(foo|bar) _/ / /
676 arg2 ___________/ /
677 arg3 _______________/
678
679When using regular expressions, it can happen that the dollar ('$') character
680appears in the expression or that a backslash ('\') is used in the replacement
681string. In this case these ones will also be processed inside the double quotes
682thus single quotes are preferred (or double escaping). Example:
683
684 http-request set-path '%[path,regsub("^/(here)(/|$)","my/\1",g)]'
685 ------------ -------- -----------------------------------------
686 word1 word2 word3=%[path,regsub("^/(here)(/|$)","my/\1",g)]
687 |-------------| |-----||-|
688 arg1=(here)(/|$) _/ / /
689 arg2=my/\1 ________________/ /
690 arg3 ______________________/
691
Daniel Corbett9f0843f2021-05-08 10:50:37 -0400692Remember that backslashes are not escape characters within single quotes and
Thayne McCombsf87e4f72021-10-04 01:02:58 -0600693that the whole word above is already protected against them using the single
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100694quotes. Conversely, if double quotes had been used around the whole expression,
695single the dollar character and the backslashes would have been resolved at top
696level, breaking the argument contents at the second level.
697
Thayne McCombsf87e4f72021-10-04 01:02:58 -0600698Unfortunately, since single quotes can't be escaped inside of strong quoting,
699if you need to include single quotes in your argument, you will need to escape
700or quote them twice. There are a few ways to do this:
701
702 http-request set-var(txn.foo) str("\\'foo\\'")
703 http-request set-var(txn.foo) str(\"\'foo\'\")
704 http-request set-var(txn.foo) str(\\\'foo\\\')
705
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100706When in doubt, simply do not use quotes anywhere, and start to place single or
707double quotes around arguments that require a comma or a closing parenthesis,
Thayne McCombsf87e4f72021-10-04 01:02:58 -0600708and think about escaping these quotes using a backslash if the string contains
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100709a dollar or a backslash. Again, this is pretty similar to what is used under
710a Bourne shell when double-escaping a command passed to "eval". For API writers
711the best is probably to place escaped quotes around each and every argument,
712regardless of their contents. Users will probably find that using single quotes
713around the whole expression and double quotes around each argument provides
714more readable configurations.
William Lallemandf9873ba2015-05-05 17:37:14 +0200715
716
William Lallemandb2f07452015-05-12 14:27:13 +02007172.3. Environment variables
718--------------------------
719
720HAProxy's configuration supports environment variables. Those variables are
721interpreted only within double quotes. Variables are expanded during the
722configuration parsing. Variable names must be preceded by a dollar ("$") and
723optionally enclosed with braces ("{}") similarly to what is done in Bourne
724shell. Variable names can contain alphanumerical characters or the character
Amaury Denoyellefa41cb62020-10-01 14:32:35 +0200725underscore ("_") but should not start with a digit. If the variable contains a
726list of several values separated by spaces, it can be expanded as individual
727arguments by enclosing the variable with braces and appending the suffix '[*]'
728before the closing brace.
William Lallemandb2f07452015-05-12 14:27:13 +0200729
730 Example:
731
732 bind "fd@${FD_APP1}"
733
734 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
735
736 user "$HAPROXY_USER"
737
William Lallemand4d03e432019-06-14 15:35:37 +0200738Some variables are defined by HAProxy, they can be used in the configuration
739file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200740
William Lallemand4d03e432019-06-14 15:35:37 +0200741* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
742 name of the local peer. (See "-L" in the management guide.)
743
744* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
745 separated by semicolons. Can be useful in the case you specified a
746 directory.
747
748* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
749
John Roeslerfb2fce12019-07-10 15:45:51 -0500750* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200751 processes, separated by semicolons.
752
John Roeslerfb2fce12019-07-10 15:45:51 -0500753* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200754 CLI, separated by semicolons.
755
Willy Tarreaua46f1af2021-05-06 10:25:11 +0200756In addition, some pseudo-variables are internally resolved and may be used as
757regular variables. Pseudo-variables always start with a dot ('.'), and are the
758only ones where the dot is permitted. The current list of pseudo-variables is:
759
760* .FILE: the name of the configuration file currently being parsed.
761
762* .LINE: the line number of the configuration file currently being parsed,
763 starting at one.
764
765* .SECTION: the name of the section currently being parsed, or its type if the
766 section doesn't have a name (e.g. "global"), or an empty string before the
767 first section.
768
769These variables are resolved at the location where they are parsed. For example
770if a ".LINE" variable is used in a "log-format" directive located in a defaults
771section, its line number will be resolved before parsing and compiling the
772"log-format" directive, so this same line number will be reused by subsequent
773proxies.
774
775This way it is possible to emit information to help locate a rule in variables,
776logs, error statuses, health checks, header values, or even to use line numbers
777to name some config objects like servers for example.
778
William Lallemand4d03e432019-06-14 15:35:37 +0200779See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200780
Willy Tarreau4b103022021-02-12 17:59:10 +0100781
7822.4. Conditional blocks
783-----------------------
784
785It may sometimes be convenient to be able to conditionally enable or disable
786some arbitrary parts of the configuration, for example to enable/disable SSL or
787ciphers, enable or disable some pre-production listeners without modifying the
788configuration, or adjust the configuration's syntax to support two distinct
789versions of HAProxy during a migration.. HAProxy brings a set of nestable
790preprocessor-like directives which allow to integrate or ignore some blocks of
791text. These directives must be placed on their own line and they act on the
792lines that follow them. Two of them support an expression, the other ones only
793switch to an alternate block or end a current level. The 4 following directives
794are defined to form conditional blocks:
795
796 - .if <condition>
797 - .elif <condition>
798 - .else
799 - .endif
800
801The ".if" directive nests a new level, ".elif" stays at the same level, ".else"
802as well, and ".endif" closes a level. Each ".if" must be terminated by a
803matching ".endif". The ".elif" may only be placed after ".if" or ".elif", and
804there is no limit to the number of ".elif" that may be chained. There may be
805only one ".else" per ".if" and it must always be after the ".if" or the last
806".elif" of a block.
807
808Comments may be placed on the same line if needed after a '#', they will be
809ignored. The directives are tokenized like other configuration directives, and
810as such it is possible to use environment variables in conditions.
811
812The conditions are currently limited to:
813
814 - an empty string, always returns "false"
815 - the integer zero ('0'), always returns "false"
816 - a non-nul integer (e.g. '1'), always returns "true".
Willy Tarreau42ed14b2021-05-06 15:55:14 +0200817 - a predicate optionally followed by argument(s) in parenthesis.
818
819The list of currently supported predicates is the following:
820
821 - defined(<name>) : returns true if an environment variable <name>
822 exists, regardless of its contents
823
Willy Tarreau58ca7062021-05-06 16:34:23 +0200824 - feature(<name>) : returns true if feature <name> is listed as present
825 in the features list reported by "haproxy -vv"
826 (which means a <name> appears after a '+')
827
Willy Tarreau6492e872021-05-06 16:10:09 +0200828 - streq(<str1>,<str2>) : returns true only if the two strings are equal
829 - strneq(<str1>,<str2>) : returns true only if the two strings differ
830
Willy Tarreau0b7c78a2021-05-06 16:53:26 +0200831 - version_atleast(<ver>): returns true if the current haproxy version is
832 at least as recent as <ver> otherwise false. The
833 version syntax is the same as shown by "haproxy -v"
834 and missing components are assumed as being zero.
835
836 - version_before(<ver>) : returns true if the current haproxy version is
837 strictly older than <ver> otherwise false. The
838 version syntax is the same as shown by "haproxy -v"
839 and missing components are assumed as being zero.
840
Willy Tarreau42ed14b2021-05-06 15:55:14 +0200841Example:
Willy Tarreau4b103022021-02-12 17:59:10 +0100842
Willy Tarreau42ed14b2021-05-06 15:55:14 +0200843 .if defined(HAPROXY_MWORKER)
844 listen mwcli_px
845 bind :1111
846 ...
847 .endif
Willy Tarreau4b103022021-02-12 17:59:10 +0100848
Willy Tarreau6492e872021-05-06 16:10:09 +0200849 .if strneq("$SSL_ONLY",yes)
850 bind :80
851 .endif
852
853 .if streq("$WITH_SSL",yes)
Willy Tarreau58ca7062021-05-06 16:34:23 +0200854 .if feature(OPENSSL)
Willy Tarreau6492e872021-05-06 16:10:09 +0200855 bind :443 ssl crt ...
Willy Tarreau58ca7062021-05-06 16:34:23 +0200856 .endif
Willy Tarreau6492e872021-05-06 16:10:09 +0200857 .endif
858
Willy Tarreau0b7c78a2021-05-06 16:53:26 +0200859 .if version_atleast(2.4-dev19)
860 profiling.memory on
861 .endif
862
Willy Tarreau7190b982021-05-07 08:59:50 +0200863Four other directives are provided to report some status:
Willy Tarreau4b103022021-02-12 17:59:10 +0100864
Willy Tarreau7190b982021-05-07 08:59:50 +0200865 - .diag "message" : emit this message only when in diagnostic mode (-dD)
Willy Tarreau4b103022021-02-12 17:59:10 +0100866 - .notice "message" : emit this message at level NOTICE
867 - .warning "message" : emit this message at level WARNING
868 - .alert "message" : emit this message at level ALERT
869
870Messages emitted at level WARNING may cause the process to fail to start if the
871"strict-mode" is enabled. Messages emitted at level ALERT will always cause a
872fatal error. These can be used to detect some inappropriate conditions and
873provide advice to the user.
874
875Example:
876
877 .if "${A}"
878 .if "${B}"
879 .notice "A=1, B=1"
880 .elif "${C}"
881 .notice "A=1, B=0, C=1"
882 .elif "${D}"
883 .warning "A=1, B=0, C=0, D=1"
884 .else
885 .alert "A=1, B=0, C=0, D=0"
886 .endif
887 .else
888 .notice "A=0"
889 .endif
890
Willy Tarreau7190b982021-05-07 08:59:50 +0200891 .diag "WTA/2021-05-07: replace 'redirect' with 'return' after switch to 2.4"
892 http-request redirect location /goaway if ABUSE
893
Willy Tarreau4b103022021-02-12 17:59:10 +0100894
8952.5. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200896----------------
897
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100898Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100899values are generally expressed in milliseconds (unless explicitly stated
900otherwise) but may be expressed in any other unit by suffixing the unit to the
901numeric value. It is important to consider this because it will not be repeated
902for every keyword. Supported units are :
903
904 - us : microseconds. 1 microsecond = 1/1000000 second
905 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
906 - s : seconds. 1s = 1000ms
907 - m : minutes. 1m = 60s = 60000ms
908 - h : hours. 1h = 60m = 3600s = 3600000ms
909 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
910
911
Willy Tarreau4b103022021-02-12 17:59:10 +01009122.6. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200913-------------
914
915 # Simple configuration for an HTTP proxy listening on port 80 on all
916 # interfaces and forwarding requests to a single backend "servers" with a
917 # single server "server1" listening on 127.0.0.1:8000
918 global
919 daemon
920 maxconn 256
921
922 defaults
923 mode http
924 timeout connect 5000ms
925 timeout client 50000ms
926 timeout server 50000ms
927
928 frontend http-in
929 bind *:80
930 default_backend servers
931
932 backend servers
933 server server1 127.0.0.1:8000 maxconn 32
934
935
936 # The same configuration defined with a single listen block. Shorter but
937 # less expressive, especially in HTTP mode.
938 global
939 daemon
940 maxconn 256
941
942 defaults
943 mode http
944 timeout connect 5000ms
945 timeout client 50000ms
946 timeout server 50000ms
947
948 listen http-in
949 bind *:80
950 server server1 127.0.0.1:8000 maxconn 32
951
952
953Assuming haproxy is in $PATH, test these configurations in a shell with:
954
Willy Tarreauccb289d2010-12-11 20:19:38 +0100955 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200956
957
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009583. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200959--------------------
960
961Parameters in the "global" section are process-wide and often OS-specific. They
962are generally set once for all and do not need being changed once correct. Some
963of them have command-line equivalents.
964
965The following keywords are supported in the "global" section :
966
967 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200968 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200969 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200970 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200971 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200972 - daemon
Willy Tarreau8a022d52021-04-27 20:29:11 +0200973 - default-path
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200974 - description
975 - deviceatlas-json-file
976 - deviceatlas-log-level
977 - deviceatlas-separator
978 - deviceatlas-properties-cookie
Amaury Denoyelled2e53cd2021-05-06 16:21:39 +0200979 - expose-experimental-directives
Simon Horman98637e52014-06-20 12:30:16 +0900980 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200981 - gid
982 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100983 - hard-stop-after
Christopher Faulet98fbe952019-07-22 16:18:24 +0200984 - h1-case-adjust
985 - h1-case-adjust-file
Willy Tarreaud96f1122019-12-03 07:07:36 +0100986 - insecure-fork-wanted
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100987 - insecure-setuid-wanted
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100988 - issuers-chain-path
Amaury Denoyelle0ea2c4f2021-07-09 17:14:30 +0200989 - h2-workaround-bogus-websocket-clients
Dragan Dosen13cd54c2020-06-18 18:24:05 +0200990 - localpeer
Willy Tarreau6a06a402007-07-15 20:15:28 +0200991 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200992 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100993 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200994 - lua-load
Thierry Fournier59f11be2020-11-29 00:37:41 +0100995 - lua-load-per-thread
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +0100996 - lua-prepend-path
William Lallemand27edc4b2019-05-07 17:49:33 +0200997 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200998 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200999 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +02001000 - node
Amaury Denoyelle0f50cb92021-03-26 18:50:33 +01001001 - numa-cpu-mapping
Willy Tarreau6a06a402007-07-15 20:15:28 +02001002 - pidfile
Willy Tarreau119e50e2020-05-22 13:53:29 +02001003 - pp2-never-send-local
Willy Tarreau1d549722016-02-16 12:41:57 +01001004 - presetenv
1005 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +02001006 - uid
1007 - ulimit-n
1008 - user
Willy Tarreau636848a2019-04-15 19:38:50 +02001009 - set-dumpable
Willy Tarreau13d2ba22021-03-26 11:38:08 +01001010 - set-var
Willy Tarreau1d549722016-02-16 12:41:57 +01001011 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +02001012 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +02001013 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001014 - ssl-default-bind-ciphersuites
Jerome Magninb203ff62020-04-03 15:28:22 +02001015 - ssl-default-bind-curves
Baptiste Assmann3493d0f2015-10-12 20:21:23 +02001016 - ssl-default-bind-options
1017 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001018 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +02001019 - ssl-default-server-options
1020 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +01001021 - ssl-server-verify
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001022 - ssl-skip-self-issued-ca
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001023 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +01001024 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +01001025 - 51degrees-data-file
1026 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +02001027 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001028 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001029 - wurfl-data-file
1030 - wurfl-information-list
1031 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001032 - wurfl-cache-size
William Dauchy0fec3ab2019-10-27 20:08:11 +01001033 - strict-limits
Willy Tarreaud72758d2010-01-12 10:42:19 +01001034
Willy Tarreau6a06a402007-07-15 20:15:28 +02001035 * Performance tuning
William Dauchy0a8824f2019-10-27 20:08:09 +01001036 - busy-polling
Willy Tarreau1746eec2014-04-25 10:46:47 +02001037 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +02001038 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +02001039 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +01001040 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +01001041 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001042 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +02001043 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +02001044 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +02001045 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +02001046 - maxzlibmem
Willy Tarreau85dd5212022-03-08 10:41:40 +01001047 - no-memory-trimming
Willy Tarreau6a06a402007-07-15 20:15:28 +02001048 - noepoll
1049 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001050 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +02001051 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001052 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001053 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001054 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +01001055 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +02001056 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +02001057 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001058 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +00001059 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001060 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +02001061 - tune.buffers.limit
1062 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +02001063 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +02001064 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +01001065 - tune.comp.maxlevel
Willy Tarreaubc52bec2020-06-18 08:58:47 +02001066 - tune.fd.edge-triggered
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001067 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +02001068 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +02001069 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +01001070 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001071 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +02001072 - tune.http.maxhdr
Willy Tarreau76cc6992020-07-01 18:49:24 +02001073 - tune.idle-pool.shared
Willy Tarreau7e312732014-02-12 16:35:14 +01001074 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001075 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +01001076 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001077 - tune.lua.session-timeout
1078 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001079 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001080 - tune.maxaccept
1081 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +02001082 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +02001083 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001084 - tune.pipesize
Willy Tarreaua8e2d972020-07-01 18:27:16 +02001085 - tune.pool-high-fd-ratio
1086 - tune.pool-low-fd-ratio
Willy Tarreaue803de22010-01-21 17:43:04 +01001087 - tune.rcvbuf.client
1088 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +01001089 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +02001090 - tune.runqueue-depth
Willy Tarreaue7723bd2020-06-24 11:11:02 +02001091 - tune.sched.low-latency
Willy Tarreaue803de22010-01-21 17:43:04 +01001092 - tune.sndbuf.client
1093 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001094 - tune.ssl.cachesize
William Lallemand7d42ef52020-07-06 11:41:30 +02001095 - tune.ssl.keylog
Willy Tarreaubfd59462013-02-21 07:46:09 +01001096 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +02001097 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +01001098 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001099 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +02001100 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001101 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001102 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +01001103 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001104 - tune.vars.reqres-max-size
1105 - tune.vars.sess-max-size
1106 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +01001107 - tune.zlib.memlevel
1108 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +01001109
Willy Tarreau6a06a402007-07-15 20:15:28 +02001110 * Debugging
Willy Tarreau6a06a402007-07-15 20:15:28 +02001111 - quiet
Willy Tarreau3eb10b82020-04-15 16:42:39 +02001112 - zero-warning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001113
1114
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011153.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +02001116------------------------------------
1117
Emeric Brunc8e8d122012-10-02 18:42:10 +02001118ca-base <dir>
1119 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +01001120 relative path is used with "ca-file", "ca-verify-file" or "crl-file"
1121 directives. Absolute locations specified in "ca-file", "ca-verify-file" and
1122 "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +02001123
Willy Tarreau6a06a402007-07-15 20:15:28 +02001124chroot <jail dir>
1125 Changes current directory to <jail dir> and performs a chroot() there before
1126 dropping privileges. This increases the security level in case an unknown
1127 vulnerability would be exploited, since it would make it very hard for the
1128 attacker to exploit the system. This only works when the process is started
1129 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +01001130 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +01001131
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001132cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
1133 On Linux 2.6 and above, it is possible to bind a process or a thread to a
1134 specific CPU set. This means that the process or the thread will never run on
1135 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
1136 sets. The first argument is a process set, eventually followed by a thread
1137 set. These sets have the format
1138
1139 all | odd | even | number[-[number]]
1140
1141 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +01001142 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001143 ignored. It is possible to specify a range with two such number delimited by
1144 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +01001145 "all", only odd numbers using "odd" or even numbers using "even", just like
1146 with the "bind-process" directive. The second and forthcoming arguments are
Amaury Denoyelle982fb532021-04-21 18:39:58 +02001147 CPU sets. Each CPU set is either a unique number starting at 0 for the first
1148 CPU or a range with two such numbers delimited by a dash ('-'). Outside of
1149 Linux and BSDs, there may be a limitation on the maximum CPU index to either
1150 31 or 63. Multiple CPU numbers or ranges may be specified, and the processes
1151 or threads will be allowed to bind to all of them. Obviously, multiple
1152 "cpu-map" directives may be specified. Each "cpu-map" directive will replace
1153 the previous ones when they overlap. A thread will be bound on the
1154 intersection of its mapping and the one of the process on which it is
1155 attached. If the intersection is null, no specific binding will be set for
1156 the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +01001157
Christopher Fauletff4121f2017-11-22 16:38:49 +01001158 Ranges can be partially defined. The higher bound can be omitted. In such
1159 case, it is replaced by the corresponding maximum value, 32 or 64 depending
1160 on the machine's word size.
1161
Christopher Faulet26028f62017-11-22 15:01:51 +01001162 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001163 automatically bind a process or a thread to a CPU by incrementing
1164 process/thread and CPU sets. To be valid, both sets must have the same
1165 size. No matter the declaration order of the CPU sets, it will be bound from
1166 the lowest to the highest bound. Having a process and a thread range with the
1167 "auto:" prefix is not supported. Only one range is supported, the other one
1168 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +01001169
1170 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001171 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
1172
1173 cpu-map 1/all 0-3 # bind all threads of the first process on the
1174 # first 4 CPUs
1175
1176 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
1177 # or "cpu-map 1-32 0-31" depending on the machine's
1178 # word size.
1179
Christopher Faulet26028f62017-11-22 15:01:51 +01001180 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001181 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +01001182 cpu-map auto:1-4 0-3
1183 cpu-map auto:1-4 0-1 2-3
1184 cpu-map auto:1-4 3 2 1 0
1185
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001186 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
1187 # and so on.
1188 cpu-map auto:1/1-4 0-3
1189 cpu-map auto:1/1-4 0-1 2-3
1190 cpu-map auto:1/1-4 3 2 1 0
1191
Davor Ocelice9ed2812017-12-25 17:49:28 +01001192 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +01001193 cpu-map auto:all 0-63
1194 cpu-map auto:even 0-31
1195 cpu-map auto:odd 32-63
1196
1197 # invalid cpu-map because process and CPU sets have different sizes.
1198 cpu-map auto:1-4 0 # invalid
1199 cpu-map auto:1 0-3 # invalid
1200
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001201 # invalid cpu-map because automatic binding is used with a process range
1202 # and a thread range.
1203 cpu-map auto:all/all 0 # invalid
1204 cpu-map auto:all/1-4 0 # invalid
1205 cpu-map auto:1-4/all 0 # invalid
1206
Emeric Brunc8e8d122012-10-02 18:42:10 +02001207crt-base <dir>
1208 Assigns a default directory to fetch SSL certificates from when a relative
William Dauchy238ea3b2020-01-11 13:09:12 +01001209 path is used with "crtfile" or "crt" directives. Absolute locations specified
1210 prevail and ignore "crt-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +02001211
Willy Tarreau6a06a402007-07-15 20:15:28 +02001212daemon
1213 Makes the process fork into background. This is the recommended mode of
1214 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +01001215 disabled by the command line "-db" argument. This option is ignored in
1216 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001217
Willy Tarreau8a022d52021-04-27 20:29:11 +02001218default-path { current | config | parent | origin <path> }
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001219 By default HAProxy loads all files designated by a relative path from the
Willy Tarreau8a022d52021-04-27 20:29:11 +02001220 location the process is started in. In some circumstances it might be
1221 desirable to force all relative paths to start from a different location
1222 just as if the process was started from such locations. This is what this
1223 directive is made for. Technically it will perform a temporary chdir() to
1224 the designated location while processing each configuration file, and will
1225 return to the original directory after processing each file. It takes an
1226 argument indicating the policy to use when loading files whose path does
1227 not start with a slash ('/'):
1228 - "current" indicates that all relative files are to be loaded from the
1229 directory the process is started in ; this is the default.
1230
1231 - "config" indicates that all relative files should be loaded from the
1232 directory containing the configuration file. More specifically, if the
1233 configuration file contains a slash ('/'), the longest part up to the
1234 last slash is used as the directory to change to, otherwise the current
1235 directory is used. This mode is convenient to bundle maps, errorfiles,
1236 certificates and Lua scripts together as relocatable packages. When
1237 multiple configuration files are loaded, the directory is updated for
1238 each of them.
1239
1240 - "parent" indicates that all relative files should be loaded from the
1241 parent of the directory containing the configuration file. More
1242 specifically, if the configuration file contains a slash ('/'), ".."
1243 is appended to the longest part up to the last slash is used as the
1244 directory to change to, otherwise the directory is "..". This mode is
1245 convenient to bundle maps, errorfiles, certificates and Lua scripts
1246 together as relocatable packages, but where each part is located in a
1247 different subdirectory (e.g. "config/", "certs/", "maps/", ...).
1248
1249 - "origin" indicates that all relative files should be loaded from the
1250 designated (mandatory) path. This may be used to ease management of
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001251 different HAProxy instances running in parallel on a system, where each
Willy Tarreau8a022d52021-04-27 20:29:11 +02001252 instance uses a different prefix but where the rest of the sections are
1253 made easily relocatable.
1254
1255 Each "default-path" directive instantly replaces any previous one and will
1256 possibly result in switching to a different directory. While this should
1257 always result in the desired behavior, it is really not a good practice to
1258 use multiple default-path directives, and if used, the policy ought to remain
1259 consistent across all configuration files.
1260
1261 Warning: some configuration elements such as maps or certificates are
1262 uniquely identified by their configured path. By using a relocatable layout,
1263 it becomes possible for several of them to end up with the same unique name,
1264 making it difficult to update them at run time, especially when multiple
1265 configuration files are loaded from different directories. It is essential to
1266 observe a strict collision-free file naming scheme before adopting relative
1267 paths. A robust approach could consist in prefixing all files names with
1268 their respective site name, or in doing so at the directory level.
1269
David Carlier8167f302015-06-01 13:50:06 +02001270deviceatlas-json-file <path>
1271 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001272 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +02001273
1274deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001275 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +02001276 optional and set to 0 by default if not set.
1277
1278deviceatlas-separator <char>
1279 Sets the character separator for the API properties results. This directive
1280 is optional and set to | by default if not set.
1281
Cyril Bonté0306c4a2015-10-26 22:37:38 +01001282deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001283 Sets the client cookie's name used for the detection if the DeviceAtlas
1284 Client-side component was used during the request. This directive is optional
1285 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +01001286
Amaury Denoyelled2e53cd2021-05-06 16:21:39 +02001287expose-experimental-directives
1288 This statement must appear before using directives tagged as experimental or
1289 the config file will be rejected.
1290
Simon Horman98637e52014-06-20 12:30:16 +09001291external-check
Willy Tarreaud96f1122019-12-03 07:07:36 +01001292 Allows the use of an external agent to perform health checks. This is
1293 disabled by default as a security precaution, and even when enabled, checks
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001294 may still fail unless "insecure-fork-wanted" is enabled as well. If the
1295 program launched makes use of a setuid executable (it should really not),
1296 you may also need to set "insecure-setuid-wanted" in the global section.
1297 See "option external-check", and "insecure-fork-wanted", and
1298 "insecure-setuid-wanted".
Simon Horman98637e52014-06-20 12:30:16 +09001299
Willy Tarreau6a06a402007-07-15 20:15:28 +02001300gid <number>
Thayne McCombscdbcca92021-01-07 21:24:41 -07001301 Changes the process's group ID to <number>. It is recommended that the group
Willy Tarreau6a06a402007-07-15 20:15:28 +02001302 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1303 be started with a user belonging to this group, or with superuser privileges.
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001304 Note that if HAProxy is started from a user having supplementary groups, it
Michael Schererab012dd2013-01-12 18:35:19 +01001305 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001306 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +01001307
Willy Tarreau11770ce2019-12-03 08:29:22 +01001308group <group name>
1309 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
1310 See also "gid" and "user".
1311
Cyril Bonté203ec5a2017-03-23 22:44:13 +01001312hard-stop-after <time>
1313 Defines the maximum time allowed to perform a clean soft-stop.
1314
1315 Arguments :
1316 <time> is the maximum time (by default in milliseconds) for which the
1317 instance will remain alive when a soft-stop is received via the
1318 SIGUSR1 signal.
1319
1320 This may be used to ensure that the instance will quit even if connections
1321 remain opened during a soft-stop (for example with long timeouts for a proxy
1322 in tcp mode). It applies both in TCP and HTTP mode.
1323
1324 Example:
1325 global
1326 hard-stop-after 30s
1327
Christopher Faulet98fbe952019-07-22 16:18:24 +02001328h1-case-adjust <from> <to>
1329 Defines the case adjustment to apply, when enabled, to the header name
1330 <from>, to change it to <to> before sending it to HTTP/1 clients or
1331 servers. <from> must be in lower case, and <from> and <to> must not differ
1332 except for their case. It may be repeated if several header names need to be
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05001333 adjusted. Duplicate entries are not allowed. If a lot of header names have to
Christopher Faulet98fbe952019-07-22 16:18:24 +02001334 be adjusted, it might be more convenient to use "h1-case-adjust-file".
1335 Please note that no transformation will be applied unless "option
1336 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
1337 specified in a proxy.
1338
1339 There is no standard case for header names because, as stated in RFC7230,
1340 they are case-insensitive. So applications must handle them in a case-
1341 insensitive manner. But some bogus applications violate the standards and
1342 erroneously rely on the cases most commonly used by browsers. This problem
1343 becomes critical with HTTP/2 because all header names must be exchanged in
1344 lower case, and HAProxy follows the same convention. All header names are
1345 sent in lower case to clients and servers, regardless of the HTTP version.
1346
1347 Applications which fail to properly process requests or responses may require
1348 to temporarily use such workarounds to adjust header names sent to them for
1349 the time it takes the application to be fixed. Please note that an
1350 application which requires such workarounds might be vulnerable to content
1351 smuggling attacks and must absolutely be fixed.
1352
1353 Example:
1354 global
1355 h1-case-adjust content-length Content-Length
1356
1357 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
1358 "option h1-case-adjust-bogus-server".
1359
1360h1-case-adjust-file <hdrs-file>
1361 Defines a file containing a list of key/value pairs used to adjust the case
1362 of some header names before sending them to HTTP/1 clients or servers. The
1363 file <hdrs-file> must contain 2 header names per line. The first one must be
1364 in lower case and both must not differ except for their case. Lines which
1365 start with '#' are ignored, just like empty lines. Leading and trailing tabs
1366 and spaces are stripped. Duplicate entries are not allowed. Please note that
1367 no transformation will be applied unless "option h1-case-adjust-bogus-client"
1368 or "option h1-case-adjust-bogus-server" is specified in a proxy.
1369
1370 If this directive is repeated, only the last one will be processed. It is an
1371 alternative to the directive "h1-case-adjust" if a lot of header names need
1372 to be adjusted. Please read the risks associated with using this.
1373
1374 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
1375 "option h1-case-adjust-bogus-server".
1376
Willy Tarreaud96f1122019-12-03 07:07:36 +01001377insecure-fork-wanted
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001378 By default HAProxy tries hard to prevent any thread and process creation
Willy Tarreaud96f1122019-12-03 07:07:36 +01001379 after it starts. Doing so is particularly important when using Lua files of
1380 uncertain origin, and when experimenting with development versions which may
1381 still contain bugs whose exploitability is uncertain. And generally speaking
1382 it's good hygiene to make sure that no unexpected background activity can be
1383 triggered by traffic. But this prevents external checks from working, and may
1384 break some very specific Lua scripts which actively rely on the ability to
1385 fork. This option is there to disable this protection. Note that it is a bad
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001386 idea to disable it, as a vulnerability in a library or within HAProxy itself
Willy Tarreaud96f1122019-12-03 07:07:36 +01001387 will be easier to exploit once disabled. In addition, forking from Lua or
1388 anywhere else is not reliable as the forked process may randomly embed a lock
1389 set by another thread and never manage to finish an operation. As such it is
1390 highly recommended that this option is never used and that any workload
1391 requiring such a fork be reconsidered and moved to a safer solution (such as
1392 agents instead of external checks). This option supports the "no" prefix to
1393 disable it.
1394
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001395insecure-setuid-wanted
1396 HAProxy doesn't need to call executables at run time (except when using
1397 external checks which are strongly recommended against), and is even expected
1398 to isolate itself into an empty chroot. As such, there basically is no valid
1399 reason to allow a setuid executable to be called without the user being fully
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001400 aware of the risks. In a situation where HAProxy would need to call external
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001401 checks and/or disable chroot, exploiting a vulnerability in a library or in
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001402 HAProxy itself could lead to the execution of an external program. On Linux
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001403 it is possible to lock the process so that any setuid bit present on such an
1404 executable is ignored. This significantly reduces the risk of privilege
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001405 escalation in such a situation. This is what HAProxy does by default. In case
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001406 this causes a problem to an external check (for example one which would need
1407 the "ping" command), then it is possible to disable this protection by
1408 explicitly adding this directive in the global section. If enabled, it is
1409 possible to turn it back off by prefixing it with the "no" keyword.
1410
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +01001411issuers-chain-path <dir>
1412 Assigns a directory to load certificate chain for issuer completion. All
1413 files must be in PEM format. For certificates loaded with "crt" or "crt-list",
1414 if certificate chain is not included in PEM (also commonly known as
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001415 intermediate certificate), HAProxy will complete chain if the issuer of the
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +01001416 certificate corresponds to the first certificate of the chain loaded with
1417 "issuers-chain-path".
1418 A "crt" file with PrivateKey+Certificate+IntermediateCA2+IntermediateCA1
1419 could be replaced with PrivateKey+Certificate. HAProxy will complete the
1420 chain if a file with IntermediateCA2+IntermediateCA1 is present in
1421 "issuers-chain-path" directory. All other certificates with the same issuer
1422 will share the chain in memory.
1423
Amaury Denoyelle0ea2c4f2021-07-09 17:14:30 +02001424h2-workaround-bogus-websocket-clients
1425 This disables the announcement of the support for h2 websockets to clients.
1426 This can be use to overcome clients which have issues when implementing the
1427 relatively fresh RFC8441, such as Firefox 88. To allow clients to
1428 automatically downgrade to http/1.1 for the websocket tunnel, specify h2
1429 support on the bind line using "alpn" without an explicit "proto" keyword. If
1430 this statement was previously activated, this can be disabled by prefixing
1431 the keyword with "no'.
1432
Dragan Dosen13cd54c2020-06-18 18:24:05 +02001433localpeer <name>
1434 Sets the local instance's peer name. It will be ignored if the "-L"
1435 command line argument is specified or if used after "peers" section
1436 definitions. In such cases, a warning message will be emitted during
1437 the configuration parsing.
1438
1439 This option will also set the HAPROXY_LOCALPEER environment variable.
1440 See also "-L" in the management guide and "peers" section below.
1441
Jan Wagner3e678602020-12-17 22:22:32 +01001442log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001443 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +01001444 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +01001445 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +01001446 configured with "log global".
1447
1448 <address> can be one of:
1449
Willy Tarreau2769aa02007-12-27 18:26:09 +01001450 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +01001451 no port is specified, 514 is used by default (the standard syslog
1452 port).
1453
David du Colombier24bb5f52011-03-17 10:40:23 +01001454 - An IPv6 address followed by a colon and optionally a UDP port. If
1455 no port is specified, 514 is used by default (the standard syslog
1456 port).
1457
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001458 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +01001459 considerations for chroot (be sure the path is accessible inside
1460 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +01001461 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +01001462
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001463 - A file descriptor number in the form "fd@<number>", which may point
1464 to a pipe, terminal, or socket. In this case unbuffered logs are used
1465 and one writev() call per log is performed. This is a bit expensive
1466 but acceptable for most workloads. Messages sent this way will not be
1467 truncated but may be dropped, in which case the DroppedLogs counter
1468 will be incremented. The writev() call is atomic even on pipes for
1469 messages up to PIPE_BUF size, which POSIX recommends to be at least
1470 512 and which is 4096 bytes on most modern operating systems. Any
1471 larger message may be interleaved with messages from other processes.
1472 Exceptionally for debugging purposes the file descriptor may also be
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001473 directed to a file, but doing so will significantly slow HAProxy down
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001474 as non-blocking calls will be ignored. Also there will be no way to
1475 purge nor rotate this file without restarting the process. Note that
1476 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +01001477 for use with a TCP syslog server. See also the "short" and "raw"
1478 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001479
1480 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
1481 "fd@2", see above.
1482
Willy Tarreauc046d162019-08-30 15:24:59 +02001483 - A ring buffer in the form "ring@<name>", which will correspond to an
1484 in-memory ring buffer accessible over the CLI using the "show events"
1485 command, which will also list existing rings and their sizes. Such
1486 buffers are lost on reload or restart but when used as a complement
1487 this can help troubleshooting by having the logs instantly available.
1488
William Lallemandb2f07452015-05-12 14:27:13 +02001489 You may want to reference some environment variables in the address
1490 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001491
Willy Tarreau18324f52014-06-27 18:10:07 +02001492 <length> is an optional maximum line length. Log lines larger than this value
1493 will be truncated before being sent. The reason is that syslog
1494 servers act differently on log line length. All servers support the
1495 default value of 1024, but some servers simply drop larger lines
1496 while others do log them. If a server supports long lines, it may
1497 make sense to set this value here in order to avoid truncating long
1498 lines. Similarly, if a server drops long lines, it is preferable to
1499 truncate them before sending them. Accepted values are 80 to 65535
1500 inclusive. The default value of 1024 is generally fine for all
1501 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +01001502 JSON-formatted logs may require larger values. You may also need to
1503 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +02001504
Dragan Dosen7ad31542015-09-28 17:16:47 +02001505 <format> is the log format used when generating syslog messages. It may be
1506 one of the following :
1507
Emeric Brun0237c4e2020-11-27 16:24:34 +01001508 local Analog to rfc3164 syslog message format except that hostname
1509 field is stripped. This is the default.
1510 Note: option "log-send-hostname" switches the default to
1511 rfc3164.
1512
1513 rfc3164 The RFC3164 syslog message format.
Dragan Dosen7ad31542015-09-28 17:16:47 +02001514 (https://tools.ietf.org/html/rfc3164)
1515
1516 rfc5424 The RFC5424 syslog message format.
1517 (https://tools.ietf.org/html/rfc5424)
1518
Emeric Brun54648852020-07-06 15:54:06 +02001519 priority A message containing only a level plus syslog facility between
1520 angle brackets such as '<63>', followed by the text. The PID,
1521 date, time, process name and system name are omitted. This is
1522 designed to be used with a local log server.
1523
Willy Tarreaue8746a02018-11-12 08:45:00 +01001524 short A message containing only a level between angle brackets such as
1525 '<3>', followed by the text. The PID, date, time, process name
1526 and system name are omitted. This is designed to be used with a
1527 local log server. This format is compatible with what the systemd
1528 logger consumes.
1529
Emeric Brun54648852020-07-06 15:54:06 +02001530 timed A message containing only a level between angle brackets such as
1531 '<3>', followed by ISO date and by the text. The PID, process
1532 name and system name are omitted. This is designed to be
1533 used with a local log server.
1534
1535 iso A message containing only the ISO date, followed by the text.
1536 The PID, process name and system name are omitted. This is
1537 designed to be used with a local log server.
1538
Willy Tarreauc1b06452018-11-12 11:57:56 +01001539 raw A message containing only the text. The level, PID, date, time,
1540 process name and system name are omitted. This is designed to be
1541 used in containers or during development, where the severity only
1542 depends on the file descriptor used (stdout/stderr).
1543
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001544 <ranges> A list of comma-separated ranges to identify the logs to sample.
1545 This is used to balance the load of the logs to send to the log
1546 server. The limits of the ranges cannot be null. They are numbered
1547 from 1. The size or period (in number of logs) of the sample must be
1548 set with <sample_size> parameter.
1549
1550 <sample_size>
1551 The size of the sample in number of logs to consider when balancing
1552 their logging loads. It is used to balance the load of the logs to
1553 send to the syslog server. This size must be greater or equal to the
1554 maximum of the high limits of the ranges.
1555 (see also <ranges> parameter).
1556
Robert Tsai81ae1952007-12-05 10:47:29 +01001557 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001558
Willy Tarreaue8746a02018-11-12 08:45:00 +01001559 kern user mail daemon auth syslog lpr news
1560 uucp cron auth2 ftp ntp audit alert cron2
1561 local0 local1 local2 local3 local4 local5 local6 local7
1562
Willy Tarreauc1b06452018-11-12 11:57:56 +01001563 Note that the facility is ignored for the "short" and "raw"
1564 formats, but still required as a positional field. It is
1565 recommended to use "daemon" in this case to make it clear that
1566 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001567
1568 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001569 all messages are sent. If a maximum level is specified, only messages with a
1570 severity at least as important as this level will be sent. An optional minimum
1571 level can be specified. If it is set, logs emitted with a more severe level
1572 than this one will be capped to this level. This is used to avoid sending
1573 "emerg" messages on all terminals on some default syslog configurations.
1574 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001575
Cyril Bontédc4d9032012-04-08 21:57:39 +02001576 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001577
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001578log-send-hostname [<string>]
1579 Sets the hostname field in the syslog header. If optional "string" parameter
1580 is set the header is set to the string contents, otherwise uses the hostname
1581 of the system. Generally used if one is not relaying logs through an
1582 intermediate syslog server or for simply customizing the hostname printed in
1583 the logs.
1584
Kevinm48936af2010-12-22 16:08:21 +00001585log-tag <string>
1586 Sets the tag field in the syslog header to this string. It defaults to the
1587 program name as launched from the command line, which usually is "haproxy".
1588 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001589 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001590
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001591lua-load <file>
Thierry Fournier59f11be2020-11-29 00:37:41 +01001592 This global directive loads and executes a Lua file in the shared context
1593 that is visible to all threads. Any variable set in such a context is visible
1594 from any thread. This is the easiest and recommended way to load Lua programs
1595 but it will not scale well if a lot of Lua calls are performed, as only one
1596 thread may be running on the global state at a time. A program loaded this
1597 way will always see 0 in the "core.thread" variable. This directive can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001598 used multiple times.
1599
Thierry Fournier59f11be2020-11-29 00:37:41 +01001600lua-load-per-thread <file>
1601 This global directive loads and executes a Lua file into each started thread.
1602 Any global variable has a thread-local visibility so that each thread could
1603 see a different value. As such it is strongly recommended not to use global
1604 variables in programs loaded this way. An independent copy is loaded and
1605 initialized for each thread, everything is done sequentially and in the
1606 thread's numeric order from 1 to nbthread. If some operations need to be
1607 performed only once, the program should check the "core.thread" variable to
1608 figure what thread is being initialized. Programs loaded this way will run
1609 concurrently on all threads and will be highly scalable. This is the
1610 recommended way to load simple functions that register sample-fetches,
1611 converters, actions or services once it is certain the program doesn't depend
1612 on global variables. For the sake of simplicity, the directive is available
1613 even if only one thread is used and even if threads are disabled (in which
1614 case it will be equivalent to lua-load). This directive can be used multiple
1615 times.
1616
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +01001617lua-prepend-path <string> [<type>]
1618 Prepends the given string followed by a semicolon to Lua's package.<type>
1619 variable.
1620 <type> must either be "path" or "cpath". If <type> is not given it defaults
1621 to "path".
1622
1623 Lua's paths are semicolon delimited lists of patterns that specify how the
1624 `require` function attempts to find the source file of a library. Question
1625 marks (?) within a pattern will be replaced by module name. The path is
1626 evaluated left to right. This implies that paths that are prepended later
1627 will be checked earlier.
1628
1629 As an example by specifying the following path:
1630
1631 lua-prepend-path /usr/share/haproxy-lua/?/init.lua
1632 lua-prepend-path /usr/share/haproxy-lua/?.lua
1633
1634 When `require "example"` is being called Lua will first attempt to load the
1635 /usr/share/haproxy-lua/example.lua script, if that does not exist the
1636 /usr/share/haproxy-lua/example/init.lua will be attempted and the default
1637 paths if that does not exist either.
1638
1639 See https://www.lua.org/pil/8.1.html for the details within the Lua
1640 documentation.
1641
William Lallemand4cfede82017-11-24 22:02:34 +01001642master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001643 Master-worker mode. It is equivalent to the command line "-W" argument.
1644 This mode will launch a "master" which will monitor the "workers". Using
1645 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001646 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001647 or daemon mode. It is recommended to use this mode with multiprocess and
1648 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001649 By default, if a worker exits with a bad return code, in the case of a
1650 segfault for example, all workers will be killed, and the master will leave.
1651 It is convenient to combine this behavior with Restart=on-failure in a
1652 systemd unit file in order to relaunch the whole process. If you don't want
1653 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001654
William Lallemand4cfede82017-11-24 22:02:34 +01001655 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001656
William Lallemand27edc4b2019-05-07 17:49:33 +02001657mworker-max-reloads <number>
1658 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001659 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001660 number of reloads is greater than this number, the worker will receive a
1661 SIGTERM. This option helps to keep under control the number of workers.
1662 See also "show proc" in the Management Guide.
1663
Willy Tarreauf42d7942020-10-20 11:54:49 +02001664nbproc <number> (deprecated)
Willy Tarreau6a06a402007-07-15 20:15:28 +02001665 Creates <number> processes when going daemon. This requires the "daemon"
1666 mode. By default, only one process is created, which is the recommended mode
1667 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001668 process, it may be needed to fork multiple daemons. When set to a value
1669 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreauf42d7942020-10-20 11:54:49 +02001670 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. This directive is deprecated
1671 and scheduled for removal in 2.5. Please use "nbthread" instead. See also
1672 "daemon" and "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001673
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001674nbthread <number>
1675 This setting is only available when support for threads was built in. It
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001676 makes HAProxy run on <number> threads. This is exclusive with "nbproc". While
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001677 "nbproc" historically used to be the only way to use multiple processors, it
1678 also involved a number of shortcomings related to the lack of synchronization
1679 between processes (health-checks, peers, stick-tables, stats, ...) which do
1680 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001681 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1682 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1683 when nbproc is not used, the default "nbthread" value is automatically set to
1684 the number of CPUs the process is bound to upon startup. This means that the
1685 thread count can easily be adjusted from the calling process using commands
1686 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1687 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001688
Amaury Denoyelle0f50cb92021-03-26 18:50:33 +01001689numa-cpu-mapping
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001690 By default, if running on Linux, HAProxy inspects on startup the CPU topology
Amaury Denoyelle0f50cb92021-03-26 18:50:33 +01001691 of the machine. If a multi-socket machine is detected, the affinity is
1692 automatically calculated to run on the CPUs of a single node. This is done in
1693 order to not suffer from the performance penalties caused by the inter-socket
1694 bus latency. However, if the applied binding is non optimal on a particular
1695 architecture, it can be disabled with the statement 'no numa-cpu-mapping'.
1696 This automatic binding is also not applied if a nbthread statement is present
1697 in the configuration, or the affinity of the process is already specified,
1698 for example via the 'cpu-map' directive or the taskset utility.
1699
Willy Tarreau6a06a402007-07-15 20:15:28 +02001700pidfile <pidfile>
MIZUTA Takeshic32f3942020-08-26 13:46:19 +09001701 Writes PIDs of all daemons into file <pidfile> when daemon mode or writes PID
1702 of master process into file <pidfile> when master-worker mode. This option is
1703 equivalent to the "-p" command line argument. The file must be accessible to
1704 the user starting the process. See also "daemon" and "master-worker".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001705
Willy Tarreau119e50e2020-05-22 13:53:29 +02001706pp2-never-send-local
1707 A bug in the PROXY protocol v2 implementation was present in HAProxy up to
1708 version 2.1, causing it to emit a PROXY command instead of a LOCAL command
1709 for health checks. This is particularly minor but confuses some servers'
1710 logs. Sadly, the bug was discovered very late and revealed that some servers
1711 which possibly only tested their PROXY protocol implementation against
1712 HAProxy fail to properly handle the LOCAL command, and permanently remain in
1713 the "down" state when HAProxy checks them. When this happens, it is possible
1714 to enable this global option to revert to the older (bogus) behavior for the
1715 time it takes to contact the affected components' vendors and get them fixed.
1716 This option is disabled by default and acts on all servers having the
1717 "send-proxy-v2" statement.
1718
Willy Tarreau1d549722016-02-16 12:41:57 +01001719presetenv <name> <value>
1720 Sets environment variable <name> to value <value>. If the variable exists, it
1721 is NOT overwritten. The changes immediately take effect so that the next line
1722 in the configuration file sees the new value. See also "setenv", "resetenv",
1723 and "unsetenv".
1724
1725resetenv [<name> ...]
1726 Removes all environment variables except the ones specified in argument. It
1727 allows to use a clean controlled environment before setting new values with
1728 setenv or unsetenv. Please note that some internal functions may make use of
1729 some environment variables, such as time manipulation functions, but also
1730 OpenSSL or even external checks. This must be used with extreme care and only
1731 after complete validation. The changes immediately take effect so that the
1732 next line in the configuration file sees the new environment. See also
1733 "setenv", "presetenv", and "unsetenv".
1734
Christopher Fauletff4121f2017-11-22 16:38:49 +01001735stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001736 Limits the stats socket to a certain set of processes numbers. By default the
1737 stats socket is bound to all processes, causing a warning to be emitted when
1738 nbproc is greater than 1 because there is no way to select the target process
1739 when connecting. However, by using this setting, it becomes possible to pin
1740 the stats socket to a specific set of processes, typically the first one. The
1741 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001742 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001743 word size (32 or 64). Ranges can be partially defined. The higher bound can
1744 be omitted. In such case, it is replaced by the corresponding maximum
1745 value. A better option consists in using the "process" setting of the "stats
1746 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001747
Baptiste Assmann5626f482015-08-23 10:00:10 +02001748server-state-base <directory>
1749 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001750 file names which do not start with a '/'. See also "server-state-file",
1751 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001752
1753server-state-file <file>
1754 Specifies the path to the file containing state of servers. If the path starts
1755 with a slash ('/'), it is considered absolute, otherwise it is considered
1756 relative to the directory specified using "server-state-base" (if set) or to
1757 the current directory. Before reloading HAProxy, it is possible to save the
1758 servers' current state using the stats command "show servers state". The
1759 output of this command must be written in the file pointed by <file>. When
1760 starting up, before handling traffic, HAProxy will read, load and apply state
1761 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001762 configuration. See also "server-state-base" and "show servers state",
1763 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001764
Willy Tarreau13d2ba22021-03-26 11:38:08 +01001765set-var <var-name> <expr>
1766 Sets the process-wide variable '<var-name>' to the result of the evaluation
1767 of the sample expression <expr>. The variable '<var-name>' may only be a
1768 process-wide variable (using the 'proc.' prefix). It works exactly like the
1769 'set-var' action in TCP or HTTP rules except that the expression is evaluated
1770 at configuration parsing time and that the variable is instantly set. The
1771 sample fetch functions and converters permitted in the expression are only
1772 those using internal data, typically 'int(value)' or 'str(value)'. It's is
1773 possible to reference previously allocated variables as well. These variables
1774 will then be readable (and modifiable) from the regular rule sets.
1775
1776 Example:
1777 global
1778 set-var proc.current_state str(primary)
1779 set-var proc.prio int(100)
1780 set-var proc.threshold int(200),sub(proc.prio)
1781
Willy Tarreau1d549722016-02-16 12:41:57 +01001782setenv <name> <value>
1783 Sets environment variable <name> to value <value>. If the variable exists, it
1784 is overwritten. The changes immediately take effect so that the next line in
1785 the configuration file sees the new value. See also "presetenv", "resetenv",
1786 and "unsetenv".
1787
Willy Tarreau636848a2019-04-15 19:38:50 +02001788set-dumpable
1789 This option is better left disabled by default and enabled only upon a
William Dauchyec730982019-10-27 20:08:10 +01001790 developer's request. If it has been enabled, it may still be forcibly
1791 disabled by prefixing it with the "no" keyword. It has no impact on
1792 performance nor stability but will try hard to re-enable core dumps that were
1793 possibly disabled by file size limitations (ulimit -f), core size limitations
1794 (ulimit -c), or "dumpability" of a process after changing its UID/GID (such
1795 as /proc/sys/fs/suid_dumpable on Linux). Core dumps might still be limited by
1796 the current directory's permissions (check what directory the file is started
1797 from), the chroot directory's permission (it may be needed to temporarily
1798 disable the chroot directive or to move it to a dedicated writable location),
1799 or any other system-specific constraint. For example, some Linux flavours are
1800 notorious for replacing the default core file with a path to an executable
1801 not even installed on the system (check /proc/sys/kernel/core_pattern). Often,
1802 simply writing "core", "core.%p" or "/var/log/core/core.%p" addresses the
1803 issue. When trying to enable this option waiting for a rare issue to
1804 re-appear, it's often a good idea to first try to obtain such a dump by
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001805 issuing, for example, "kill -11" to the "haproxy" process and verify that it
William Dauchyec730982019-10-27 20:08:10 +01001806 leaves a core where expected when dying.
Willy Tarreau636848a2019-04-15 19:38:50 +02001807
Willy Tarreau610f04b2014-02-13 11:36:41 +01001808ssl-default-bind-ciphers <ciphers>
1809 This setting is only available when support for OpenSSL was built in. It sets
1810 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001811 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001812 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001813 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1814 information and recommendations see e.g.
1815 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1816 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1817 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1818 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001819
1820ssl-default-bind-ciphersuites <ciphersuites>
1821 This setting is only available when support for OpenSSL was built in and
1822 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1823 describing the list of cipher algorithms ("cipher suite") that are negotiated
1824 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1825 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001826 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1827 cipher configuration for TLSv1.2 and earlier, please check the
1828 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001829 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001830
Jerome Magninb203ff62020-04-03 15:28:22 +02001831ssl-default-bind-curves <curves>
1832 This setting is only available when support for OpenSSL was built in. It sets
1833 the default string describing the list of elliptic curves algorithms ("curve
1834 suite") that are negotiated during the SSL/TLS handshake with ECDHE. The format
1835 of the string is a colon-delimited list of curve name.
1836 Please check the "bind" keyword for more information.
1837
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001838ssl-default-bind-options [<option>]...
1839 This setting is only available when support for OpenSSL was built in. It sets
1840 default ssl-options to force on all "bind" lines. Please check the "bind"
1841 keyword to see available options.
1842
1843 Example:
1844 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001845 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001846
Willy Tarreau610f04b2014-02-13 11:36:41 +01001847ssl-default-server-ciphers <ciphers>
1848 This setting is only available when support for OpenSSL was built in. It
1849 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001850 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001851 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001852 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1853 information and recommendations see e.g.
1854 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1855 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1856 For TLSv1.3 cipher configuration, please check the
1857 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1858 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001859
1860ssl-default-server-ciphersuites <ciphersuites>
1861 This setting is only available when support for OpenSSL was built in and
1862 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1863 string describing the list of cipher algorithms that are negotiated during
1864 the TLSv1.3 handshake with the server, for all "server" lines which do not
1865 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001866 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1867 cipher configuration for TLSv1.2 and earlier, please check the
1868 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1869 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001870
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001871ssl-default-server-options [<option>]...
1872 This setting is only available when support for OpenSSL was built in. It sets
1873 default ssl-options to force on all "server" lines. Please check the "server"
1874 keyword to see available options.
1875
Remi Gacogne47783ef2015-05-29 15:53:22 +02001876ssl-dh-param-file <file>
1877 This setting is only available when support for OpenSSL was built in. It sets
1878 the default DH parameters that are used during the SSL/TLS handshake when
1879 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001880 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001881 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001882 are not specified either by using ssl-dh-param-file or by setting them
1883 directly in the certificate file, pre-generated DH parameters of the size
1884 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1885 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001886 Custom DH parameters may be generated by using the OpenSSL command
1887 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1888 parameters should not be considered secure anymore.
1889
William Lallemand8e8581e2020-10-20 17:36:46 +02001890ssl-load-extra-del-ext
1891 This setting allows to configure the way HAProxy does the lookup for the
1892 extra SSL files. By default HAProxy adds a new extension to the filename.
William Lallemand089c1382020-10-23 17:35:12 +02001893 (ex: with "foobar.crt" load "foobar.crt.key"). With this option enabled,
William Lallemand8e8581e2020-10-20 17:36:46 +02001894 HAProxy removes the extension before adding the new one (ex: with
William Lallemand089c1382020-10-23 17:35:12 +02001895 "foobar.crt" load "foobar.key").
1896
1897 Your crt file must have a ".crt" extension for this option to work.
William Lallemand8e8581e2020-10-20 17:36:46 +02001898
1899 This option is not compatible with bundle extensions (.ecdsa, .rsa. .dsa)
1900 and won't try to remove them.
1901
1902 This option is disabled by default. See also "ssl-load-extra-files".
1903
William Lallemand4c5adbf2020-02-24 14:23:22 +01001904ssl-load-extra-files <none|all|bundle|sctl|ocsp|issuer|key>*
William Lallemand3af48e72020-02-03 17:15:52 +01001905 This setting alters the way HAProxy will look for unspecified files during
Remi Tricot-Le Breton7c980df2021-05-07 15:28:08 +02001906 the loading of the SSL certificates. This option applies to certificates
1907 associated to "bind" lines as well as "server" lines but some of the extra
1908 files will not have any functional impact for "server" line certificates.
William Lallemand3af48e72020-02-03 17:15:52 +01001909
1910 By default, HAProxy discovers automatically a lot of files not specified in
1911 the configuration, and you may want to disable this behavior if you want to
1912 optimize the startup time.
1913
1914 "none": Only load the files specified in the configuration. Don't try to load
1915 a certificate bundle if the file does not exist. In the case of a directory,
1916 it won't try to bundle the certificates if they have the same basename.
1917
1918 "all": This is the default behavior, it will try to load everything,
William Lallemand4c5adbf2020-02-24 14:23:22 +01001919 bundles, sctl, ocsp, issuer, key.
William Lallemand3af48e72020-02-03 17:15:52 +01001920
1921 "bundle": When a file specified in the configuration does not exist, HAProxy
Remi Tricot-Le Breton7c980df2021-05-07 15:28:08 +02001922 will try to load a "cert bundle". Certificate bundles are only managed on the
1923 frontend side and will not work for backend certificates.
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001924
1925 Starting from HAProxy 2.3, the bundles are not loaded in the same OpenSSL
1926 certificate store, instead it will loads each certificate in a separate
1927 store which is equivalent to declaring multiple "crt". OpenSSL 1.1.1 is
1928 required to achieve this. Which means that bundles are now used only for
1929 backward compatibility and are not mandatory anymore to do an hybrid RSA/ECC
Remi Tricot-Le Breton7c980df2021-05-07 15:28:08 +02001930 bind configuration.
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001931
1932 To associate these PEM files into a "cert bundle" that is recognized by
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001933 HAProxy, they must be named in the following way: All PEM files that are to
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001934 be bundled must have the same base name, with a suffix indicating the key
1935 type. Currently, three suffixes are supported: rsa, dsa and ecdsa. For
1936 example, if www.example.com has two PEM files, an RSA file and an ECDSA
1937 file, they must be named: "example.pem.rsa" and "example.pem.ecdsa". The
1938 first part of the filename is arbitrary; only the suffix matters. To load
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001939 this bundle into HAProxy, specify the base name only:
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001940
1941 Example : bind :8443 ssl crt example.pem
1942
Daniel Corbett9f0843f2021-05-08 10:50:37 -04001943 Note that the suffix is not given to HAProxy; this tells HAProxy to look for
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001944 a cert bundle.
1945
1946 HAProxy will load all PEM files in the bundle as if they were configured
1947 separately in several "crt".
1948
1949 The bundle loading does not have an impact anymore on the directory loading
1950 since files are loading separately.
1951
1952 On the CLI, bundles are seen as separate files, and the bundle extension is
1953 required to commit them.
1954
William Dauchy57dd6f12020-10-06 15:22:37 +02001955 OCSP files (.ocsp), issuer files (.issuer), Certificate Transparency (.sctl)
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001956 as well as private keys (.key) are supported with multi-cert bundling.
William Lallemand3af48e72020-02-03 17:15:52 +01001957
Remi Tricot-Le Breton7c980df2021-05-07 15:28:08 +02001958 "sctl": Try to load "<basename>.sctl" for each crt keyword. If provided for
1959 a backend certificate, it will be loaded but will not have any functional
1960 impact.
William Lallemand3af48e72020-02-03 17:15:52 +01001961
Remi Tricot-Le Breton7c980df2021-05-07 15:28:08 +02001962 "ocsp": Try to load "<basename>.ocsp" for each crt keyword. If provided for
1963 a backend certificate, it will be loaded but will not have any functional
1964 impact.
William Lallemand3af48e72020-02-03 17:15:52 +01001965
1966 "issuer": Try to load "<basename>.issuer" if the issuer of the OCSP file is
Remi Tricot-Le Breton7c980df2021-05-07 15:28:08 +02001967 not provided in the PEM file. If provided for a backend certificate, it will
1968 be loaded but will not have any functional impact.
William Lallemand3af48e72020-02-03 17:15:52 +01001969
William Lallemand4c5adbf2020-02-24 14:23:22 +01001970 "key": If the private key was not provided by the PEM file, try to load a
1971 file "<basename>.key" containing a private key.
1972
William Lallemand3af48e72020-02-03 17:15:52 +01001973 The default behavior is "all".
1974
1975 Example:
1976 ssl-load-extra-files bundle sctl
1977 ssl-load-extra-files sctl ocsp issuer
1978 ssl-load-extra-files none
1979
Remi Tricot-Le Breton7c980df2021-05-07 15:28:08 +02001980 See also: "crt", section 5.1 about bind options and section 5.2 about server
1981 options.
William Lallemand3af48e72020-02-03 17:15:52 +01001982
Emeric Brun850efd52014-01-29 12:24:34 +01001983ssl-server-verify [none|required]
1984 The default behavior for SSL verify on servers side. If specified to 'none',
1985 servers certificates are not verified. The default is 'required' except if
1986 forced using cmdline option '-dV'.
1987
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001988ssl-skip-self-issued-ca
Daniel Corbett67a82712020-07-06 23:01:19 -04001989 Self issued CA, aka x509 root CA, is the anchor for chain validation: as a
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001990 server is useless to send it, client must have it. Standard configuration
1991 need to not include such CA in PEM file. This option allows you to keep such
1992 CA in PEM file without sending it to the client. Use case is to provide
1993 issuer for ocsp without the need for '.issuer' file and be able to share it
1994 with 'issuers-chain-path'. This concerns all certificates without intermediate
1995 certificates. It's useless for BoringSSL, .issuer is ignored because ocsp
William Lallemand9a1d8392020-08-10 17:28:23 +02001996 bits does not need it. Requires at least OpenSSL 1.0.2.
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001997
Willy Tarreauabb175f2012-09-24 12:43:26 +02001998stats socket [<address:port>|<path>] [param*]
1999 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
2000 Connections to this socket will return various statistics outputs and even
2001 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02002002 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02002003 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02002004
Willy Tarreauabb175f2012-09-24 12:43:26 +02002005 All parameters supported by "bind" lines are supported, for instance to
2006 restrict access to some users or their access rights. Please consult
2007 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02002008
2009stats timeout <timeout, in milliseconds>
2010 The default timeout on the stats socket is set to 10 seconds. It is possible
2011 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01002012 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02002013
2014stats maxconn <connections>
2015 By default, the stats socket is limited to 10 concurrent connections. It is
2016 possible to change this value with "stats maxconn".
2017
Willy Tarreau6a06a402007-07-15 20:15:28 +02002018uid <number>
Thayne McCombscdbcca92021-01-07 21:24:41 -07002019 Changes the process's user ID to <number>. It is recommended that the user ID
Willy Tarreau6a06a402007-07-15 20:15:28 +02002020 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
2021 be started with superuser privileges in order to be able to switch to another
2022 one. See also "gid" and "user".
2023
2024ulimit-n <number>
2025 Sets the maximum number of per-process file-descriptors to <number>. By
2026 default, it is automatically computed, so it is recommended not to use this
2027 option.
2028
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002029unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
2030 [ group <group> ] [ gid <gid> ]
2031
2032 Fixes common settings to UNIX listening sockets declared in "bind" statements.
2033 This is mainly used to simplify declaration of those UNIX sockets and reduce
2034 the risk of errors, since those settings are most commonly required but are
2035 also process-specific. The <prefix> setting can be used to force all socket
2036 path to be relative to that directory. This might be needed to access another
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002037 component's chroot. Note that those paths are resolved before HAProxy chroots
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002038 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
2039 all have the same meaning as their homonyms used by the "bind" statement. If
2040 both are specified, the "bind" statement has priority, meaning that the
2041 "unix-bind" settings may be seen as process-wide default settings.
2042
Willy Tarreau1d549722016-02-16 12:41:57 +01002043unsetenv [<name> ...]
2044 Removes environment variables specified in arguments. This can be useful to
2045 hide some sensitive information that are occasionally inherited from the
2046 user's environment during some operations. Variables which did not exist are
2047 silently ignored so that after the operation, it is certain that none of
2048 these variables remain. The changes immediately take effect so that the next
2049 line in the configuration file will not see these variables. See also
2050 "setenv", "presetenv", and "resetenv".
2051
Willy Tarreau6a06a402007-07-15 20:15:28 +02002052user <user name>
2053 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
2054 See also "uid" and "group".
2055
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02002056node <name>
2057 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
2058
2059 This statement is useful in HA configurations where two or more processes or
2060 servers share the same IP address. By setting a different node-name on all
2061 nodes, it becomes easy to immediately spot what server is handling the
2062 traffic.
2063
2064description <text>
2065 Add a text that describes the instance.
2066
2067 Please note that it is required to escape certain characters (# for example)
2068 and this text is inserted into a html page so you should avoid using
2069 "<" and ">" characters.
2070
Thomas Holmesdb04f192015-05-18 13:21:39 +0100207151degrees-data-file <file path>
2072 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01002073 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01002074
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002075 Please note that this option is only available when HAProxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01002076 compiled with USE_51DEGREES.
2077
Ben Shillitof25e8e52016-12-02 14:25:37 +0000207851degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01002079 A list of 51Degrees property names to be load from the dataset. A full list
2080 of names is available on the 51Degrees website:
2081 https://51degrees.com/resources/property-dictionary
2082
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002083 Please note that this option is only available when HAProxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01002084 compiled with USE_51DEGREES.
2085
Dragan Dosen93b38d92015-06-29 16:43:25 +0200208651degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01002087 A char that will be appended to every property value in a response header
2088 containing 51Degrees results. If not set that will be set as ','.
2089
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002090 Please note that this option is only available when HAProxy has been
Dragan Dosenae6d39a2015-06-29 16:43:27 +02002091 compiled with USE_51DEGREES.
2092
209351degrees-cache-size <number>
2094 Sets the size of the 51Degrees converter cache to <number> entries. This
2095 is an LRU cache which reminds previous device detections and their results.
2096 By default, this cache is disabled.
2097
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002098 Please note that this option is only available when HAProxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01002099 compiled with USE_51DEGREES.
2100
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02002101wurfl-data-file <file path>
2102 The path of the WURFL data file to provide device detection services. The
2103 file should be accessible by HAProxy with relevant permissions.
2104
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002105 Please note that this option is only available when HAProxy has been compiled
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02002106 with USE_WURFL=1.
2107
2108wurfl-information-list [<capability>]*
2109 A space-delimited list of WURFL capabilities, virtual capabilities, property
2110 names we plan to use in injected headers. A full list of capability and
2111 virtual capability names is available on the Scientiamobile website :
2112
2113 https://www.scientiamobile.com/wurflCapability
2114
2115 Valid WURFL properties are:
2116 - wurfl_id Contains the device ID of the matched device.
2117
2118 - wurfl_root_id Contains the device root ID of the matched
2119 device.
2120
2121 - wurfl_isdevroot Tells if the matched device is a root device.
2122 Possible values are "TRUE" or "FALSE".
2123
2124 - wurfl_useragent The original useragent coming with this
2125 particular web request.
2126
2127 - wurfl_api_version Contains a string representing the currently
2128 used Libwurfl API version.
2129
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02002130 - wurfl_info A string containing information on the parsed
2131 wurfl.xml and its full path.
2132
2133 - wurfl_last_load_time Contains the UNIX timestamp of the last time
2134 WURFL has been loaded successfully.
2135
2136 - wurfl_normalized_useragent The normalized useragent.
2137
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002138 Please note that this option is only available when HAProxy has been compiled
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02002139 with USE_WURFL=1.
2140
2141wurfl-information-list-separator <char>
2142 A char that will be used to separate values in a response header containing
2143 WURFL results. If not set that a comma (',') will be used by default.
2144
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002145 Please note that this option is only available when HAProxy has been compiled
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02002146 with USE_WURFL=1.
2147
2148wurfl-patch-file [<file path>]
2149 A list of WURFL patch file paths. Note that patches are loaded during startup
2150 thus before the chroot.
2151
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002152 Please note that this option is only available when HAProxy has been compiled
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02002153 with USE_WURFL=1.
2154
paulborilebad132c2019-04-18 11:57:04 +02002155wurfl-cache-size <size>
2156 Sets the WURFL Useragent cache size. For faster lookups, already processed user
2157 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02002158 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02002159 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02002160
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002161 Please note that this option is only available when HAProxy has been compiled
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02002162 with USE_WURFL=1.
2163
William Dauchy0fec3ab2019-10-27 20:08:11 +01002164strict-limits
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002165 Makes process fail at startup when a setrlimit fails. HAProxy tries to set the
William Dauchya5194602020-03-28 19:29:58 +01002166 best setrlimit according to what has been calculated. If it fails, it will
2167 emit a warning. This option is here to guarantee an explicit failure of
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002168 HAProxy when those limits fail. It is enabled by default. It may still be
William Dauchya5194602020-03-28 19:29:58 +01002169 forcibly disabled by prefixing it with the "no" keyword.
William Dauchy0fec3ab2019-10-27 20:08:11 +01002170
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021713.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02002172-----------------------
2173
Willy Tarreaubeb859a2018-11-22 18:07:59 +01002174busy-polling
2175 In some situations, especially when dealing with low latency on processors
2176 supporting a variable frequency or when running inside virtual machines, each
2177 time the process waits for an I/O using the poller, the processor goes back
2178 to sleep or is offered to another VM for a long time, and it causes
2179 excessively high latencies. This option provides a solution preventing the
2180 processor from sleeping by always using a null timeout on the pollers. This
2181 results in a significant latency reduction (30 to 100 microseconds observed)
2182 at the expense of a risk to overheat the processor. It may even be used with
2183 threads, in which case improperly bound threads may heavily conflict,
2184 resulting in a worse performance and high values for the CPU stolen fields
2185 in "show info" output, indicating which threads are misconfigured. It is
2186 important not to let the process run on the same processor as the network
2187 interrupts when this option is used. It is also better to avoid using it on
2188 multiple CPU threads sharing the same core. This option is disabled by
2189 default. If it has been enabled, it may still be forcibly disabled by
2190 prefixing it with the "no" keyword. It is ignored by the "select" and
2191 "poll" pollers.
2192
William Dauchy3894d972019-12-28 15:36:02 +01002193 This option is automatically disabled on old processes in the context of
2194 seamless reload; it avoids too much cpu conflicts when multiple processes
2195 stay around for some time waiting for the end of their current connections.
2196
Willy Tarreau1746eec2014-04-25 10:46:47 +02002197max-spread-checks <delay in milliseconds>
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002198 By default, HAProxy tries to spread the start of health checks across the
Willy Tarreau1746eec2014-04-25 10:46:47 +02002199 smallest health check interval of all the servers in a farm. The principle is
2200 to avoid hammering services running on the same server. But when using large
2201 check intervals (10 seconds or more), the last servers in the farm take some
2202 time before starting to be tested, which can be a problem. This parameter is
2203 used to enforce an upper bound on delay between the first and the last check,
2204 even if the servers' check intervals are larger. When servers run with
2205 shorter intervals, their intervals will be respected though.
2206
Willy Tarreau6a06a402007-07-15 20:15:28 +02002207maxconn <number>
2208 Sets the maximum per-process number of concurrent connections to <number>. It
2209 is equivalent to the command-line argument "-n". Proxies will stop accepting
2210 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02002211 automatically adjusted according to this value. See also "ulimit-n". Note:
2212 the "select" poller cannot reliably use more than 1024 file descriptors on
2213 some platforms. If your platform only supports select and reports "select
2214 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01002215 below 500 in general). If this value is not set, it will automatically be
2216 calculated based on the current file descriptors limit reported by the
2217 "ulimit -n" command, possibly reduced to a lower value if a memory limit
2218 is enforced, based on the buffer size, memory allocated to compression, SSL
2219 cache size, and use or not of SSL and the associated maxsslconn (which can
2220 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02002221
Willy Tarreau81c25d02011-09-07 15:17:21 +02002222maxconnrate <number>
2223 Sets the maximum per-process number of connections per second to <number>.
2224 Proxies will stop accepting connections when this limit is reached. It can be
2225 used to limit the global capacity regardless of each frontend capacity. It is
2226 important to note that this can only be used as a service protection measure,
2227 as there will not necessarily be a fair share between frontends when the
2228 limit is reached, so it's a good idea to also limit each frontend to some
2229 value close to its expected share. Also, lowering tune.maxaccept can improve
2230 fairness.
2231
William Lallemandd85f9172012-11-09 17:05:39 +01002232maxcomprate <number>
2233 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01002234 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01002235 level will be decreased during the session. If the maximum is reached at the
2236 beginning of a session, the session will not compress at all. If the maximum
2237 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01002238 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01002239 default value.
2240
William Lallemand072a2bf2012-11-20 17:01:01 +01002241maxcompcpuusage <number>
2242 Sets the maximum CPU usage HAProxy can reach before stopping the compression
2243 for new requests or decreasing the compression level of current requests.
2244 It works like 'maxcomprate' but measures CPU usage instead of incoming data
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002245 bandwidth. The value is expressed in percent of the CPU used by HAProxy. In
William Lallemand072a2bf2012-11-20 17:01:01 +01002246 case of multiple processes (nbproc > 1), each process manages its individual
2247 usage. A value of 100 disable the limit. The default value is 100. Setting
2248 a lower value will prevent the compression work from slowing the whole
2249 process down and from introducing high latencies.
2250
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002251maxpipes <number>
2252 Sets the maximum per-process number of pipes to <number>. Currently, pipes
2253 are only used by kernel-based tcp splicing. Since a pipe contains two file
2254 descriptors, the "ulimit-n" value will be increased accordingly. The default
2255 value is maxconn/4, which seems to be more than enough for most heavy usages.
2256 The splice code dynamically allocates and releases pipes, and can fall back
2257 to standard copy, so setting this value too low may only impact performance.
2258
Willy Tarreau93e7c002013-10-07 18:51:07 +02002259maxsessrate <number>
2260 Sets the maximum per-process number of sessions per second to <number>.
2261 Proxies will stop accepting connections when this limit is reached. It can be
2262 used to limit the global capacity regardless of each frontend capacity. It is
2263 important to note that this can only be used as a service protection measure,
2264 as there will not necessarily be a fair share between frontends when the
2265 limit is reached, so it's a good idea to also limit each frontend to some
2266 value close to its expected share. Also, lowering tune.maxaccept can improve
2267 fairness.
2268
Willy Tarreau403edff2012-09-06 11:58:37 +02002269maxsslconn <number>
2270 Sets the maximum per-process number of concurrent SSL connections to
2271 <number>. By default there is no SSL-specific limit, which means that the
2272 global maxconn setting will apply to all connections. Setting this limit
2273 avoids having openssl use too much memory and crash when malloc returns NULL
2274 (since it unfortunately does not reliably check for such conditions). Note
2275 that the limit applies both to incoming and outgoing connections, so one
2276 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01002277 If this value is not set, but a memory limit is enforced, this value will be
2278 automatically computed based on the memory limit, maxconn, the buffer size,
2279 memory allocated to compression, SSL cache size, and use of SSL in either
2280 frontends, backends or both. If neither maxconn nor maxsslconn are specified
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002281 when there is a memory limit, HAProxy will automatically adjust these values
Willy Tarreaud0256482015-01-15 21:45:22 +01002282 so that 100% of the connections can be made over SSL with no risk, and will
2283 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02002284
Willy Tarreaue43d5322013-10-07 20:01:52 +02002285maxsslrate <number>
2286 Sets the maximum per-process number of SSL sessions per second to <number>.
2287 SSL listeners will stop accepting connections when this limit is reached. It
2288 can be used to limit the global SSL CPU usage regardless of each frontend
2289 capacity. It is important to note that this can only be used as a service
2290 protection measure, as there will not necessarily be a fair share between
2291 frontends when the limit is reached, so it's a good idea to also limit each
2292 frontend to some value close to its expected share. It is also important to
2293 note that the sessions are accounted before they enter the SSL stack and not
2294 after, which also protects the stack against bad handshakes. Also, lowering
2295 tune.maxaccept can improve fairness.
2296
William Lallemand9d5f5482012-11-07 16:12:57 +01002297maxzlibmem <number>
2298 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
2299 When the maximum amount is reached, future sessions will not compress as long
2300 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01002301 The default value is 0. The value is available in bytes on the UNIX socket
2302 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
2303 "ZlibMemUsage" in bytes.
2304
Willy Tarreau85dd5212022-03-08 10:41:40 +01002305no-memory-trimming
2306 Disables memory trimming ("malloc_trim") at a few moments where attempts are
2307 made to reclaim lots of memory (on memory shortage or on reload). Trimming
2308 memory forces the system's allocator to scan all unused areas and to release
2309 them. This is generally seen as nice action to leave more available memory to
2310 a new process while the old one is unlikely to make significant use of it.
2311 But some systems dealing with tens to hundreds of thousands of concurrent
2312 connections may experience a lot of memory fragmentation, that may render
2313 this release operation extremely long. During this time, no more traffic
2314 passes through the process, new connections are not accepted anymore, some
2315 health checks may even fail, and the watchdog may even trigger and kill the
2316 unresponsive process, leaving a huge core dump. If this ever happens, then it
2317 is suggested to use this option to disable trimming and stop trying to be
2318 nice with the new process. Note that advanced memory allocators usually do
2319 not suffer from such a problem.
2320
Willy Tarreau6a06a402007-07-15 20:15:28 +02002321noepoll
2322 Disables the use of the "epoll" event polling system on Linux. It is
2323 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01002324 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02002325
2326nokqueue
2327 Disables the use of the "kqueue" event polling system on BSD. It is
2328 equivalent to the command-line argument "-dk". The next polling system
2329 used will generally be "poll". See also "nopoll".
2330
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00002331noevports
2332 Disables the use of the event ports event polling system on SunOS systems
2333 derived from Solaris 10 and later. It is equivalent to the command-line
2334 argument "-dv". The next polling system used will generally be "poll". See
2335 also "nopoll".
2336
Willy Tarreau6a06a402007-07-15 20:15:28 +02002337nopoll
2338 Disables the use of the "poll" event polling system. It is equivalent to the
2339 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002340 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00002341 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
2342 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02002343
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002344nosplice
2345 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002346 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002347 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002348 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002349 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
2350 be used. This option makes it easier to globally disable kernel splicing in
2351 case of doubt. See also "option splice-auto", "option splice-request" and
2352 "option splice-response".
2353
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002354nogetaddrinfo
2355 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
2356 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
2357
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00002358noreuseport
2359 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
2360 command line argument "-dR".
2361
Willy Tarreauca3afc22021-05-05 18:33:19 +02002362profiling.memory { on | off }
2363 Enables ('on') or disables ('off') per-function memory profiling. This will
2364 keep usage statistics of malloc/calloc/realloc/free calls anywhere in the
2365 process (including libraries) which will be reported on the CLI using the
2366 "show profiling" command. This is essentially meant to be used when an
2367 abnormal memory usage is observed that cannot be explained by the pools and
2368 other info are required. The performance hit will typically be around 1%,
2369 maybe a bit more on highly threaded machines, so it is normally suitable for
2370 use in production. The same may be achieved at run time on the CLI using the
2371 "set profiling memory" command, please consult the management manual.
2372
Willy Tarreaud2d33482019-04-25 17:09:07 +02002373profiling.tasks { auto | on | off }
2374 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
2375 the profiling automatically turns on a thread when it starts to suffer from
2376 an average latency of 1000 microseconds or higher as reported in the
2377 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05002378 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02002379 loops so it does not vary quickly and tends to significantly smooth short
2380 spikes). It may also spontaneously trigger from time to time on overloaded
2381 systems, containers, or virtual machines, or when the system swaps (which
2382 must absolutely never happen on a load balancer).
2383
2384 CPU profiling per task can be very convenient to report where the time is
2385 spent and which requests have what effect on which other request. Enabling
2386 it will typically affect the overall's performance by less than 1%, thus it
2387 is recommended to leave it to the default 'auto' value so that it only
2388 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01002389 supporting the clock_gettime(2) syscall with clock identifiers
2390 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
2391 be zero. This option may be changed at run time using "set profiling" on the
2392 CLI.
2393
Willy Tarreaufe255b72007-10-14 23:09:26 +02002394spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09002395 Sometimes it is desirable to avoid sending agent and health checks to
2396 servers at exact intervals, for instance when many logical servers are
2397 located on the same physical server. With the help of this parameter, it
2398 becomes possible to add some randomness in the check interval between 0
2399 and +/- 50%. A value between 2 and 5 seems to show good results. The
2400 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02002401
Davor Ocelice9ed2812017-12-25 17:49:28 +01002402ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00002403 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01002404 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00002405 multiple times, it will simply enable multiple crypto engines. Referencing an
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002406 unsupported engine will prevent HAProxy from starting. Note that many engines
Grant Zhang872f9c22017-01-21 01:10:18 +00002407 will lead to lower HTTPS performance than pure software with recent
2408 processors. The optional command "algo" sets the default algorithms an ENGINE
2409 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01002410 of "ALL" uses the engine for all cryptographic operations. If no list of
2411 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00002412 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
2413 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
2414 openssl configuration file uses:
2415 https://www.openssl.org/docs/man1.0.2/apps/config.html
2416
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00002417ssl-mode-async
2418 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02002419 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00002420 implementation supports a maximum of 32 engines. The Openssl ASYNC API
2421 doesn't support moving read/write buffers and is not compliant with
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002422 HAProxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05002423 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00002424 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00002425
Willy Tarreau33cb0652014-12-23 22:52:37 +01002426tune.buffers.limit <number>
2427 Sets a hard limit on the number of buffers which may be allocated per process.
2428 The default value is zero which means unlimited. The minimum non-zero value
2429 will always be greater than "tune.buffers.reserve" and should ideally always
2430 be about twice as large. Forcing this value can be particularly useful to
2431 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01002432 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01002433 another one to be released by another session. Since buffers are dynamically
2434 allocated and released, the waiting time is very short and not perceptible
2435 provided that limits remain reasonable. In fact sometimes reducing the limit
2436 may even increase performance by increasing the CPU cache's efficiency. Tests
2437 have shown good results on average HTTP traffic with a limit to 1/10 of the
2438 expected global maxconn setting, which also significantly reduces memory
2439 usage. The memory savings come from the fact that a number of connections
2440 will not allocate 2*tune.bufsize. It is best not to touch this value unless
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002441 advised to do so by an HAProxy core developer.
Willy Tarreau33cb0652014-12-23 22:52:37 +01002442
Willy Tarreau1058ae72014-12-23 22:40:40 +01002443tune.buffers.reserve <number>
2444 Sets the number of buffers which are pre-allocated and reserved for use only
2445 during memory shortage conditions resulting in failed memory allocations. The
2446 minimum value is 2 and is also the default. There is no reason a user would
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002447 want to change this value, it's mostly aimed at HAProxy core developers.
Willy Tarreau1058ae72014-12-23 22:40:40 +01002448
Willy Tarreau27a674e2009-08-17 07:23:33 +02002449tune.bufsize <number>
2450 Sets the buffer size to this size (in bytes). Lower values allow more
2451 sessions to coexist in the same amount of RAM, and higher values allow some
2452 applications with very large cookies to work. The default value is 16384 and
2453 can be changed at build time. It is strongly recommended not to change this
2454 from the default value, as very low values will break some services such as
2455 statistics, and values larger than default size will increase memory usage,
2456 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01002457 parameter should be decreased by the same factor as this one is increased. In
2458 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002459 HTTP request is larger than (tune.bufsize - tune.maxrewrite), HAProxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04002460 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002461 than this size, HAProxy will return HTTP 502 (Bad Gateway). Note that the
Willy Tarreauc77d3642018-12-12 06:19:42 +01002462 value set using this parameter will automatically be rounded up to the next
2463 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02002464
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +01002465tune.chksize <number> (deprecated)
2466 This option is deprecated and ignored.
Willy Tarreau43961d52010-10-04 20:39:20 +02002467
William Lallemandf3747832012-11-09 12:33:10 +01002468tune.comp.maxlevel <number>
2469 Sets the maximum compression level. The compression level affects CPU
2470 usage during compression. This value affects CPU usage during compression.
2471 Each session using compression initializes the compression algorithm with
2472 this value. The default value is 1.
2473
Willy Tarreauc299e1e2019-02-27 11:35:12 +01002474tune.fail-alloc
2475 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
2476 allocation attempt fails. Must be between 0 (no failure) and 100 (no
2477 success). This is useful to debug and make sure memory failures are handled
2478 gracefully.
2479
Willy Tarreaubc52bec2020-06-18 08:58:47 +02002480tune.fd.edge-triggered { on | off } [ EXPERIMENTAL ]
2481 Enables ('on') or disables ('off') the edge-triggered polling mode for FDs
2482 that support it. This is currently only support with epoll. It may noticeably
2483 reduce the number of epoll_ctl() calls and slightly improve performance in
2484 certain scenarios. This is still experimental, it may result in frozen
2485 connections if bugs are still present, and is disabled by default.
2486
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02002487tune.h2.header-table-size <number>
2488 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
2489 cannot be larger than 65536 bytes. A larger value may help certain clients
2490 send more compact requests, depending on their capabilities. This amount of
2491 memory is consumed for each HTTP/2 connection. It is recommended not to
2492 change it.
2493
Willy Tarreaue6baec02017-07-27 11:45:11 +02002494tune.h2.initial-window-size <number>
2495 Sets the HTTP/2 initial window size, which is the number of bytes the client
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002496 can upload before waiting for an acknowledgment from HAProxy. This setting
Davor Ocelice9ed2812017-12-25 17:49:28 +01002497 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02002498 The default value is 65535, which roughly allows up to 5 Mbps of upload
2499 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
2500 over a 1-ms local network. It can make sense to increase this value to allow
2501 faster uploads, or to reduce it to increase fairness when dealing with many
2502 clients. It doesn't affect resource usage.
2503
Willy Tarreau5242ef82017-07-27 11:47:28 +02002504tune.h2.max-concurrent-streams <number>
2505 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
2506 number of outstanding requests on a single connection). The default value is
2507 100. A larger one may slightly improve page load time for complex sites when
2508 visited over high latency networks, but increases the amount of resources a
2509 single client may allocate. A value of zero disables the limit so a single
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002510 client may create as many streams as allocatable by HAProxy. It is highly
Willy Tarreau5242ef82017-07-27 11:47:28 +02002511 recommended not to change this value.
2512
Willy Tarreaua24b35c2019-02-21 13:24:36 +01002513tune.h2.max-frame-size <number>
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002514 Sets the HTTP/2 maximum frame size that HAProxy announces it is willing to
Willy Tarreaua24b35c2019-02-21 13:24:36 +01002515 receive to its peers. The default value is the largest between 16384 and the
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002516 buffer size (tune.bufsize). In any case, HAProxy will not announce support
Willy Tarreaua24b35c2019-02-21 13:24:36 +01002517 for frame sizes larger than buffers. The main purpose of this setting is to
2518 allow to limit the maximum frame size setting when using large buffers. Too
2519 large frame sizes might have performance impact or cause some peers to
2520 misbehave. It is highly recommended not to change this value.
2521
Willy Tarreau193b8c62012-11-22 00:17:38 +01002522tune.http.cookielen <number>
2523 Sets the maximum length of captured cookies. This is the maximum value that
2524 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
2525 will automatically be truncated to this one. It is important not to set too
2526 high a value because all cookie captures still allocate this size whatever
2527 their configured value (they share a same pool). This value is per request
2528 per response, so the memory allocated is twice this value per connection.
2529 When not specified, the limit is set to 63 characters. It is recommended not
2530 to change this value.
2531
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002532tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002533 Sets the maximum length of request URI in logs. This prevents truncating long
2534 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002535 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002536 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002537 configuration directives too.
2538 The default value is 1024.
2539
Willy Tarreauac1932d2011-10-24 19:14:41 +02002540tune.http.maxhdr <number>
2541 Sets the maximum number of headers in a request. When a request comes with a
2542 number of headers greater than this value (including the first line), it is
2543 rejected with a "400 Bad Request" status code. Similarly, too large responses
2544 are blocked with "502 Bad Gateway". The default value is 101, which is enough
2545 for all usages, considering that the widely deployed Apache server uses the
2546 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02002547 a buggy application to work by the time it gets fixed. The accepted range is
2548 1..32767. Keep in mind that each new header consumes 32bits of memory for
2549 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02002550
Willy Tarreau76cc6992020-07-01 18:49:24 +02002551tune.idle-pool.shared { on | off }
2552 Enables ('on') or disables ('off') sharing of idle connection pools between
2553 threads for a same server. The default is to share them between threads in
2554 order to minimize the number of persistent connections to a server, and to
2555 optimize the connection reuse rate. But to help with debugging or when
2556 suspecting a bug in HAProxy around connection reuse, it can be convenient to
2557 forcefully disable this idle pool sharing between multiple threads, and force
Willy Tarreau0784db82021-02-19 11:45:22 +01002558 this option to "off". The default is on. It is strongly recommended against
2559 disabling this option without setting a conservative value on "pool-low-conn"
2560 for all servers relying on connection reuse to achieve a high performance
2561 level, otherwise connections might be closed very often as the thread count
2562 increases.
Willy Tarreau76cc6992020-07-01 18:49:24 +02002563
Willy Tarreau7e312732014-02-12 16:35:14 +01002564tune.idletimer <timeout>
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002565 Sets the duration after which HAProxy will consider that an empty buffer is
Willy Tarreau7e312732014-02-12 16:35:14 +01002566 probably associated with an idle stream. This is used to optimally adjust
2567 some packet sizes while forwarding large and small data alternatively. The
2568 decision to use splice() or to send large buffers in SSL is modulated by this
2569 parameter. The value is in milliseconds between 0 and 65535. A value of zero
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002570 means that HAProxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01002571 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05002572 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01002573 tune.ssl.maxrecord below.
2574
Willy Tarreau7ac908b2019-02-27 12:02:18 +01002575tune.listener.multi-queue { on | off }
2576 Enables ('on') or disables ('off') the listener's multi-queue accept which
2577 spreads the incoming traffic to all threads a "bind" line is allowed to run
2578 on instead of taking them for itself. This provides a smoother traffic
2579 distribution and scales much better, especially in environments where threads
2580 may be unevenly loaded due to external activity (network interrupts colliding
2581 with one thread for example). This option is enabled by default, but it may
2582 be forcefully disabled for troubleshooting or for situations where it is
2583 estimated that the operating system already provides a good enough
2584 distribution and connections are extremely short-lived.
2585
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002586tune.lua.forced-yield <number>
2587 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01002588 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002589 HAProxy scheduler to process other tasks like accepting connections or
2590 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01002591 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002592 lowered. If the Lua code is quite long and its result is absolutely required
2593 to process the data, the <number> can be increased.
2594
Willy Tarreau32f61e22015-03-18 17:54:59 +01002595tune.lua.maxmem
2596 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
2597 default it is zero which means unlimited. It is important to set a limit to
2598 ensure that a bug in a script will not result in the system running out of
2599 memory.
2600
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002601tune.lua.session-timeout <timeout>
2602 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002603 preventing infinite loops or spending too much time in Lua. This timeout
2604 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002605 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002606
2607tune.lua.task-timeout <timeout>
2608 Purpose is the same as "tune.lua.session-timeout", but this timeout is
2609 dedicated to the tasks. By default, this timeout isn't set because a task may
2610 remain alive during of the lifetime of HAProxy. For example, a task used to
2611 check servers.
2612
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002613tune.lua.service-timeout <timeout>
2614 This is the execution timeout for the Lua services. This is useful for
2615 preventing infinite loops or spending too much time in Lua. This timeout
2616 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002617 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002618
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002619tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01002620 Sets the maximum number of consecutive connections a process may accept in a
2621 row before switching to other work. In single process mode, higher numbers
Willy Tarreau66161322021-02-19 15:50:27 +01002622 used to give better performance at high connection rates, though this is not
2623 the case anymore with the multi-queue. This value applies individually to
2624 each listener, so that the number of processes a listener is bound to is
2625 taken into account. This value defaults to 4 which showed best results. If a
2626 significantly higher value was inherited from an ancient config, it might be
2627 worth removing it as it will both increase performance and lower response
2628 time. In multi-process mode, it is divided by twice the number of processes
2629 the listener is bound to. Setting this value to -1 completely disables the
2630 limitation. It should normally not be needed to tweak this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002631
2632tune.maxpollevents <number>
2633 Sets the maximum amount of events that can be processed at once in a call to
2634 the polling system. The default value is adapted to the operating system. It
2635 has been noticed that reducing it below 200 tends to slightly decrease
2636 latency at the expense of network bandwidth, and increasing it above 200
2637 tends to trade latency for slightly increased bandwidth.
2638
Willy Tarreau27a674e2009-08-17 07:23:33 +02002639tune.maxrewrite <number>
2640 Sets the reserved buffer space to this size in bytes. The reserved space is
2641 used for header rewriting or appending. The first reads on sockets will never
2642 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
2643 bufsize, though that does not make much sense since there are rarely large
2644 numbers of headers to add. Setting it too high prevents processing of large
2645 requests or responses. Setting it too low prevents addition of new headers
2646 to already large requests or to POST requests. It is generally wise to set it
2647 to about 1024. It is automatically readjusted to half of bufsize if it is
2648 larger than that. This means you don't have to worry about it when changing
2649 bufsize.
2650
Willy Tarreauf3045d22015-04-29 16:24:50 +02002651tune.pattern.cache-size <number>
2652 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
2653 cache which reminds previous lookups and their results. It is used by ACLs
2654 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
2655 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
2656 strings. It applies to pattern expressions which means that it will be able
2657 to memorize the result of a lookup among all the patterns specified on a
2658 configuration line (including all those loaded from files). It automatically
2659 invalidates entries which are updated using HTTP actions or on the CLI. The
2660 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau403bfbb2019-10-23 06:59:31 +02002661 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
2662 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02002663 risk of collision in this cache, which is in the order of the size of the
2664 cache divided by 2^64. Typically, at 10000 requests per second with the
2665 default cache size of 10000 entries, there's 1% chance that a brute force
2666 attack could cause a single collision after 60 years, or 0.1% after 6 years.
2667 This is considered much lower than the risk of a memory corruption caused by
2668 aging components. If this is not acceptable, the cache can be disabled by
2669 setting this parameter to 0.
2670
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02002671tune.pipesize <number>
2672 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
2673 are the default size for the system. But sometimes when using TCP splicing,
2674 it can improve performance to increase pipe sizes, especially if it is
2675 suspected that pipes are not filled and that many calls to splice() are
2676 performed. This has an impact on the kernel's memory footprint, so this must
2677 not be changed if impacts are not understood.
2678
Olivier Houchard88698d92019-04-16 19:07:22 +02002679tune.pool-high-fd-ratio <number>
2680 This setting sets the max number of file descriptors (in percentage) used by
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002681 HAProxy globally against the maximum number of file descriptors HAProxy can
Olivier Houchard88698d92019-04-16 19:07:22 +02002682 use before we start killing idle connections when we can't reuse a connection
2683 and we have to create a new one. The default is 25 (one quarter of the file
2684 descriptor will mean that roughly half of the maximum front connections can
2685 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05002686 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02002687
Willy Tarreau83ca3052020-07-01 18:30:16 +02002688tune.pool-low-fd-ratio <number>
2689 This setting sets the max number of file descriptors (in percentage) used by
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002690 HAProxy globally against the maximum number of file descriptors HAProxy can
Willy Tarreau83ca3052020-07-01 18:30:16 +02002691 use before we stop putting connection into the idle pool for reuse. The
2692 default is 20.
2693
Willy Tarreaue803de22010-01-21 17:43:04 +01002694tune.rcvbuf.client <number>
2695tune.rcvbuf.server <number>
2696 Forces the kernel socket receive buffer size on the client or the server side
2697 to the specified value in bytes. This value applies to all TCP/HTTP frontends
2698 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002699 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002700 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002701 order to save kernel memory by preventing it from buffering too large amounts
2702 of received data. Lower values will significantly increase CPU usage though.
2703
Willy Tarreaub22fc302015-12-14 12:04:35 +01002704tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002705 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01002706 socket buffers. One of them is that a read returns more than <recv_enough>
2707 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
2708 may be changed by this setting to better deal with workloads involving lots
2709 of short messages such as telnet or SSH sessions.
2710
Olivier Houchard1599b802018-05-24 18:59:04 +02002711tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05002712 Sets the maximum amount of task that can be processed at once when running
Willy Tarreau060a7612021-03-10 11:06:26 +01002713 tasks. The default value depends on the number of threads but sits between 35
2714 and 280, which tend to show the highest request rates and lowest latencies.
2715 Increasing it may incur latency when dealing with I/Os, making it too small
2716 can incur extra overhead. Higher thread counts benefit from lower values.
2717 When experimenting with much larger values, it may be useful to also enable
2718 tune.sched.low-latency and possibly tune.fd.edge-triggered to limit the
2719 maximum latency to the lowest possible.
Willy Tarreaue7723bd2020-06-24 11:11:02 +02002720
2721tune.sched.low-latency { on | off }
2722 Enables ('on') or disables ('off') the low-latency task scheduler. By default
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002723 HAProxy processes tasks from several classes one class at a time as this is
Willy Tarreaue7723bd2020-06-24 11:11:02 +02002724 the most efficient. But when running with large values of tune.runqueue-depth
2725 this can have a measurable effect on request or connection latency. When this
2726 low-latency setting is enabled, tasks of lower priority classes will always
2727 be executed before other ones if they exist. This will permit to lower the
2728 maximum latency experienced by new requests or connections in the middle of
2729 massive traffic, at the expense of a higher impact on this large traffic.
2730 For regular usage it is better to leave this off. The default value is off.
Olivier Houchard1599b802018-05-24 18:59:04 +02002731
Willy Tarreaue803de22010-01-21 17:43:04 +01002732tune.sndbuf.client <number>
2733tune.sndbuf.server <number>
2734 Forces the kernel socket send buffer size on the client or the server side to
2735 the specified value in bytes. This value applies to all TCP/HTTP frontends
2736 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002737 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002738 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002739 order to save kernel memory by preventing it from buffering too large amounts
2740 of received data. Lower values will significantly increase CPU usage though.
2741 Another use case is to prevent write timeouts with extremely slow clients due
2742 to the kernel waiting for a large part of the buffer to be read before
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002743 notifying HAProxy again.
Willy Tarreaue803de22010-01-21 17:43:04 +01002744
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002745tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01002746 Sets the size of the global SSL session cache, in a number of blocks. A block
William Dauchy9a4bbfe2021-02-12 15:58:46 +01002747 is large enough to contain an encoded session without peer certificate. An
2748 encoded session with peer certificate is stored in multiple blocks depending
2749 on the size of the peer certificate. A block uses approximately 200 bytes of
2750 memory (based on `sizeof(struct sh_ssl_sess_hdr) + SHSESS_BLOCK_MIN_SIZE`
2751 calculation used for `shctx_init` function). The default value may be forced
2752 at build time, otherwise defaults to 20000. When the cache is full, the most
2753 idle entries are purged and reassigned. Higher values reduce the occurrence
2754 of such a purge, hence the number of CPU-intensive SSL handshakes by ensuring
2755 that all users keep their session as long as possible. All entries are
2756 pre-allocated upon startup and are shared between all processes if "nbproc"
2757 is greater than 1. Setting this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002758
Emeric Brun8dc60392014-05-09 13:52:00 +02002759tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02002760 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02002761 should normally not be used since it will force many renegotiations due to
2762 clients hitting a random process. But it may be required on some operating
2763 systems where none of the SSL cache synchronization method may be used. In
2764 this case, adding a first layer of hash-based load balancing before the SSL
2765 layer might limit the impact of the lack of session sharing.
2766
William Lallemand7d42ef52020-07-06 11:41:30 +02002767tune.ssl.keylog { on | off }
2768 This option activates the logging of the TLS keys. It should be used with
2769 care as it will consume more memory per SSL session and could decrease
2770 performances. This is disabled by default.
2771
2772 These sample fetches should be used to generate the SSLKEYLOGFILE that is
2773 required to decipher traffic with wireshark.
2774
2775 https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/Key_Log_Format
2776
2777 The SSLKEYLOG is a series of lines which are formatted this way:
2778
2779 <Label> <space> <ClientRandom> <space> <Secret>
2780
2781 The ClientRandom is provided by the %[ssl_fc_client_random,hex] sample
2782 fetch, the secret and the Label could be find in the array below. You need
2783 to generate a SSLKEYLOGFILE with all the labels in this array.
2784
2785 The following sample fetches are hexadecimal strings and does not need to be
2786 converted.
2787
2788 SSLKEYLOGFILE Label | Sample fetches for the Secrets
2789 --------------------------------|-----------------------------------------
2790 CLIENT_EARLY_TRAFFIC_SECRET | %[ssl_fc_client_early_traffic_secret]
2791 CLIENT_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_client_handshake_traffic_secret]
2792 SERVER_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_server_handshake_traffic_secret]
2793 CLIENT_TRAFFIC_SECRET_0 | %[ssl_fc_client_traffic_secret_0]
2794 SERVER_TRAFFIC_SECRET_0 | %[ssl_fc_server_traffic_secret_0]
William Lallemandd742b6c2020-07-07 10:14:56 +02002795 EXPORTER_SECRET | %[ssl_fc_exporter_secret]
2796 EARLY_EXPORTER_SECRET | %[ssl_fc_early_exporter_secret]
William Lallemand7d42ef52020-07-06 11:41:30 +02002797
2798 This is only available with OpenSSL 1.1.1, and useful with TLS1.3 session.
2799
2800 If you want to generate the content of a SSLKEYLOGFILE with TLS < 1.3, you
2801 only need this line:
2802
2803 "CLIENT_RANDOM %[ssl_fc_client_random,hex] %[ssl_fc_session_key,hex]"
2804
Emeric Brun4f65bff2012-11-16 15:11:00 +01002805tune.ssl.lifetime <timeout>
2806 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002807 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01002808 does not guarantee that sessions will last that long, because if the cache is
2809 full, the longest idle sessions will be purged despite their configured
2810 lifetime. The real usefulness of this setting is to prevent sessions from
2811 being used for too long.
2812
Willy Tarreaubfd59462013-02-21 07:46:09 +01002813tune.ssl.maxrecord <number>
2814 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
2815 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
2816 data only once it has received a full record. With large records, it means
2817 that clients might have to download up to 16kB of data before starting to
2818 process them. Limiting the value can improve page load times on browsers
2819 located over high latency or low bandwidth networks. It is suggested to find
2820 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
2821 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
2822 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
2823 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002824 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01002825 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01002826
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002827tune.ssl.default-dh-param <number>
2828 Sets the maximum size of the Diffie-Hellman parameters used for generating
2829 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
2830 final size will try to match the size of the server's RSA (or DSA) key (e.g,
2831 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
Willy Tarreau3ba77d22020-05-08 09:31:18 +02002832 this maximum value. Default value if 2048. Only 1024 or higher values are
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002833 allowed. Higher values will increase the CPU load, and values greater than
2834 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02002835 used if static Diffie-Hellman parameters are supplied either directly
2836 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002837
Christopher Faulet31af49d2015-06-09 17:29:50 +02002838tune.ssl.ssl-ctx-cache-size <number>
2839 Sets the size of the cache used to store generated certificates to <number>
2840 entries. This is a LRU cache. Because generating a SSL certificate
2841 dynamically is expensive, they are cached. The default cache size is set to
2842 1000 entries.
2843
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01002844tune.ssl.capture-cipherlist-size <number>
2845 Sets the maximum size of the buffer used for capturing client-hello cipher
2846 list. If the value is 0 (default value) the capture is disabled, otherwise
2847 a buffer is allocated for each SSL/TLS connection.
2848
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002849tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002850tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002851tune.vars.reqres-max-size <size>
2852tune.vars.sess-max-size <size>
2853tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002854 These five tunes help to manage the maximum amount of memory used by the
2855 variables system. "global" limits the overall amount of memory available for
2856 all scopes. "proc" limits the memory for the process scope, "sess" limits the
2857 memory for the session scope, "txn" for the transaction scope, and "reqres"
2858 limits the memory for each request or response processing.
2859 Memory accounting is hierarchical, meaning more coarse grained limits include
2860 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
2861 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002862
Daniel Schneller0b547052016-03-21 20:46:57 +01002863 For example, when "tune.vars.sess-max-size" is limited to 100,
2864 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
2865 100 either. If we create a variable "txn.var" that contains 100 bytes,
2866 all available space is consumed.
2867 Notice that exceeding the limits at runtime will not result in an error
2868 message, but values might be cut off or corrupted. So make sure to accurately
2869 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002870
William Lallemanda509e4c2012-11-07 16:54:34 +01002871tune.zlib.memlevel <number>
2872 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002873 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002874 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002875 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002876 between 1 and 9. The default value is 8.
2877
2878tune.zlib.windowsize <number>
2879 Sets the window size (the size of the history buffer) as a parameter of the
2880 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002881 in better compression at the expense of memory usage. Can be a value between
2882 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002883
Willy Tarreauc57f0e22009-05-10 13:12:33 +020028843.3. Debugging
2885--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002886
Willy Tarreau6a06a402007-07-15 20:15:28 +02002887quiet
2888 Do not display any message during startup. It is equivalent to the command-
2889 line argument "-q".
2890
Willy Tarreau3eb10b82020-04-15 16:42:39 +02002891zero-warning
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002892 When this option is set, HAProxy will refuse to start if any warning was
Willy Tarreau3eb10b82020-04-15 16:42:39 +02002893 emitted while processing the configuration. It is highly recommended to set
2894 this option on configurations that are not changed often, as it helps detect
2895 subtle mistakes and keep the configuration clean and forward-compatible. Note
2896 that "haproxy -c" will also report errors in such a case. This option is
2897 equivalent to command line argument "-dW".
2898
Emeric Brunf099e792010-09-27 12:05:28 +02002899
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010029003.4. Userlists
2901--------------
2902It is possible to control access to frontend/backend/listen sections or to
2903http stats by allowing only authenticated and authorized users. To do this,
2904it is required to create at least one userlist and to define users.
2905
2906userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002907 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002908 used to store authentication & authorization data for independent customers.
2909
2910group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002911 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002912 attach users to this group by using a comma separated list of names
2913 proceeded by "users" keyword.
2914
Cyril Bontéf0c60612010-02-06 14:44:47 +01002915user <username> [password|insecure-password <password>]
2916 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002917 Adds user <username> to the current userlist. Both secure (encrypted) and
2918 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002919 evaluated using the crypt(3) function, so depending on the system's
2920 capabilities, different algorithms are supported. For example, modern Glibc
2921 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2922 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002923
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002924 Attention: Be aware that using encrypted passwords might cause significantly
2925 increased CPU usage, depending on the number of requests, and the algorithm
2926 used. For any of the hashed variants, the password for each request must
2927 be processed through the chosen algorithm, before it can be compared to the
2928 value specified in the config file. Most current algorithms are deliberately
2929 designed to be expensive to compute to achieve resistance against brute
2930 force attacks. They do not simply salt/hash the clear text password once,
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002931 but thousands of times. This can quickly become a major factor in HAProxy's
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002932 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002933
2934 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002935 userlist L1
2936 group G1 users tiger,scott
2937 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002938
Cyril Bontéf0c60612010-02-06 14:44:47 +01002939 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2940 user scott insecure-password elgato
2941 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002942
Cyril Bontéf0c60612010-02-06 14:44:47 +01002943 userlist L2
2944 group G1
2945 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002946
Cyril Bontéf0c60612010-02-06 14:44:47 +01002947 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2948 user scott insecure-password elgato groups G1,G2
2949 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002950
2951 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002952
Emeric Brunf099e792010-09-27 12:05:28 +02002953
29543.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002955----------
Emeric Brun94900952015-06-11 18:25:54 +02002956It is possible to propagate entries of any data-types in stick-tables between
Daniel Corbett9f0843f2021-05-08 10:50:37 -04002957several HAProxy instances over TCP connections in a multi-master fashion. Each
Emeric Brun94900952015-06-11 18:25:54 +02002958instance pushes its local updates and insertions to remote peers. The pushed
2959values overwrite remote ones without aggregation. Interrupted exchanges are
2960automatically detected and recovered from the last known point.
2961In addition, during a soft restart, the old process connects to the new one
2962using such a TCP connection to push all its entries before the new process
2963tries to connect to other peers. That ensures very fast replication during a
2964reload, it typically takes a fraction of a second even for large tables.
2965Note that Server IDs are used to identify servers remotely, so it is important
2966that configurations look similar or at least that the same IDs are forced on
2967each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002968
2969peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002970 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002971 which is referenced by one or more stick-tables.
2972
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002973bind [<address>]:<port_range> [, ...] [param*]
2974 Defines the binding parameters of the local peer of this "peers" section.
2975 Such lines are not supported with "peer" line in the same "peers" section.
2976
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002977disabled
2978 Disables a peers section. It disables both listening and any synchronization
2979 related to this section. This is provided to disable synchronization of stick
2980 tables without having to comment out all "peers" references.
2981
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002982default-bind [param*]
2983 Defines the binding parameters for the local peer, excepted its address.
2984
2985default-server [param*]
2986 Change default options for a server in a "peers" section.
2987
2988 Arguments:
2989 <param*> is a list of parameters for this server. The "default-server"
2990 keyword accepts an important number of options and has a complete
2991 section dedicated to it. Please refer to section 5 for more
2992 details.
2993
2994
2995 See also: "server" and section 5 about server options
2996
Emeric Brun7214dcf2021-09-29 10:29:52 +02002997enabled
2998 This re-enables a peers section which was previously disabled via the
2999 "disabled" keyword.
Willy Tarreau77e4bd12015-05-01 20:02:17 +02003000
Jan Wagner3e678602020-12-17 22:22:32 +01003001log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailleb6f759b2019-11-05 09:57:45 +01003002 <facility> [<level> [<minlevel>]]
3003 "peers" sections support the same "log" keyword as for the proxies to
3004 log information about the "peers" listener. See "log" option for proxies for
3005 more details.
3006
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01003007peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02003008 Defines a peer inside a peers section.
3009 If <peername> is set to the local peer name (by default hostname, or forced
Dragan Dosen13cd54c2020-06-18 18:24:05 +02003010 using "-L" command line option or "localpeer" global configuration setting),
Daniel Corbett9f0843f2021-05-08 10:50:37 -04003011 HAProxy will listen for incoming remote peer connection on <ip>:<port>.
Dragan Dosen13cd54c2020-06-18 18:24:05 +02003012 Otherwise, <ip>:<port> defines where to connect to in order to join the
3013 remote peer, and <peername> is used at the protocol level to identify and
3014 validate the remote peer on the server side.
Emeric Brunf099e792010-09-27 12:05:28 +02003015
3016 During a soft restart, local peer <ip>:<port> is used by the old instance to
3017 connect the new one and initiate a complete replication (teaching process).
3018
3019 It is strongly recommended to have the exact same peers declaration on all
Dragan Dosen13cd54c2020-06-18 18:24:05 +02003020 peers and to only rely on the "-L" command line argument or the "localpeer"
3021 global configuration setting to change the local peer name. This makes it
3022 easier to maintain coherent configuration files across all peers.
Emeric Brunf099e792010-09-27 12:05:28 +02003023
William Lallemandb2f07452015-05-12 14:27:13 +02003024 You may want to reference some environment variables in the address
3025 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01003026
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01003027 Note: "peer" keyword may transparently be replaced by "server" keyword (see
3028 "server" keyword explanation below).
3029
3030server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02003031 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01003032 with a support for all "server" parameters found in 5.2 paragraph.
3033 If the underlying peer is local, <ip>:<port> parameters must not be present.
3034 These parameters must be provided on a "bind" line (see "bind" keyword
3035 of this "peers" section).
3036 Some of these parameters are irrelevant for "peers" sections.
3037
3038
Cyril Bontédc4d9032012-04-08 21:57:39 +02003039 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01003040 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02003041 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01003042 peer haproxy1 192.168.0.1:1024
3043 peer haproxy2 192.168.0.2:1024
3044 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02003045
3046 backend mybackend
3047 mode tcp
3048 balance roundrobin
3049 stick-table type ip size 20k peers mypeers
3050 stick on src
3051
Willy Tarreauf7b30a92010-12-06 22:59:17 +01003052 server srv1 192.168.0.30:80
3053 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02003054
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01003055 Example:
3056 peers mypeers
3057 bind 127.0.0.11:10001 ssl crt mycerts/pem
3058 default-server ssl verify none
3059 server hostA 127.0.0.10:10000
3060 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02003061
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01003062
3063table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
3064 size <size> [expire <expire>] [nopurge] [store <data_type>]*
3065
3066 Configure a stickiness table for the current section. This line is parsed
3067 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05003068 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01003069 mandatory first parameter to designate the stick-table. Contrary to others
3070 sections, there may be several "table" lines in "peers" sections (see also
3071 "stick-table" keyword).
3072
3073 Also be aware of the fact that "peers" sections have their own stick-table
3074 namespaces to avoid collisions between stick-table names identical in
3075 different "peers" section. This is internally handled prepending the "peers"
3076 sections names to the name of the stick-tables followed by a '/' character.
3077 If somewhere else in the configuration file you have to refer to such
3078 stick-tables declared in "peers" sections you must use the prefixed version
3079 of the stick-table name as follows:
3080
3081 peers mypeers
3082 peer A ...
3083 peer B ...
3084 table t1 ...
3085
3086 frontend fe1
3087 tcp-request content track-sc0 src table mypeers/t1
3088
3089 This is also this prefixed version of the stick-table names which must be
3090 used to refer to stick-tables through the CLI.
3091
3092 About "peers" protocol, as only "peers" belonging to the same section may
3093 communicate with each others, there is no need to do such a distinction.
3094 Several "peers" sections may declare stick-tables with the same name.
3095 This is shorter version of the stick-table name which is sent over the network.
3096 There is only a '/' character as prefix to avoid stick-table name collisions between
3097 stick-tables declared as backends and stick-table declared in "peers" sections
3098 as follows in this weird but supported configuration:
3099
3100 peers mypeers
3101 peer A ...
3102 peer B ...
3103 table t1 type string size 10m store gpc0
3104
3105 backend t1
3106 stick-table type string size 10m store gpc0 peers mypeers
3107
Daniel Corbett67a82712020-07-06 23:01:19 -04003108 Here "t1" table declared in "mypeers" section has "mypeers/t1" as global name.
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01003109 "t1" table declared as a backend as "t1" as global name. But at peer protocol
3110 level the former table is named "/t1", the latter is again named "t1".
3111
Simon Horman51a1cf62015-02-03 13:00:44 +090031123.6. Mailers
3113------------
3114It is possible to send email alerts when the state of servers changes.
3115If configured email alerts are sent to each mailer that is configured
3116in a mailers section. Email is sent to mailers using SMTP.
3117
Pieter Baauw386a1272015-08-16 15:26:24 +02003118mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09003119 Creates a new mailer list with the name <mailersect>. It is an
3120 independent section which is referenced by one or more proxies.
3121
3122mailer <mailername> <ip>:<port>
3123 Defines a mailer inside a mailers section.
3124
3125 Example:
3126 mailers mymailers
3127 mailer smtp1 192.168.0.1:587
3128 mailer smtp2 192.168.0.2:587
3129
3130 backend mybackend
3131 mode tcp
3132 balance roundrobin
3133
3134 email-alert mailers mymailers
3135 email-alert from test1@horms.org
3136 email-alert to test2@horms.org
3137
3138 server srv1 192.168.0.30:80
3139 server srv2 192.168.0.31:80
3140
Pieter Baauw235fcfc2016-02-13 15:33:40 +01003141timeout mail <time>
3142 Defines the time available for a mail/connection to be made and send to
3143 the mail-server. If not defined the default value is 10 seconds. To allow
3144 for at least two SYN-ACK packets to be send during initial TCP handshake it
3145 is advised to keep this value above 4 seconds.
3146
3147 Example:
3148 mailers mymailers
3149 timeout mail 20s
3150 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09003151
William Lallemandc9515522019-06-12 16:32:11 +020031523.7. Programs
3153-------------
3154In master-worker mode, it is possible to launch external binaries with the
3155master, these processes are called programs. These programs are launched and
3156managed the same way as the workers.
3157
3158During a reload of HAProxy, those processes are dealing with the same
3159sequence as a worker:
3160
3161 - the master is re-executed
3162 - the master sends a SIGUSR1 signal to the program
3163 - if "option start-on-reload" is not disabled, the master launches a new
3164 instance of the program
3165
3166During a stop, or restart, a SIGTERM is sent to the programs.
3167
3168program <name>
3169 This is a new program section, this section will create an instance <name>
3170 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
3171 the management guide).
3172
3173command <command> [arguments*]
3174 Define the command to start with optional arguments. The command is looked
3175 up in the current PATH if it does not include an absolute path. This is a
3176 mandatory option of the program section. Arguments containing spaces must
3177 be enclosed in quotes or double quotes or be prefixed by a backslash.
3178
Andrew Heberle97236962019-07-12 11:50:26 +08003179user <user name>
3180 Changes the executed command user ID to the <user name> from /etc/passwd.
3181 See also "group".
3182
3183group <group name>
3184 Changes the executed command group ID to the <group name> from /etc/group.
3185 See also "user".
3186
William Lallemandc9515522019-06-12 16:32:11 +02003187option start-on-reload
3188no option start-on-reload
3189 Start (or not) a new instance of the program upon a reload of the master.
3190 The default is to start a new instance. This option may only be used in a
3191 program section.
3192
3193
Christopher Faulet76edc0f2020-01-13 15:52:01 +010031943.8. HTTP-errors
3195----------------
3196
3197It is possible to globally declare several groups of HTTP errors, to be
3198imported afterwards in any proxy section. Same group may be referenced at
3199several places and can be fully or partially imported.
3200
3201http-errors <name>
3202 Create a new http-errors group with the name <name>. It is an independent
3203 section that may be referenced by one or more proxies using its name.
3204
3205errorfile <code> <file>
3206 Associate a file contents to an HTTP error code
3207
3208 Arguments :
3209 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02003210 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01003211 425, 429, 500, 501, 502, 503, and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003212
3213 <file> designates a file containing the full HTTP response. It is
3214 recommended to follow the common practice of appending ".http" to
3215 the filename so that people do not confuse the response with HTML
3216 error pages, and to use absolute paths, since files are read
3217 before any chroot is performed.
3218
3219 Please referrers to "errorfile" keyword in section 4 for details.
3220
3221 Example:
3222 http-errors website-1
3223 errorfile 400 /etc/haproxy/errorfiles/site1/400.http
3224 errorfile 404 /etc/haproxy/errorfiles/site1/404.http
3225 errorfile 408 /dev/null # work around Chrome pre-connect bug
3226
3227 http-errors website-2
3228 errorfile 400 /etc/haproxy/errorfiles/site2/400.http
3229 errorfile 404 /etc/haproxy/errorfiles/site2/404.http
3230 errorfile 408 /dev/null # work around Chrome pre-connect bug
3231
Emeric Brun99c453d2020-05-25 15:01:04 +020032323.9. Rings
3233----------
3234
3235It is possible to globally declare ring-buffers, to be used as target for log
3236servers or traces.
3237
3238ring <ringname>
3239 Creates a new ring-buffer with name <ringname>.
3240
3241description <text>
Daniel Corbett67a82712020-07-06 23:01:19 -04003242 The description is an optional description string of the ring. It will
Emeric Brun99c453d2020-05-25 15:01:04 +02003243 appear on CLI. By default, <name> is reused to fill this field.
3244
3245format <format>
3246 Format used to store events into the ring buffer.
3247
3248 Arguments:
3249 <format> is the log format used when generating syslog messages. It may be
3250 one of the following :
3251
3252 iso A message containing only the ISO date, followed by the text.
3253 The PID, process name and system name are omitted. This is
3254 designed to be used with a local log server.
3255
Emeric Brun0237c4e2020-11-27 16:24:34 +01003256 local Analog to rfc3164 syslog message format except that hostname
3257 field is stripped. This is the default.
3258 Note: option "log-send-hostname" switches the default to
3259 rfc3164.
3260
Emeric Brun99c453d2020-05-25 15:01:04 +02003261 raw A message containing only the text. The level, PID, date, time,
3262 process name and system name are omitted. This is designed to be
3263 used in containers or during development, where the severity
3264 only depends on the file descriptor used (stdout/stderr). This
3265 is the default.
3266
Emeric Brun0237c4e2020-11-27 16:24:34 +01003267 rfc3164 The RFC3164 syslog message format.
Emeric Brun99c453d2020-05-25 15:01:04 +02003268 (https://tools.ietf.org/html/rfc3164)
3269
3270 rfc5424 The RFC5424 syslog message format.
3271 (https://tools.ietf.org/html/rfc5424)
3272
3273 short A message containing only a level between angle brackets such as
3274 '<3>', followed by the text. The PID, date, time, process name
3275 and system name are omitted. This is designed to be used with a
3276 local log server. This format is compatible with what the systemd
3277 logger consumes.
3278
Emeric Brun54648852020-07-06 15:54:06 +02003279 priority A message containing only a level plus syslog facility between angle
3280 brackets such as '<63>', followed by the text. The PID, date, time,
3281 process name and system name are omitted. This is designed to be used
3282 with a local log server.
3283
Emeric Brun99c453d2020-05-25 15:01:04 +02003284 timed A message containing only a level between angle brackets such as
3285 '<3>', followed by ISO date and by the text. The PID, process
3286 name and system name are omitted. This is designed to be
3287 used with a local log server.
3288
3289maxlen <length>
3290 The maximum length of an event message stored into the ring,
3291 including formatted header. If an event message is longer than
3292 <length>, it will be truncated to this length.
3293
Emeric Brun494c5052020-05-28 11:13:15 +02003294server <name> <address> [param*]
3295 Used to configure a syslog tcp server to forward messages from ring buffer.
3296 This supports for all "server" parameters found in 5.2 paragraph. Some of
3297 these parameters are irrelevant for "ring" sections. Important point: there
3298 is little reason to add more than one server to a ring, because all servers
3299 will receive the exact same copy of the ring contents, and as such the ring
3300 will progress at the speed of the slowest server. If one server does not
3301 respond, it will prevent old messages from being purged and may block new
3302 messages from being inserted into the ring. The proper way to send messages
3303 to multiple servers is to use one distinct ring per log server, not to
Emeric Brun97556472020-05-30 01:42:45 +02003304 attach multiple servers to the same ring. Note that specific server directive
3305 "log-proto" is used to set the protocol used to send messages.
Emeric Brun494c5052020-05-28 11:13:15 +02003306
Emeric Brun99c453d2020-05-25 15:01:04 +02003307size <size>
3308 This is the optional size in bytes for the ring-buffer. Default value is
3309 set to BUFSIZE.
3310
Emeric Brun494c5052020-05-28 11:13:15 +02003311timeout connect <timeout>
3312 Set the maximum time to wait for a connection attempt to a server to succeed.
3313
3314 Arguments :
3315 <timeout> is the timeout value specified in milliseconds by default, but
3316 can be in any other unit if the number is suffixed by the unit,
3317 as explained at the top of this document.
3318
3319timeout server <timeout>
3320 Set the maximum time for pending data staying into output buffer.
3321
3322 Arguments :
3323 <timeout> is the timeout value specified in milliseconds by default, but
3324 can be in any other unit if the number is suffixed by the unit,
3325 as explained at the top of this document.
3326
Emeric Brun99c453d2020-05-25 15:01:04 +02003327 Example:
3328 global
3329 log ring@myring local7
3330
3331 ring myring
3332 description "My local buffer"
3333 format rfc3164
3334 maxlen 1200
3335 size 32764
Emeric Brun494c5052020-05-28 11:13:15 +02003336 timeout connect 5s
3337 timeout server 10s
Emeric Brun97556472020-05-30 01:42:45 +02003338 server mysyslogsrv 127.0.0.1:6514 log-proto octet-count
Emeric Brun99c453d2020-05-25 15:01:04 +02003339
Emeric Brun12941c82020-07-07 14:19:42 +020033403.10. Log forwarding
3341-------------------
3342
3343It is possible to declare one or multiple log forwarding section,
Daniel Corbett9f0843f2021-05-08 10:50:37 -04003344HAProxy will forward all received log messages to a log servers list.
Emeric Brun12941c82020-07-07 14:19:42 +02003345
3346log-forward <name>
3347 Creates a new log forwarder proxy identified as <name>.
3348
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003349backlog <conns>
3350 Give hints to the system about the approximate listen backlog desired size
3351 on connections accept.
3352
3353bind <addr> [param*]
3354 Used to configure a stream log listener to receive messages to forward.
Emeric Brunda46c1c2020-10-08 08:39:02 +02003355 This supports the "bind" parameters found in 5.1 paragraph including
3356 those about ssl but some statements such as "alpn" may be irrelevant for
3357 syslog protocol over TCP.
3358 Those listeners support both "Octet Counting" and "Non-Transparent-Framing"
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003359 modes as defined in rfc-6587.
3360
Willy Tarreau76aaa7f2020-09-16 15:07:22 +02003361dgram-bind <addr> [param*]
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003362 Used to configure a datagram log listener to receive messages to forward.
3363 Addresses must be in IPv4 or IPv6 form,followed by a port. This supports
3364 for some of the "bind" parameters found in 5.1 paragraph among which
3365 "interface", "namespace" or "transparent", the other ones being
Willy Tarreau26ff5da2020-09-16 15:22:19 +02003366 silently ignored as irrelevant for UDP/syslog case.
Emeric Brun12941c82020-07-07 14:19:42 +02003367
3368log global
Jan Wagner3e678602020-12-17 22:22:32 +01003369log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Emeric Brun12941c82020-07-07 14:19:42 +02003370 <facility> [<level> [<minlevel>]]
3371 Used to configure target log servers. See more details on proxies
3372 documentation.
Daniel Corbett9f0843f2021-05-08 10:50:37 -04003373 If no format specified, HAProxy tries to keep the incoming log format.
Emeric Brun12941c82020-07-07 14:19:42 +02003374 Configured facility is ignored, except if incoming message does not
3375 present a facility but one is mandatory on the outgoing format.
3376 If there is no timestamp available in the input format, but the field
Daniel Corbett9f0843f2021-05-08 10:50:37 -04003377 exists in output format, HAProxy will use the local date.
Emeric Brun12941c82020-07-07 14:19:42 +02003378
3379 Example:
3380 global
3381 log stderr format iso local7
3382
3383 ring myring
3384 description "My local buffer"
3385 format rfc5424
3386 maxlen 1200
3387 size 32764
3388 timeout connect 5s
3389 timeout server 10s
3390 # syslog tcp server
3391 server mysyslogsrv 127.0.0.1:514 log-proto octet-count
3392
3393 log-forward sylog-loadb
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003394 dgram-bind 127.0.0.1:1514
3395 bind 127.0.0.1:1514
Emeric Brun12941c82020-07-07 14:19:42 +02003396 # all messages on stderr
3397 log global
3398 # all messages on local tcp syslog server
3399 log ring@myring local0
3400 # load balance messages on 4 udp syslog servers
3401 log 127.0.0.1:10001 sample 1:4 local0
3402 log 127.0.0.1:10002 sample 2:4 local0
3403 log 127.0.0.1:10003 sample 3:4 local0
3404 log 127.0.0.1:10004 sample 4:4 local0
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003405
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003406maxconn <conns>
3407 Fix the maximum number of concurrent connections on a log forwarder.
3408 10 is the default.
3409
3410timeout client <timeout>
3411 Set the maximum inactivity time on the client side.
3412
Willy Tarreauc57f0e22009-05-10 13:12:33 +020034134. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02003414----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003415
Willy Tarreau6a06a402007-07-15 20:15:28 +02003416Proxy configuration can be located in a set of sections :
Willy Tarreau7c0b4d82021-02-12 14:58:08 +01003417 - defaults [<name>] [ from <defaults_name> ]
3418 - frontend <name> [ from <defaults_name> ]
3419 - backend <name> [ from <defaults_name> ]
3420 - listen <name> [ from <defaults_name> ]
Willy Tarreau6a06a402007-07-15 20:15:28 +02003421
3422A "frontend" section describes a set of listening sockets accepting client
3423connections.
3424
3425A "backend" section describes a set of servers to which the proxy will connect
3426to forward incoming connections.
3427
3428A "listen" section defines a complete proxy with its frontend and backend
3429parts combined in one section. It is generally useful for TCP-only traffic.
3430
Willy Tarreau7c0b4d82021-02-12 14:58:08 +01003431A "defaults" section resets all settings to the documented ones and presets new
3432ones for use by subsequent sections. All of "frontend", "backend" and "listen"
3433sections always take their initial settings from a defaults section, by default
3434the latest one that appears before the newly created section. It is possible to
3435explicitly designate a specific "defaults" section to load the initial settings
3436from by indicating its name on the section line after the optional keyword
3437"from". While "defaults" section do not impose a name, this use is encouraged
3438for better readability. It is also the only way to designate a specific section
3439to use instead of the default previous one. Since "defaults" section names are
3440optional, by default a very permissive check is applied on their name and these
3441are even permitted to overlap. However if a "defaults" section is referenced by
3442any other section, its name must comply with the syntax imposed on all proxy
3443names, and this name must be unique among the defaults sections. Please note
3444that regardless of what is currently permitted, it is recommended to avoid
3445duplicate section names in general and to respect the same syntax as for proxy
3446names. This rule might be enforced in a future version.
3447
3448Note that it is even possible for a defaults section to take its initial
3449settings from another one, and as such, inherit settings across multiple levels
3450of defaults sections. This can be convenient to establish certain configuration
3451profiles to carry groups of default settings (e.g. TCP vs HTTP or short vs long
3452timeouts) but can quickly become confusing to follow.
3453
Willy Tarreau0ba27502007-12-24 16:55:16 +01003454All proxy names must be formed from upper and lower case letters, digits,
3455'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
3456case-sensitive, which means that "www" and "WWW" are two different proxies.
3457
3458Historically, all proxy names could overlap, it just caused troubles in the
3459logs. Since the introduction of content switching, it is mandatory that two
3460proxies with overlapping capabilities (frontend/backend) have different names.
3461However, it is still permitted that a frontend and a backend share the same
3462name, as this configuration seems to be commonly encountered.
3463
3464Right now, two major proxy modes are supported : "tcp", also known as layer 4,
3465and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003466bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003467protocol, and can interact with it by allowing, blocking, switching, adding,
3468modifying, or removing arbitrary contents in requests or responses, based on
3469arbitrary criteria.
3470
Willy Tarreau70dffda2014-01-30 03:07:23 +01003471In HTTP mode, the processing applied to requests and responses flowing over
3472a connection depends in the combination of the frontend's HTTP options and
Julien Pivotto21ad3152019-12-10 13:11:17 +01003473the backend's. HAProxy supports 3 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01003474
3475 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
3476 requests and responses are processed, and connections remain open but idle
3477 between responses and new requests.
3478
Willy Tarreau70dffda2014-01-30 03:07:23 +01003479 - SCL: server close ("option http-server-close") : the server-facing
3480 connection is closed after the end of the response is received, but the
3481 client-facing connection remains open.
3482
Christopher Faulet315b39c2018-09-21 16:26:19 +02003483 - CLO: close ("option httpclose"): the connection is closed after the end of
3484 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003485
3486The effective mode that will be applied to a connection passing through a
3487frontend and a backend can be determined by both proxy modes according to the
3488following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02003489weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003490
Christopher Faulet315b39c2018-09-21 16:26:19 +02003491 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01003492
Christopher Faulet315b39c2018-09-21 16:26:19 +02003493 | KAL | SCL | CLO
3494 ----+-----+-----+----
3495 KAL | KAL | SCL | CLO
3496 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02003497 mode SCL | SCL | SCL | CLO
3498 ----+-----+-----+----
3499 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01003500
Christopher Faulet4d37e532021-03-26 14:44:00 +01003501It is possible to chain a TCP frontend to an HTTP backend. It is pointless if
Daniel Corbett9f0843f2021-05-08 10:50:37 -04003502only HTTP traffic is handled. But it may be used to handle several protocols
3503within the same frontend. In this case, the client's connection is first handled
Christopher Faulet4d37e532021-03-26 14:44:00 +01003504as a raw tcp connection before being upgraded to HTTP. Before the upgrade, the
Daniel Corbett9f0843f2021-05-08 10:50:37 -04003505content processings are performend on raw data. Once upgraded, data is parsed
Christopher Faulet4d37e532021-03-26 14:44:00 +01003506and stored using an internal representation called HTX and it is no longer
3507possible to rely on raw representation. There is no way to go back.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003508
Christopher Faulet4d37e532021-03-26 14:44:00 +01003509There are two kind of upgrades, in-place upgrades and destructive upgrades. The
Daniel Corbett9f0843f2021-05-08 10:50:37 -04003510first ones involves a TCP to HTTP/1 upgrade. In HTTP/1, the request
Christopher Faulet4d37e532021-03-26 14:44:00 +01003511processings are serialized, thus the applicative stream can be preserved. The
Daniel Corbett9f0843f2021-05-08 10:50:37 -04003512second one involves a TCP to HTTP/2 upgrade. Because it is a multiplexed
Christopher Faulet4d37e532021-03-26 14:44:00 +01003513protocol, the applicative stream cannot be associated to any HTTP/2 stream and
3514is destroyed. New applicative streams are then created when HAProxy receives
3515new HTTP/2 streams at the lower level, in the H2 multiplexer. It is important
Daniel Corbett9f0843f2021-05-08 10:50:37 -04003516to understand this difference because that drastically changes the way to
Christopher Faulet4d37e532021-03-26 14:44:00 +01003517process data. When an HTTP/1 upgrade is performed, the content processings
3518already performed on raw data are neither lost nor reexecuted while for an
3519HTTP/2 upgrade, applicative streams are distinct and all frontend rules are
3520evaluated systematically on each one. And as said, the first stream, the TCP
3521one, is destroyed, but only after the frontend rules were evaluated.
3522
3523There is another importnat point to understand when HTTP processings are
3524performed from a TCP proxy. While HAProxy is able to parse HTTP/1 in-fly from
3525tcp-request content rules, it is not possible for HTTP/2. Only the HTTP/2
3526preface can be parsed. This is a huge limitation regarding the HTTP content
3527analysis in TCP. Concretely it is only possible to know if received data are
3528HTTP. For instance, it is not possible to choose a backend based on the Host
3529header value while it is trivial in HTTP/1. Hopefully, there is a solution to
3530mitigate this drawback.
3531
Daniel Corbett9f0843f2021-05-08 10:50:37 -04003532There are two ways to perform an HTTP upgrade. The first one, the historical
Christopher Faulet4d37e532021-03-26 14:44:00 +01003533method, is to select an HTTP backend. The upgrade happens when the backend is
3534set. Thus, for in-place upgrades, only the backend configuration is considered
3535in the HTTP data processing. For destructive upgrades, the applicative stream
3536is destroyed, thus its processing is stopped. With this method, possibilities
3537to choose a backend with an HTTP/2 connection are really limited, as mentioned
3538above, and a bit useless because the stream is destroyed. The second method is
3539to upgrade during the tcp-request content rules evaluation, thanks to the
3540"switch-mode http" action. In this case, the upgrade is performed in the
3541frontend context and it is possible to define HTTP directives in this
3542frontend. For in-place upgrades, it offers all the power of the HTTP analysis
3543as soon as possible. It is not that far from an HTTP frontend. For destructive
3544upgrades, it does not change anything except it is useless to choose a backend
3545on limited information. It is of course the recommended method. Thus, testing
3546the request protocol from the tcp-request content rules to perform an HTTP
3547upgrade is enough. All the remaining HTTP manipulation may be moved to the
3548frontend http-request ruleset. But keep in mind that tcp-request content rules
3549remains evaluated on each streams, that can't be changed.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003550
Willy Tarreauc57f0e22009-05-10 13:12:33 +020035514.1. Proxy keywords matrix
3552--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003553
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003554The following list of keywords is supported. Most of them may only be used in a
3555limited set of section types. Some of them are marked as "deprecated" because
3556they are inherited from an old syntax which may be confusing or functionally
3557limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01003558marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003559option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02003560and must be disabled for a specific instance. Such options may also be prefixed
3561with "default" in order to restore default settings regardless of what has been
3562specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003563
Willy Tarreau6a06a402007-07-15 20:15:28 +02003564
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003565 keyword defaults frontend listen backend
3566------------------------------------+----------+----------+---------+---------
3567acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003568backlog X X X -
3569balance X - X X
3570bind - X X -
3571bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003572capture cookie - X X -
3573capture request header - X X -
3574capture response header - X X -
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003575clitcpka-cnt X X X -
3576clitcpka-idle X X X -
3577clitcpka-intvl X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02003578compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003579cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003580declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003581default-server X - X X
3582default_backend X X X -
3583description - X X X
3584disabled X X X X
3585dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09003586email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09003587email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09003588email-alert mailers X X X X
3589email-alert myhostname X X X X
3590email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003591enabled X X X X
3592errorfile X X X X
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003593errorfiles X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003594errorloc X X X X
3595errorloc302 X X X X
3596-- keyword -------------------------- defaults - frontend - listen -- backend -
3597errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003598force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003599filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003600fullconn X - X X
3601grace X X X X
3602hash-type X - X X
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01003603http-after-response - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003604http-check comment X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003605http-check connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003606http-check disable-on-404 X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003607http-check expect X - X X
Peter Gervai8912ae62020-06-11 18:26:36 +02003608http-check send X - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02003609http-check send-state X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003610http-check set-var X - X X
3611http-check unset-var X - X X
Christopher Faulet3b967c12020-05-15 15:47:44 +02003612http-error X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003613http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003614http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02003615http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02003616http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003617id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003618ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003619load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02003620log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01003621log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02003622log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01003623log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02003624max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003625maxconn X X X -
3626mode X X X X
3627monitor fail - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003628monitor-uri X X X -
3629option abortonclose (*) X - X X
3630option accept-invalid-http-request (*) X X X -
3631option accept-invalid-http-response (*) X - X X
3632option allbackups (*) X - X X
3633option checkcache (*) X - X X
3634option clitcpka (*) X X X -
3635option contstats (*) X X X -
Christopher Faulet89aed322020-06-02 17:33:56 +02003636option disable-h2-upgrade (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003637option dontlog-normal (*) X X X -
3638option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003639-- keyword -------------------------- defaults - frontend - listen -- backend -
3640option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02003641option h1-case-adjust-bogus-client (*) X X X -
3642option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02003643option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02003644option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01003645option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02003646option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02003647option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003648option http-server-close (*) X X X X
3649option http-use-proxy-header (*) X X X -
3650option httpchk X - X X
3651option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01003652option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003653option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003654option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02003655option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003656option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003657option log-health-checks (*) X - X X
3658option log-separate-errors (*) X X X -
3659option logasap (*) X X X -
3660option mysql-check X - X X
3661option nolinger (*) X X X X
3662option originalto X X X X
3663option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02003664option pgsql-check X - X X
3665option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003666option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02003667option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003668option smtpchk X - X X
3669option socket-stats (*) X X X -
3670option splice-auto (*) X X X X
3671option splice-request (*) X X X X
3672option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01003673option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003674option srvtcpka (*) X - X X
3675option ssl-hello-chk X - X X
3676-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01003677option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003678option tcp-smart-accept (*) X X X -
3679option tcp-smart-connect (*) X - X X
3680option tcpka X X X X
3681option tcplog X X X X
3682option transparent (*) X - X X
William Dauchy4711a892022-01-05 22:53:24 +01003683option idle-close-on-response (*) X X X -
Simon Horman98637e52014-06-20 12:30:16 +09003684external-check command X - X X
3685external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003686persist rdp-cookie X - X X
3687rate-limit sessions X X X -
3688redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003689-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003690retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02003691retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003692server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003693server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02003694server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003695source X - X X
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003696srvtcpka-cnt X - X X
3697srvtcpka-idle X - X X
3698srvtcpka-intvl X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02003699stats admin - X X X
3700stats auth X X X X
3701stats enable X X X X
3702stats hide-version X X X X
3703stats http-request - X X X
3704stats realm X X X X
3705stats refresh X X X X
3706stats scope X X X X
3707stats show-desc X X X X
3708stats show-legends X X X X
3709stats show-node X X X X
3710stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003711-- keyword -------------------------- defaults - frontend - listen -- backend -
3712stick match - - X X
3713stick on - - X X
3714stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02003715stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01003716stick-table - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003717tcp-check comment X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003718tcp-check connect X - X X
3719tcp-check expect X - X X
3720tcp-check send X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003721tcp-check send-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003722tcp-check send-binary X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003723tcp-check send-binary-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003724tcp-check set-var X - X X
3725tcp-check unset-var X - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02003726tcp-request connection - X X -
3727tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02003728tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02003729tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02003730tcp-response content - - X X
3731tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003732timeout check X - X X
3733timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02003734timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003735timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003736timeout http-keep-alive X X X X
3737timeout http-request X X X X
3738timeout queue X - X X
3739timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02003740timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003741timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02003742timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003743transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01003744unique-id-format X X X -
3745unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003746use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02003747use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02003748use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003749------------------------------------+----------+----------+---------+---------
3750 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02003751
Willy Tarreau0ba27502007-12-24 16:55:16 +01003752
Willy Tarreauc57f0e22009-05-10 13:12:33 +020037534.2. Alphabetically sorted keywords reference
3754---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003755
3756This section provides a description of each keyword and its usage.
3757
3758
3759acl <aclname> <criterion> [flags] [operator] <value> ...
3760 Declare or complete an access list.
3761 May be used in sections : defaults | frontend | listen | backend
3762 no | yes | yes | yes
3763 Example:
3764 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
3765 acl invalid_src src_port 0:1023
3766 acl local_dst hdr(host) -i localhost
3767
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003768 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003769
3770
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003771backlog <conns>
3772 Give hints to the system about the approximate listen backlog desired size
3773 May be used in sections : defaults | frontend | listen | backend
3774 yes | yes | yes | no
3775 Arguments :
3776 <conns> is the number of pending connections. Depending on the operating
3777 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02003778 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003779
3780 In order to protect against SYN flood attacks, one solution is to increase
3781 the system's SYN backlog size. Depending on the system, sometimes it is just
3782 tunable via a system parameter, sometimes it is not adjustable at all, and
3783 sometimes the system relies on hints given by the application at the time of
3784 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
3785 to the listen() syscall. On systems which can make use of this value, it can
3786 sometimes be useful to be able to specify a different value, hence this
3787 backlog parameter.
3788
3789 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
3790 used as a hint and the system accepts up to the smallest greater power of
3791 two, and never more than some limits (usually 32768).
3792
3793 See also : "maxconn" and the target operating system's tuning guide.
3794
3795
Willy Tarreau0ba27502007-12-24 16:55:16 +01003796balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02003797balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003798 Define the load balancing algorithm to be used in a backend.
3799 May be used in sections : defaults | frontend | listen | backend
3800 yes | no | yes | yes
3801 Arguments :
3802 <algorithm> is the algorithm used to select a server when doing load
3803 balancing. This only applies when no persistence information
3804 is available, or when a connection is redispatched to another
3805 server. <algorithm> may be one of the following :
3806
3807 roundrobin Each server is used in turns, according to their weights.
3808 This is the smoothest and fairest algorithm when the server's
3809 processing time remains equally distributed. This algorithm
3810 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02003811 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08003812 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02003813 large farms, when a server becomes up after having been down
3814 for a very short time, it may sometimes take a few hundreds
3815 requests for it to be re-integrated into the farm and start
3816 receiving traffic. This is normal, though very rare. It is
3817 indicated here in case you would have the chance to observe
3818 it, so that you don't worry.
3819
3820 static-rr Each server is used in turns, according to their weights.
3821 This algorithm is as similar to roundrobin except that it is
3822 static, which means that changing a server's weight on the
3823 fly will have no effect. On the other hand, it has no design
3824 limitation on the number of servers, and when a server goes
3825 up, it is always immediately reintroduced into the farm, once
3826 the full map is recomputed. It also uses slightly less CPU to
3827 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003828
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003829 leastconn The server with the lowest number of connections receives the
3830 connection. Round-robin is performed within groups of servers
3831 of the same load to ensure that all servers will be used. Use
3832 of this algorithm is recommended where very long sessions are
3833 expected, such as LDAP, SQL, TSE, etc... but is not very well
3834 suited for protocols using short sessions such as HTTP. This
3835 algorithm is dynamic, which means that server weights may be
Willy Tarreau8c855f62020-10-22 17:41:45 +02003836 adjusted on the fly for slow starts for instance. It will
3837 also consider the number of queued connections in addition to
3838 the established ones in order to minimize queuing.
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003839
Willy Tarreauf09c6602012-02-13 17:12:08 +01003840 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003841 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01003842 identifier to the highest (see server parameter "id"), which
3843 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02003844 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01003845 not make sense to use this algorithm without setting maxconn.
3846 The purpose of this algorithm is to always use the smallest
3847 number of servers so that extra servers can be powered off
3848 during non-intensive hours. This algorithm ignores the server
3849 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02003850 or IMAP than HTTP, though it can be useful there too. In
3851 order to use this algorithm efficiently, it is recommended
3852 that a cloud controller regularly checks server usage to turn
3853 them off when unused, and regularly checks backend queue to
3854 turn new servers on when the queue inflates. Alternatively,
3855 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01003856
Willy Tarreau0ba27502007-12-24 16:55:16 +01003857 source The source IP address is hashed and divided by the total
3858 weight of the running servers to designate which server will
3859 receive the request. This ensures that the same client IP
3860 address will always reach the same server as long as no
3861 server goes down or up. If the hash result changes due to the
3862 number of running servers changing, many clients will be
3863 directed to a different server. This algorithm is generally
3864 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003865 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01003866 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003867 static by default, which means that changing a server's
3868 weight on the fly will have no effect, but this can be
3869 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003870
Oskar Stolc8dc41842012-05-19 10:19:54 +01003871 uri This algorithm hashes either the left part of the URI (before
3872 the question mark) or the whole URI (if the "whole" parameter
3873 is present) and divides the hash value by the total weight of
3874 the running servers. The result designates which server will
3875 receive the request. This ensures that the same URI will
3876 always be directed to the same server as long as no server
3877 goes up or down. This is used with proxy caches and
3878 anti-virus proxies in order to maximize the cache hit rate.
3879 Note that this algorithm may only be used in an HTTP backend.
3880 This algorithm is static by default, which means that
3881 changing a server's weight on the fly will have no effect,
3882 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003883
Oskar Stolc8dc41842012-05-19 10:19:54 +01003884 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003885 "depth", both followed by a positive integer number. These
3886 options may be helpful when it is needed to balance servers
3887 based on the beginning of the URI only. The "len" parameter
3888 indicates that the algorithm should only consider that many
3889 characters at the beginning of the URI to compute the hash.
3890 Note that having "len" set to 1 rarely makes sense since most
3891 URIs start with a leading "/".
3892
3893 The "depth" parameter indicates the maximum directory depth
3894 to be used to compute the hash. One level is counted for each
3895 slash in the request. If both parameters are specified, the
3896 evaluation stops when either is reached.
3897
Willy Tarreau57a37412020-09-23 08:56:29 +02003898 A "path-only" parameter indicates that the hashing key starts
3899 at the first '/' of the path. This can be used to ignore the
3900 authority part of absolute URIs, and to make sure that HTTP/1
3901 and HTTP/2 URIs will provide the same hash.
3902
Willy Tarreau0ba27502007-12-24 16:55:16 +01003903 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003904 the query string of each HTTP GET request.
3905
3906 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02003907 request entity will be searched for the parameter argument,
3908 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02003909 ('?') in the URL. The message body will only start to be
3910 analyzed once either the advertised amount of data has been
3911 received or the request buffer is full. In the unlikely event
3912 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02003913 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02003914 be randomly balanced if at all. This keyword used to support
3915 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003916
3917 If the parameter is found followed by an equal sign ('=') and
3918 a value, then the value is hashed and divided by the total
3919 weight of the running servers. The result designates which
3920 server will receive the request.
3921
3922 This is used to track user identifiers in requests and ensure
3923 that a same user ID will always be sent to the same server as
3924 long as no server goes up or down. If no value is found or if
3925 the parameter is not found, then a round robin algorithm is
3926 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003927 backend. This algorithm is static by default, which means
3928 that changing a server's weight on the fly will have no
3929 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003930
Cyril Bontédc4d9032012-04-08 21:57:39 +02003931 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
3932 request. Just as with the equivalent ACL 'hdr()' function,
3933 the header name in parenthesis is not case sensitive. If the
3934 header is absent or if it does not contain any value, the
3935 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01003936
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003937 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01003938 reducing the hash algorithm to the main domain part with some
3939 specific headers such as 'Host'. For instance, in the Host
3940 value "haproxy.1wt.eu", only "1wt" will be considered.
3941
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003942 This algorithm is static by default, which means that
3943 changing a server's weight on the fly will have no effect,
3944 but this can be changed using "hash-type".
3945
Willy Tarreau21c741a2019-01-14 18:14:27 +01003946 random
3947 random(<draws>)
3948 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02003949 hashing function. This means that the servers' weights are
3950 respected, dynamic weight changes immediately take effect, as
3951 well as new server additions. Random load balancing can be
3952 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01003953 or removed as it may avoid the hammering effect that could
3954 result from roundrobin or leastconn in this situation. The
3955 hash-balance-factor directive can be used to further improve
3956 fairness of the load balancing, especially in situations
3957 where servers show highly variable response times. When an
3958 argument <draws> is present, it must be an integer value one
3959 or greater, indicating the number of draws before selecting
3960 the least loaded of these servers. It was indeed demonstrated
3961 that picking the least loaded of two servers is enough to
3962 significantly improve the fairness of the algorithm, by
3963 always avoiding to pick the most loaded server within a farm
3964 and getting rid of any bias that could be induced by the
3965 unfair distribution of the consistent list. Higher values N
3966 will take away N-1 of the highest loaded servers at the
3967 expense of performance. With very high values, the algorithm
3968 will converge towards the leastconn's result but much slower.
3969 The default value is 2, which generally shows very good
3970 distribution and performance. This algorithm is also known as
3971 the Power of Two Random Choices and is described here :
3972 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02003973
Emeric Brun736aa232009-06-30 17:56:00 +02003974 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02003975 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02003976 The RDP cookie <name> (or "mstshash" if omitted) will be
3977 looked up and hashed for each incoming TCP request. Just as
Christian Ruppertfd3184c2022-02-20 22:54:01 +01003978 with the equivalent ACL 'req.rdp_cookie()' function, the name
Emeric Brun736aa232009-06-30 17:56:00 +02003979 is not case-sensitive. This mechanism is useful as a degraded
3980 persistence mode, as it makes it possible to always send the
3981 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003982 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02003983 used instead.
3984
3985 Note that for this to work, the frontend must ensure that an
3986 RDP cookie is already present in the request buffer. For this
3987 you must use 'tcp-request content accept' rule combined with
Christian Ruppertfd3184c2022-02-20 22:54:01 +01003988 a 'req.rdp_cookie_cnt' ACL.
Emeric Brun736aa232009-06-30 17:56:00 +02003989
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003990 This algorithm is static by default, which means that
3991 changing a server's weight on the fly will have no effect,
3992 but this can be changed using "hash-type".
3993
Willy Tarreau0ba27502007-12-24 16:55:16 +01003994 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003995 algorithms. Right now, only "url_param" and "uri" support an
3996 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003997
Willy Tarreau3cd9af22009-03-15 14:06:41 +01003998 The load balancing algorithm of a backend is set to roundrobin when no other
3999 algorithm, mode nor option have been set. The algorithm may only be set once
4000 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004001
Lukas Tribus80512b12018-10-27 20:07:40 +02004002 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05004003 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02004004 to be routed to different backend servers, breaking the invalid assumptions
4005 NTLM relies on.
4006
Willy Tarreau0ba27502007-12-24 16:55:16 +01004007 Examples :
4008 balance roundrobin
4009 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02004010 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01004011 balance hdr(User-Agent)
4012 balance hdr(host)
4013 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02004014
4015 Note: the following caveats and limitations on using the "check_post"
4016 extension with "url_param" must be considered :
4017
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004018 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02004019 to determine if the parameters will be found in the body or entity which
4020 may contain binary data. Therefore another method may be required to
4021 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02004022 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02004023
4024 - using a <max_wait> value larger than the request buffer size does not
4025 make sense and is useless. The buffer size is set at build time, and
4026 defaults to 16 kB.
4027
4028 - Content-Encoding is not supported, the parameter search will probably
4029 fail; and load balancing will fall back to Round Robin.
4030
4031 - Expect: 100-continue is not supported, load balancing will fall back to
4032 Round Robin.
4033
Lukas Tribus23953682017-04-28 13:24:30 +00004034 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02004035 If the entire parameter value is not present in the first chunk, the
4036 selection of server is undefined (actually, defined by how little
4037 actually appeared in the first chunk).
4038
4039 - This feature does not support generation of a 100, 411 or 501 response.
4040
4041 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004042 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02004043 white space or control characters are found, indicating the end of what
4044 might be a URL parameter list. This is probably not a concern with SGML
4045 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004046
Willy Tarreau294d0f02015-08-10 19:40:12 +02004047 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004048
4049
Willy Tarreaub6205fd2012-09-24 12:27:33 +02004050bind [<address>]:<port_range> [, ...] [param*]
4051bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01004052 Define one or several listening addresses and/or ports in a frontend.
4053 May be used in sections : defaults | frontend | listen | backend
4054 no | yes | yes | no
4055 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01004056 <address> is optional and can be a host name, an IPv4 address, an IPv6
4057 address, or '*'. It designates the address the frontend will
4058 listen on. If unset, all IPv4 addresses of the system will be
4059 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01004060 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01004061 Optionally, an address family prefix may be used before the
4062 address to force the family regardless of the address format,
4063 which can be useful to specify a path to a unix socket with
4064 no slash ('/'). Currently supported prefixes are :
4065 - 'ipv4@' -> address is always IPv4
4066 - 'ipv6@' -> address is always IPv6
Emeric Brun3835c0d2020-07-07 09:46:09 +02004067 - 'udp@' -> address is resolved as IPv4 or IPv6 and
Emeric Brun12941c82020-07-07 14:19:42 +02004068 protocol UDP is used. Currently those listeners are
4069 supported only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02004070 - 'udp4@' -> address is always IPv4 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02004071 is used. Currently those listeners are supported
4072 only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02004073 - 'udp6@' -> address is always IPv6 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02004074 is used. Currently those listeners are supported
4075 only in log-forward sections.
Willy Tarreau24709282013-03-10 21:32:12 +01004076 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02004077 - 'abns@' -> address is in abstract namespace (Linux only).
4078 Note: since abstract sockets are not "rebindable", they
4079 do not cope well with multi-process mode during
4080 soft-restart, so it is better to avoid them if
4081 nbproc is greater than 1. The effect is that if the
4082 new process fails to start, only one of the old ones
4083 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01004084 - 'fd@<n>' -> use file descriptor <n> inherited from the
4085 parent. The fd must be bound and may or may not already
4086 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02004087 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
4088 connected unix socket or of a socketpair. The bind waits
4089 to receive a FD over the unix socket and uses it as if it
4090 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02004091 You may want to reference some environment variables in the
4092 address parameter, see section 2.3 about environment
4093 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01004094
Willy Tarreauc5011ca2010-03-22 11:53:56 +01004095 <port_range> is either a unique TCP port, or a port range for which the
4096 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01004097 above. The port is mandatory for TCP listeners. Note that in
4098 the case of an IPv6 address, the port is always the number
4099 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01004100 - a numerical port (ex: '80')
4101 - a dash-delimited ports range explicitly stating the lower
4102 and upper bounds (ex: '2000-2100') which are included in
4103 the range.
4104
4105 Particular care must be taken against port ranges, because
4106 every <address:port> couple consumes one socket (= a file
4107 descriptor), so it's easy to consume lots of descriptors
4108 with a simple range, and to run out of sockets. Also, each
4109 <address:port> couple must be used only once among all
4110 instances running on a same system. Please note that binding
4111 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04004112 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01004113 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004114
Willy Tarreauceb24bc2010-11-09 12:46:41 +01004115 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01004116 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01004117 receive UNIX connections on the socket located at this place.
4118 The path must begin with a slash and by default is absolute.
4119 It can be relative to the prefix defined by "unix-bind" in
4120 the global section. Note that the total length of the prefix
4121 followed by the socket path cannot exceed some system limits
4122 for UNIX sockets, which commonly are set to 107 characters.
4123
Willy Tarreaub6205fd2012-09-24 12:27:33 +02004124 <param*> is a list of parameters common to all sockets declared on the
4125 same line. These numerous parameters depend on OS and build
4126 options and have a complete section dedicated to them. Please
4127 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02004128
Willy Tarreau0ba27502007-12-24 16:55:16 +01004129 It is possible to specify a list of address:port combinations delimited by
4130 commas. The frontend will then listen on all of these addresses. There is no
4131 fixed limit to the number of addresses and ports which can be listened on in
4132 a frontend, as well as there is no limit to the number of "bind" statements
4133 in a frontend.
4134
4135 Example :
4136 listen http_proxy
4137 bind :80,:443
4138 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01004139 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01004140
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02004141 listen http_https_proxy
4142 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02004143 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02004144
Willy Tarreau24709282013-03-10 21:32:12 +01004145 listen http_https_proxy_explicit
4146 bind ipv6@:80
4147 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
4148 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
4149
Willy Tarreaudad36a32013-03-11 01:20:04 +01004150 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02004151 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01004152
Willy Tarreau55dcaf62015-09-27 15:03:15 +02004153 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
4154 sun_path length is used for the address length. Some other programs
4155 such as socat use the string length only by default. Pass the option
4156 ",unix-tightsocklen=0" to any abstract socket definition in socat to
4157 make it compatible with HAProxy's.
4158
Willy Tarreauceb24bc2010-11-09 12:46:41 +01004159 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02004160 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004161
4162
Christopher Fauletff4121f2017-11-22 16:38:49 +01004163bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004164 Limit visibility of an instance to a certain set of processes numbers.
4165 May be used in sections : defaults | frontend | listen | backend
4166 yes | yes | yes | yes
4167 Arguments :
4168 all All process will see this instance. This is the default. It
4169 may be used to override a default value.
4170
Willy Tarreaua9db57e2013-01-18 11:29:29 +01004171 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004172 option may be combined with other numbers.
4173
Willy Tarreaua9db57e2013-01-18 11:29:29 +01004174 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004175 option may be combined with other numbers. Do not use it
4176 with less than 2 processes otherwise some instances might be
4177 missing from all processes.
4178
Christopher Fauletff4121f2017-11-22 16:38:49 +01004179 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01004180 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01004181 the machine's word size. Ranges can be partially defined. The
4182 higher bound can be omitted. In such case, it is replaced by
4183 the corresponding maximum value. If a proxy is bound to
4184 process numbers greater than the configured global.nbproc, it
4185 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02004186 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004187
4188 This keyword limits binding of certain instances to certain processes. This
4189 is useful in order not to have too many processes listening to the same
4190 ports. For instance, on a dual-core machine, it might make sense to set
4191 'nbproc 2' in the global section, then distributes the listeners among 'odd'
4192 and 'even' instances.
4193
Willy Tarreaua9db57e2013-01-18 11:29:29 +01004194 At the moment, it is not possible to reference more than 32 or 64 processes
4195 using this keyword, but this should be more than enough for most setups.
4196 Please note that 'all' really means all processes regardless of the machine's
4197 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004198
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02004199 Each "bind" line may further be limited to a subset of the proxy's processes,
4200 please consult the "process" bind keyword in section 5.1.
4201
Willy Tarreaub369a042014-09-16 13:21:03 +02004202 When a frontend has no explicit "bind-process" line, it tries to bind to all
4203 the processes referenced by its "bind" lines. That means that frontends can
4204 easily adapt to their listeners' processes.
4205
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004206 If some backends are referenced by frontends bound to other processes, the
4207 backend automatically inherits the frontend's processes.
4208
4209 Example :
4210 listen app_ip1
4211 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02004212 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004213
4214 listen app_ip2
4215 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02004216 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004217
4218 listen management
4219 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02004220 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004221
Willy Tarreau110ecc12012-11-15 17:50:01 +01004222 listen management
4223 bind 10.0.0.4:80
4224 bind-process 1-4
4225
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02004226 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004227
4228
Willy Tarreau0ba27502007-12-24 16:55:16 +01004229capture cookie <name> len <length>
4230 Capture and log a cookie in the request and in the response.
4231 May be used in sections : defaults | frontend | listen | backend
4232 no | yes | yes | no
4233 Arguments :
4234 <name> is the beginning of the name of the cookie to capture. In order
4235 to match the exact name, simply suffix the name with an equal
4236 sign ('='). The full name will appear in the logs, which is
4237 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01004238 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01004239
4240 <length> is the maximum number of characters to report in the logs, which
4241 include the cookie name, the equal sign and the value, all in the
4242 standard "name=value" form. The string will be truncated on the
4243 right if it exceeds <length>.
4244
4245 Only the first cookie is captured. Both the "cookie" request headers and the
4246 "set-cookie" response headers are monitored. This is particularly useful to
4247 check for application bugs causing session crossing or stealing between
4248 users, because generally the user's cookies can only change on a login page.
4249
4250 When the cookie was not presented by the client, the associated log column
4251 will report "-". When a request does not cause a cookie to be assigned by the
4252 server, a "-" is reported in the response column.
4253
4254 The capture is performed in the frontend only because it is necessary that
4255 the log format does not change for a given frontend depending on the
4256 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01004257 "capture cookie" statement in a frontend. The maximum capture length is set
4258 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
4259 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004260
4261 Example:
4262 capture cookie ASPSESSION len 32
4263
4264 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004265 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004266
4267
4268capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01004269 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004270 May be used in sections : defaults | frontend | listen | backend
4271 no | yes | yes | no
4272 Arguments :
4273 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004274 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01004275 appear in the requests, with the first letter of each word in
4276 upper case. The header name will not appear in the logs, only the
4277 value is reported, but the position in the logs is respected.
4278
4279 <length> is the maximum number of characters to extract from the value and
4280 report in the logs. The string will be truncated on the right if
4281 it exceeds <length>.
4282
Willy Tarreau4460d032012-11-21 23:37:37 +01004283 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01004284 value will be added to the logs between braces ('{}'). If multiple headers
4285 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004286 in the same order they were declared in the configuration. Non-existent
4287 headers will be logged just as an empty string. Common uses for request
4288 header captures include the "Host" field in virtual hosting environments, the
4289 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004290 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004291 environments to find where the request came from.
4292
4293 Note that when capturing headers such as "User-agent", some spaces may be
4294 logged, making the log analysis more difficult. Thus be careful about what
4295 you log if you know your log parser is not smart enough to rely on the
4296 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004297
Willy Tarreau0900abb2012-11-22 00:21:46 +01004298 There is no limit to the number of captured request headers nor to their
4299 length, though it is wise to keep them low to limit memory usage per session.
4300 In order to keep log format consistent for a same frontend, header captures
4301 can only be declared in a frontend. It is not possible to specify a capture
4302 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004303
4304 Example:
4305 capture request header Host len 15
4306 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01004307 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01004308
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004309 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01004310 about logging.
4311
4312
4313capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01004314 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004315 May be used in sections : defaults | frontend | listen | backend
4316 no | yes | yes | no
4317 Arguments :
4318 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004319 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01004320 appear in the response, with the first letter of each word in
4321 upper case. The header name will not appear in the logs, only the
4322 value is reported, but the position in the logs is respected.
4323
4324 <length> is the maximum number of characters to extract from the value and
4325 report in the logs. The string will be truncated on the right if
4326 it exceeds <length>.
4327
Willy Tarreau4460d032012-11-21 23:37:37 +01004328 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01004329 result will be added to the logs between braces ('{}') after the captured
4330 request headers. If multiple headers are captured, they will be delimited by
4331 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004332 the configuration. Non-existent headers will be logged just as an empty
4333 string. Common uses for response header captures include the "Content-length"
4334 header which indicates how many bytes are expected to be returned, the
4335 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004336
Willy Tarreau0900abb2012-11-22 00:21:46 +01004337 There is no limit to the number of captured response headers nor to their
4338 length, though it is wise to keep them low to limit memory usage per session.
4339 In order to keep log format consistent for a same frontend, header captures
4340 can only be declared in a frontend. It is not possible to specify a capture
4341 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004342
4343 Example:
4344 capture response header Content-length len 9
4345 capture response header Location len 15
4346
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004347 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01004348 about logging.
4349
4350
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004351clitcpka-cnt <count>
4352 Sets the maximum number of keepalive probes TCP should send before dropping
4353 the connection on the client side.
4354 May be used in sections : defaults | frontend | listen | backend
4355 yes | yes | yes | no
4356 Arguments :
4357 <count> is the maximum number of keepalive probes.
4358
4359 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
4360 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004361 The availability of this setting depends on the operating system. It is
4362 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004363
4364 See also : "option clitcpka", "clitcpka-idle", "clitcpka-intvl".
4365
4366
4367clitcpka-idle <timeout>
4368 Sets the time the connection needs to remain idle before TCP starts sending
4369 keepalive probes, if enabled the sending of TCP keepalive packets on the
4370 client side.
4371 May be used in sections : defaults | frontend | listen | backend
4372 yes | yes | yes | no
4373 Arguments :
4374 <timeout> is the time the connection needs to remain idle before TCP starts
4375 sending keepalive probes. It is specified in seconds by default,
4376 but can be in any other unit if the number is suffixed by the
4377 unit, as explained at the top of this document.
4378
4379 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
4380 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004381 The availability of this setting depends on the operating system. It is
4382 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004383
4384 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-intvl".
4385
4386
4387clitcpka-intvl <timeout>
4388 Sets the time between individual keepalive probes on the client side.
4389 May be used in sections : defaults | frontend | listen | backend
4390 yes | yes | yes | no
4391 Arguments :
4392 <timeout> is the time between individual keepalive probes. It is specified
4393 in seconds by default, but can be in any other unit if the number
4394 is suffixed by the unit, as explained at the top of this
4395 document.
4396
4397 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
4398 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004399 The availability of this setting depends on the operating system. It is
4400 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004401
4402 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-idle".
4403
4404
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004405compression algo <algorithm> ...
4406compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02004407compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02004408 Enable HTTP compression.
4409 May be used in sections : defaults | frontend | listen | backend
4410 yes | yes | yes | yes
4411 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004412 algo is followed by the list of supported compression algorithms.
4413 type is followed by the list of MIME types that will be compressed.
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004414 offload makes HAProxy work as a compression offloader only (see notes).
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004415
4416 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01004417 identity this is mostly for debugging, and it was useful for developing
4418 the compression feature. Identity does not apply any change on
4419 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004420
Willy Tarreauc91840a2015-03-28 17:00:39 +01004421 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01004422 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01004423
4424 deflate same as "gzip", but with deflate algorithm and zlib format.
4425 Note that this algorithm has ambiguous support on many
4426 browsers and no support at all from recent ones. It is
4427 strongly recommended not to use it for anything else than
4428 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01004429 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004430
Willy Tarreauc91840a2015-03-28 17:00:39 +01004431 raw-deflate same as "deflate" without the zlib wrapper, and used as an
4432 alternative when the browser wants "deflate". All major
4433 browsers understand it and despite violating the standards,
4434 it is known to work better than "deflate", at least on MSIE
4435 and some versions of Safari. Do not use it in conjunction
4436 with "deflate", use either one or the other since both react
4437 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01004438 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004439
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04004440 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004441 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004442 If backend servers support HTTP compression, these directives
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004443 will be no-op: HAProxy will see the compressed response and will not
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004444 compress again. If backend servers do not support HTTP compression and
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004445 there is Accept-Encoding header in request, HAProxy will compress the
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004446 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02004447
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004448 The "offload" setting makes HAProxy remove the Accept-Encoding header to
Willy Tarreau70737d12012-10-27 00:34:28 +02004449 prevent backend servers from compressing responses. It is strongly
4450 recommended not to do this because this means that all the compression work
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004451 will be done on the single point where HAProxy is located. However in some
4452 deployment scenarios, HAProxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004453 with broken HTTP compression implementation which can't be turned off.
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004454 In that case HAProxy can be used to prevent that gateway from emitting
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004455 invalid payloads. In this case, simply removing the header in the
4456 configuration does not work because it applies before the header is parsed,
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004457 so that prevents HAProxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02004458 then be used for such scenarios. Note: for now, the "offload" setting is
4459 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02004460
William Lallemand05097442012-11-20 12:14:28 +01004461 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01004462 * the request does not advertise a supported compression algorithm in the
4463 "Accept-Encoding" header
Julien Pivottoff80c822021-03-29 12:41:40 +02004464 * the response message is not HTTP/1.1 or above
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01004465 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01004466 * response contain neither a "Content-Length" header nor a
4467 "Transfer-Encoding" whose last value is "chunked"
4468 * response contains a "Content-Type" header whose first value starts with
4469 "multipart"
4470 * the response contains the "no-transform" value in the "Cache-control"
4471 header
4472 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
4473 and later
4474 * The response contains a "Content-Encoding" header, indicating that the
4475 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01004476 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01004477
Tim Duesterhusb229f012019-01-29 16:38:56 +01004478 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01004479
William Lallemand82fe75c2012-10-23 10:25:10 +02004480 Examples :
4481 compression algo gzip
4482 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01004483
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004484
Willy Tarreau55165fe2009-05-10 12:02:55 +02004485cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02004486 [ postonly ] [ preserve ] [ httponly ] [ secure ]
4487 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Christopher Faulet2f533902020-01-21 11:06:48 +01004488 [ dynamic ] [ attr <value> ]*
Willy Tarreau0ba27502007-12-24 16:55:16 +01004489 Enable cookie-based persistence in a backend.
4490 May be used in sections : defaults | frontend | listen | backend
4491 yes | no | yes | yes
4492 Arguments :
4493 <name> is the name of the cookie which will be monitored, modified or
4494 inserted in order to bring persistence. This cookie is sent to
4495 the client via a "Set-Cookie" header in the response, and is
4496 brought back by the client in a "Cookie" header in all requests.
4497 Special care should be taken to choose a name which does not
4498 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01004499 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004500 HTTP/HTTPS), care should be taken to use different cookie names
4501 between all backends if persistence between them is not desired.
4502
4503 rewrite This keyword indicates that the cookie will be provided by the
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004504 server and that HAProxy will have to modify its value to set the
Willy Tarreau0ba27502007-12-24 16:55:16 +01004505 server's identifier in it. This mode is handy when the management
4506 of complex combinations of "Set-cookie" and "Cache-control"
4507 headers is left to the application. The application can then
4508 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004509 cookie. Since all responses should be monitored, this mode
4510 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01004511 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004512 start with this mode for new deployments. This keyword is
4513 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004514
4515 insert This keyword indicates that the persistence cookie will have to
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004516 be inserted by HAProxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004517
Willy Tarreaua79094d2010-08-31 22:54:15 +02004518 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004519 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02004520 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01004521 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004522 existing configurations running in the "rewrite" mode. The cookie
4523 will only be a session cookie and will not be stored on the
4524 client's disk. By default, unless the "indirect" option is added,
4525 the server will see the cookies emitted by the client. Due to
4526 caching effects, it is generally wise to add the "nocache" or
4527 "postonly" keywords (see below). The "insert" keyword is not
4528 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004529
4530 prefix This keyword indicates that instead of relying on a dedicated
4531 cookie for the persistence, an existing one will be completed.
4532 This may be needed in some specific environments where the client
4533 does not support more than one single cookie and the application
4534 already needs it. In this case, whenever the server sets a cookie
4535 named <name>, it will be prefixed with the server's identifier
4536 and a delimiter. The prefix will be removed from all client
4537 requests so that the server still finds the cookie it emitted.
4538 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004539 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02004540 not compatible with "rewrite" and "insert". Note: it is highly
4541 recommended not to use "indirect" with "prefix", otherwise server
4542 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004543
Willy Tarreaua79094d2010-08-31 22:54:15 +02004544 indirect When this option is specified, no cookie will be emitted to a
4545 client which already has a valid one for the server which has
4546 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004547 it will be removed, unless the "preserve" option is also set. In
4548 "insert" mode, this will additionally remove cookies from the
4549 requests transmitted to the server, making the persistence
4550 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02004551 Note: it is highly recommended not to use "indirect" with
4552 "prefix", otherwise server cookie updates would not be sent to
4553 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004554
4555 nocache This option is recommended in conjunction with the insert mode
4556 when there is a cache between the client and HAProxy, as it
4557 ensures that a cacheable response will be tagged non-cacheable if
4558 a cookie needs to be inserted. This is important because if all
4559 persistence cookies are added on a cacheable home page for
4560 instance, then all customers will then fetch the page from an
4561 outer cache and will all share the same persistence cookie,
4562 leading to one server receiving much more traffic than others.
4563 See also the "insert" and "postonly" options.
4564
4565 postonly This option ensures that cookie insertion will only be performed
4566 on responses to POST requests. It is an alternative to the
4567 "nocache" option, because POST responses are not cacheable, so
4568 this ensures that the persistence cookie will never get cached.
4569 Since most sites do not need any sort of persistence before the
4570 first POST which generally is a login request, this is a very
4571 efficient method to optimize caching without risking to find a
4572 persistence cookie in the cache.
4573 See also the "insert" and "nocache" options.
4574
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004575 preserve This option may only be used with "insert" and/or "indirect". It
4576 allows the server to emit the persistence cookie itself. In this
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004577 case, if a cookie is found in the response, HAProxy will leave it
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004578 untouched. This is useful in order to end persistence after a
4579 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01004580 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004581 the past. By combining this mechanism with the "disable-on-404"
4582 check option, it is possible to perform a completely graceful
4583 shutdown because users will definitely leave the server after
4584 they logout.
4585
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004586 httponly This option tells HAProxy to add an "HttpOnly" cookie attribute
Willy Tarreau4992dd22012-05-31 21:02:17 +02004587 when a cookie is inserted. This attribute is used so that a
4588 user agent doesn't share the cookie with non-HTTP components.
4589 Please check RFC6265 for more information on this attribute.
4590
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004591 secure This option tells HAProxy to add a "Secure" cookie attribute when
Willy Tarreau4992dd22012-05-31 21:02:17 +02004592 a cookie is inserted. This attribute is used so that a user agent
4593 never emits this cookie over non-secure channels, which means
4594 that a cookie learned with this flag will be presented only over
4595 SSL/TLS connections. Please check RFC6265 for more information on
4596 this attribute.
4597
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02004598 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004599 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01004600 name. If the domain begins with a dot, the browser is allowed to
4601 use it for any host ending with that name. It is also possible to
4602 specify several domain names by invoking this option multiple
4603 times. Some browsers might have small limits on the number of
4604 domains, so be careful when doing that. For the record, sending
4605 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02004606
Willy Tarreau996a92c2010-10-13 19:30:47 +02004607 maxidle This option allows inserted cookies to be ignored after some idle
4608 time. It only works with insert-mode cookies. When a cookie is
4609 sent to the client, the date this cookie was emitted is sent too.
4610 Upon further presentations of this cookie, if the date is older
4611 than the delay indicated by the parameter (in seconds), it will
4612 be ignored. Otherwise, it will be refreshed if needed when the
4613 response is sent to the client. This is particularly useful to
4614 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004615 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02004616 this option is set and a cookie has no date, it is always
4617 accepted, but gets refreshed in the response. This maintains the
4618 ability for admins to access their sites. Cookies that have a
4619 date in the future further than 24 hours are ignored. Doing so
4620 lets admins fix timezone issues without risking kicking users off
4621 the site.
4622
4623 maxlife This option allows inserted cookies to be ignored after some life
4624 time, whether they're in use or not. It only works with insert
4625 mode cookies. When a cookie is first sent to the client, the date
4626 this cookie was emitted is sent too. Upon further presentations
4627 of this cookie, if the date is older than the delay indicated by
4628 the parameter (in seconds), it will be ignored. If the cookie in
4629 the request has no date, it is accepted and a date will be set.
4630 Cookies that have a date in the future further than 24 hours are
4631 ignored. Doing so lets admins fix timezone issues without risking
4632 kicking users off the site. Contrary to maxidle, this value is
4633 not refreshed, only the first visit date counts. Both maxidle and
4634 maxlife may be used at the time. This is particularly useful to
4635 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004636 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02004637 is stronger than the maxidle method in that it forces a
4638 redispatch after some absolute delay.
4639
Olivier Houchard4e694042017-03-14 20:01:29 +01004640 dynamic Activate dynamic cookies. When used, a session cookie is
4641 dynamically created for each server, based on the IP and port
4642 of the server, and a secret key, specified in the
4643 "dynamic-cookie-key" backend directive.
4644 The cookie will be regenerated each time the IP address change,
4645 and is only generated for IPv4/IPv6.
4646
Daniel Corbett9f0843f2021-05-08 10:50:37 -04004647 attr This option tells HAProxy to add an extra attribute when a
Christopher Faulet2f533902020-01-21 11:06:48 +01004648 cookie is inserted. The attribute value can contain any
4649 characters except control ones or ";". This option may be
4650 repeated.
4651
Willy Tarreau0ba27502007-12-24 16:55:16 +01004652 There can be only one persistence cookie per HTTP backend, and it can be
4653 declared in a defaults section. The value of the cookie will be the value
4654 indicated after the "cookie" keyword in a "server" statement. If no cookie
4655 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02004656
Willy Tarreau0ba27502007-12-24 16:55:16 +01004657 Examples :
4658 cookie JSESSIONID prefix
4659 cookie SRV insert indirect nocache
4660 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02004661 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01004662
Willy Tarreau294d0f02015-08-10 19:40:12 +02004663 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004664
Willy Tarreau983e01e2010-01-11 18:42:06 +01004665
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004666declare capture [ request | response ] len <length>
4667 Declares a capture slot.
4668 May be used in sections : defaults | frontend | listen | backend
4669 no | yes | yes | no
4670 Arguments:
4671 <length> is the length allowed for the capture.
4672
4673 This declaration is only available in the frontend or listen section, but the
4674 reserved slot can be used in the backends. The "request" keyword allocates a
4675 capture slot for use in the request, and "response" allocates a capture slot
4676 for use in the response.
4677
4678 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02004679 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004680 "http-request capture" and "http-response capture".
4681
4682
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004683default-server [param*]
4684 Change default options for a server in a backend
4685 May be used in sections : defaults | frontend | listen | backend
4686 yes | no | yes | yes
4687 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01004688 <param*> is a list of parameters for this server. The "default-server"
4689 keyword accepts an important number of options and has a complete
4690 section dedicated to it. Please refer to section 5 for more
4691 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004692
Willy Tarreau983e01e2010-01-11 18:42:06 +01004693 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004694 default-server inter 1000 weight 13
4695
4696 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01004697
Willy Tarreau983e01e2010-01-11 18:42:06 +01004698
Willy Tarreau0ba27502007-12-24 16:55:16 +01004699default_backend <backend>
4700 Specify the backend to use when no "use_backend" rule has been matched.
4701 May be used in sections : defaults | frontend | listen | backend
4702 yes | yes | yes | no
4703 Arguments :
4704 <backend> is the name of the backend to use.
4705
4706 When doing content-switching between frontend and backends using the
4707 "use_backend" keyword, it is often useful to indicate which backend will be
4708 used when no rule has matched. It generally is the dynamic backend which
4709 will catch all undetermined requests.
4710
Willy Tarreau0ba27502007-12-24 16:55:16 +01004711 Example :
4712
4713 use_backend dynamic if url_dyn
4714 use_backend static if url_css url_img extension_img
4715 default_backend dynamic
4716
Willy Tarreau98d04852015-05-26 12:18:29 +02004717 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004718
Willy Tarreau0ba27502007-12-24 16:55:16 +01004719
Baptiste Assmann27f51342013-10-09 06:51:49 +02004720description <string>
4721 Describe a listen, frontend or backend.
4722 May be used in sections : defaults | frontend | listen | backend
4723 no | yes | yes | yes
4724 Arguments : string
4725
4726 Allows to add a sentence to describe the related object in the HAProxy HTML
4727 stats page. The description will be printed on the right of the object name
4728 it describes.
4729 No need to backslash spaces in the <string> arguments.
4730
4731
Willy Tarreau0ba27502007-12-24 16:55:16 +01004732disabled
4733 Disable a proxy, frontend or backend.
4734 May be used in sections : defaults | frontend | listen | backend
4735 yes | yes | yes | yes
4736 Arguments : none
4737
4738 The "disabled" keyword is used to disable an instance, mainly in order to
4739 liberate a listening port or to temporarily disable a service. The instance
4740 will still be created and its configuration will be checked, but it will be
4741 created in the "stopped" state and will appear as such in the statistics. It
4742 will not receive any traffic nor will it send any health-checks or logs. It
4743 is possible to disable many instances at once by adding the "disabled"
4744 keyword in a "defaults" section.
4745
4746 See also : "enabled"
4747
4748
Willy Tarreau5ce94572010-06-07 14:35:41 +02004749dispatch <address>:<port>
4750 Set a default server address
4751 May be used in sections : defaults | frontend | listen | backend
4752 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004753 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02004754
4755 <address> is the IPv4 address of the default server. Alternatively, a
4756 resolvable hostname is supported, but this name will be resolved
4757 during start-up.
4758
4759 <ports> is a mandatory port specification. All connections will be sent
4760 to this port, and it is not permitted to use port offsets as is
4761 possible with normal servers.
4762
Willy Tarreau787aed52011-04-15 06:45:37 +02004763 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02004764 server can take the connection. In the past it was used to forward non
4765 persistent connections to an auxiliary load balancer. Due to its simple
4766 syntax, it has also been used for simple TCP relays. It is recommended not to
4767 use it for more clarity, and to use the "server" directive instead.
4768
4769 See also : "server"
4770
Olivier Houchard4e694042017-03-14 20:01:29 +01004771
4772dynamic-cookie-key <string>
4773 Set the dynamic cookie secret key for a backend.
4774 May be used in sections : defaults | frontend | listen | backend
4775 yes | no | yes | yes
4776 Arguments : The secret key to be used.
4777
4778 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01004779 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01004780 specified on the "server" line), using a hash of the IP address of the
4781 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004782 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01004783 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02004784
Willy Tarreau0ba27502007-12-24 16:55:16 +01004785enabled
4786 Enable a proxy, frontend or backend.
4787 May be used in sections : defaults | frontend | listen | backend
4788 yes | yes | yes | yes
4789 Arguments : none
4790
4791 The "enabled" keyword is used to explicitly enable an instance, when the
4792 defaults has been set to "disabled". This is very rarely used.
4793
4794 See also : "disabled"
4795
4796
4797errorfile <code> <file>
4798 Return a file contents instead of errors generated by HAProxy
4799 May be used in sections : defaults | frontend | listen | backend
4800 yes | yes | yes | yes
4801 Arguments :
4802 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004803 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004804 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004805
4806 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004807 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01004808 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01004809 error pages, and to use absolute paths, since files are read
4810 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004811
4812 It is important to understand that this keyword is not meant to rewrite
4813 errors returned by the server, but errors detected and returned by HAProxy.
4814 This is why the list of supported errors is limited to a small set.
4815
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004816 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4817
Christopher Faulet70170672020-05-18 17:42:48 +02004818 The files are parsed when HAProxy starts and must be valid according to the
4819 HTTP specification. They should not exceed the configured buffer size
4820 (BUFSIZE), which generally is 16 kB, otherwise an internal error will be
4821 returned. It is also wise not to put any reference to local contents
4822 (e.g. images) in order to avoid loops between the client and HAProxy when all
4823 servers are down, causing an error to be returned instead of an
4824 image. Finally, The response cannot exceed (tune.bufsize - tune.maxrewrite)
4825 so that "http-after-response" rules still have room to operate (see
4826 "tune.maxrewrite").
Willy Tarreau59140a22009-02-22 12:02:30 +01004827
Willy Tarreau0ba27502007-12-24 16:55:16 +01004828 The files are read at the same time as the configuration and kept in memory.
4829 For this reason, the errors continue to be returned even when the process is
4830 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01004831 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01004832 403 status code and interrogating a blocked URL.
4833
Christopher Faulet3b967c12020-05-15 15:47:44 +02004834 See also : "http-error", "errorloc", "errorloc302", "errorloc303"
Willy Tarreau0ba27502007-12-24 16:55:16 +01004835
Willy Tarreau59140a22009-02-22 12:02:30 +01004836 Example :
4837 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01004838 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01004839 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
4840 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
4841
Willy Tarreau2769aa02007-12-27 18:26:09 +01004842
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004843errorfiles <name> [<code> ...]
4844 Import, fully or partially, the error files defined in the <name> http-errors
4845 section.
4846 May be used in sections : defaults | frontend | listen | backend
4847 yes | yes | yes | yes
4848 Arguments :
4849 <name> is the name of an existing http-errors section.
4850
4851 <code> is a HTTP status code. Several status code may be listed.
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004852 Currently, HAProxy is capable of generating codes 200, 400, 401,
Christopher Faulete095f312020-12-07 11:22:24 +01004853 403, 404, 405, 407, 408, 410, 413, 425, 429, 500, 501, 502, 503,
4854 and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004855
4856 Errors defined in the http-errors section with the name <name> are imported
4857 in the current proxy. If no status code is specified, all error files of the
4858 http-errors section are imported. Otherwise, only error files associated to
4859 the listed status code are imported. Those error files override the already
4860 defined custom errors for the proxy. And they may be overridden by following
Daniel Corbett67a82712020-07-06 23:01:19 -04004861 ones. Functionally, it is exactly the same as declaring all error files by
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004862 hand using "errorfile" directives.
4863
Christopher Faulet3b967c12020-05-15 15:47:44 +02004864 See also : "http-error", "errorfile", "errorloc", "errorloc302" ,
4865 "errorloc303" and section 3.8 about http-errors.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004866
4867 Example :
4868 errorfiles generic
4869 errorfiles site-1 403 404
4870
4871
Willy Tarreau2769aa02007-12-27 18:26:09 +01004872errorloc <code> <url>
4873errorloc302 <code> <url>
4874 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4875 May be used in sections : defaults | frontend | listen | backend
4876 yes | yes | yes | yes
4877 Arguments :
4878 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004879 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004880 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004881
4882 <url> it is the exact contents of the "Location" header. It may contain
4883 either a relative URI to an error page hosted on the same site,
4884 or an absolute URI designating an error page on another site.
4885 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004886 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004887
4888 It is important to understand that this keyword is not meant to rewrite
4889 errors returned by the server, but errors detected and returned by HAProxy.
4890 This is why the list of supported errors is limited to a small set.
4891
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004892 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4893
Willy Tarreau2769aa02007-12-27 18:26:09 +01004894 Note that both keyword return the HTTP 302 status code, which tells the
4895 client to fetch the designated URL using the same HTTP method. This can be
4896 quite problematic in case of non-GET methods such as POST, because the URL
4897 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01004898 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01004899 status code, indicating to the client that the URL must be fetched with a GET
4900 request.
4901
Christopher Faulet3b967c12020-05-15 15:47:44 +02004902 See also : "http-error", "errorfile", "errorloc303"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004903
4904
4905errorloc303 <code> <url>
4906 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4907 May be used in sections : defaults | frontend | listen | backend
4908 yes | yes | yes | yes
4909 Arguments :
4910 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004911 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004912 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004913
4914 <url> it is the exact contents of the "Location" header. It may contain
4915 either a relative URI to an error page hosted on the same site,
4916 or an absolute URI designating an error page on another site.
4917 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004918 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004919
4920 It is important to understand that this keyword is not meant to rewrite
4921 errors returned by the server, but errors detected and returned by HAProxy.
4922 This is why the list of supported errors is limited to a small set.
4923
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004924 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4925
Willy Tarreau2769aa02007-12-27 18:26:09 +01004926 Note that both keyword return the HTTP 303 status code, which tells the
4927 client to fetch the designated URL using the same HTTP GET method. This
4928 solves the usual problems associated with "errorloc" and the 302 code. It is
4929 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004930 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004931
Christopher Faulet3b967c12020-05-15 15:47:44 +02004932 See also : "http-error", "errorfile", "errorloc", "errorloc302"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004933
4934
Simon Horman51a1cf62015-02-03 13:00:44 +09004935email-alert from <emailaddr>
4936 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01004937 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09004938 May be used in sections: defaults | frontend | listen | backend
4939 yes | yes | yes | yes
4940
4941 Arguments :
4942
4943 <emailaddr> is the from email address to use when sending email alerts
4944
4945 Also requires "email-alert mailers" and "email-alert to" to be set
4946 and if so sending email alerts is enabled for the proxy.
4947
Simon Horman64e34162015-02-06 11:11:57 +09004948 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02004949 "email-alert myhostname", "email-alert to", section 3.6 about
4950 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09004951
4952
4953email-alert level <level>
4954 Declare the maximum log level of messages for which email alerts will be
4955 sent. This acts as a filter on the sending of email alerts.
4956 May be used in sections: defaults | frontend | listen | backend
4957 yes | yes | yes | yes
4958
4959 Arguments :
4960
4961 <level> One of the 8 syslog levels:
4962 emerg alert crit err warning notice info debug
4963 The above syslog levels are ordered from lowest to highest.
4964
4965 By default level is alert
4966
4967 Also requires "email-alert from", "email-alert mailers" and
4968 "email-alert to" to be set and if so sending email alerts is enabled
4969 for the proxy.
4970
Simon Horman1421e212015-04-30 13:10:35 +09004971 Alerts are sent when :
4972
4973 * An un-paused server is marked as down and <level> is alert or lower
4974 * A paused server is marked as down and <level> is notice or lower
4975 * A server is marked as up or enters the drain state and <level>
4976 is notice or lower
4977 * "option log-health-checks" is enabled, <level> is info or lower,
4978 and a health check status update occurs
4979
Simon Horman64e34162015-02-06 11:11:57 +09004980 See also : "email-alert from", "email-alert mailers",
4981 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09004982 section 3.6 about mailers.
4983
4984
4985email-alert mailers <mailersect>
4986 Declare the mailers to be used when sending email alerts
4987 May be used in sections: defaults | frontend | listen | backend
4988 yes | yes | yes | yes
4989
4990 Arguments :
4991
4992 <mailersect> is the name of the mailers section to send email alerts.
4993
4994 Also requires "email-alert from" and "email-alert to" to be set
4995 and if so sending email alerts is enabled for the proxy.
4996
Simon Horman64e34162015-02-06 11:11:57 +09004997 See also : "email-alert from", "email-alert level", "email-alert myhostname",
4998 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004999
5000
5001email-alert myhostname <hostname>
5002 Declare the to hostname address to be used when communicating with
5003 mailers.
5004 May be used in sections: defaults | frontend | listen | backend
5005 yes | yes | yes | yes
5006
5007 Arguments :
5008
Baptiste Assmann738bad92015-12-21 15:27:53 +01005009 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09005010
5011 By default the systems hostname is used.
5012
5013 Also requires "email-alert from", "email-alert mailers" and
5014 "email-alert to" to be set and if so sending email alerts is enabled
5015 for the proxy.
5016
Simon Horman64e34162015-02-06 11:11:57 +09005017 See also : "email-alert from", "email-alert level", "email-alert mailers",
5018 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09005019
5020
5021email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01005022 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09005023 header of email alerts. This is the address that email alerts are sent to.
5024 May be used in sections: defaults | frontend | listen | backend
5025 yes | yes | yes | yes
5026
5027 Arguments :
5028
5029 <emailaddr> is the to email address to use when sending email alerts
5030
5031 Also requires "email-alert mailers" and "email-alert to" to be set
5032 and if so sending email alerts is enabled for the proxy.
5033
Simon Horman64e34162015-02-06 11:11:57 +09005034 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09005035 "email-alert myhostname", section 3.6 about mailers.
5036
5037
Willy Tarreau4de91492010-01-22 19:10:05 +01005038force-persist { if | unless } <condition>
5039 Declare a condition to force persistence on down servers
5040 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005041 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01005042
5043 By default, requests are not dispatched to down servers. It is possible to
5044 force this using "option persist", but it is unconditional and redispatches
5045 to a valid server if "option redispatch" is set. That leaves with very little
5046 possibilities to force some requests to reach a server which is artificially
5047 marked down for maintenance operations.
5048
5049 The "force-persist" statement allows one to declare various ACL-based
5050 conditions which, when met, will cause a request to ignore the down status of
5051 a server and still try to connect to it. That makes it possible to start a
5052 server, still replying an error to the health checks, and run a specially
5053 configured browser to test the service. Among the handy methods, one could
5054 use a specific source IP address, or a specific cookie. The cookie also has
5055 the advantage that it can easily be added/removed on the browser from a test
5056 page. Once the service is validated, it is then possible to open the service
5057 to the world by returning a valid response to health checks.
5058
5059 The forced persistence is enabled when an "if" condition is met, or unless an
5060 "unless" condition is met. The final redispatch is always disabled when this
5061 is used.
5062
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005063 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02005064 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01005065
Christopher Fauletc3fe5332016-04-07 15:30:10 +02005066
5067filter <name> [param*]
5068 Add the filter <name> in the filter list attached to the proxy.
5069 May be used in sections : defaults | frontend | listen | backend
5070 no | yes | yes | yes
5071 Arguments :
5072 <name> is the name of the filter. Officially supported filters are
5073 referenced in section 9.
5074
Tim Düsterhus4896c442016-11-29 02:15:19 +01005075 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02005076 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01005077 filter. Please refer to the documentation of the corresponding
5078 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02005079
5080 Multiple occurrences of the filter line can be used for the same proxy. The
5081 same filter can be referenced many times if needed.
5082
5083 Example:
5084 listen
5085 bind *:80
5086
5087 filter trace name BEFORE-HTTP-COMP
5088 filter compression
5089 filter trace name AFTER-HTTP-COMP
5090
5091 compression algo gzip
5092 compression offload
5093
5094 server srv1 192.168.0.1:80
5095
5096 See also : section 9.
5097
Willy Tarreau4de91492010-01-22 19:10:05 +01005098
Willy Tarreau2769aa02007-12-27 18:26:09 +01005099fullconn <conns>
5100 Specify at what backend load the servers will reach their maxconn
5101 May be used in sections : defaults | frontend | listen | backend
5102 yes | no | yes | yes
5103 Arguments :
5104 <conns> is the number of connections on the backend which will make the
5105 servers use the maximal number of connections.
5106
Willy Tarreau198a7442008-01-17 12:05:32 +01005107 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01005108 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01005109 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01005110 load. The server will then always accept at least <minconn> connections,
5111 never more than <maxconn>, and the limit will be on the ramp between both
5112 values when the backend has less than <conns> concurrent connections. This
5113 makes it possible to limit the load on the servers during normal loads, but
5114 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005115 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005116
Daniel Corbett9f0843f2021-05-08 10:50:37 -04005117 Since it's hard to get this value right, HAProxy automatically sets it to
Willy Tarreaufbb78422011-06-05 15:38:35 +02005118 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01005119 backend (based on "use_backend" and "default_backend" rules). That way it's
5120 safe to leave it unset. However, "use_backend" involving dynamic names are
5121 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02005122
Willy Tarreau2769aa02007-12-27 18:26:09 +01005123 Example :
5124 # The servers will accept between 100 and 1000 concurrent connections each
5125 # and the maximum of 1000 will be reached when the backend reaches 10000
5126 # connections.
5127 backend dynamic
5128 fullconn 10000
5129 server srv1 dyn1:80 minconn 100 maxconn 1000
5130 server srv2 dyn2:80 minconn 100 maxconn 1000
5131
5132 See also : "maxconn", "server"
5133
5134
Willy Tarreauab0a5192020-10-09 19:07:01 +02005135grace <time> (deprecated)
Willy Tarreau2769aa02007-12-27 18:26:09 +01005136 Maintain a proxy operational for some time after a soft stop
5137 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01005138 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01005139 Arguments :
5140 <time> is the time (by default in milliseconds) for which the instance
5141 will remain operational with the frontend sockets still listening
5142 when a soft-stop is received via the SIGUSR1 signal.
5143
5144 This may be used to ensure that the services disappear in a certain order.
5145 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005146 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01005147 needed by the equipment to detect the failure.
5148
5149 Note that currently, there is very little benefit in using this parameter,
5150 and it may in fact complicate the soft-reconfiguration process more than
5151 simplify it.
5152
Willy Tarreau0ba27502007-12-24 16:55:16 +01005153
Andrew Rodland17be45e2016-10-25 17:04:12 -04005154hash-balance-factor <factor>
5155 Specify the balancing factor for bounded-load consistent hashing
5156 May be used in sections : defaults | frontend | listen | backend
5157 yes | no | no | yes
5158 Arguments :
5159 <factor> is the control for the maximum number of concurrent requests to
5160 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01005161 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04005162
5163 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
5164 enables an algorithm that prevents any one server from getting too many
5165 requests at once, even if some hash buckets receive many more requests than
5166 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
5167 <factor> is a percentage greater than 100. For example, if <factor> is 150,
5168 then no server will be allowed to have a load more than 1.5 times the average.
5169 If server weights are used, they will be respected.
5170
5171 If the first-choice server is disqualified, the algorithm will choose another
5172 server based on the request hash, until a server with additional capacity is
5173 found. A higher <factor> allows more imbalance between the servers, while a
5174 lower <factor> means that more servers will be checked on average, affecting
5175 performance. Reasonable values are from 125 to 200.
5176
Willy Tarreau760e81d2018-05-03 07:20:40 +02005177 This setting is also used by "balance random" which internally relies on the
5178 consistent hashing mechanism.
5179
Andrew Rodland17be45e2016-10-25 17:04:12 -04005180 See also : "balance" and "hash-type".
5181
5182
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005183hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005184 Specify a method to use for mapping hashes to servers
5185 May be used in sections : defaults | frontend | listen | backend
5186 yes | no | yes | yes
5187 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04005188 <method> is the method used to select a server from the hash computed by
5189 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005190
Bhaskar98634f02013-10-29 23:30:51 -04005191 map-based the hash table is a static array containing all alive servers.
5192 The hashes will be very smooth, will consider weights, but
5193 will be static in that weight changes while a server is up
5194 will be ignored. This means that there will be no slow start.
5195 Also, since a server is selected by its position in the array,
5196 most mappings are changed when the server count changes. This
5197 means that when a server goes up or down, or when a server is
5198 added to a farm, most connections will be redistributed to
5199 different servers. This can be inconvenient with caches for
5200 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01005201
Bhaskar98634f02013-10-29 23:30:51 -04005202 consistent the hash table is a tree filled with many occurrences of each
5203 server. The hash key is looked up in the tree and the closest
5204 server is chosen. This hash is dynamic, it supports changing
5205 weights while the servers are up, so it is compatible with the
5206 slow start feature. It has the advantage that when a server
5207 goes up or down, only its associations are moved. When a
5208 server is added to the farm, only a few part of the mappings
5209 are redistributed, making it an ideal method for caches.
5210 However, due to its principle, the distribution will never be
5211 very smooth and it may sometimes be necessary to adjust a
5212 server's weight or its ID to get a more balanced distribution.
5213 In order to get the same distribution on multiple load
5214 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005215 same IDs. Note: consistent hash uses sdbm and avalanche if no
5216 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04005217
5218 <function> is the hash function to be used :
5219
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005220 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04005221 reimplementation of ndbm) database library. It was found to do
5222 well in scrambling bits, causing better distribution of the keys
5223 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005224 function with good distribution, unless the total server weight
5225 is a multiple of 64, in which case applying the avalanche
5226 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04005227
5228 djb2 this function was first proposed by Dan Bernstein many years ago
5229 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005230 function provides a better distribution than sdbm. It generally
5231 works well with text-based inputs though it can perform extremely
5232 poorly with numeric-only input or when the total server weight is
5233 a multiple of 33, unless the avalanche modifier is also used.
5234
Daniel Corbett9f0843f2021-05-08 10:50:37 -04005235 wt6 this function was designed for HAProxy while testing other
Willy Tarreaua0f42712013-11-14 14:30:35 +01005236 functions in the past. It is not as smooth as the other ones, but
5237 is much less sensible to the input data set or to the number of
5238 servers. It can make sense as an alternative to sdbm+avalanche or
5239 djb2+avalanche for consistent hashing or when hashing on numeric
5240 data such as a source IP address or a visitor identifier in a URL
5241 parameter.
5242
Willy Tarreau324f07f2015-01-20 19:44:50 +01005243 crc32 this is the most common CRC32 implementation as used in Ethernet,
5244 gzip, PNG, etc. It is slower than the other ones but may provide
5245 a better distribution or less predictable results especially when
5246 used on strings.
5247
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005248 <modifier> indicates an optional method applied after hashing the key :
5249
5250 avalanche This directive indicates that the result from the hash
5251 function above should not be used in its raw form but that
5252 a 4-byte full avalanche hash must be applied first. The
5253 purpose of this step is to mix the resulting bits from the
5254 previous hash in order to avoid any undesired effect when
5255 the input contains some limited values or when the number of
5256 servers is a multiple of one of the hash's components (64
5257 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
5258 result less predictable, but it's also not as smooth as when
5259 using the original function. Some testing might be needed
5260 with some workloads. This hash is one of the many proposed
5261 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005262
Bhaskar98634f02013-10-29 23:30:51 -04005263 The default hash type is "map-based" and is recommended for most usages. The
5264 default function is "sdbm", the selection of a function should be based on
5265 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005266
Andrew Rodland17be45e2016-10-25 17:04:12 -04005267 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005268
5269
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005270http-after-response <action> <options...> [ { if | unless } <condition> ]
5271 Access control for all Layer 7 responses (server, applet/service and internal
5272 ones).
5273
5274 May be used in sections: defaults | frontend | listen | backend
5275 no | yes | yes | yes
5276
5277 The http-after-response statement defines a set of rules which apply to layer
5278 7 processing. The rules are evaluated in their declaration order when they
5279 are met in a frontend, listen or backend section. Any rule may optionally be
5280 followed by an ACL-based condition, in which case it will only be evaluated
5281 if the condition is true. Since these rules apply on responses, the backend
5282 rules are applied first, followed by the frontend's rules.
5283
5284 Unlike http-response rules, these ones are applied on all responses, the
5285 server ones but also to all responses generated by HAProxy. These rules are
5286 evaluated at the end of the responses analysis, before the data forwarding.
5287
5288 The first keyword is the rule's action. The supported actions are described
5289 below.
5290
5291 There is no limit to the number of http-after-response statements per
5292 instance.
5293
Christopher Fauletd5ac6de2020-12-02 08:40:14 +01005294 Note: Errors emitted in early stage of the request parsing are handled by the
5295 multiplexer at a lower level, before any http analysis. Thus no
5296 http-after-response ruleset is evaluated on these errors.
5297
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005298 Example:
5299 http-after-response set-header Strict-Transport-Security "max-age=31536000"
5300 http-after-response set-header Cache-Control "no-store,no-cache,private"
5301 http-after-response set-header Pragma "no-cache"
5302
5303http-after-response add-header <name> <fmt> [ { if | unless } <condition> ]
5304
5305 This appends an HTTP header field whose name is specified in <name> and whose
5306 value is defined by <fmt> which follows the log-format rules (see Custom Log
5307 Format in section 8.2.4). This may be used to send a cookie to a client for
5308 example, or to pass some internal information.
5309 This rule is not final, so it is possible to add other similar rules.
5310 Note that header addition is performed immediately, so one rule might reuse
5311 the resulting header from a previous rule.
5312
5313http-after-response allow [ { if | unless } <condition> ]
5314
5315 This stops the evaluation of the rules and lets the response pass the check.
5316 No further "http-after-response" rules are evaluated.
5317
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005318http-after-response del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005319
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005320 This removes all HTTP header fields whose name is specified in <name>. <meth>
5321 is the matching method, applied on the header name. Supported matching methods
5322 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
5323 (substring match) and "reg" (regex match). If not specified, exact matching
5324 method is used.
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005325
5326http-after-response replace-header <name> <regex-match> <replace-fmt>
5327 [ { if | unless } <condition> ]
5328
5329 This works like "http-response replace-header".
5330
5331 Example:
5332 http-after-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
5333
5334 # applied to:
5335 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
5336
5337 # outputs:
5338 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
5339
5340 # assuming the backend IP is 192.168.1.20.
5341
5342http-after-response replace-value <name> <regex-match> <replace-fmt>
5343 [ { if | unless } <condition> ]
5344
5345 This works like "http-response replace-value".
5346
5347 Example:
5348 http-after-response replace-value Cache-control ^public$ private
5349
5350 # applied to:
5351 Cache-Control: max-age=3600, public
5352
5353 # outputs:
5354 Cache-Control: max-age=3600, private
5355
5356http-after-response set-header <name> <fmt> [ { if | unless } <condition> ]
5357
5358 This does the same as "add-header" except that the header name is first
5359 removed if it existed. This is useful when passing security information to
5360 the server, where the header must not be manipulated by external users.
5361
5362http-after-response set-status <status> [reason <str>]
5363 [ { if | unless } <condition> ]
5364
5365 This replaces the response status code with <status> which must be an integer
5366 between 100 and 999. Optionally, a custom reason text can be provided defined
5367 by <str>, or the default reason for the specified code will be used as a
5368 fallback.
5369
5370 Example:
5371 # return "431 Request Header Fields Too Large"
5372 http-response set-status 431
5373 # return "503 Slow Down", custom reason
5374 http-response set-status 503 reason "Slow Down"
5375
5376http-after-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5377
5378 This is used to set the contents of a variable. The variable is declared
5379 inline.
5380
5381 Arguments:
5382 <var-name> The name of the variable starts with an indication about its
5383 scope. The scopes allowed are:
5384 "proc" : the variable is shared with the whole process
5385 "sess" : the variable is shared with the whole session
5386 "txn" : the variable is shared with the transaction
5387 (request and response)
5388 "req" : the variable is shared only during request
5389 processing
5390 "res" : the variable is shared only during response
5391 processing
5392 This prefix is followed by a name. The separator is a '.'.
5393 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5394 and '_'.
5395
5396 <expr> Is a standard HAProxy expression formed by a sample-fetch
5397 followed by some converters.
5398
5399 Example:
5400 http-after-response set-var(sess.last_redir) res.hdr(location)
5401
5402http-after-response strict-mode { on | off }
5403
5404 This enables or disables the strict rewriting mode for following rules. It
5405 does not affect rules declared before it and it is only applicable on rules
5406 performing a rewrite on the responses. When the strict mode is enabled, any
5407 rewrite failure triggers an internal error. Otherwise, such errors are
5408 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005409 rewrites optional while others must be performed to continue the response
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005410 processing.
5411
5412 By default, the strict rewriting mode is enabled. Its value is also reset
5413 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04005414 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005415 rules evaluation.
5416
5417http-after-response unset-var(<var-name>) [ { if | unless } <condition> ]
5418
5419 This is used to unset a variable. See "http-after-response set-var" for
5420 details about <var-name>.
5421
5422 Example:
5423 http-after-response unset-var(sess.last_redir)
5424
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005425
5426http-check comment <string>
5427 Defines a comment for the following the http-check rule, reported in logs if
5428 it fails.
5429 May be used in sections : defaults | frontend | listen | backend
5430 yes | no | yes | yes
5431
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005432 Arguments :
5433 <string> is the comment message to add in logs if the following http-check
5434 rule fails.
5435
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005436 It only works for connect, send and expect rules. It is useful to make
5437 user-friendly error reporting.
5438
Daniel Corbett67a82712020-07-06 23:01:19 -04005439 See also : "option httpchk", "http-check connect", "http-check send" and
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005440 "http-check expect".
5441
5442
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005443http-check connect [default] [port <expr>] [addr <ip>] [send-proxy]
5444 [via-socks4] [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +02005445 [proto <name>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02005446 Opens a new connection to perform an HTTP health check
5447 May be used in sections : defaults | frontend | listen | backend
5448 yes | no | yes | yes
5449
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005450 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005451 comment <msg> defines a message to report if the rule evaluation fails.
5452
Christopher Faulete5870d82020-04-15 11:32:03 +02005453 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -04005454 checks. The server options are used only if not redefined.
Christopher Faulete5870d82020-04-15 11:32:03 +02005455
5456 port <expr> if not set, check port or server port is used.
5457 It tells HAProxy where to open the connection to.
5458 <port> must be a valid TCP port source integer, from 1 to
5459 65535 or an sample-fetch expression.
5460
5461 addr <ip> defines the IP address to do the health check.
5462
5463 send-proxy send a PROXY protocol string
5464
5465 via-socks4 enables outgoing health checks using upstream socks4 proxy.
5466
5467 ssl opens a ciphered connection
5468
5469 sni <sni> specifies the SNI to use to do health checks over SSL.
5470
5471 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
5472 list consists in a comma-delimited list of protocol names,
5473 for instance: "h2,http/1.1". If it is not set, the server ALPN
5474 is used.
5475
Christopher Fauletedc6ed92020-04-23 16:27:59 +02005476 proto <name> forces the multiplexer's protocol to use for this connection.
5477 It must be an HTTP mux protocol and it must be usable on the
5478 backend side. The list of available protocols is reported in
5479 haproxy -vv.
5480
Christopher Faulete5870d82020-04-15 11:32:03 +02005481 linger cleanly close the connection instead of using a single RST.
5482
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005483 Just like tcp-check health checks, it is possible to configure the connection
5484 to use to perform HTTP health check. This directive should also be used to
5485 describe a scenario involving several request/response exchanges, possibly on
5486 different ports or with different servers.
5487
5488 When there are no TCP port configured on the server line neither server port
5489 directive, then the first step of the http-check sequence must be to specify
5490 the port with a "http-check connect".
5491
5492 In an http-check ruleset a 'connect' is required, it is also mandatory to start
5493 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
5494 do.
5495
5496 When a connect must start the ruleset, if may still be preceded by set-var,
5497 unset-var or comment rules.
5498
5499 Examples :
Christopher Faulete5870d82020-04-15 11:32:03 +02005500 # check HTTP and HTTPs services on a server.
5501 # first open port 80 thanks to server line port directive, then
5502 # tcp-check opens port 443, ciphered and run a request on it:
5503 option httpchk
5504
5505 http-check connect
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02005506 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005507 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02005508 http-check connect port 443 ssl sni haproxy.1wt.eu
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02005509 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005510 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02005511
5512 server www 10.0.0.1 check port 80
5513
5514 See also : "option httpchk", "http-check send", "http-check expect"
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005515
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005516
Willy Tarreau0ba27502007-12-24 16:55:16 +01005517http-check disable-on-404
5518 Enable a maintenance mode upon HTTP/404 response to health-checks
5519 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01005520 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01005521 Arguments : none
5522
5523 When this option is set, a server which returns an HTTP code 404 will be
5524 excluded from further load-balancing, but will still receive persistent
5525 connections. This provides a very convenient method for Web administrators
5526 to perform a graceful shutdown of their servers. It is also important to note
5527 that a server which is detected as failed while it was in this mode will not
5528 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
5529 will immediately be reinserted into the farm. The status on the stats page
5530 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01005531 option only works in conjunction with the "httpchk" option. If this option
5532 is used with "http-check expect", then it has precedence over it so that 404
Christopher Fauletfa8b89a2020-11-20 18:54:13 +01005533 responses will still be considered as soft-stop. Note also that a stopped
5534 server will stay stopped even if it replies 404s. This option is only
5535 evaluated for running servers.
Willy Tarreaubd741542010-03-16 18:46:54 +01005536
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005537 See also : "option httpchk" and "http-check expect".
Willy Tarreaubd741542010-03-16 18:46:54 +01005538
5539
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005540http-check expect [min-recv <int>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02005541 [ok-status <st>] [error-status <st>] [tout-status <st>]
5542 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
5543 [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005544 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01005545 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02005546 yes | no | yes | yes
Christopher Faulete5870d82020-04-15 11:32:03 +02005547
Willy Tarreaubd741542010-03-16 18:46:54 +01005548 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005549 comment <msg> defines a message to report if the rule evaluation fails.
5550
Christopher Faulete5870d82020-04-15 11:32:03 +02005551 min-recv is optional and can define the minimum amount of data required to
5552 evaluate the current expect rule. If the number of received bytes
5553 is under this limit, the check will wait for more data. This
5554 option can be used to resolve some ambiguous matching rules or to
5555 avoid executing costly regex matches on content known to be still
5556 incomplete. If an exact string is used, the minimum between the
5557 string length and this parameter is used. This parameter is
5558 ignored if it is set to -1. If the expect rule does not match,
5559 the check will wait for more data. If set to 0, the evaluation
5560 result is always conclusive.
5561
5562 ok-status <st> is optional and can be used to set the check status if
5563 the expect rule is successfully evaluated and if it is
5564 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005565 "L6OK" and "L4OK" are supported :
5566 - L7OK : check passed on layer 7
Christopher Faulet83662b52020-11-20 17:47:47 +01005567 - L7OKC : check conditionally passed on layer 7, set
5568 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005569 - L6OK : check passed on layer 6
5570 - L4OK : check passed on layer 4
5571 By default "L7OK" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02005572
5573 error-status <st> is optional and can be used to set the check status if
5574 an error occurred during the expect rule evaluation.
Christopher Faulet83662b52020-11-20 17:47:47 +01005575 "L7OKC", "L7RSP", "L7STS", "L6RSP" and "L4CON" are
5576 supported :
5577 - L7OKC : check conditionally passed on layer 7, set
5578 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005579 - L7RSP : layer 7 invalid response - protocol error
5580 - L7STS : layer 7 response error, for example HTTP 5xx
5581 - L6RSP : layer 6 invalid response - protocol error
5582 - L4CON : layer 1-4 connection problem
5583 By default "L7RSP" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02005584
5585 tout-status <st> is optional and can be used to set the check status if
5586 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005587 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
5588 - L7TOUT : layer 7 (HTTP/SMTP) timeout
5589 - L6TOUT : layer 6 (SSL) timeout
5590 - L4TOUT : layer 1-4 timeout
Christopher Faulete5870d82020-04-15 11:32:03 +02005591 By default "L7TOUT" is used.
5592
5593 on-success <fmt> is optional and can be used to customize the
5594 informational message reported in logs if the expect
5595 rule is successfully evaluated and if it is the last rule
5596 in the tcp-check ruleset. <fmt> is a log-format string.
5597
5598 on-error <fmt> is optional and can be used to customize the
5599 informational message reported in logs if an error
5600 occurred during the expect rule evaluation. <fmt> is a
5601 log-format string.
5602
Willy Tarreaubd741542010-03-16 18:46:54 +01005603 <match> is a keyword indicating how to look for a specific pattern in the
Christopher Fauletb5594262020-05-05 20:23:13 +02005604 response. The keyword may be one of "status", "rstatus", "hdr",
5605 "fhdr", "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01005606 exclamation mark ("!") to negate the match. Spaces are allowed
5607 between the exclamation mark and the keyword. See below for more
5608 details on the supported keywords.
5609
Christopher Faulet39708192020-05-05 10:47:36 +02005610 <pattern> is the pattern to look for. It may be a string, a regular
5611 expression or a more complex pattern with several arguments. If
5612 the string pattern contains spaces, they must be escaped with the
5613 usual backslash ('\').
Willy Tarreaubd741542010-03-16 18:46:54 +01005614
5615 By default, "option httpchk" considers that response statuses 2xx and 3xx
5616 are valid, and that others are invalid. When "http-check expect" is used,
5617 it defines what is considered valid or invalid. Only one "http-check"
5618 statement is supported in a backend. If a server fails to respond or times
5619 out, the check obviously fails. The available matches are :
5620
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005621 status <codes> : test the status codes found parsing <codes> string. it
5622 must be a comma-separated list of status codes or range
5623 codes. A health check response will be considered as
5624 valid if the response's status code matches any status
5625 code or is inside any range of the list. If the "status"
5626 keyword is prefixed with "!", then the response will be
5627 considered invalid if the status code matches.
Willy Tarreaubd741542010-03-16 18:46:54 +01005628
5629 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005630 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005631 response's status code matches the expression. If the
5632 "rstatus" keyword is prefixed with "!", then the response
5633 will be considered invalid if the status code matches.
5634 This is mostly used to check for multiple codes.
5635
Christopher Fauletb5594262020-05-05 20:23:13 +02005636 hdr { name | name-lf } [ -m <meth> ] <name>
5637 [ { value | value-lf } [ -m <meth> ] <value> :
Christopher Faulet39708192020-05-05 10:47:36 +02005638 test the specified header pattern on the HTTP response
5639 headers. The name pattern is mandatory but the value
5640 pattern is optional. If not specified, only the header
5641 presence is verified. <meth> is the matching method,
5642 applied on the header name or the header value. Supported
5643 matching methods are "str" (exact match), "beg" (prefix
5644 match), "end" (suffix match), "sub" (substring match) or
5645 "reg" (regex match). If not specified, exact matching
Christopher Fauletb5594262020-05-05 20:23:13 +02005646 method is used. If the "name-lf" parameter is used,
5647 <name> is evaluated as a log-format string. If "value-lf"
5648 parameter is used, <value> is evaluated as a log-format
5649 string. These parameters cannot be used with the regex
5650 matching method. Finally, the header value is considered
5651 as comma-separated list. Note that matchings are case
5652 insensitive on the header names.
5653
5654 fhdr { name | name-lf } [ -m <meth> ] <name>
5655 [ { value | value-lf } [ -m <meth> ] <value> :
5656 test the specified full header pattern on the HTTP
5657 response headers. It does exactly the same than "hdr"
5658 keyword, except the full header value is tested, commas
5659 are not considered as delimiters.
Christopher Faulet39708192020-05-05 10:47:36 +02005660
Willy Tarreaubd741542010-03-16 18:46:54 +01005661 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005662 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005663 response's body contains this exact string. If the
5664 "string" keyword is prefixed with "!", then the response
5665 will be considered invalid if the body contains this
5666 string. This can be used to look for a mandatory word at
5667 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005668 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01005669 trace).
5670
5671 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005672 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005673 response's body matches this expression. If the "rstring"
5674 keyword is prefixed with "!", then the response will be
5675 considered invalid if the body matches the expression.
5676 This can be used to look for a mandatory word at the end
5677 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01005678 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01005679
Christopher Fauletaaab0832020-05-05 15:54:22 +02005680 string-lf <fmt> : test a log-format string match in the HTTP response body.
5681 A health check response will be considered valid if the
5682 response's body contains the string resulting of the
5683 evaluation of <fmt>, which follows the log-format rules.
5684 If prefixed with "!", then the response will be
5685 considered invalid if the body contains the string.
5686
Willy Tarreaubd741542010-03-16 18:46:54 +01005687 It is important to note that the responses will be limited to a certain size
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +01005688 defined by the global "tune.bufsize" option, which defaults to 16384 bytes.
Willy Tarreaubd741542010-03-16 18:46:54 +01005689 Thus, too large responses may not contain the mandatory pattern when using
5690 "string" or "rstring". If a large response is absolutely required, it is
5691 possible to change the default max size by setting the global variable.
5692 However, it is worth keeping in mind that parsing very large responses can
5693 waste some CPU cycles, especially when regular expressions are used, and that
5694 it is always better to focus the checks on smaller resources.
5695
Christopher Faulete5870d82020-04-15 11:32:03 +02005696 In an http-check ruleset, the last expect rule may be implicit. If no expect
5697 rule is specified after the last "http-check send", an implicit expect rule
5698 is defined to match on 2xx or 3xx status codes. It means this rule is also
5699 defined if there is no "http-check" rule at all, when only "option httpchk"
5700 is set.
Cyril Bonté32602d22015-01-30 00:07:07 +01005701
Willy Tarreaubd741542010-03-16 18:46:54 +01005702 Last, if "http-check expect" is combined with "http-check disable-on-404",
5703 then this last one has precedence when the server responds with 404.
5704
5705 Examples :
5706 # only accept status 200 as valid
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005707 http-check expect status 200,201,300-310
Willy Tarreaubd741542010-03-16 18:46:54 +01005708
Christopher Faulet39708192020-05-05 10:47:36 +02005709 # be sure a sessid coookie is set
5710 http-check expect header name "set-cookie" value -m beg "sessid="
5711
Willy Tarreaubd741542010-03-16 18:46:54 +01005712 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005713 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01005714
5715 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005716 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01005717
5718 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005719 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01005720
Christopher Faulete5870d82020-04-15 11:32:03 +02005721 See also : "option httpchk", "http-check connect", "http-check disable-on-404"
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005722 and "http-check send".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005723
5724
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005725http-check send [meth <method>] [{ uri <uri> | uri-lf <fmt> }>] [ver <version>]
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005726 [hdr <name> <fmt>]* [{ body <string> | body-lf <fmt> }]
5727 [comment <msg>]
Christopher Faulet8acb1282020-04-09 08:44:06 +02005728 Add a possible list of headers and/or a body to the request sent during HTTP
5729 health checks.
5730 May be used in sections : defaults | frontend | listen | backend
5731 yes | no | yes | yes
5732 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005733 comment <msg> defines a message to report if the rule evaluation fails.
5734
Christopher Faulete5870d82020-04-15 11:32:03 +02005735 meth <method> is the optional HTTP method used with the requests. When not
5736 set, the "OPTIONS" method is used, as it generally requires
5737 low server processing and is easy to filter out from the
5738 logs. Any method may be used, though it is not recommended
5739 to invent non-standard ones.
5740
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005741 uri <uri> is optional and set the URI referenced in the HTTP requests
5742 to the string <uri>. It defaults to "/" which is accessible
5743 by default on almost any server, but may be changed to any
5744 other URI. Query strings are permitted.
5745
5746 uri-lf <fmt> is optional and set the URI referenced in the HTTP requests
5747 using the log-format string <fmt>. It defaults to "/" which
5748 is accessible by default on almost any server, but may be
5749 changed to any other URI. Query strings are permitted.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005750
Christopher Faulet907701b2020-04-28 09:37:00 +02005751 ver <version> is the optional HTTP version string. It defaults to
Christopher Faulete5870d82020-04-15 11:32:03 +02005752 "HTTP/1.0" but some servers might behave incorrectly in HTTP
Daniel Corbett67a82712020-07-06 23:01:19 -04005753 1.0, so turning it to HTTP/1.1 may sometimes help. Note that
Christopher Faulete5870d82020-04-15 11:32:03 +02005754 the Host field is mandatory in HTTP/1.1, use "hdr" argument
5755 to add it.
5756
5757 hdr <name> <fmt> adds the HTTP header field whose name is specified in
5758 <name> and whose value is defined by <fmt>, which follows
5759 to the log-format rules.
5760
5761 body <string> add the body defined by <string> to the request sent during
5762 HTTP health checks. If defined, the "Content-Length" header
5763 is thus automatically added to the request.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005764
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005765 body-lf <fmt> add the body defined by the log-format string <fmt> to the
5766 request sent during HTTP health checks. If defined, the
5767 "Content-Length" header is thus automatically added to the
5768 request.
5769
Christopher Faulet8acb1282020-04-09 08:44:06 +02005770 In addition to the request line defined by the "option httpchk" directive,
5771 this one is the valid way to add some headers and optionally a body to the
5772 request sent during HTTP health checks. If a body is defined, the associate
Christopher Faulet9df910c2020-04-29 14:20:47 +02005773 "Content-Length" header is automatically added. Thus, this header or
5774 "Transfer-encoding" header should not be present in the request provided by
5775 "http-check send". If so, it will be ignored. The old trick consisting to add
5776 headers after the version string on the "option httpchk" line is now
Amaury Denoyelle6d975f02020-12-22 14:08:52 +01005777 deprecated.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005778
Christopher Faulete5870d82020-04-15 11:32:03 +02005779 Also "http-check send" doesn't support HTTP keep-alive. Keep in mind that it
Amaury Denoyelle6d975f02020-12-22 14:08:52 +01005780 will automatically append a "Connection: close" header, unless a Connection
5781 header has already already been configured via a hdr entry.
Christopher Faulet9df910c2020-04-29 14:20:47 +02005782
5783 Note that the Host header and the request authority, when both defined, are
5784 automatically synchronized. It means when the HTTP request is sent, when a
5785 Host is inserted in the request, the request authority is accordingly
5786 updated. Thus, don't be surprised if the Host header value overwrites the
5787 configured request authority.
5788
5789 Note also for now, no Host header is automatically added in HTTP/1.1 or above
5790 requests. You should add it explicitly.
Christopher Faulete5870d82020-04-15 11:32:03 +02005791
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005792 See also : "option httpchk", "http-check send-state" and "http-check expect".
Christopher Faulet8acb1282020-04-09 08:44:06 +02005793
5794
Willy Tarreauef781042010-01-27 11:53:01 +01005795http-check send-state
5796 Enable emission of a state header with HTTP health checks
5797 May be used in sections : defaults | frontend | listen | backend
5798 yes | no | yes | yes
5799 Arguments : none
5800
Daniel Corbett9f0843f2021-05-08 10:50:37 -04005801 When this option is set, HAProxy will systematically send a special header
Willy Tarreauef781042010-01-27 11:53:01 +01005802 "X-Haproxy-Server-State" with a list of parameters indicating to each server
Daniel Corbett9f0843f2021-05-08 10:50:37 -04005803 how they are seen by HAProxy. This can be used for instance when a server is
5804 manipulated without access to HAProxy and the operator needs to know whether
5805 HAProxy still sees it up or not, or if the server is the last one in a farm.
Willy Tarreauef781042010-01-27 11:53:01 +01005806
5807 The header is composed of fields delimited by semi-colons, the first of which
5808 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
5809 checks on the total number before transition, just as appears in the stats
5810 interface. Next headers are in the form "<variable>=<value>", indicating in
5811 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08005812 - a variable "address", containing the address of the backend server.
5813 This corresponds to the <address> field in the server declaration. For
5814 unix domain sockets, it will read "unix".
5815
5816 - a variable "port", containing the port of the backend server. This
5817 corresponds to the <port> field in the server declaration. For unix
5818 domain sockets, it will read "unix".
5819
Willy Tarreauef781042010-01-27 11:53:01 +01005820 - a variable "name", containing the name of the backend followed by a slash
5821 ("/") then the name of the server. This can be used when a server is
5822 checked in multiple backends.
5823
Daniel Corbett9f0843f2021-05-08 10:50:37 -04005824 - a variable "node" containing the name of the HAProxy node, as set in the
Willy Tarreauef781042010-01-27 11:53:01 +01005825 global "node" variable, otherwise the system's hostname if unspecified.
5826
5827 - a variable "weight" indicating the weight of the server, a slash ("/")
5828 and the total weight of the farm (just counting usable servers). This
5829 helps to know if other servers are available to handle the load when this
5830 one fails.
5831
5832 - a variable "scur" indicating the current number of concurrent connections
5833 on the server, followed by a slash ("/") then the total number of
5834 connections on all servers of the same backend.
5835
5836 - a variable "qcur" indicating the current number of requests in the
5837 server's queue.
5838
5839 Example of a header received by the application server :
5840 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
5841 scur=13/22; qcur=0
5842
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005843 See also : "option httpchk", "http-check disable-on-404" and
5844 "http-check send".
Willy Tarreauef781042010-01-27 11:53:01 +01005845
Christopher Faulete5870d82020-04-15 11:32:03 +02005846
5847http-check set-var(<var-name>) <expr>
Christopher Faulete5870d82020-04-15 11:32:03 +02005848 This operation sets the content of a variable. The variable is declared inline.
Christopher Faulete5870d82020-04-15 11:32:03 +02005849 May be used in sections: defaults | frontend | listen | backend
5850 yes | no | yes | yes
5851
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005852 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005853 <var-name> The name of the variable starts with an indication about its
5854 scope. The scopes allowed for http-check are:
5855 "proc" : the variable is shared with the whole process.
5856 "sess" : the variable is shared with the tcp-check session.
5857 "check": the variable is declared for the lifetime of the tcp-check.
5858 This prefix is followed by a name. The separator is a '.'.
5859 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5860 and '-'.
5861
5862 <expr> Is a sample-fetch expression potentially followed by converters.
5863
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005864 Examples :
5865 http-check set-var(check.port) int(1234)
Christopher Faulete5870d82020-04-15 11:32:03 +02005866
5867
5868http-check unset-var(<var-name>)
Christopher Faulete5870d82020-04-15 11:32:03 +02005869 Free a reference to a variable within its scope.
Christopher Faulete5870d82020-04-15 11:32:03 +02005870 May be used in sections: defaults | frontend | listen | backend
5871 yes | no | yes | yes
5872
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005873 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005874 <var-name> The name of the variable starts with an indication about its
5875 scope. The scopes allowed for http-check are:
5876 "proc" : the variable is shared with the whole process.
5877 "sess" : the variable is shared with the tcp-check session.
5878 "check": the variable is declared for the lifetime of the tcp-check.
5879 This prefix is followed by a name. The separator is a '.'.
5880 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5881 and '-'.
5882
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005883 Examples :
5884 http-check unset-var(check.port)
Christopher Faulete5870d82020-04-15 11:32:03 +02005885
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005886
Christopher Faulet3b967c12020-05-15 15:47:44 +02005887http-error status <code> [content-type <type>]
5888 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5889 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5890 [ hdr <name> <fmt> ]*
5891 Defines a custom error message to use instead of errors generated by HAProxy.
5892 May be used in sections : defaults | frontend | listen | backend
5893 yes | yes | yes | yes
5894 Arguments :
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005895 status <code> is the HTTP status code. It must be specified.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005896 Currently, HAProxy is capable of generating codes
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005897 200, 400, 401, 403, 404, 405, 407, 408, 410, 413, 425,
Christopher Faulete095f312020-12-07 11:22:24 +01005898 429, 500, 501, 502, 503, and 504.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005899
5900 content-type <type> is the response content type, for instance
5901 "text/plain". This parameter is ignored and should be
5902 omitted when an errorfile is configured or when the
5903 payload is empty. Otherwise, it must be defined.
5904
5905 default-errorfiles Reset the previously defined error message for current
5906 proxy for the status <code>. If used on a backend, the
5907 frontend error message is used, if defined. If used on
5908 a frontend, the default error message is used.
5909
5910 errorfile <file> designates a file containing the full HTTP response.
5911 It is recommended to follow the common practice of
5912 appending ".http" to the filename so that people do
5913 not confuse the response with HTML error pages, and to
5914 use absolute paths, since files are read before any
5915 chroot is performed.
5916
5917 errorfiles <name> designates the http-errors section to use to import
5918 the error message with the status code <code>. If no
5919 such message is found, the proxy's error messages are
5920 considered.
5921
5922 file <file> specifies the file to use as response payload. If the
5923 file is not empty, its content-type must be set as
5924 argument to "content-type", otherwise, any
5925 "content-type" argument is ignored. <file> is
5926 considered as a raw string.
5927
5928 string <str> specifies the raw string to use as response payload.
5929 The content-type must always be set as argument to
5930 "content-type".
5931
5932 lf-file <file> specifies the file to use as response payload. If the
5933 file is not empty, its content-type must be set as
5934 argument to "content-type", otherwise, any
5935 "content-type" argument is ignored. <file> is
5936 evaluated as a log-format string.
5937
5938 lf-string <str> specifies the log-format string to use as response
5939 payload. The content-type must always be set as
5940 argument to "content-type".
5941
5942 hdr <name> <fmt> adds to the response the HTTP header field whose name
5943 is specified in <name> and whose value is defined by
5944 <fmt>, which follows to the log-format rules.
5945 This parameter is ignored if an errorfile is used.
5946
5947 This directive may be used instead of "errorfile", to define a custom error
5948 message. As "errorfile" directive, it is used for errors detected and
5949 returned by HAProxy. If an errorfile is defined, it is parsed when HAProxy
5950 starts and must be valid according to the HTTP standards. The generated
5951 response must not exceed the configured buffer size (BUFFSIZE), otherwise an
5952 internal error will be returned. Finally, if you consider to use some
5953 http-after-response rules to rewrite these errors, the reserved buffer space
5954 should be available (see "tune.maxrewrite").
5955
5956 The files are read at the same time as the configuration and kept in memory.
5957 For this reason, the errors continue to be returned even when the process is
5958 chrooted, and no file change is considered while the process is running.
5959
Christopher Fauletd5ac6de2020-12-02 08:40:14 +01005960 Note: 400/408/500 errors emitted in early stage of the request parsing are
5961 handled by the multiplexer at a lower level. No custom formatting is
5962 supported at this level. Thus only static error messages, defined with
5963 "errorfile" directive, are supported. However, this limitation only
5964 exists during the request headers parsing or between two transactions.
5965
Christopher Faulet3b967c12020-05-15 15:47:44 +02005966 See also : "errorfile", "errorfiles", "errorloc", "errorloc302",
5967 "errorloc303" and section 3.8 about http-errors.
5968
5969
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005970http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005971 Access control for Layer 7 requests
5972
5973 May be used in sections: defaults | frontend | listen | backend
5974 no | yes | yes | yes
5975
Willy Tarreau20b0de52012-12-24 15:45:22 +01005976 The http-request statement defines a set of rules which apply to layer 7
5977 processing. The rules are evaluated in their declaration order when they are
5978 met in a frontend, listen or backend section. Any rule may optionally be
5979 followed by an ACL-based condition, in which case it will only be evaluated
5980 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005981
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005982 The first keyword is the rule's action. The supported actions are described
5983 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005984
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005985 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005986
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005987 Example:
5988 acl nagios src 192.168.129.3
5989 acl local_net src 192.168.0.0/16
5990 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01005991
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005992 http-request allow if nagios
5993 http-request allow if local_net auth_ok
5994 http-request auth realm Gimme if local_net auth_ok
5995 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01005996
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005997 Example:
5998 acl key req.hdr(X-Add-Acl-Key) -m found
5999 acl add path /addacl
6000 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01006001
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006002 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01006003
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006004 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
6005 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02006006
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006007 Example:
6008 acl value req.hdr(X-Value) -m found
6009 acl setmap path /setmap
6010 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06006011
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006012 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06006013
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006014 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
6015 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06006016
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006017 See also : "stats http-request", section 3.4 about userlists and section 7
6018 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06006019
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006020http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006021
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006022 This is used to add a new entry into an ACL. The ACL must be loaded from a
6023 file (even a dummy empty file). The file name of the ACL to be updated is
6024 passed between parentheses. It takes one argument: <key fmt>, which follows
6025 log-format rules, to collect content of the new entry. It performs a lookup
6026 in the ACL before insertion, to avoid duplicated (or more) values. This
6027 lookup is done by a linear search and can be expensive with large lists!
6028 It is the equivalent of the "add acl" command from the stats socket, but can
6029 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06006030
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006031http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006032
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006033 This appends an HTTP header field whose name is specified in <name> and
6034 whose value is defined by <fmt> which follows the log-format rules (see
6035 Custom Log Format in section 8.2.4). This is particularly useful to pass
6036 connection-specific information to the server (e.g. the client's SSL
6037 certificate), or to combine several headers into one. This rule is not
6038 final, so it is possible to add other similar rules. Note that header
6039 addition is performed immediately, so one rule might reuse the resulting
6040 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06006041
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006042http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006043
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006044 This stops the evaluation of the rules and lets the request pass the check.
6045 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06006046
Sasha Pachev218f0642014-06-16 12:05:59 -06006047
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006048http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006049
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006050 This stops the evaluation of the rules and immediately responds with an
6051 HTTP 401 or 407 error code to invite the user to present a valid user name
6052 and password. No further "http-request" rules are evaluated. An optional
6053 "realm" parameter is supported, it sets the authentication realm that is
6054 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06006055
Christopher Faulet612f2ea2020-05-27 09:57:28 +02006056 The corresponding proxy's error message is used. It may be customized using
6057 an "errorfile" or an "http-error" directive. For 401 responses, all
6058 occurrences of the WWW-Authenticate header are removed and replaced by a new
6059 one with a basic authentication challenge for realm "<realm>". For 407
6060 responses, the same is done on the Proxy-Authenticate header. If the error
6061 message must not be altered, consider to use "http-request return" rule
6062 instead.
6063
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006064 Example:
6065 acl auth_ok http_auth_group(L1) G1
6066 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06006067
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02006068http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006069
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006070 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01006071
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006072http-request capture <sample> [ len <length> | id <id> ]
6073 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01006074
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006075 This captures sample expression <sample> from the request buffer, and
6076 converts it to a string of at most <len> characters. The resulting string is
6077 stored into the next request "capture" slot, so it will possibly appear next
6078 to some captured HTTP headers. It will then automatically appear in the logs,
6079 and it will be possible to extract it using sample fetch rules to feed it
6080 into headers or anything. The length should be limited given that this size
6081 will be allocated for each capture during the whole session life.
6082 Please check section 7.3 (Fetching samples) and "capture request header" for
6083 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01006084
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006085 If the keyword "id" is used instead of "len", the action tries to store the
6086 captured string in a previously declared capture slot. This is useful to run
6087 captures in backends. The slot id can be declared by a previous directive
Baptiste Assmann19a69b32020-01-16 14:34:22 +01006088 "http-request capture" or with the "declare capture" keyword.
6089
6090 When using this action in a backend, double check that the relevant
6091 frontend(s) have the required capture slots otherwise, this rule will be
6092 ignored at run time. This can't be detected at configuration parsing time
6093 due to HAProxy's ability to dynamically resolve backend name at runtime.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01006094
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006095http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01006096
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006097 This is used to delete an entry from an ACL. The ACL must be loaded from a
6098 file (even a dummy empty file). The file name of the ACL to be updated is
6099 passed between parentheses. It takes one argument: <key fmt>, which follows
6100 log-format rules, to collect content of the entry to delete.
6101 It is the equivalent of the "del acl" command from the stats socket, but can
6102 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01006103
Maciej Zdebebdd4c52020-11-20 13:58:48 +00006104http-request del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02006105
Maciej Zdebebdd4c52020-11-20 13:58:48 +00006106 This removes all HTTP header fields whose name is specified in <name>. <meth>
6107 is the matching method, applied on the header name. Supported matching methods
6108 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
6109 (substring match) and "reg" (regex match). If not specified, exact matching
6110 method is used.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02006111
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006112http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02006113
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006114 This is used to delete an entry from a MAP. The MAP must be loaded from a
6115 file (even a dummy empty file). The file name of the MAP to be updated is
6116 passed between parentheses. It takes one argument: <key fmt>, which follows
6117 log-format rules, to collect content of the entry to delete.
6118 It takes one argument: "file name" It is the equivalent of the "del map"
6119 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02006120
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006121http-request deny [deny_status <status>] [ { if | unless } <condition> ]
6122http-request deny [ { status | deny_status } <code>] [content-type <type>]
6123 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6124 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6125 [ hdr <name> <fmt> ]*
6126 [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04006127
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006128 This stops the evaluation of the rules and immediately rejects the request.
6129 By default an HTTP 403 error is returned. But the response may be customized
6130 using same syntax than "http-request return" rules. Thus, see "http-request
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006131 return" for details. For compatibility purpose, when no argument is defined,
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006132 or only "deny_status", the argument "default-errorfiles" is implied. It means
6133 "http-request deny [deny_status <status>]" is an alias of
6134 "http-request deny [status <status>] default-errorfiles".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006135 No further "http-request" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006136 See also "http-request return".
Patrick Hemmer268a7072018-05-11 12:52:31 -04006137
Olivier Houchard602bf7d2019-05-10 13:59:15 +02006138http-request disable-l7-retry [ { if | unless } <condition> ]
6139 This disables any attempt to retry the request if it fails for any other
6140 reason than a connection failure. This can be useful for example to make
6141 sure POST requests aren't retried on failure.
6142
Baptiste Assmann333939c2019-01-21 08:34:50 +01006143http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
6144
6145 This action performs a DNS resolution of the output of <expr> and stores
6146 the result in the variable <var>. It uses the DNS resolvers section
6147 pointed by <resolvers>.
6148 It is possible to choose a resolution preference using the optional
6149 arguments 'ipv4' or 'ipv6'.
6150 When performing the DNS resolution, the client side connection is on
6151 pause waiting till the end of the resolution.
6152 If an IP address can be found, it is stored into <var>. If any kind of
6153 error occurs, then <var> is not set.
6154 One can use this action to discover a server IP address at run time and
6155 based on information found in the request (IE a Host header).
6156 If this action is used to find the server's IP address (using the
6157 "set-dst" action), then the server IP address in the backend must be set
6158 to 0.0.0.0.
6159
6160 Example:
6161 resolvers mydns
6162 nameserver local 127.0.0.53:53
6163 nameserver google 8.8.8.8:53
6164 timeout retry 1s
6165 hold valid 10s
6166 hold nx 3s
6167 hold other 3s
6168 hold obsolete 0s
6169 accepted_payload_size 8192
6170
6171 frontend fe
6172 bind 10.42.0.1:80
6173 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
6174 http-request capture var(txn.myip) len 40
6175
6176 # return 503 when the variable is not set,
6177 # which mean DNS resolution error
6178 use_backend b_503 unless { var(txn.myip) -m found }
6179
6180 default_backend be
6181
6182 backend b_503
6183 # dummy backend used to return 503.
6184 # one can use the errorfile directive to send a nice
6185 # 503 error page to end users
6186
6187 backend be
6188 # rule to prevent HAProxy from reconnecting to services
6189 # on the local network (forged DNS name used to scan the network)
6190 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
6191 http-request set-dst var(txn.myip)
6192 server clear 0.0.0.0:0
6193
6194 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
6195 be used to scan the network or worst won't loop over itself...
6196
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01006197http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
6198
6199 This is used to build an HTTP 103 Early Hints response prior to any other one.
6200 This appends an HTTP header field to this response whose name is specified in
6201 <name> and whose value is defined by <fmt> which follows the log-format rules
6202 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01006203 to the client some Link headers to preload resources required to render the
6204 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01006205
6206 See RFC 8297 for more information.
6207
Tim Duesterhusd371e992021-04-15 21:45:58 +02006208http-request normalize-uri <normalizer> [ { if | unless } <condition> ]
Tim Duesterhusdec1c362021-05-10 17:28:26 +02006209http-request normalize-uri fragment-encode [ { if | unless } <condition> ]
Tim Duesterhusc9e05ab2021-05-10 17:28:25 +02006210http-request normalize-uri fragment-strip [ { if | unless } <condition> ]
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006211http-request normalize-uri path-merge-slashes [ { if | unless } <condition> ]
Maximilian Maderff3bb8b2021-04-21 00:22:50 +02006212http-request normalize-uri path-strip-dot [ { if | unless } <condition> ]
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006213http-request normalize-uri path-strip-dotdot [ full ] [ { if | unless } <condition> ]
Tim Duesterhus2e4a18e2021-04-21 21:20:36 +02006214http-request normalize-uri percent-decode-unreserved [ strict ] [ { if | unless } <condition> ]
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006215http-request normalize-uri percent-to-uppercase [ strict ] [ { if | unless } <condition> ]
6216http-request normalize-uri query-sort-by-name [ { if | unless } <condition> ]
Tim Duesterhusd371e992021-04-15 21:45:58 +02006217
Tim Duesterhusb918a4a2021-04-16 23:52:29 +02006218 Performs normalization of the request's URI.
6219
Tim Duesterhus2963fd32021-04-17 00:24:56 +02006220 URI normalization in HAProxy 2.4 is currently available as an experimental
Amaury Denoyellea9e639a2021-05-06 15:50:12 +02006221 technical preview. As such, it requires the global directive
6222 'expose-experimental-directives' first to be able to invoke it. You should be
6223 prepared that the behavior of normalizers might change to fix possible
6224 issues, possibly breaking proper request processing in your infrastructure.
Tim Duesterhus2963fd32021-04-17 00:24:56 +02006225
Tim Duesterhusb918a4a2021-04-16 23:52:29 +02006226 Each normalizer handles a single type of normalization to allow for a
6227 fine-grained selection of the level of normalization that is appropriate for
6228 the supported backend.
6229
6230 As an example the "path-strip-dotdot" normalizer might be useful for a static
6231 fileserver that directly maps the requested URI to the path within the local
6232 filesystem. However it might break routing of an API that expects a specific
6233 number of segments in the path.
6234
Tim Duesterhus2e4a18e2021-04-21 21:20:36 +02006235 It is important to note that some normalizers might result in unsafe
6236 transformations for broken URIs. It might also be possible that a combination
6237 of normalizers that are safe by themselves results in unsafe transformations
6238 when improperly combined.
6239
6240 As an example the "percent-decode-unreserved" normalizer might result in
6241 unexpected results when a broken URI includes bare percent characters. One
6242 such a broken URI is "/%%36%36" which would be decoded to "/%66" which in
6243 turn is equivalent to "/f". By specifying the "strict" option requests to
6244 such a broken URI would safely be rejected.
6245
Tim Duesterhusb918a4a2021-04-16 23:52:29 +02006246 The following normalizers are available:
Tim Duesterhusd371e992021-04-15 21:45:58 +02006247
Tim Duesterhusdec1c362021-05-10 17:28:26 +02006248 - fragment-encode: Encodes "#" as "%23".
6249
6250 The "fragment-strip" normalizer should be preferred, unless it is known
6251 that broken clients do not correctly encode '#' within the path component.
6252
6253 Example:
6254 - /#foo -> /%23foo
6255
Tim Duesterhusc9e05ab2021-05-10 17:28:25 +02006256 - fragment-strip: Removes the URI's "fragment" component.
6257
6258 According to RFC 3986#3.5 the "fragment" component of an URI should not
6259 be sent, but handled by the User Agent after retrieving a resource.
6260
6261 This normalizer should be applied first to ensure that the fragment is
6262 not interpreted as part of the request's path component.
6263
6264 Example:
6265 - /#foo -> /
6266
Tim Duesterhusd6d33de2021-04-21 21:20:35 +02006267 - path-strip-dot: Removes "/./" segments within the "path" component
6268 (RFC 3986#6.2.2.3).
Maximilian Maderff3bb8b2021-04-21 00:22:50 +02006269
Tim Duesterhus2e4a18e2021-04-21 21:20:36 +02006270 Segments including percent encoded dots ("%2E") will not be detected. Use
6271 the "percent-decode-unreserved" normalizer first if this is undesired.
6272
Tim Duesterhus7a95f412021-04-21 21:20:33 +02006273 Example:
6274 - /. -> /
6275 - /./bar/ -> /bar/
6276 - /a/./a -> /a/a
6277 - /.well-known/ -> /.well-known/ (no change)
Maximilian Maderff3bb8b2021-04-21 00:22:50 +02006278
Tim Duesterhusd6d33de2021-04-21 21:20:35 +02006279 - path-strip-dotdot: Normalizes "/../" segments within the "path" component
6280 (RFC 3986#6.2.2.3).
6281
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006282 This merges segments that attempt to access the parent directory with
Tim Duesterhus2e4a18e2021-04-21 21:20:36 +02006283 their preceding segment.
6284
6285 Empty segments do not receive special treatment. Use the "merge-slashes"
6286 normalizer first if this is undesired.
6287
6288 Segments including percent encoded dots ("%2E") will not be detected. Use
6289 the "percent-decode-unreserved" normalizer first if this is undesired.
Tim Duesterhus9982fc22021-04-15 21:45:59 +02006290
6291 Example:
6292 - /foo/../ -> /
6293 - /foo/../bar/ -> /bar/
6294 - /foo/bar/../ -> /foo/
6295 - /../bar/ -> /../bar/
Tim Duesterhus560e1a62021-04-15 21:46:00 +02006296 - /bar/../../ -> /../
Tim Duesterhus9982fc22021-04-15 21:45:59 +02006297 - /foo//../ -> /foo/
Tim Duesterhus2e4a18e2021-04-21 21:20:36 +02006298 - /foo/%2E%2E/ -> /foo/%2E%2E/
Tim Duesterhus9982fc22021-04-15 21:45:59 +02006299
Tim Duesterhus560e1a62021-04-15 21:46:00 +02006300 If the "full" option is specified then "../" at the beginning will be
6301 removed as well:
6302
6303 Example:
6304 - /../bar/ -> /bar/
6305 - /bar/../../ -> /
6306
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006307 - path-merge-slashes: Merges adjacent slashes within the "path" component
6308 into a single slash.
Tim Duesterhusd371e992021-04-15 21:45:58 +02006309
6310 Example:
6311 - // -> /
6312 - /foo//bar -> /foo/bar
6313
Tim Duesterhus2e4a18e2021-04-21 21:20:36 +02006314 - percent-decode-unreserved: Decodes unreserved percent encoded characters to
6315 their representation as a regular character (RFC 3986#6.2.2.2).
6316
6317 The set of unreserved characters includes all letters, all digits, "-",
6318 ".", "_", and "~".
6319
6320 Example:
6321 - /%61dmin -> /admin
6322 - /foo%3Fbar=baz -> /foo%3Fbar=baz (no change)
6323 - /%%36%36 -> /%66 (unsafe)
6324 - /%ZZ -> /%ZZ
6325
6326 If the "strict" option is specified then invalid sequences will result
6327 in a HTTP 400 Bad Request being returned.
6328
6329 Example:
6330 - /%%36%36 -> HTTP 400
6331 - /%ZZ -> HTTP 400
6332
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006333 - percent-to-uppercase: Uppercases letters within percent-encoded sequences
Tim Duesterhusc315efd2021-04-21 21:20:34 +02006334 (RFC 3986#6.2.2.1).
Tim Duesterhusa4071932021-04-15 21:46:02 +02006335
6336 Example:
6337 - /%6f -> /%6F
6338 - /%zz -> /%zz
6339
6340 If the "strict" option is specified then invalid sequences will result
6341 in a HTTP 400 Bad Request being returned.
6342
6343 Example:
6344 - /%zz -> HTTP 400
6345
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006346 - query-sort-by-name: Sorts the query string parameters by parameter name.
Tim Duesterhusd7b89be2021-04-15 21:46:01 +02006347 Parameters are assumed to be delimited by '&'. Shorter names sort before
6348 longer names and identical parameter names maintain their relative order.
6349
6350 Example:
6351 - /?c=3&a=1&b=2 -> /?a=1&b=2&c=3
6352 - /?aaa=3&a=1&aa=2 -> /?a=1&aa=2&aaa=3
6353 - /?a=3&b=4&a=1&b=5&a=2 -> /?a=3&a=1&a=2&b=4&b=5
6354
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006355http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006356
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006357 This performs an HTTP redirection based on a redirect rule. This is exactly
6358 the same as the "redirect" statement except that it inserts a redirect rule
6359 which can be processed in the middle of other "http-request" rules and that
6360 these rules use the "log-format" strings. See the "redirect" keyword for the
6361 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006362
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006363http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006364
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006365 This stops the evaluation of the rules and immediately closes the connection
6366 without sending any response. It acts similarly to the
6367 "tcp-request content reject" rules. It can be useful to force an immediate
6368 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006369
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006370http-request replace-header <name> <match-regex> <replace-fmt>
6371 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02006372
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006373 This matches the value of all occurrences of header field <name> against
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006374 <match-regex>. Matching is performed case-sensitively. Matching values are
6375 completely replaced by <replace-fmt>. Format characters are allowed in
6376 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
6377 Standard back-references using the backslash ('\') followed by a number are
6378 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02006379
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006380 This action acts on whole header lines, regardless of the number of values
6381 they may contain. Thus it is well-suited to process headers naturally
6382 containing commas in their value, such as If-Modified-Since. Headers that
6383 contain a comma-separated list of values, such as Accept, should be processed
6384 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01006385
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006386 Example:
6387 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
6388
6389 # applied to:
6390 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
6391
6392 # outputs:
6393 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
6394
6395 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02006396
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006397 http-request replace-header User-Agent curl foo
6398
6399 # applied to:
6400 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02006401
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006402 # outputs:
6403 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02006404
Willy Tarreau262c3f12019-12-17 06:52:51 +01006405http-request replace-path <match-regex> <replace-fmt>
6406 [ { if | unless } <condition> ]
6407
6408 This works like "replace-header" except that it works on the request's path
6409 component instead of a header. The path component starts at the first '/'
Christopher Faulet82c83322020-09-02 14:16:59 +02006410 after an optional scheme+authority and ends before the question mark. Thus,
6411 the replacement does not modify the scheme, the authority and the
6412 query-string.
Willy Tarreau262c3f12019-12-17 06:52:51 +01006413
6414 It is worth noting that regular expressions may be more expensive to evaluate
6415 than certain ACLs, so rare replacements may benefit from a condition to avoid
6416 performing the evaluation at all if it does not match.
6417
6418 Example:
6419 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
6420 http-request replace-path (.*) /foo\1
6421
Willy Tarreau262c3f12019-12-17 06:52:51 +01006422 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
6423 http-request replace-path /foo/(.*) /\1
6424 # or more efficient if only some requests match :
6425 http-request replace-path /foo/(.*) /\1 if { url_beg /foo/ }
6426
Christopher Faulet312294f2020-09-02 17:17:44 +02006427http-request replace-pathq <match-regex> <replace-fmt>
6428 [ { if | unless } <condition> ]
6429
6430 This does the same as "http-request replace-path" except that the path
6431 contains the query-string if any is present. Thus, the path and the
6432 query-string are replaced.
6433
6434 Example:
6435 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
6436 http-request replace-pathq ([^?]*)(\?(.*))? \1/foo\2
6437
Willy Tarreau33810222019-06-12 17:44:02 +02006438http-request replace-uri <match-regex> <replace-fmt>
6439 [ { if | unless } <condition> ]
6440
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006441 This works like "replace-header" except that it works on the request's URI part
6442 instead of a header. The URI part may contain an optional scheme, authority or
6443 query string. These are considered to be part of the value that is matched
6444 against.
6445
6446 It is worth noting that regular expressions may be more expensive to evaluate
6447 than certain ACLs, so rare replacements may benefit from a condition to avoid
6448 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02006449
Willy Tarreau62b59132019-12-17 06:51:20 +01006450 IMPORTANT NOTE: historically in HTTP/1.x, the vast majority of requests sent
6451 by browsers use the "origin form", which differs from the "absolute form" in
6452 that they do not contain a scheme nor authority in the URI portion. Mostly
6453 only requests sent to proxies, those forged by hand and some emitted by
6454 certain applications use the absolute form. As such, "replace-uri" usually
6455 works fine most of the time in HTTP/1.x with rules starting with a "/". But
6456 with HTTP/2, clients are encouraged to send absolute URIs only, which look
6457 like the ones HTTP/1 clients use to talk to proxies. Such partial replace-uri
6458 rules may then fail in HTTP/2 when they work in HTTP/1. Either the rules need
Willy Tarreau262c3f12019-12-17 06:52:51 +01006459 to be adapted to optionally match a scheme and authority, or replace-path
6460 should be used.
Willy Tarreau33810222019-06-12 17:44:02 +02006461
Willy Tarreau62b59132019-12-17 06:51:20 +01006462 Example:
6463 # rewrite all "http" absolute requests to "https":
6464 http-request replace-uri ^http://(.*) https://\1
Willy Tarreau33810222019-06-12 17:44:02 +02006465
Willy Tarreau62b59132019-12-17 06:51:20 +01006466 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
6467 http-request replace-uri ([^/:]*://[^/]*)?(.*) \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02006468
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006469http-request replace-value <name> <match-regex> <replace-fmt>
6470 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02006471
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006472 This works like "replace-header" except that it matches the regex against
6473 every comma-delimited value of the header field <name> instead of the
6474 entire header. This is suited for all headers which are allowed to carry
6475 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02006476
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006477 Example:
6478 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02006479
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006480 # applied to:
6481 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02006482
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006483 # outputs:
6484 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01006485
Christopher Faulet24231ab2020-01-24 17:44:23 +01006486http-request return [status <code>] [content-type <type>]
6487 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6488 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006489 [ hdr <name> <fmt> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01006490 [ { if | unless } <condition> ]
6491
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006492 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01006493 default status code used for the response is 200. It can be optionally
6494 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04006495 specified as an argument to "content-type". Finally the response itself may
Sébastien Grossab877122020-10-08 10:06:03 +02006496 be defined. It can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006497 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01006498 are followed to create the response :
6499
6500 * If neither the errorfile nor the payload to use is defined, a dummy
6501 response is returned. Only the "status" argument is considered. It can be
6502 any code in the range [200, 599]. The "content-type" argument, if any, is
6503 ignored.
6504
6505 * If "default-errorfiles" argument is set, the proxy's errorfiles are
6506 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett9f0843f2021-05-08 10:50:37 -04006507 status code handled by HAProxy (200, 400, 403, 404, 405, 408, 410, 413,
Christopher Faulete095f312020-12-07 11:22:24 +01006508 425, 429, 500, 501, 502, 503, and 504). The "content-type" argument, if
6509 any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006510
6511 * If a specific errorfile is defined, with an "errorfile" argument, the
6512 corresponding file, containing a full HTTP response, is returned. Only the
6513 "status" argument is considered. It must be one of the status code handled
Daniel Corbett9f0843f2021-05-08 10:50:37 -04006514 by HAProxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 501,
Christopher Faulete095f312020-12-07 11:22:24 +01006515 502, 503, and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006516
6517 * If an http-errors section is defined, with an "errorfiles" argument, the
6518 corresponding file in the specified http-errors section, containing a full
6519 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett9f0843f2021-05-08 10:50:37 -04006520 must be one of the status code handled by HAProxy (200, 400, 403, 404, 405,
Christopher Faulete095f312020-12-07 11:22:24 +01006521 408, 410, 413, 425, 429, 500, 501, 502, 503, and 504). The "content-type"
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006522 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006523
6524 * If a "file" or a "lf-file" argument is specified, the file's content is
6525 used as the response payload. If the file is not empty, its content-type
6526 must be set as argument to "content-type". Otherwise, any "content-type"
6527 argument is ignored. With a "lf-file" argument, the file's content is
6528 evaluated as a log-format string. With a "file" argument, it is considered
6529 as a raw content.
6530
6531 * If a "string" or "lf-string" argument is specified, the defined string is
6532 used as the response payload. The content-type must always be set as
6533 argument to "content-type". With a "lf-string" argument, the string is
6534 evaluated as a log-format string. With a "string" argument, it is
6535 considered as a raw string.
6536
Sébastien Grossab877122020-10-08 10:06:03 +02006537 When the response is not based on an errorfile, it is possible to append HTTP
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006538 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
6539 arguments are ignored. For each one, the header name is specified in <name>
6540 and its value is defined by <fmt> which follows the log-format rules.
6541
Christopher Faulet24231ab2020-01-24 17:44:23 +01006542 Note that the generated response must be smaller than a buffer. And to avoid
6543 any warning, when an errorfile or a raw file is loaded, the buffer space
Sébastien Grossab877122020-10-08 10:06:03 +02006544 reserved for the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006545
6546 No further "http-request" rules are evaluated.
6547
6548 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006549 http-request return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006550 if { path /ping }
6551
6552 http-request return content-type image/x-icon file /var/www/favicon.ico \
6553 if { path /favicon.ico }
6554
6555 http-request return status 403 content-type text/plain \
6556 lf-string "Access denied. IP %[src] is blacklisted." \
6557 if { src -f /etc/haproxy/blacklist.lst }
6558
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006559http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6560http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006561
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006562 This actions increments the GPC0 or GPC1 counter according with the sticky
6563 counter designated by <sc-id>. If an error occurs, this action silently fails
6564 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006565
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006566http-request sc-set-gpt0(<sc-id>) { <int> | <expr> }
6567 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006568
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006569 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6570 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
6571 boolean. If an error occurs, this action silently fails and the actions
6572 evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006573
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006574http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006575
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006576 This is used to set the destination IP address to the value of specified
6577 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
6578 but provides the correct IP in a HTTP header; or you want to mask the IP for
6579 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
6580 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01006581
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006582 Arguments:
6583 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6584 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01006585
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006586 Example:
6587 http-request set-dst hdr(x-dst)
6588 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01006589
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006590 When possible, set-dst preserves the original destination port as long as the
6591 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006592
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006593http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006594
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006595 This is used to set the destination port address to the value of specified
6596 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
6597 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006598
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006599 Arguments:
6600 <expr> Is a standard HAProxy expression formed by a sample-fetch
6601 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006602
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006603 Example:
6604 http-request set-dst-port hdr(x-port)
6605 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006606
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006607 When possible, set-dst-port preserves the original destination address as
6608 long as the address family supports a port, otherwise it forces the
6609 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02006610
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006611http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02006612
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006613 This does the same as "http-request add-header" except that the header name
6614 is first removed if it existed. This is useful when passing security
6615 information to the server, where the header must not be manipulated by
6616 external users. Note that the new value is computed before the removal so it
6617 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02006618
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006619 Example:
6620 http-request set-header X-Haproxy-Current-Date %T
6621 http-request set-header X-SSL %[ssl_fc]
6622 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
6623 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
6624 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
6625 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
6626 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
6627 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
6628 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02006629
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006630http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02006631
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006632 This is used to change the log level of the current request when a certain
6633 condition is met. Valid levels are the 8 syslog levels (see the "log"
6634 keyword) plus the special level "silent" which disables logging for this
6635 request. This rule is not final so the last matching rule wins. This rule
6636 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006637
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006638http-request set-map(<file-name>) <key fmt> <value fmt>
6639 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006640
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006641 This is used to add a new entry into a MAP. The MAP must be loaded from a
6642 file (even a dummy empty file). The file name of the MAP to be updated is
6643 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
6644 log-format rules, used to collect MAP key, and <value fmt>, which follows
6645 log-format rules, used to collect content for the new entry.
6646 It performs a lookup in the MAP before insertion, to avoid duplicated (or
6647 more) values. This lookup is done by a linear search and can be expensive
6648 with large lists! It is the equivalent of the "set map" command from the
6649 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006650
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006651http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006652
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006653 This is used to set the Netfilter MARK on all packets sent to the client to
6654 the value passed in <mark> on platforms which support it. This value is an
6655 unsigned 32 bit value which can be matched by netfilter and by the routing
6656 table. It can be expressed both in decimal or hexadecimal format (prefixed by
6657 "0x"). This can be useful to force certain packets to take a different route
6658 (for example a cheaper network path for bulk downloads). This works on Linux
6659 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02006660
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006661http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006662
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006663 This rewrites the request method with the result of the evaluation of format
6664 string <fmt>. There should be very few valid reasons for having to do so as
6665 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006666
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006667http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006668
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006669 This sets the "nice" factor of the current request being processed. It only
6670 has effect against the other requests being processed at the same time.
6671 The default value is 0, unless altered by the "nice" setting on the "bind"
6672 line. The accepted range is -1024..1024. The higher the value, the nicest
6673 the request will be. Lower values will make the request more important than
6674 other ones. This can be useful to improve the speed of some requests, or
6675 lower the priority of non-important requests. Using this setting without
6676 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006677
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006678http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02006679
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006680 This rewrites the request path with the result of the evaluation of format
6681 string <fmt>. The query string, if any, is left intact. If a scheme and
6682 authority is found before the path, they are left intact as well. If the
6683 request doesn't have a path ("*"), this one is replaced with the format.
6684 This can be used to prepend a directory component in front of a path for
6685 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02006686
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006687 Example :
6688 # prepend the host name before the path
6689 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006690
Christopher Faulet312294f2020-09-02 17:17:44 +02006691http-request set-pathq <fmt> [ { if | unless } <condition> ]
6692
6693 This does the same as "http-request set-path" except that the query-string is
6694 also rewritten. It may be used to remove the query-string, including the
6695 question mark (it is not possible using "http-request set-query").
6696
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006697http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02006698
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006699 This is used to set the queue priority class of the current request.
6700 The value must be a sample expression which converts to an integer in the
6701 range -2047..2047. Results outside this range will be truncated.
6702 The priority class determines the order in which queued requests are
6703 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006704
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006705http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006706
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006707 This is used to set the queue priority timestamp offset of the current
6708 request. The value must be a sample expression which converts to an integer
6709 in the range -524287..524287. Results outside this range will be truncated.
6710 When a request is queued, it is ordered first by the priority class, then by
6711 the current timestamp adjusted by the given offset in milliseconds. Lower
6712 values have higher priority.
6713 Note that the resulting timestamp is is only tracked with enough precision
6714 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
6715 adjusted timestamp exceeds this value, it will be misidentified as highest
6716 priority. Thus it is important to set "timeout queue" to a value, where when
6717 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006718
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006719http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01006720
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006721 This rewrites the request's query string which appears after the first
6722 question mark ("?") with the result of the evaluation of format string <fmt>.
6723 The part prior to the question mark is left intact. If the request doesn't
6724 contain a question mark and the new value is not empty, then one is added at
6725 the end of the URI, followed by the new value. If a question mark was
6726 present, it will never be removed even if the value is empty. This can be
6727 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08006728
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006729 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006730
6731 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006732 # replace "%3D" with "=" in the query string
6733 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006734
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006735http-request set-src <expr> [ { if | unless } <condition> ]
6736 This is used to set the source IP address to the value of specified
6737 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
6738 provides the correct IP in a HTTP header; or you want to mask source IP for
Olivier Doucet56e31202020-04-21 09:32:56 +02006739 privacy. All subsequent calls to "src" fetch will return this value
6740 (see example).
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006741
6742 Arguments :
6743 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6744 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006745
Olivier Doucet56e31202020-04-21 09:32:56 +02006746 See also "option forwardfor".
6747
Cyril Bonté78caf842010-03-10 22:41:43 +01006748 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006749 http-request set-src hdr(x-forwarded-for)
6750 http-request set-src src,ipmask(24)
6751
Olivier Doucet56e31202020-04-21 09:32:56 +02006752 # After the masking this will track connections
6753 # based on the IP address with the last byte zeroed out.
6754 http-request track-sc0 src
6755
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006756 When possible, set-src preserves the original source port as long as the
6757 address family allows it, otherwise the source port is set to 0.
6758
6759http-request set-src-port <expr> [ { if | unless } <condition> ]
6760
6761 This is used to set the source port address to the value of specified
6762 expression.
6763
6764 Arguments:
6765 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6766 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006767
Willy Tarreau20b0de52012-12-24 15:45:22 +01006768 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006769 http-request set-src-port hdr(x-port)
6770 http-request set-src-port int(4000)
6771
6772 When possible, set-src-port preserves the original source address as long as
6773 the address family supports a port, otherwise it forces the source address to
6774 IPv4 "0.0.0.0" before rewriting the port.
6775
Alex59c53352021-04-27 12:57:07 +02006776http-request set-timeout { server | tunnel } { <timeout> | <expr> }
Amaury Denoyelle8d228232020-12-10 13:43:54 +01006777 [ { if | unless } <condition> ]
6778
6779 This action overrides the specified "server" or "tunnel" timeout for the
6780 current stream only. The timeout can be specified in millisecond or with any
6781 other unit if the number is suffixed by the unit as explained at the top of
6782 this document. It is also possible to write an expression which must returns
6783 a number interpreted as a timeout in millisecond.
6784
6785 Note that the server/tunnel timeouts are only relevant on the backend side
6786 and thus this rule is only available for the proxies with backend
6787 capabilities. Also the timeout value must be non-null to obtain the expected
6788 results.
6789
6790 Example:
Alex59c53352021-04-27 12:57:07 +02006791 http-request set-timeout tunnel 5s
6792 http-request set-timeout server req.hdr(host),map_int(host.lst)
Amaury Denoyelle8d228232020-12-10 13:43:54 +01006793
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006794http-request set-tos <tos> [ { if | unless } <condition> ]
6795
6796 This is used to set the TOS or DSCP field value of packets sent to the client
6797 to the value passed in <tos> on platforms which support this. This value
6798 represents the whole 8 bits of the IP TOS field, and can be expressed both in
6799 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
6800 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
6801 be used to adjust some routing behavior on border routers based on some
6802 information from the request.
6803
6804 See RFC 2474, 2597, 3260 and 4594 for more information.
6805
6806http-request set-uri <fmt> [ { if | unless } <condition> ]
6807
6808 This rewrites the request URI with the result of the evaluation of format
6809 string <fmt>. The scheme, authority, path and query string are all replaced
6810 at once. This can be used to rewrite hosts in front of proxies, or to
6811 perform complex modifications to the URI such as moving parts between the
6812 path and the query string.
6813 See also "http-request set-path" and "http-request set-query".
6814
6815http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
6816
6817 This is used to set the contents of a variable. The variable is declared
6818 inline.
6819
6820 Arguments:
6821 <var-name> The name of the variable starts with an indication about its
6822 scope. The scopes allowed are:
6823 "proc" : the variable is shared with the whole process
6824 "sess" : the variable is shared with the whole session
6825 "txn" : the variable is shared with the transaction
6826 (request and response)
6827 "req" : the variable is shared only during request
6828 processing
6829 "res" : the variable is shared only during response
6830 processing
6831 This prefix is followed by a name. The separator is a '.'.
6832 The name may only contain characters 'a-z', 'A-Z', '0-9'
6833 and '_'.
6834
6835 <expr> Is a standard HAProxy expression formed by a sample-fetch
6836 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01006837
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006838 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006839 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006840
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006841http-request send-spoe-group <engine-name> <group-name>
6842 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006843
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006844 This action is used to trigger sending of a group of SPOE messages. To do so,
6845 the SPOE engine used to send messages must be defined, as well as the SPOE
6846 group to send. Of course, the SPOE engine must refer to an existing SPOE
6847 filter. If not engine name is provided on the SPOE filter line, the SPOE
6848 agent name must be used.
6849
6850 Arguments:
6851 <engine-name> The SPOE engine name.
6852
6853 <group-name> The SPOE group name as specified in the engine
6854 configuration.
6855
6856http-request silent-drop [ { if | unless } <condition> ]
6857
6858 This stops the evaluation of the rules and makes the client-facing connection
6859 suddenly disappear using a system-dependent way that tries to prevent the
6860 client from being notified. The effect it then that the client still sees an
6861 established connection while there's none on HAProxy. The purpose is to
6862 achieve a comparable effect to "tarpit" except that it doesn't use any local
6863 resource at all on the machine running HAProxy. It can resist much higher
6864 loads than "tarpit", and slow down stronger attackers. It is important to
6865 understand the impact of using this mechanism. All stateful equipment placed
6866 between the client and HAProxy (firewalls, proxies, load balancers) will also
6867 keep the established connection for a long time and may suffer from this
6868 action.
6869 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6870 option is used to block the emission of a TCP reset. On other systems, the
6871 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6872 router, though it's still delivered to local networks. Do not use it unless
6873 you fully understand how it works.
6874
Christopher Faulet46f95542019-12-20 10:07:22 +01006875http-request strict-mode { on | off }
6876
6877 This enables or disables the strict rewriting mode for following rules. It
6878 does not affect rules declared before it and it is only applicable on rules
6879 performing a rewrite on the requests. When the strict mode is enabled, any
6880 rewrite failure triggers an internal error. Otherwise, such errors are
6881 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006882 rewrites optional while others must be performed to continue the request
Christopher Faulet46f95542019-12-20 10:07:22 +01006883 processing.
6884
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006885 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006886 when a ruleset evaluation ends. So, for instance, if you change the mode on
6887 the frontend, the default mode is restored when HAProxy starts the backend
6888 rules evaluation.
6889
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006890http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
6891http-request tarpit [ { status | deny_status } <code>] [content-type <type>]
6892 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6893 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6894 [ hdr <name> <fmt> ]*
6895 [ { if | unless } <condition> ]
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006896
6897 This stops the evaluation of the rules and immediately blocks the request
6898 without responding for a delay specified by "timeout tarpit" or
6899 "timeout connect" if the former is not set. After that delay, if the client
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006900 is still connected, a response is returned so that the client does not
6901 suspect it has been tarpitted. Logs will report the flags "PT". The goal of
6902 the tarpit rule is to slow down robots during an attack when they're limited
6903 on the number of concurrent requests. It can be very efficient against very
6904 dumb robots, and will significantly reduce the load on firewalls compared to
6905 a "deny" rule. But when facing "correctly" developed robots, it can make
Daniel Corbett9f0843f2021-05-08 10:50:37 -04006906 things worse by forcing HAProxy and the front firewall to support insane
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006907 number of concurrent connections. By default an HTTP error 500 is returned.
6908 But the response may be customized using same syntax than
6909 "http-request return" rules. Thus, see "http-request return" for details.
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006910 For compatibility purpose, when no argument is defined, or only "deny_status",
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006911 the argument "default-errorfiles" is implied. It means
6912 "http-request tarpit [deny_status <status>]" is an alias of
6913 "http-request tarpit [status <status>] default-errorfiles".
6914 No further "http-request" rules are evaluated.
6915 See also "http-request return" and "http-request silent-drop".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006916
6917http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6918http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6919http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
6920
6921 This enables tracking of sticky counters from current request. These rules do
6922 not stop evaluation and do not change default action. The number of counters
6923 that may be simultaneously tracked by the same connection is set in
6924 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
Matteo Contrini1857b8c2020-10-16 17:35:54 +02006925 so the track-sc number is between 0 and (MAX_SESS_STKCTR-1). The first
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006926 "track-sc0" rule executed enables tracking of the counters of the specified
6927 table as the first set. The first "track-sc1" rule executed enables tracking
6928 of the counters of the specified table as the second set. The first
6929 "track-sc2" rule executed enables tracking of the counters of the specified
6930 table as the third set. It is a recommended practice to use the first set of
6931 counters for the per-frontend counters and the second set for the per-backend
6932 ones. But this is just a guideline, all may be used everywhere.
6933
6934 Arguments :
6935 <key> is mandatory, and is a sample expression rule as described in
6936 section 7.3. It describes what elements of the incoming request or
6937 connection will be analyzed, extracted, combined, and used to
6938 select which table entry to update the counters.
6939
6940 <table> is an optional table to be used instead of the default one, which
6941 is the stick-table declared in the current proxy. All the counters
6942 for the matches and updates for the key will then be performed in
6943 that table until the session ends.
6944
6945 Once a "track-sc*" rule is executed, the key is looked up in the table and if
6946 it is not found, an entry is allocated for it. Then a pointer to that entry
6947 is kept during all the session's life, and this entry's counters are updated
6948 as often as possible, every time the session's counters are updated, and also
6949 systematically when the session ends. Counters are only updated for events
6950 that happen after the tracking has been started. As an exception, connection
6951 counters and request counters are systematically updated so that they reflect
6952 useful information.
6953
6954 If the entry tracks concurrent connection counters, one connection is counted
6955 for as long as the entry is tracked, and the entry will not expire during
6956 that time. Tracking counters also provides a performance advantage over just
6957 checking the keys, because only one table lookup is performed for all ACL
6958 checks that make use of it.
6959
6960http-request unset-var(<var-name>) [ { if | unless } <condition> ]
6961
6962 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006963
6964 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006965 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006966
Christopher Faulet579d83b2019-11-22 15:34:17 +01006967http-request use-service <service-name> [ { if | unless } <condition> ]
6968
6969 This directive executes the configured HTTP service to reply to the request
6970 and stops the evaluation of the rules. An HTTP service may choose to reply by
6971 sending any valid HTTP response or it may immediately close the connection
6972 without sending any response. Outside natives services, for instance the
6973 Prometheus exporter, it is possible to write your own services in Lua. No
6974 further "http-request" rules are evaluated.
6975
6976 Arguments :
6977 <service-name> is mandatory. It is the service to call
6978
6979 Example:
6980 http-request use-service prometheus-exporter if { path /metrics }
6981
Christopher Faulet021a8e42021-03-29 10:46:38 +02006982http-request wait-for-body time <time> [ at-least <bytes> ]
6983 [ { if | unless } <condition> ]
6984
6985 This will delay the processing of the request waiting for the payload for at
6986 most <time> milliseconds. if "at-least" argument is specified, HAProxy stops
6987 to wait the payload when the first <bytes> bytes are received. 0 means no
6988 limit, it is the default value. Regardless the "at-least" argument value,
6989 HAProxy stops to wait if the whole payload is received or if the request
6990 buffer is full. This action may be used as a replacement to "option
6991 http-buffer-request".
6992
6993 Arguments :
6994
6995 <time> is mandatory. It is the maximum time to wait for the body. It
6996 follows the HAProxy time format and is expressed in milliseconds.
6997
6998 <bytes> is optional. It is the minimum payload size to receive to stop to
Ilya Shipitsinb2be9a12021-04-24 13:25:42 +05006999 wait. It follows the HAProxy size format and is expressed in
Christopher Faulet021a8e42021-03-29 10:46:38 +02007000 bytes.
7001
7002 Example:
7003 http-request wait-for-body time 1s at-least 1k if METH_POST
7004
7005 See also : "option http-buffer-request"
7006
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02007007http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007008
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02007009 This will delay the processing of the request until the SSL handshake
7010 happened. This is mostly useful to delay processing early data until we're
7011 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007012
Willy Tarreauef781042010-01-27 11:53:01 +01007013
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007014http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02007015 Access control for Layer 7 responses
7016
7017 May be used in sections: defaults | frontend | listen | backend
7018 no | yes | yes | yes
7019
7020 The http-response statement defines a set of rules which apply to layer 7
7021 processing. The rules are evaluated in their declaration order when they are
7022 met in a frontend, listen or backend section. Any rule may optionally be
7023 followed by an ACL-based condition, in which case it will only be evaluated
7024 if the condition is true. Since these rules apply on responses, the backend
7025 rules are applied first, followed by the frontend's rules.
7026
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007027 The first keyword is the rule's action. The supported actions are described
7028 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02007029
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007030 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02007031
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007032 Example:
7033 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02007034
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007035 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06007036
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007037 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
7038 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06007039
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007040 Example:
7041 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06007042
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007043 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06007044
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007045 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
7046 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06007047
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007048 See also : "http-request", section 3.4 about userlists and section 7 about
7049 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06007050
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007051http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06007052
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007053 This is used to add a new entry into an ACL. The ACL must be loaded from a
7054 file (even a dummy empty file). The file name of the ACL to be updated is
7055 passed between parentheses. It takes one argument: <key fmt>, which follows
7056 log-format rules, to collect content of the new entry. It performs a lookup
7057 in the ACL before insertion, to avoid duplicated (or more) values.
7058 This lookup is done by a linear search and can be expensive with large lists!
7059 It is the equivalent of the "add acl" command from the stats socket, but can
7060 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06007061
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007062http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06007063
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007064 This appends an HTTP header field whose name is specified in <name> and whose
7065 value is defined by <fmt> which follows the log-format rules (see Custom Log
7066 Format in section 8.2.4). This may be used to send a cookie to a client for
7067 example, or to pass some internal information.
7068 This rule is not final, so it is possible to add other similar rules.
7069 Note that header addition is performed immediately, so one rule might reuse
7070 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06007071
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007072http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06007073
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007074 This stops the evaluation of the rules and lets the response pass the check.
7075 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06007076
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02007077http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06007078
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007079 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06007080
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007081http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06007082
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007083 This captures sample expression <sample> from the response buffer, and
7084 converts it to a string. The resulting string is stored into the next request
7085 "capture" slot, so it will possibly appear next to some captured HTTP
7086 headers. It will then automatically appear in the logs, and it will be
7087 possible to extract it using sample fetch rules to feed it into headers or
7088 anything. Please check section 7.3 (Fetching samples) and
7089 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02007090
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007091 The keyword "id" is the id of the capture slot which is used for storing the
7092 string. The capture slot must be defined in an associated frontend.
7093 This is useful to run captures in backends. The slot id can be declared by a
7094 previous directive "http-response capture" or with the "declare capture"
7095 keyword.
Baptiste Assmann19a69b32020-01-16 14:34:22 +01007096
7097 When using this action in a backend, double check that the relevant
7098 frontend(s) have the required capture slots otherwise, this rule will be
7099 ignored at run time. This can't be detected at configuration parsing time
7100 due to HAProxy's ability to dynamically resolve backend name at runtime.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02007101
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007102http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02007103
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007104 This is used to delete an entry from an ACL. The ACL must be loaded from a
7105 file (even a dummy empty file). The file name of the ACL to be updated is
7106 passed between parentheses. It takes one argument: <key fmt>, which follows
7107 log-format rules, to collect content of the entry to delete.
7108 It is the equivalent of the "del acl" command from the stats socket, but can
7109 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02007110
Maciej Zdebebdd4c52020-11-20 13:58:48 +00007111http-response del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02007112
Maciej Zdebebdd4c52020-11-20 13:58:48 +00007113 This removes all HTTP header fields whose name is specified in <name>. <meth>
7114 is the matching method, applied on the header name. Supported matching methods
7115 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
7116 (substring match) and "reg" (regex match). If not specified, exact matching
7117 method is used.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02007118
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007119http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02007120
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007121 This is used to delete an entry from a MAP. The MAP must be loaded from a
7122 file (even a dummy empty file). The file name of the MAP to be updated is
7123 passed between parentheses. It takes one argument: <key fmt>, which follows
7124 log-format rules, to collect content of the entry to delete.
7125 It takes one argument: "file name" It is the equivalent of the "del map"
7126 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007127
Christopher Faulet5cb513a2020-05-13 17:56:56 +02007128http-response deny [deny_status <status>] [ { if | unless } <condition> ]
7129http-response deny [ { status | deny_status } <code>] [content-type <type>]
7130 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
7131 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
7132 [ hdr <name> <fmt> ]*
7133 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007134
Christopher Faulet5cb513a2020-05-13 17:56:56 +02007135 This stops the evaluation of the rules and immediately rejects the response.
7136 By default an HTTP 502 error is returned. But the response may be customized
7137 using same syntax than "http-response return" rules. Thus, see
Ilya Shipitsin11057a32020-06-21 21:18:27 +05007138 "http-response return" for details. For compatibility purpose, when no
Christopher Faulet5cb513a2020-05-13 17:56:56 +02007139 argument is defined, or only "deny_status", the argument "default-errorfiles"
7140 is implied. It means "http-response deny [deny_status <status>]" is an alias
7141 of "http-response deny [status <status>] default-errorfiles".
Christopher Faulet040c8cd2020-01-13 16:43:45 +01007142 No further "http-response" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02007143 See also "http-response return".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007144
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007145http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007146
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007147 This performs an HTTP redirection based on a redirect rule.
7148 This supports a format string similarly to "http-request redirect" rules,
7149 with the exception that only the "location" type of redirect is possible on
7150 the response. See the "redirect" keyword for the rule's syntax. When a
7151 redirect rule is applied during a response, connections to the server are
7152 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02007153
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007154http-response replace-header <name> <regex-match> <replace-fmt>
7155 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02007156
Tim Duesterhus2252beb2019-10-29 00:05:13 +01007157 This works like "http-request replace-header" except that it works on the
7158 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01007159
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007160 Example:
7161 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02007162
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007163 # applied to:
7164 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02007165
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007166 # outputs:
7167 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02007168
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007169 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02007170
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007171http-response replace-value <name> <regex-match> <replace-fmt>
7172 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02007173
Tim Duesterhus6bd909b2020-01-17 15:53:18 +01007174 This works like "http-request replace-value" except that it works on the
Tim Duesterhus2252beb2019-10-29 00:05:13 +01007175 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02007176
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007177 Example:
7178 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01007179
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007180 # applied to:
7181 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01007182
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007183 # outputs:
7184 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01007185
Christopher Faulet24231ab2020-01-24 17:44:23 +01007186http-response return [status <code>] [content-type <type>]
7187 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
7188 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01007189 [ hdr <name> <value> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01007190 [ { if | unless } <condition> ]
7191
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05007192 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01007193 default status code used for the response is 200. It can be optionally
7194 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04007195 specified as an argument to "content-type". Finally the response itself may
Christopher Faulet24231ab2020-01-24 17:44:23 +01007196 be defined. If can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05007197 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01007198 are followed to create the response :
7199
7200 * If neither the errorfile nor the payload to use is defined, a dummy
7201 response is returned. Only the "status" argument is considered. It can be
7202 any code in the range [200, 599]. The "content-type" argument, if any, is
7203 ignored.
7204
7205 * If "default-errorfiles" argument is set, the proxy's errorfiles are
7206 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007207 status code handled by HAProxy (200, 400, 403, 404, 405, 408, 410, 413,
Christopher Faulete095f312020-12-07 11:22:24 +01007208 425, 429, 500, 501, 502, 503, and 504). The "content-type" argument, if
7209 any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01007210
7211 * If a specific errorfile is defined, with an "errorfile" argument, the
7212 corresponding file, containing a full HTTP response, is returned. Only the
7213 "status" argument is considered. It must be one of the status code handled
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007214 by HAProxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 501,
Christopher Faulete095f312020-12-07 11:22:24 +01007215 502, 503, and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01007216
7217 * If an http-errors section is defined, with an "errorfiles" argument, the
7218 corresponding file in the specified http-errors section, containing a full
7219 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007220 must be one of the status code handled by HAProxy (200, 400, 403, 404, 405,
Christopher Faulete095f312020-12-07 11:22:24 +01007221 408, 410, 413, 425, 429, 500, 501, 502, 503, and 504). The "content-type"
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02007222 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01007223
7224 * If a "file" or a "lf-file" argument is specified, the file's content is
7225 used as the response payload. If the file is not empty, its content-type
7226 must be set as argument to "content-type". Otherwise, any "content-type"
7227 argument is ignored. With a "lf-file" argument, the file's content is
7228 evaluated as a log-format string. With a "file" argument, it is considered
7229 as a raw content.
7230
7231 * If a "string" or "lf-string" argument is specified, the defined string is
7232 used as the response payload. The content-type must always be set as
7233 argument to "content-type". With a "lf-string" argument, the string is
7234 evaluated as a log-format string. With a "string" argument, it is
7235 considered as a raw string.
7236
Christopher Faulet4a2c1422020-01-31 17:36:01 +01007237 When the response is not based an errorfile, it is possible to appends HTTP
7238 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
7239 arguments are ignored. For each one, the header name is specified in <name>
7240 and its value is defined by <fmt> which follows the log-format rules.
7241
Christopher Faulet24231ab2020-01-24 17:44:23 +01007242 Note that the generated response must be smaller than a buffer. And to avoid
7243 any warning, when an errorfile or a raw file is loaded, the buffer space
Ilya Shipitsin11057a32020-06-21 21:18:27 +05007244 reserved to the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01007245
7246 No further "http-response" rules are evaluated.
7247
7248 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04007249 http-response return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01007250 if { status eq 404 }
7251
7252 http-response return content-type text/plain \
7253 string "This is the end !" \
7254 if { status eq 500 }
7255
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007256http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
7257http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08007258
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007259 This action increments the GPC0 or GPC1 counter according with the sticky
7260 counter designated by <sc-id>. If an error occurs, this action silently fails
7261 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02007262
Cédric Dufour0d7712d2019-11-06 18:38:53 +01007263http-response sc-set-gpt0(<sc-id>) { <int> | <expr> }
7264 [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02007265
Cédric Dufour0d7712d2019-11-06 18:38:53 +01007266 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
7267 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
7268 boolean. If an error occurs, this action silently fails and the actions
7269 evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01007270
Christopher Faulet68fc3a12021-08-12 09:32:07 +02007271http-response send-spoe-group <engine-name> <group-name>
7272 [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02007273
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007274 This action is used to trigger sending of a group of SPOE messages. To do so,
7275 the SPOE engine used to send messages must be defined, as well as the SPOE
7276 group to send. Of course, the SPOE engine must refer to an existing SPOE
7277 filter. If not engine name is provided on the SPOE filter line, the SPOE
7278 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02007279
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007280 Arguments:
7281 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02007282
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007283 <group-name> The SPOE group name as specified in the engine
7284 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02007285
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007286http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02007287
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007288 This does the same as "add-header" except that the header name is first
7289 removed if it existed. This is useful when passing security information to
7290 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02007291
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007292http-response set-log-level <level> [ { if | unless } <condition> ]
7293
7294 This is used to change the log level of the current request when a certain
7295 condition is met. Valid levels are the 8 syslog levels (see the "log"
7296 keyword) plus the special level "silent" which disables logging for this
7297 request. This rule is not final so the last matching rule wins. This rule can
7298 be useful to disable health checks coming from another equipment.
7299
7300http-response set-map(<file-name>) <key fmt> <value fmt>
7301
7302 This is used to add a new entry into a MAP. The MAP must be loaded from a
7303 file (even a dummy empty file). The file name of the MAP to be updated is
7304 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
7305 log-format rules, used to collect MAP key, and <value fmt>, which follows
7306 log-format rules, used to collect content for the new entry. It performs a
7307 lookup in the MAP before insertion, to avoid duplicated (or more) values.
7308 This lookup is done by a linear search and can be expensive with large lists!
7309 It is the equivalent of the "set map" command from the stats socket, but can
7310 be triggered by an HTTP response.
7311
7312http-response set-mark <mark> [ { if | unless } <condition> ]
7313
7314 This is used to set the Netfilter MARK on all packets sent to the client to
7315 the value passed in <mark> on platforms which support it. This value is an
7316 unsigned 32 bit value which can be matched by netfilter and by the routing
7317 table. It can be expressed both in decimal or hexadecimal format (prefixed
7318 by "0x"). This can be useful to force certain packets to take a different
7319 route (for example a cheaper network path for bulk downloads). This works on
7320 Linux kernels 2.6.32 and above and requires admin privileges.
7321
7322http-response set-nice <nice> [ { if | unless } <condition> ]
7323
7324 This sets the "nice" factor of the current request being processed.
7325 It only has effect against the other requests being processed at the same
7326 time. The default value is 0, unless altered by the "nice" setting on the
7327 "bind" line. The accepted range is -1024..1024. The higher the value, the
7328 nicest the request will be. Lower values will make the request more important
7329 than other ones. This can be useful to improve the speed of some requests, or
7330 lower the priority of non-important requests. Using this setting without
7331 prior experimentation can cause some major slowdown.
7332
7333http-response set-status <status> [reason <str>]
7334 [ { if | unless } <condition> ]
7335
7336 This replaces the response status code with <status> which must be an integer
7337 between 100 and 999. Optionally, a custom reason text can be provided defined
7338 by <str>, or the default reason for the specified code will be used as a
7339 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007340
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007341 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007342 # return "431 Request Header Fields Too Large"
7343 http-response set-status 431
7344 # return "503 Slow Down", custom reason
7345 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007346
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007347http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007348
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007349 This is used to set the TOS or DSCP field value of packets sent to the client
7350 to the value passed in <tos> on platforms which support this.
7351 This value represents the whole 8 bits of the IP TOS field, and can be
7352 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
7353 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
7354 always 0. This can be used to adjust some routing behavior on border routers
7355 based on some information from the request.
7356
7357 See RFC 2474, 2597, 3260 and 4594 for more information.
7358
7359http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
7360
7361 This is used to set the contents of a variable. The variable is declared
7362 inline.
7363
7364 Arguments:
7365 <var-name> The name of the variable starts with an indication about its
7366 scope. The scopes allowed are:
7367 "proc" : the variable is shared with the whole process
7368 "sess" : the variable is shared with the whole session
7369 "txn" : the variable is shared with the transaction
7370 (request and response)
7371 "req" : the variable is shared only during request
7372 processing
7373 "res" : the variable is shared only during response
7374 processing
7375 This prefix is followed by a name. The separator is a '.'.
7376 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
7377 and '_'.
7378
7379 <expr> Is a standard HAProxy expression formed by a sample-fetch
7380 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007381
7382 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007383 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007384
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007385http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007386
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007387 This stops the evaluation of the rules and makes the client-facing connection
7388 suddenly disappear using a system-dependent way that tries to prevent the
7389 client from being notified. The effect it then that the client still sees an
7390 established connection while there's none on HAProxy. The purpose is to
7391 achieve a comparable effect to "tarpit" except that it doesn't use any local
7392 resource at all on the machine running HAProxy. It can resist much higher
7393 loads than "tarpit", and slow down stronger attackers. It is important to
7394 understand the impact of using this mechanism. All stateful equipment placed
7395 between the client and HAProxy (firewalls, proxies, load balancers) will also
7396 keep the established connection for a long time and may suffer from this
7397 action.
7398 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
7399 option is used to block the emission of a TCP reset. On other systems, the
7400 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
7401 router, though it's still delivered to local networks. Do not use it unless
7402 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007403
Christopher Faulet46f95542019-12-20 10:07:22 +01007404http-response strict-mode { on | off }
7405
7406 This enables or disables the strict rewriting mode for following rules. It
7407 does not affect rules declared before it and it is only applicable on rules
7408 performing a rewrite on the responses. When the strict mode is enabled, any
7409 rewrite failure triggers an internal error. Otherwise, such errors are
7410 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05007411 rewrites optional while others must be performed to continue the response
Christopher Faulet46f95542019-12-20 10:07:22 +01007412 processing.
7413
Christopher Faulet1aea50e2020-01-17 16:03:53 +01007414 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01007415 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04007416 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet46f95542019-12-20 10:07:22 +01007417 rules evaluation.
7418
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007419http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
7420http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
7421http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02007422
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007423 This enables tracking of sticky counters from current response. Please refer
7424 to "http-request track-sc" for a complete description. The only difference
7425 from "http-request track-sc" is the <key> sample expression can only make use
7426 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
7427 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007428 supported, HAProxy will fail and warn while parsing the config.
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007429
7430http-response unset-var(<var-name>) [ { if | unless } <condition> ]
7431
7432 This is used to unset a variable. See "http-response set-var" for details
7433 about <var-name>.
7434
7435 Example:
7436 http-response unset-var(sess.last_redir)
7437
Christopher Faulet021a8e42021-03-29 10:46:38 +02007438http-response wait-for-body time <time> [ at-least <bytes> ]
7439 [ { if | unless } <condition> ]
7440
7441 This will delay the processing of the response waiting for the payload for at
7442 most <time> milliseconds. if "at-least" argument is specified, HAProxy stops
7443 to wait the payload when the first <bytes> bytes are received. 0 means no
7444 limit, it is the default value. Regardless the "at-least" argument value,
7445 HAProxy stops to wait if the whole payload is received or if the response
7446 buffer is full.
7447
7448 Arguments :
7449
7450 <time> is mandatory. It is the maximum time to wait for the body. It
7451 follows the HAProxy time format and is expressed in milliseconds.
7452
7453 <bytes> is optional. It is the minimum payload size to receive to stop to
Ilya Shipitsinb2be9a12021-04-24 13:25:42 +05007454 wait. It follows the HAProxy size format and is expressed in
Christopher Faulet021a8e42021-03-29 10:46:38 +02007455 bytes.
7456
7457 Example:
7458 http-response wait-for-body time 1s at-least 10k
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02007459
Willy Tarreau30631952015-08-06 15:05:24 +02007460http-reuse { never | safe | aggressive | always }
7461 Declare how idle HTTP connections may be shared between requests
7462
7463 May be used in sections: defaults | frontend | listen | backend
7464 yes | no | yes | yes
7465
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007466 By default, a connection established between HAProxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01007467 which is considered safe for reuse is moved back to the server's idle
7468 connections pool so that any other request can make use of it. This is the
7469 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02007470
7471 The argument indicates the desired connection reuse strategy :
7472
Olivier Houchard86006a52018-12-14 19:37:49 +01007473 - "never" : idle connections are never shared between sessions. This mode
7474 may be enforced to cancel a different strategy inherited from
7475 a defaults section or for troubleshooting. For example, if an
7476 old bogus application considers that multiple requests over
7477 the same connection come from the same client and it is not
7478 possible to fix the application, it may be desirable to
7479 disable connection sharing in a single backend. An example of
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007480 such an application could be an old HAProxy using cookie
Olivier Houchard86006a52018-12-14 19:37:49 +01007481 insertion in tunnel mode and not checking any request past the
7482 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02007483
Olivier Houchard86006a52018-12-14 19:37:49 +01007484 - "safe" : this is the default and the recommended strategy. The first
7485 request of a session is always sent over its own connection,
7486 and only subsequent requests may be dispatched over other
7487 existing connections. This ensures that in case the server
7488 closes the connection when the request is being sent, the
7489 browser can decide to silently retry it. Since it is exactly
7490 equivalent to regular keep-alive, there should be no side
Amaury Denoyelle27179652020-10-14 18:17:12 +02007491 effects. There is also a special handling for the connections
7492 using protocols subject to Head-of-line blocking (backend with
7493 h2 or fcgi). In this case, when at least one stream is
7494 processed, the used connection is reserved to handle streams
7495 of the same session. When no more streams are processed, the
7496 connection is released and can be reused.
Willy Tarreau30631952015-08-06 15:05:24 +02007497
7498 - "aggressive" : this mode may be useful in webservices environments where
7499 all servers are not necessarily known and where it would be
7500 appreciable to deliver most first requests over existing
7501 connections. In this case, first requests are only delivered
7502 over existing connections that have been reused at least once,
7503 proving that the server correctly supports connection reuse.
7504 It should only be used when it's sure that the client can
7505 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02007506 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02007507 downsides of rare connection failures.
7508
7509 - "always" : this mode is only recommended when the path to the server is
7510 known for never breaking existing connections quickly after
7511 releasing them. It allows the first request of a session to be
7512 sent to an existing connection. This can provide a significant
7513 performance increase over the "safe" strategy when the backend
7514 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01007515 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02007516 sharing. It is recommended that the "http-keep-alive" timeout
7517 remains low in this mode so that no dead connections remain
7518 usable. In most cases, this will lead to the same performance
7519 gains as "aggressive" but with more risks. It should only be
7520 used when it improves the situation over "aggressive".
7521
7522 When http connection sharing is enabled, a great care is taken to respect the
Amaury Denoyelled773a4e2021-01-29 15:18:49 +01007523 connection properties and compatibility. Indeed, some properties are specific
7524 and it is not possibly to reuse it blindly. Those are the SSL SNI, source
7525 and destination address and proxy protocol block. A connection is reused only
7526 if it shares the same set of properties with the request.
Willy Tarreau30631952015-08-06 15:05:24 +02007527
Amaury Denoyelled773a4e2021-01-29 15:18:49 +01007528 Also note that connections with certain bogus authentication schemes (relying
7529 on the connection) like NTLM are marked private and never shared.
Willy Tarreau30631952015-08-06 15:05:24 +02007530
Lukas Tribuse8adfeb2019-11-06 11:50:25 +01007531 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02007532
7533 Note: connection reuse improves the accuracy of the "server maxconn" setting,
7534 because almost no new connection will be established while idle connections
7535 remain available. This is particularly true with the "always" strategy.
7536
7537 See also : "option http-keep-alive", "server maxconn"
7538
7539
Mark Lamourinec2247f02012-01-04 13:02:01 -05007540http-send-name-header [<header>]
7541 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05007542 May be used in sections: defaults | frontend | listen | backend
7543 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05007544 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05007545 <header> The header string to use to send the server name
7546
Willy Tarreau81bef7e2019-10-07 14:58:02 +02007547 The "http-send-name-header" statement causes the header field named <header>
7548 to be set to the name of the target server at the moment the request is about
7549 to be sent on the wire. Any existing occurrences of this header are removed.
7550 Upon retries and redispatches, the header field is updated to always reflect
7551 the server being attempted to connect to. Given that this header is modified
7552 very late in the connection setup, it may have unexpected effects on already
7553 modified headers. For example using it with transport-level header such as
7554 connection, content-length, transfer-encoding and so on will likely result in
7555 invalid requests being sent to the server. Additionally it has been reported
7556 that this directive is currently being used as a way to overwrite the Host
7557 header field in outgoing requests; while this trick has been known to work
7558 as a side effect of the feature for some time, it is not officially supported
7559 and might possibly not work anymore in a future version depending on the
7560 technical difficulties this feature induces. A long-term solution instead
7561 consists in fixing the application which required this trick so that it binds
7562 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05007563
7564 See also : "server"
7565
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01007566id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02007567 Set a persistent ID to a proxy.
7568 May be used in sections : defaults | frontend | listen | backend
7569 no | yes | yes | yes
7570 Arguments : none
7571
7572 Set a persistent ID for the proxy. This ID must be unique and positive.
7573 An unused ID will automatically be assigned if unset. The first assigned
7574 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01007575
7576
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007577ignore-persist { if | unless } <condition>
7578 Declare a condition to ignore persistence
7579 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01007580 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007581
7582 By default, when cookie persistence is enabled, every requests containing
7583 the cookie are unconditionally persistent (assuming the target server is up
7584 and running).
7585
7586 The "ignore-persist" statement allows one to declare various ACL-based
7587 conditions which, when met, will cause a request to ignore persistence.
7588 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007589 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007590 persistence for a specific User-Agent (for example, some web crawler bots).
7591
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007592 The persistence is ignored when an "if" condition is met, or unless an
7593 "unless" condition is met.
7594
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03007595 Example:
7596 acl url_static path_beg /static /images /img /css
7597 acl url_static path_end .gif .png .jpg .css .js
7598 ignore-persist if url_static
7599
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007600 See also : "force-persist", "cookie", and section 7 about ACL usage.
7601
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007602load-server-state-from-file { global | local | none }
7603 Allow seamless reload of HAProxy
7604 May be used in sections: defaults | frontend | listen | backend
7605 yes | no | yes | yes
7606
7607 This directive points HAProxy to a file where server state from previous
7608 running process has been saved. That way, when starting up, before handling
7609 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01007610 reload occurred. The purpose of the "load-server-state-from-file" directive is
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007611 to tell HAProxy which file to use. For now, only 2 arguments to either prevent
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007612 loading state or load states from a file containing all backends and servers.
7613 The state file can be generated by running the command "show servers state"
7614 over the stats socket and redirect output.
7615
Davor Ocelice9ed2812017-12-25 17:49:28 +01007616 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007617 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02007618 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007619
7620 Arguments:
7621 global load the content of the file pointed by the global directive
7622 named "server-state-file".
7623
7624 local load the content of the file pointed by the directive
7625 "server-state-file-name" if set. If not set, then the backend
7626 name is used as a file name.
7627
7628 none don't load any stat for this backend
7629
7630 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01007631 - server's IP address is preserved across reloads by default, but the
7632 order can be changed thanks to the server's "init-addr" setting. This
7633 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007634 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01007635 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007636
7637 - server's weight is applied from previous running process unless it has
7638 has changed between previous and new configuration files.
7639
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007640 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007641
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007642 global
7643 stats socket /tmp/socket
7644 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007645
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007646 defaults
7647 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007648
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007649 backend bk
7650 server s1 127.0.0.1:22 check weight 11
7651 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007652
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007653
7654 Then one can run :
7655
7656 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
7657
7658 Content of the file /tmp/server_state would be like this:
7659
7660 1
7661 # <field names skipped for the doc example>
7662 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
7663 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
7664
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007665 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007666
7667 global
7668 stats socket /tmp/socket
7669 server-state-base /etc/haproxy/states
7670
7671 defaults
7672 load-server-state-from-file local
7673
7674 backend bk
7675 server s1 127.0.0.1:22 check weight 11
7676 server s2 127.0.0.1:22 check weight 12
7677
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007678
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007679 Then one can run :
7680
7681 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
7682
7683 Content of the file /etc/haproxy/states/bk would be like this:
7684
7685 1
7686 # <field names skipped for the doc example>
7687 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
7688 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
7689
7690 See also: "server-state-file", "server-state-file-name", and
7691 "show servers state"
7692
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007693
Willy Tarreau2769aa02007-12-27 18:26:09 +01007694log global
Jan Wagner3e678602020-12-17 22:22:32 +01007695log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02007696 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02007697no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01007698 Enable per-instance logging of events and traffic.
7699 May be used in sections : defaults | frontend | listen | backend
7700 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02007701
7702 Prefix :
7703 no should be used when the logger list must be flushed. For example,
7704 if you don't want to inherit from the default logger list. This
7705 prefix does not allow arguments.
7706
Willy Tarreau2769aa02007-12-27 18:26:09 +01007707 Arguments :
7708 global should be used when the instance's logging parameters are the
7709 same as the global ones. This is the most common usage. "global"
7710 replaces <address>, <facility> and <level> with those of the log
7711 entries found in the "global" section. Only one "log global"
7712 statement may be used per instance, and this form takes no other
7713 parameter.
7714
7715 <address> indicates where to send the logs. It takes the same format as
7716 for the "global" section's logs, and can be one of :
7717
7718 - An IPv4 address optionally followed by a colon (':') and a UDP
7719 port. If no port is specified, 514 is used by default (the
7720 standard syslog port).
7721
David du Colombier24bb5f52011-03-17 10:40:23 +01007722 - An IPv6 address followed by a colon (':') and optionally a UDP
7723 port. If no port is specified, 514 is used by default (the
7724 standard syslog port).
7725
Willy Tarreau2769aa02007-12-27 18:26:09 +01007726 - A filesystem path to a UNIX domain socket, keeping in mind
7727 considerations for chroot (be sure the path is accessible
7728 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007729 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007730
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007731 - A file descriptor number in the form "fd@<number>", which may
7732 point to a pipe, terminal, or socket. In this case unbuffered
7733 logs are used and one writev() call per log is performed. This
7734 is a bit expensive but acceptable for most workloads. Messages
7735 sent this way will not be truncated but may be dropped, in
7736 which case the DroppedLogs counter will be incremented. The
7737 writev() call is atomic even on pipes for messages up to
7738 PIPE_BUF size, which POSIX recommends to be at least 512 and
7739 which is 4096 bytes on most modern operating systems. Any
7740 larger message may be interleaved with messages from other
7741 processes. Exceptionally for debugging purposes the file
7742 descriptor may also be directed to a file, but doing so will
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007743 significantly slow HAProxy down as non-blocking calls will be
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007744 ignored. Also there will be no way to purge nor rotate this
7745 file without restarting the process. Note that the configured
7746 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01007747 with a TCP syslog server. See also the "short" and "raw"
7748 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007749
7750 - "stdout" / "stderr", which are respectively aliases for "fd@1"
7751 and "fd@2", see above.
7752
Willy Tarreauc046d162019-08-30 15:24:59 +02007753 - A ring buffer in the form "ring@<name>", which will correspond
7754 to an in-memory ring buffer accessible over the CLI using the
7755 "show events" command, which will also list existing rings and
7756 their sizes. Such buffers are lost on reload or restart but
7757 when used as a complement this can help troubleshooting by
7758 having the logs instantly available.
7759
Emeric Brun94aab062021-04-02 10:41:36 +02007760 - An explicit stream address prefix such as "tcp@","tcp6@",
7761 "tcp4@" or "uxst@" will allocate an implicit ring buffer with
7762 a stream forward server targeting the given address.
7763
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007764 You may want to reference some environment variables in the
7765 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01007766
Willy Tarreau18324f52014-06-27 18:10:07 +02007767 <length> is an optional maximum line length. Log lines larger than this
7768 value will be truncated before being sent. The reason is that
7769 syslog servers act differently on log line length. All servers
7770 support the default value of 1024, but some servers simply drop
7771 larger lines while others do log them. If a server supports long
7772 lines, it may make sense to set this value here in order to avoid
7773 truncating long lines. Similarly, if a server drops long lines,
7774 it is preferable to truncate them before sending them. Accepted
7775 values are 80 to 65535 inclusive. The default value of 1024 is
7776 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01007777 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02007778
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02007779 <ranges> A list of comma-separated ranges to identify the logs to sample.
7780 This is used to balance the load of the logs to send to the log
7781 server. The limits of the ranges cannot be null. They are numbered
7782 from 1. The size or period (in number of logs) of the sample must
7783 be set with <sample_size> parameter.
7784
7785 <sample_size>
7786 The size of the sample in number of logs to consider when balancing
7787 their logging loads. It is used to balance the load of the logs to
7788 send to the syslog server. This size must be greater or equal to the
7789 maximum of the high limits of the ranges.
7790 (see also <ranges> parameter).
7791
Willy Tarreauadb345d2018-11-12 07:56:13 +01007792 <format> is the log format used when generating syslog messages. It may be
7793 one of the following :
7794
Emeric Brun0237c4e2020-11-27 16:24:34 +01007795 local Analog to rfc3164 syslog message format except that hostname
7796 field is stripped. This is the default.
7797 Note: option "log-send-hostname" switches the default to
7798 rfc3164.
7799
7800 rfc3164 The RFC3164 syslog message format.
Willy Tarreauadb345d2018-11-12 07:56:13 +01007801 (https://tools.ietf.org/html/rfc3164)
7802
7803 rfc5424 The RFC5424 syslog message format.
7804 (https://tools.ietf.org/html/rfc5424)
7805
Emeric Brun54648852020-07-06 15:54:06 +02007806 priority A message containing only a level plus syslog facility between
7807 angle brackets such as '<63>', followed by the text. The PID,
7808 date, time, process name and system name are omitted. This is
7809 designed to be used with a local log server.
7810
Willy Tarreaue8746a02018-11-12 08:45:00 +01007811 short A message containing only a level between angle brackets such as
7812 '<3>', followed by the text. The PID, date, time, process name
7813 and system name are omitted. This is designed to be used with a
7814 local log server. This format is compatible with what the
7815 systemd logger consumes.
7816
Emeric Brun54648852020-07-06 15:54:06 +02007817 timed A message containing only a level between angle brackets such as
7818 '<3>', followed by ISO date and by the text. The PID, process
7819 name and system name are omitted. This is designed to be
7820 used with a local log server.
7821
7822 iso A message containing only the ISO date, followed by the text.
7823 The PID, process name and system name are omitted. This is
7824 designed to be used with a local log server.
7825
Willy Tarreauc1b06452018-11-12 11:57:56 +01007826 raw A message containing only the text. The level, PID, date, time,
7827 process name and system name are omitted. This is designed to
7828 be used in containers or during development, where the severity
7829 only depends on the file descriptor used (stdout/stderr).
7830
Willy Tarreau2769aa02007-12-27 18:26:09 +01007831 <facility> must be one of the 24 standard syslog facilities :
7832
Willy Tarreaue8746a02018-11-12 08:45:00 +01007833 kern user mail daemon auth syslog lpr news
7834 uucp cron auth2 ftp ntp audit alert cron2
7835 local0 local1 local2 local3 local4 local5 local6 local7
7836
Willy Tarreauc1b06452018-11-12 11:57:56 +01007837 Note that the facility is ignored for the "short" and "raw"
7838 formats, but still required as a positional field. It is
7839 recommended to use "daemon" in this case to make it clear that
7840 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007841
7842 <level> is optional and can be specified to filter outgoing messages. By
7843 default, all messages are sent. If a level is specified, only
7844 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007845 will be sent. An optional minimum level can be specified. If it
7846 is set, logs emitted with a more severe level than this one will
7847 be capped to this level. This is used to avoid sending "emerg"
7848 messages on all terminals on some default syslog configurations.
7849 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007850
7851 emerg alert crit err warning notice info debug
7852
William Lallemand0f99e342011-10-12 17:50:54 +02007853 It is important to keep in mind that it is the frontend which decides what to
7854 log from a connection, and that in case of content switching, the log entries
7855 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01007856
7857 However, backend log declaration define how and where servers status changes
7858 will be logged. Level "notice" will be used to indicate a server going up,
7859 "warning" will be used for termination signals and definitive service
7860 termination, and "alert" will be used for when a server goes down.
7861
7862 Note : According to RFC3164, messages are truncated to 1024 bytes before
7863 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007864
7865 Example :
7866 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01007867 log stdout format short daemon # send log to systemd
7868 log stdout format raw daemon # send everything to stdout
7869 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007870 log 127.0.0.1:514 local0 notice # only send important events
Emeric Brun94aab062021-04-02 10:41:36 +02007871 log tcp@127.0.0.1:514 local0 notice notice # same but limit output
7872 # level and send in tcp
William Lallemandb2f07452015-05-12 14:27:13 +02007873 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01007874
Willy Tarreau2769aa02007-12-27 18:26:09 +01007875
William Lallemand48940402012-01-30 16:47:22 +01007876log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007877 Specifies the log format string to use for traffic logs
7878 May be used in sections: defaults | frontend | listen | backend
7879 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01007880
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007881 This directive specifies the log format string that will be used for all logs
7882 resulting from traffic passing through the frontend using this line. If the
7883 directive is used in a defaults section, all subsequent frontends will use
7884 the same log format. Please see section 8.2.4 which covers the log format
7885 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01007886
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007887 "log-format" directive overrides previous "option tcplog", "log-format" and
7888 "option httplog" directives.
7889
Dragan Dosen7ad31542015-09-28 17:16:47 +02007890log-format-sd <string>
7891 Specifies the RFC5424 structured-data log format string
7892 May be used in sections: defaults | frontend | listen | backend
7893 yes | yes | yes | no
7894
7895 This directive specifies the RFC5424 structured-data log format string that
7896 will be used for all logs resulting from traffic passing through the frontend
7897 using this line. If the directive is used in a defaults section, all
7898 subsequent frontends will use the same log format. Please see section 8.2.4
7899 which covers the log format string in depth.
7900
7901 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
7902 about the RFC5424 structured-data part.
7903
7904 Note : This log format string will be used only for loggers that have set
7905 log format to "rfc5424".
7906
7907 Example :
7908 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
7909
7910
Willy Tarreau094af4e2015-01-07 15:03:42 +01007911log-tag <string>
7912 Specifies the log tag to use for all outgoing logs
7913 May be used in sections: defaults | frontend | listen | backend
7914 yes | yes | yes | yes
7915
7916 Sets the tag field in the syslog header to this string. It defaults to the
7917 log-tag set in the global section, otherwise the program name as launched
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007918 from the command line, which usually is "HAProxy". Sometimes it can be useful
Willy Tarreau094af4e2015-01-07 15:03:42 +01007919 to differentiate between multiple processes running on the same host, or to
7920 differentiate customer instances running in the same process. In the backend,
7921 logs about servers up/down will use this tag. As a hint, it can be convenient
7922 to set a log-tag related to a hosted customer in a defaults section then put
7923 all the frontends and backends for that customer, then start another customer
7924 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007925
Willy Tarreauc35362a2014-04-25 13:58:37 +02007926max-keep-alive-queue <value>
7927 Set the maximum server queue size for maintaining keep-alive connections
7928 May be used in sections: defaults | frontend | listen | backend
7929 yes | no | yes | yes
7930
7931 HTTP keep-alive tries to reuse the same server connection whenever possible,
7932 but sometimes it can be counter-productive, for example if a server has a lot
7933 of connections while other ones are idle. This is especially true for static
7934 servers.
7935
7936 The purpose of this setting is to set a threshold on the number of queued
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007937 connections at which HAProxy stops trying to reuse the same server and prefers
Willy Tarreauc35362a2014-04-25 13:58:37 +02007938 to find another one. The default value, -1, means there is no limit. A value
7939 of zero means that keep-alive requests will never be queued. For very close
7940 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007941 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02007942 use a value of 10 or less). For remote servers suffering from a high latency,
7943 higher values might be needed to cover for the latency and/or the cost of
7944 picking a different server.
7945
7946 Note that this has no impact on responses which are maintained to the same
7947 server consecutively to a 401 response. They will still go to the same server
7948 even if they have to be queued.
7949
7950 See also : "option http-server-close", "option prefer-last-server", server
7951 "maxconn" and cookie persistence.
7952
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01007953max-session-srv-conns <nb>
7954 Set the maximum number of outgoing connections we can keep idling for a given
7955 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
7956 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02007957
Willy Tarreau2769aa02007-12-27 18:26:09 +01007958maxconn <conns>
7959 Fix the maximum number of concurrent connections on a frontend
7960 May be used in sections : defaults | frontend | listen | backend
7961 yes | yes | yes | no
7962 Arguments :
7963 <conns> is the maximum number of concurrent connections the frontend will
7964 accept to serve. Excess connections will be queued by the system
7965 in the socket's listen queue and will be served once a connection
7966 closes.
7967
7968 If the system supports it, it can be useful on big sites to raise this limit
Daniel Corbett9f0843f2021-05-08 10:50:37 -04007969 very high so that HAProxy manages connection queues, instead of leaving the
Willy Tarreau2769aa02007-12-27 18:26:09 +01007970 clients with unanswered connection attempts. This value should not exceed the
7971 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01007972 of tune.bufsize (16kB by default) each, as well as some other data resulting
7973 in about 33 kB of RAM being consumed per established connection. That means
7974 that a medium system equipped with 1GB of RAM can withstand around
7975 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007976
7977 Also, when <conns> is set to large values, it is possible that the servers
7978 are not sized to accept such loads, and for this reason it is generally wise
7979 to assign them some reasonable connection limits.
7980
Willy Tarreauc8d5b952019-02-27 17:25:52 +01007981 When this value is set to zero, which is the default, the global "maxconn"
7982 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02007983
Willy Tarreau2769aa02007-12-27 18:26:09 +01007984 See also : "server", global section's "maxconn", "fullconn"
7985
7986
Willy Tarreau77e0dae2020-10-14 15:44:27 +02007987mode { tcp|http }
Willy Tarreau2769aa02007-12-27 18:26:09 +01007988 Set the running mode or protocol of the instance
7989 May be used in sections : defaults | frontend | listen | backend
7990 yes | yes | yes | yes
7991 Arguments :
7992 tcp The instance will work in pure TCP mode. A full-duplex connection
7993 will be established between clients and servers, and no layer 7
7994 examination will be performed. This is the default mode. It
7995 should be used for SSL, SSH, SMTP, ...
7996
7997 http The instance will work in HTTP mode. The client request will be
7998 analyzed in depth before connecting to any server. Any request
7999 which is not RFC-compliant will be rejected. Layer 7 filtering,
8000 processing and switching will be possible. This is the mode which
8001 brings HAProxy most of its value.
8002
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008003 When doing content switching, it is mandatory that the frontend and the
8004 backend are in the same mode (generally HTTP), otherwise the configuration
8005 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01008006
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008007 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01008008 defaults http_instances
8009 mode http
8010
Willy Tarreau0ba27502007-12-24 16:55:16 +01008011
Cyril Bontéf0c60612010-02-06 14:44:47 +01008012monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01008013 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008014 May be used in sections : defaults | frontend | listen | backend
8015 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01008016 Arguments :
8017 if <cond> the monitor request will fail if the condition is satisfied,
8018 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008019 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01008020 are met, for instance a low number of servers both in a
8021 backend and its backup.
8022
8023 unless <cond> the monitor request will succeed only if the condition is
8024 satisfied, and will fail otherwise. Such a condition may be
8025 based on a test on the presence of a minimum number of active
8026 servers in a list of backends.
8027
8028 This statement adds a condition which can force the response to a monitor
8029 request to report a failure. By default, when an external component queries
8030 the URI dedicated to monitoring, a 200 response is returned. When one of the
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008031 conditions above is met, HAProxy will return 503 instead of 200. This is
Willy Tarreau0ba27502007-12-24 16:55:16 +01008032 very useful to report a site failure to an external component which may base
8033 routing advertisements between multiple sites on the availability reported by
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008034 HAProxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02008035 criterion. Note that "monitor fail" only works in HTTP mode. Both status
8036 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008037
8038 Example:
8039 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01008040 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01008041 acl site_dead nbsrv(dynamic) lt 2
8042 acl site_dead nbsrv(static) lt 2
8043 monitor-uri /site_alive
8044 monitor fail if site_dead
8045
Willy Tarreau9e9919d2020-10-14 15:55:23 +02008046 See also : "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01008047
8048
Willy Tarreau2769aa02007-12-27 18:26:09 +01008049monitor-uri <uri>
8050 Intercept a URI used by external components' monitor requests
8051 May be used in sections : defaults | frontend | listen | backend
8052 yes | yes | yes | no
8053 Arguments :
8054 <uri> is the exact URI which we want to intercept to return HAProxy's
8055 health status instead of forwarding the request.
8056
8057 When an HTTP request referencing <uri> will be received on a frontend,
8058 HAProxy will not forward it nor log it, but instead will return either
8059 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
8060 conditions defined with "monitor fail". This is normally enough for any
8061 front-end HTTP probe to detect that the service is UP and running without
8062 forwarding the request to a backend server. Note that the HTTP method, the
8063 version and all headers are ignored, but the request must at least be valid
8064 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
8065
Willy Tarreau721d8e02017-12-01 18:25:08 +01008066 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02008067 and even before any "http-request". The only rulesets applied before are the
8068 tcp-request ones. They cannot be logged either, and it is the intended
8069 purpose. They are only used to report HAProxy's health to an upper component,
8070 nothing more. However, it is possible to add any number of conditions using
8071 "monitor fail" and ACLs so that the result can be adjusted to whatever check
8072 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01008073
Christopher Faulet6072beb2020-02-18 15:34:58 +01008074 Note: if <uri> starts by a slash ('/'), the matching is performed against the
8075 request's path instead of the request's uri. It is a workaround to let
8076 the HTTP/2 requests match the monitor-uri. Indeed, in HTTP/2, clients
8077 are encouraged to send absolute URIs only.
8078
Willy Tarreau2769aa02007-12-27 18:26:09 +01008079 Example :
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008080 # Use /haproxy_test to report HAProxy's status
Willy Tarreau2769aa02007-12-27 18:26:09 +01008081 frontend www
8082 mode http
8083 monitor-uri /haproxy_test
8084
Willy Tarreau9e9919d2020-10-14 15:55:23 +02008085 See also : "monitor fail"
Willy Tarreau2769aa02007-12-27 18:26:09 +01008086
Willy Tarreau0ba27502007-12-24 16:55:16 +01008087
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008088option abortonclose
8089no option abortonclose
8090 Enable or disable early dropping of aborted requests pending in queues.
8091 May be used in sections : defaults | frontend | listen | backend
8092 yes | no | yes | yes
8093 Arguments : none
8094
8095 In presence of very high loads, the servers will take some time to respond.
8096 The per-instance connection queue will inflate, and the response time will
8097 increase respective to the size of the queue times the average per-session
8098 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01008099 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008100 the queue, and slowing down other users, and the servers as well, because the
8101 request will eventually be served, then aborted at the first error
8102 encountered while delivering the response.
8103
8104 As there is no way to distinguish between a full STOP and a simple output
8105 close on the client side, HTTP agents should be conservative and consider
8106 that the client might only have closed its output channel while waiting for
8107 the response. However, this introduces risks of congestion when lots of users
8108 do the same, and is completely useless nowadays because probably no client at
8109 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01008110 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008111 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01008112 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008113 of being the single component to break rare but valid traffic is extremely
8114 low, which adds to the temptation to be able to abort a session early while
8115 still not served and not pollute the servers.
8116
Davor Ocelice9ed2812017-12-25 17:49:28 +01008117 In HAProxy, the user can choose the desired behavior using the option
8118 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008119 compliant and aborted requests will be served. But when the option is
8120 specified, a session with an incoming channel closed will be aborted while
8121 it is still possible, either pending in the queue for a connection slot, or
8122 during the connection establishment if the server has not yet acknowledged
8123 the connection request. This considerably reduces the queue size and the load
8124 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01008125 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008126
8127 If this option has been enabled in a "defaults" section, it can be disabled
8128 in a specific instance by prepending the "no" keyword before it.
8129
8130 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
8131
8132
Willy Tarreau4076a152009-04-02 15:18:36 +02008133option accept-invalid-http-request
8134no option accept-invalid-http-request
8135 Enable or disable relaxing of HTTP request parsing
8136 May be used in sections : defaults | frontend | listen | backend
8137 yes | yes | yes | no
8138 Arguments : none
8139
Willy Tarreau91852eb2015-05-01 13:26:00 +02008140 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02008141 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01008142 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02008143 forbidden characters are essentially used to build attacks exploiting server
8144 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
8145 server will emit invalid header names for whatever reason (configuration,
8146 implementation) and the issue will not be immediately fixed. In such a case,
8147 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01008148 even if that does not make sense, by specifying this option. Similarly, the
8149 list of characters allowed to appear in a URI is well defined by RFC3986, and
8150 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
8151 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01008152 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02008153 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02008154 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
8155 to pass through (no version specified) and multiple digits for both the major
8156 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02008157
8158 This option should never be enabled by default as it hides application bugs
8159 and open security breaches. It should only be deployed after a problem has
8160 been confirmed.
8161
8162 When this option is enabled, erroneous header names will still be accepted in
8163 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01008164 analysis using the "show errors" request on the UNIX stats socket. Similarly,
8165 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02008166 also helps confirming that the issue has been solved.
8167
8168 If this option has been enabled in a "defaults" section, it can be disabled
8169 in a specific instance by prepending the "no" keyword before it.
8170
8171 See also : "option accept-invalid-http-response" and "show errors" on the
8172 stats socket.
8173
8174
8175option accept-invalid-http-response
8176no option accept-invalid-http-response
8177 Enable or disable relaxing of HTTP response parsing
8178 May be used in sections : defaults | frontend | listen | backend
8179 yes | no | yes | yes
8180 Arguments : none
8181
Willy Tarreau91852eb2015-05-01 13:26:00 +02008182 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02008183 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01008184 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02008185 forbidden characters are essentially used to build attacks exploiting server
8186 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
8187 server will emit invalid header names for whatever reason (configuration,
8188 implementation) and the issue will not be immediately fixed. In such a case,
8189 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02008190 even if that does not make sense, by specifying this option. This option also
8191 relaxes the test on the HTTP version format, it allows multiple digits for
8192 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02008193
8194 This option should never be enabled by default as it hides application bugs
8195 and open security breaches. It should only be deployed after a problem has
8196 been confirmed.
8197
8198 When this option is enabled, erroneous header names will still be accepted in
8199 responses, but the complete response will be captured in order to permit
8200 later analysis using the "show errors" request on the UNIX stats socket.
8201 Doing this also helps confirming that the issue has been solved.
8202
8203 If this option has been enabled in a "defaults" section, it can be disabled
8204 in a specific instance by prepending the "no" keyword before it.
8205
8206 See also : "option accept-invalid-http-request" and "show errors" on the
8207 stats socket.
8208
8209
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008210option allbackups
8211no option allbackups
8212 Use either all backup servers at a time or only the first one
8213 May be used in sections : defaults | frontend | listen | backend
8214 yes | no | yes | yes
8215 Arguments : none
8216
8217 By default, the first operational backup server gets all traffic when normal
8218 servers are all down. Sometimes, it may be preferred to use multiple backups
8219 at once, because one will not be enough. When "option allbackups" is enabled,
8220 the load balancing will be performed among all backup servers when all normal
8221 ones are unavailable. The same load balancing algorithm will be used and the
8222 servers' weights will be respected. Thus, there will not be any priority
8223 order between the backup servers anymore.
8224
8225 This option is mostly used with static server farms dedicated to return a
8226 "sorry" page when an application is completely offline.
8227
8228 If this option has been enabled in a "defaults" section, it can be disabled
8229 in a specific instance by prepending the "no" keyword before it.
8230
8231
8232option checkcache
8233no option checkcache
Godbach7056a352013-12-11 20:01:07 +08008234 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008235 May be used in sections : defaults | frontend | listen | backend
8236 yes | no | yes | yes
8237 Arguments : none
8238
8239 Some high-level frameworks set application cookies everywhere and do not
8240 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008241 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008242 high risk of session crossing or stealing between users traversing the same
8243 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008244 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008245
8246 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008247 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01008248 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008249 response to check if there's a risk of caching a cookie on a client-side
8250 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01008251 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01008252 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01008253 - all those with a return code other than 200, 203, 204, 206, 300, 301,
8254 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01008255 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01008256 - all those that result from a request using a method other than GET, HEAD,
8257 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01008258 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008259 - those with a 'Pragma: no-cache' header
8260 - those with a 'Cache-control: private' header
8261 - those with a 'Cache-control: no-store' header
8262 - those with a 'Cache-control: max-age=0' header
8263 - those with a 'Cache-control: s-maxage=0' header
8264 - those with a 'Cache-control: no-cache' header
8265 - those with a 'Cache-control: no-cache="set-cookie"' header
8266 - those with a 'Cache-control: no-cache="set-cookie,' header
8267 (allowing other fields after set-cookie)
8268
8269 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02008270 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
8271 gateway". The session state shows "PH--" meaning that the proxy blocked the
8272 response during headers processing. Additionally, an alert will be sent in
8273 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008274
8275 Due to the high impact on the application, the application should be tested
8276 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008277 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01008278 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008279
8280 If this option has been enabled in a "defaults" section, it can be disabled
8281 in a specific instance by prepending the "no" keyword before it.
8282
8283
8284option clitcpka
8285no option clitcpka
8286 Enable or disable the sending of TCP keepalive packets on the client side
8287 May be used in sections : defaults | frontend | listen | backend
8288 yes | yes | yes | no
8289 Arguments : none
8290
8291 When there is a firewall or any session-aware component between a client and
8292 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01008293 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008294 components decides to expire a session which has remained idle for too long.
8295
8296 Enabling socket-level TCP keep-alives makes the system regularly send packets
8297 to the other end of the connection, leaving it active. The delay between
8298 keep-alive probes is controlled by the system only and depends both on the
8299 operating system and its tuning parameters.
8300
8301 It is important to understand that keep-alive packets are neither emitted nor
8302 received at the application level. It is only the network stacks which sees
8303 them. For this reason, even if one side of the proxy already uses keep-alives
8304 to maintain its connection alive, those keep-alive packets will not be
8305 forwarded to the other side of the proxy.
8306
8307 Please note that this has nothing to do with HTTP keep-alive.
8308
8309 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
8310 client side of a connection, which should help when session expirations are
8311 noticed between HAProxy and a client.
8312
8313 If this option has been enabled in a "defaults" section, it can be disabled
8314 in a specific instance by prepending the "no" keyword before it.
8315
8316 See also : "option srvtcpka", "option tcpka"
8317
8318
Willy Tarreau0ba27502007-12-24 16:55:16 +01008319option contstats
8320 Enable continuous traffic statistics updates
8321 May be used in sections : defaults | frontend | listen | backend
8322 yes | yes | yes | no
8323 Arguments : none
8324
8325 By default, counters used for statistics calculation are incremented
8326 only when a session finishes. It works quite well when serving small
8327 objects, but with big ones (for example large images or archives) or
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008328 with A/V streaming, a graph generated from HAProxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01008329 a hedgehog. With this option enabled counters get incremented frequently
8330 along the session, typically every 5 seconds, which is often enough to
8331 produce clean graphs. Recounting touches a hotpath directly so it is not
8332 not enabled by default, as it can cause a lot of wakeups for very large
8333 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008334
Christopher Faulet89aed322020-06-02 17:33:56 +02008335option disable-h2-upgrade
8336no option disable-h2-upgrade
8337 Enable or disable the implicit HTTP/2 upgrade from an HTTP/1.x client
8338 connection.
8339 May be used in sections : defaults | frontend | listen | backend
8340 yes | yes | yes | no
8341 Arguments : none
8342
8343 By default, HAProxy is able to implicitly upgrade an HTTP/1.x client
8344 connection to an HTTP/2 connection if the first request it receives from a
8345 given HTTP connection matches the HTTP/2 connection preface (i.e. the string
8346 "PRI * HTTP/2.0\r\n\r\nSM\r\n\r\n"). This way, it is possible to support
Christopher Faulet982e17d2021-03-26 14:44:18 +01008347 HTTP/1.x and HTTP/2 clients on a non-SSL connections. This option must be
8348 used to disable the implicit upgrade. Note this implicit upgrade is only
8349 supported for HTTP proxies, thus this option too. Note also it is possible to
8350 force the HTTP/2 on clear connections by specifying "proto h2" on the bind
8351 line. Finally, this option is applied on all bind lines. To disable implicit
8352 HTTP/2 upgrades for a specific bind line, it is possible to use "proto h1".
Christopher Faulet89aed322020-06-02 17:33:56 +02008353
8354 If this option has been enabled in a "defaults" section, it can be disabled
8355 in a specific instance by prepending the "no" keyword before it.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008356
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008357option dontlog-normal
8358no option dontlog-normal
8359 Enable or disable logging of normal, successful connections
8360 May be used in sections : defaults | frontend | listen | backend
8361 yes | yes | yes | no
8362 Arguments : none
8363
8364 There are large sites dealing with several thousand connections per second
8365 and for which logging is a major pain. Some of them are even forced to turn
8366 logs off and cannot debug production issues. Setting this option ensures that
8367 normal connections, those which experience no error, no timeout, no retry nor
8368 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
8369 mode, the response status code is checked and return codes 5xx will still be
8370 logged.
8371
8372 It is strongly discouraged to use this option as most of the time, the key to
8373 complex issues is in the normal logs which will not be logged here. If you
8374 need to separate logs, see the "log-separate-errors" option instead.
8375
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008376 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008377 logging.
8378
8379
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008380option dontlognull
8381no option dontlognull
8382 Enable or disable logging of null connections
8383 May be used in sections : defaults | frontend | listen | backend
8384 yes | yes | yes | no
8385 Arguments : none
8386
8387 In certain environments, there are components which will regularly connect to
8388 various systems to ensure that they are still alive. It can be the case from
8389 another load balancer as well as from monitoring systems. By default, even a
8390 simple port probe or scan will produce a log. If those connections pollute
8391 the logs too much, it is possible to enable option "dontlognull" to indicate
8392 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02008393 which typically corresponds to those probes. Note that errors will still be
8394 returned to the client and accounted for in the stats. If this is not what is
8395 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008396
8397 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01008398 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008399 would not be logged.
8400
8401 If this option has been enabled in a "defaults" section, it can be disabled
8402 in a specific instance by prepending the "no" keyword before it.
8403
Willy Tarreau9e9919d2020-10-14 15:55:23 +02008404 See also : "log", "http-ignore-probes", "monitor-uri", and
Willy Tarreau0f228a02015-05-01 15:37:53 +02008405 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008406
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008407
Willy Tarreau87cf5142011-08-19 22:57:24 +02008408option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01008409 Enable insertion of the X-Forwarded-For header to requests sent to servers
8410 May be used in sections : defaults | frontend | listen | backend
8411 yes | yes | yes | yes
8412 Arguments :
8413 <network> is an optional argument used to disable this option for sources
8414 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02008415 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01008416 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008417
8418 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
8419 their client address. This is sometimes annoying when the client's IP address
8420 is expected in server logs. To solve this problem, the well-known HTTP header
8421 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
8422 This header contains a value representing the client's IP address. Since this
8423 header is always appended at the end of the existing header list, the server
8424 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02008425 the server's manual to find how to enable use of this standard header. Note
8426 that only the last occurrence of the header must be used, since it is really
8427 possible that the client has already brought one.
8428
Willy Tarreaud72758d2010-01-12 10:42:19 +01008429 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02008430 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01008431 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01008432 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008433 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02008434 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01008435
8436 Sometimes, a same HAProxy instance may be shared between a direct client
8437 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
8438 used to decrypt HTTPS traffic). It is possible to disable the addition of the
8439 header for a known source address or network by adding the "except" keyword
8440 followed by the network address. In this case, any source IP matching the
8441 network will not cause an addition of this header. Most common uses are with
Christopher Faulet5d1def62021-02-26 09:19:15 +01008442 private networks or 127.0.0.1. IPv4 and IPv6 are both supported.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008443
Willy Tarreau87cf5142011-08-19 22:57:24 +02008444 Alternatively, the keyword "if-none" states that the header will only be
8445 added if it is not present. This should only be used in perfectly trusted
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008446 environment, as this might cause a security issue if headers reaching HAProxy
Willy Tarreau87cf5142011-08-19 22:57:24 +02008447 are under the control of the end-user.
8448
Willy Tarreauc27debf2008-01-06 08:57:02 +01008449 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02008450 least one of them uses it, the header will be added. Note that the backend's
8451 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02008452 both are defined. In the case of the "if-none" argument, if at least one of
8453 the frontend or the backend does not specify it, it wants the addition to be
8454 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008455
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02008456 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01008457 # Public HTTP address also used by stunnel on the same machine
8458 frontend www
8459 mode http
8460 option forwardfor except 127.0.0.1 # stunnel already adds the header
8461
Ross Westaf72a1d2008-08-03 10:51:45 +02008462 # Those servers want the IP Address in X-Client
8463 backend www
8464 mode http
8465 option forwardfor header X-Client
8466
Willy Tarreau87cf5142011-08-19 22:57:24 +02008467 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02008468 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01008469
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008470
Christopher Faulet98fbe952019-07-22 16:18:24 +02008471option h1-case-adjust-bogus-client
8472no option h1-case-adjust-bogus-client
8473 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
8474 May be used in sections : defaults | frontend | listen | backend
8475 yes | yes | yes | no
8476 Arguments : none
8477
8478 There is no standard case for header names because, as stated in RFC7230,
8479 they are case-insensitive. So applications must handle them in a case-
8480 insensitive manner. But some bogus applications violate the standards and
8481 erroneously rely on the cases most commonly used by browsers. This problem
8482 becomes critical with HTTP/2 because all header names must be exchanged in
8483 lower case, and HAProxy follows the same convention. All header names are
8484 sent in lower case to clients and servers, regardless of the HTTP version.
8485
8486 When HAProxy receives an HTTP/1 response, its header names are converted to
8487 lower case and manipulated and sent this way to the clients. If a client is
8488 known to violate the HTTP standards and to fail to process a response coming
8489 from HAProxy, it is possible to transform the lower case header names to a
8490 different format when the response is formatted and sent to the client, by
8491 enabling this option and specifying the list of headers to be reformatted
8492 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
8493 must only be a temporary workaround for the time it takes the client to be
8494 fixed, because clients which require such workarounds might be vulnerable to
8495 content smuggling attacks and must absolutely be fixed.
8496
8497 Please note that this option will not affect standards-compliant clients.
8498
8499 If this option has been enabled in a "defaults" section, it can be disabled
8500 in a specific instance by prepending the "no" keyword before it.
8501
8502 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
8503 "h1-case-adjust-file".
8504
8505
8506option h1-case-adjust-bogus-server
8507no option h1-case-adjust-bogus-server
8508 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
8509 May be used in sections : defaults | frontend | listen | backend
8510 yes | no | yes | yes
8511 Arguments : none
8512
8513 There is no standard case for header names because, as stated in RFC7230,
8514 they are case-insensitive. So applications must handle them in a case-
8515 insensitive manner. But some bogus applications violate the standards and
8516 erroneously rely on the cases most commonly used by browsers. This problem
8517 becomes critical with HTTP/2 because all header names must be exchanged in
8518 lower case, and HAProxy follows the same convention. All header names are
8519 sent in lower case to clients and servers, regardless of the HTTP version.
8520
8521 When HAProxy receives an HTTP/1 request, its header names are converted to
8522 lower case and manipulated and sent this way to the servers. If a server is
8523 known to violate the HTTP standards and to fail to process a request coming
8524 from HAProxy, it is possible to transform the lower case header names to a
8525 different format when the request is formatted and sent to the server, by
8526 enabling this option and specifying the list of headers to be reformatted
8527 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
8528 must only be a temporary workaround for the time it takes the server to be
8529 fixed, because servers which require such workarounds might be vulnerable to
8530 content smuggling attacks and must absolutely be fixed.
8531
8532 Please note that this option will not affect standards-compliant servers.
8533
8534 If this option has been enabled in a "defaults" section, it can be disabled
8535 in a specific instance by prepending the "no" keyword before it.
8536
8537 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
8538 "h1-case-adjust-file".
8539
8540
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008541option http-buffer-request
8542no option http-buffer-request
8543 Enable or disable waiting for whole HTTP request body before proceeding
8544 May be used in sections : defaults | frontend | listen | backend
8545 yes | yes | yes | yes
8546 Arguments : none
8547
8548 It is sometimes desirable to wait for the body of an HTTP request before
8549 taking a decision. This is what is being done by "balance url_param" for
8550 example. The first use case is to buffer requests from slow clients before
8551 connecting to the server. Another use case consists in taking the routing
8552 decision based on the request body's contents. This option placed in a
8553 frontend or backend forces the HTTP processing to wait until either the whole
Christopher Faulet6db8a2e2019-11-19 16:27:25 +01008554 body is received or the request buffer is full. It can have undesired side
8555 effects with some applications abusing HTTP by expecting unbuffered
8556 transmissions between the frontend and the backend, so this should definitely
8557 not be used by default.
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008558
Christopher Faulet021a8e42021-03-29 10:46:38 +02008559 See also : "option http-no-delay", "timeout http-request",
8560 "http-request wait-for-body"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008561
8562
Willy Tarreau0f228a02015-05-01 15:37:53 +02008563option http-ignore-probes
8564no option http-ignore-probes
8565 Enable or disable logging of null connections and request timeouts
8566 May be used in sections : defaults | frontend | listen | backend
8567 yes | yes | yes | no
8568 Arguments : none
8569
8570 Recently some browsers started to implement a "pre-connect" feature
8571 consisting in speculatively connecting to some recently visited web sites
8572 just in case the user would like to visit them. This results in many
8573 connections being established to web sites, which end up in 408 Request
8574 Timeout if the timeout strikes first, or 400 Bad Request when the browser
8575 decides to close them first. These ones pollute the log and feed the error
8576 counters. There was already "option dontlognull" but it's insufficient in
8577 this case. Instead, this option does the following things :
8578 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01008579 was received over a connection before it was closed;
8580 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02008581 - prevent any error counter from being incremented
8582
8583 That way the empty connection is silently ignored. Note that it is better
8584 not to use this unless it is clear that it is needed, because it will hide
8585 real problems. The most common reason for not receiving a request and seeing
8586 a 408 is due to an MTU inconsistency between the client and an intermediary
8587 element such as a VPN, which blocks too large packets. These issues are
8588 generally seen with POST requests as well as GET with large cookies. The logs
8589 are often the only way to detect them.
8590
8591 If this option has been enabled in a "defaults" section, it can be disabled
8592 in a specific instance by prepending the "no" keyword before it.
8593
8594 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
8595
8596
Willy Tarreau16bfb022010-01-16 19:48:41 +01008597option http-keep-alive
8598no option http-keep-alive
8599 Enable or disable HTTP keep-alive from client to server
8600 May be used in sections : defaults | frontend | listen | backend
8601 yes | yes | yes | yes
8602 Arguments : none
8603
Willy Tarreau70dffda2014-01-30 03:07:23 +01008604 By default HAProxy operates in keep-alive mode with regards to persistent
8605 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02008606 leaves the connection idle on both sides between the end of a response and
8607 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008608 as "option http-server-close" or "option httpclose". This option allows to
8609 set back the keep-alive mode, which can be useful when another mode was used
8610 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01008611
8612 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
8613 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01008614 network) and the fastest session reuse on the server side at the expense
8615 of maintaining idle connections to the servers. In general, it is possible
8616 with this option to achieve approximately twice the request rate that the
8617 "http-server-close" option achieves on small objects. There are mainly two
8618 situations where this option may be useful :
8619
8620 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008621 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01008622
8623 - when the cost of establishing the connection to the server is significant
8624 compared to the cost of retrieving the associated object from the server.
8625
8626 This last case can happen when the server is a fast static server of cache.
8627 In this case, the server will need to be properly tuned to support high enough
8628 connection counts because connections will last until the client sends another
8629 request.
8630
8631 If the client request has to go to another backend or another server due to
8632 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01008633 immediately be closed and a new one re-opened. Option "prefer-last-server" is
8634 available to try optimize server selection so that if the server currently
8635 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01008636
Willy Tarreau16bfb022010-01-16 19:48:41 +01008637 At the moment, logs will not indicate whether requests came from the same
8638 session or not. The accept date reported in the logs corresponds to the end
8639 of the previous request, and the request time corresponds to the time spent
8640 waiting for a new request. The keep-alive request time is still bound to the
8641 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
8642 not set.
8643
Christopher Faulet159e6672019-07-16 15:09:52 +02008644 This option disables and replaces any previous "option httpclose" or "option
8645 http-server-close". When backend and frontend options differ, all of these 4
8646 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01008647
Christopher Faulet315b39c2018-09-21 16:26:19 +02008648 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01008649 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01008650 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01008651
8652
Willy Tarreau96e31212011-05-30 18:10:30 +02008653option http-no-delay
8654no option http-no-delay
8655 Instruct the system to favor low interactive delays over performance in HTTP
8656 May be used in sections : defaults | frontend | listen | backend
8657 yes | yes | yes | yes
8658 Arguments : none
8659
8660 In HTTP, each payload is unidirectional and has no notion of interactivity.
8661 Any agent is expected to queue data somewhat for a reasonably low delay.
8662 There are some very rare server-to-server applications that abuse the HTTP
8663 protocol and expect the payload phase to be highly interactive, with many
8664 interleaved data chunks in both directions within a single request. This is
8665 absolutely not supported by the HTTP specification and will not work across
8666 most proxies or servers. When such applications attempt to do this through
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008667 HAProxy, it works but they will experience high delays due to the network
Willy Tarreau96e31212011-05-30 18:10:30 +02008668 optimizations which favor performance by instructing the system to wait for
8669 enough data to be available in order to only send full packets. Typical
8670 delays are around 200 ms per round trip. Note that this only happens with
8671 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
8672 affected.
8673
8674 When "option http-no-delay" is present in either the frontend or the backend
8675 used by a connection, all such optimizations will be disabled in order to
8676 make the exchanges as fast as possible. Of course this offers no guarantee on
8677 the functionality, as it may break at any other place. But if it works via
8678 HAProxy, it will work as fast as possible. This option should never be used
8679 by default, and should never be used at all unless such a buggy application
8680 is discovered. The impact of using this option is an increase of bandwidth
8681 usage and CPU usage, which may significantly lower performance in high
8682 latency environments.
8683
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008684 See also : "option http-buffer-request"
8685
Willy Tarreau96e31212011-05-30 18:10:30 +02008686
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008687option http-pretend-keepalive
8688no option http-pretend-keepalive
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008689 Define whether HAProxy will announce keepalive to the server or not
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008690 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02008691 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008692 Arguments : none
8693
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008694 When running with "option http-server-close" or "option httpclose", HAProxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008695 adds a "Connection: close" header to the request forwarded to the server.
8696 Unfortunately, when some servers see this header, they automatically refrain
8697 from using the chunked encoding for responses of unknown length, while this
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008698 is totally unrelated. The immediate effect is that this prevents HAProxy from
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008699 maintaining the client connection alive. A second effect is that a client or
8700 a cache could receive an incomplete response without being aware of it, and
8701 consider the response complete.
8702
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008703 By setting "option http-pretend-keepalive", HAProxy will make the server
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008704 believe it will keep the connection alive. The server will then not fall back
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008705 to the abnormal undesired above. When HAProxy gets the whole response, it
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008706 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02008707 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008708 connection is correctly closed on the server side.
8709
8710 It is recommended not to enable this option by default, because most servers
8711 will more efficiently close the connection themselves after the last packet,
8712 and release its buffers slightly earlier. Also, the added packet on the
8713 network could slightly reduce the overall peak performance. However it is
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008714 worth noting that when this option is enabled, HAProxy will have slightly
8715 less work to do. So if HAProxy is the bottleneck on the whole architecture,
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008716 enabling this option might save a few CPU cycles.
8717
Christopher Faulet98db9762018-09-21 10:25:19 +02008718 This option may be set in backend and listen sections. Using it in a frontend
8719 section will be ignored and a warning will be reported during startup. It is
8720 a backend related option, so there is no real reason to set it on a
8721 frontend. This option may be combined with "option httpclose", which will
8722 cause keepalive to be announced to the server and close to be announced to
8723 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008724
8725 If this option has been enabled in a "defaults" section, it can be disabled
8726 in a specific instance by prepending the "no" keyword before it.
8727
Christopher Faulet315b39c2018-09-21 16:26:19 +02008728 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01008729 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008730
Willy Tarreauc27debf2008-01-06 08:57:02 +01008731
Willy Tarreaub608feb2010-01-02 22:47:18 +01008732option http-server-close
8733no option http-server-close
8734 Enable or disable HTTP connection closing on the server side
8735 May be used in sections : defaults | frontend | listen | backend
8736 yes | yes | yes | yes
8737 Arguments : none
8738
Willy Tarreau70dffda2014-01-30 03:07:23 +01008739 By default HAProxy operates in keep-alive mode with regards to persistent
8740 connections: for each connection it processes each request and response, and
8741 leaves the connection idle on both sides between the end of a response and
8742 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008743 as "option http-server-close" or "option httpclose". Setting "option
8744 http-server-close" enables HTTP connection-close mode on the server side
8745 while keeping the ability to support HTTP keep-alive and pipelining on the
8746 client side. This provides the lowest latency on the client side (slow
8747 network) and the fastest session reuse on the server side to save server
8748 resources, similarly to "option httpclose". It also permits non-keepalive
8749 capable servers to be served in keep-alive mode to the clients if they
8750 conform to the requirements of RFC7230. Please note that some servers do not
8751 always conform to those requirements when they see "Connection: close" in the
8752 request. The effect will be that keep-alive will never be used. A workaround
8753 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01008754
8755 At the moment, logs will not indicate whether requests came from the same
8756 session or not. The accept date reported in the logs corresponds to the end
8757 of the previous request, and the request time corresponds to the time spent
8758 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01008759 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
8760 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01008761
8762 This option may be set both in a frontend and in a backend. It is enabled if
8763 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008764 It disables and replaces any previous "option httpclose" or "option
8765 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8766 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01008767
8768 If this option has been enabled in a "defaults" section, it can be disabled
8769 in a specific instance by prepending the "no" keyword before it.
8770
Christopher Faulet315b39c2018-09-21 16:26:19 +02008771 See also : "option httpclose", "option http-pretend-keepalive",
8772 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01008773
Willy Tarreau88d349d2010-01-25 12:15:43 +01008774option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01008775no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01008776 Make use of non-standard Proxy-Connection header instead of Connection
8777 May be used in sections : defaults | frontend | listen | backend
8778 yes | yes | yes | no
8779 Arguments : none
8780
Lukas Tribus23953682017-04-28 13:24:30 +00008781 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01008782 Connection header to indicate their wish of persistent or non-persistent
8783 connections, both browsers and proxies ignore this header for proxied
8784 connections and make use of the undocumented, non-standard Proxy-Connection
8785 header instead. The issue begins when trying to put a load balancer between
8786 browsers and such proxies, because there will be a difference between what
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008787 HAProxy understands and what the client and the proxy agree on.
Willy Tarreau88d349d2010-01-25 12:15:43 +01008788
Daniel Corbett9f0843f2021-05-08 10:50:37 -04008789 By setting this option in a frontend, HAProxy can automatically switch to use
Willy Tarreau88d349d2010-01-25 12:15:43 +01008790 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008791 defined here as one where the URI begins with neither a '/' nor a '*'. This
8792 is incompatible with the HTTP tunnel mode. Note that this option can only be
8793 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01008794
Willy Tarreau844a7e72010-01-31 21:46:18 +01008795 Also, when this option is set, a request which requires authentication will
8796 automatically switch to use proxy authentication headers if it is itself a
8797 proxied request. That makes it possible to check or enforce authentication in
8798 front of an existing proxy.
8799
Willy Tarreau88d349d2010-01-25 12:15:43 +01008800 This option should normally never be used, except in front of a proxy.
8801
Christopher Faulet315b39c2018-09-21 16:26:19 +02008802 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01008803
Willy Tarreaud63335a2010-02-26 12:56:52 +01008804option httpchk
8805option httpchk <uri>
8806option httpchk <method> <uri>
8807option httpchk <method> <uri> <version>
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008808 Enables HTTP protocol to check on the servers health
Willy Tarreaud63335a2010-02-26 12:56:52 +01008809 May be used in sections : defaults | frontend | listen | backend
8810 yes | no | yes | yes
8811 Arguments :
8812 <method> is the optional HTTP method used with the requests. When not set,
8813 the "OPTIONS" method is used, as it generally requires low server
8814 processing and is easy to filter out from the logs. Any method
8815 may be used, though it is not recommended to invent non-standard
8816 ones.
8817
8818 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
8819 which is accessible by default on almost any server, but may be
8820 changed to any other URI. Query strings are permitted.
8821
8822 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
8823 but some servers might behave incorrectly in HTTP 1.0, so turning
8824 it to HTTP/1.1 may sometimes help. Note that the Host field is
Christopher Faulet8acb1282020-04-09 08:44:06 +02008825 mandatory in HTTP/1.1, use "http-check send" directive to add it.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008826
8827 By default, server health checks only consist in trying to establish a TCP
8828 connection. When "option httpchk" is specified, a complete HTTP request is
8829 sent once the TCP connection is established, and responses 2xx and 3xx are
8830 considered valid, while all other ones indicate a server failure, including
8831 the lack of any response.
8832
Christopher Faulete5870d82020-04-15 11:32:03 +02008833 Combined with "http-check" directives, it is possible to customize the
8834 request sent during the HTTP health checks or the matching rules on the
8835 response. It is also possible to configure a send/expect sequence, just like
8836 with the directive "tcp-check" for TCP health checks.
8837
8838 The server configuration is used by default to open connections to perform
8839 HTTP health checks. By it is also possible to overwrite server parameters
8840 using "http-check connect" rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008841
Christopher Faulete5870d82020-04-15 11:32:03 +02008842 "httpchk" option does not necessarily require an HTTP backend, it also works
8843 with plain TCP backends. This is particularly useful to check simple scripts
Christopher Faulet14cd3162020-04-16 14:50:06 +02008844 bound to some dedicated ports using the inetd daemon. However, it will always
Daniel Corbett67a82712020-07-06 23:01:19 -04008845 internally relies on an HTX multiplexer. Thus, it means the request
Christopher Faulet14cd3162020-04-16 14:50:06 +02008846 formatting and the response parsing will be strict.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008847
Christopher Faulet8acb1282020-04-09 08:44:06 +02008848 Note : For a while, there was no way to add headers or body in the request
8849 used for HTTP health checks. So a workaround was to hide it at the end
8850 of the version string with a "\r\n" after the version. It is now
8851 deprecated. The directive "http-check send" must be used instead.
8852
Willy Tarreaud63335a2010-02-26 12:56:52 +01008853 Examples :
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008854 # Relay HTTPS traffic to Apache instance and check service availability
8855 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
8856 backend https_relay
8857 mode tcp
8858 option httpchk OPTIONS * HTTP/1.1
8859 http-check send hdr Host www
8860 server apache1 192.168.1.1:443 check port 80
Willy Tarreaud63335a2010-02-26 12:56:52 +01008861
Simon Hormanafc47ee2013-11-25 10:46:35 +09008862 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
8863 "option pgsql-check", "http-check" and the "check", "port" and
8864 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008865
8866
Willy Tarreauc27debf2008-01-06 08:57:02 +01008867option httpclose
8868no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02008869 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01008870 May be used in sections : defaults | frontend | listen | backend
8871 yes | yes | yes | yes
8872 Arguments : none
8873
Willy Tarreau70dffda2014-01-30 03:07:23 +01008874 By default HAProxy operates in keep-alive mode with regards to persistent
8875 connections: for each connection it processes each request and response, and
8876 leaves the connection idle on both sides between the end of a response and
8877 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008878 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01008879
Christopher Faulet315b39c2018-09-21 16:26:19 +02008880 If "option httpclose" is set, HAProxy will close connections with the server
8881 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05008882 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02008883 and will add one if missing. Any "Connection" header different from "close"
8884 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008885
Christopher Faulet315b39c2018-09-21 16:26:19 +02008886 This option may also be combined with "option http-pretend-keepalive", which
8887 will disable sending of the "Connection: close" header, but will still cause
8888 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008889
8890 This option may be set both in a frontend and in a backend. It is enabled if
8891 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008892 It disables and replaces any previous "option http-server-close" or "option
8893 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8894 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008895
8896 If this option has been enabled in a "defaults" section, it can be disabled
8897 in a specific instance by prepending the "no" keyword before it.
8898
Christopher Faulet315b39c2018-09-21 16:26:19 +02008899 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01008900
8901
Emeric Brun3a058f32009-06-30 18:26:00 +02008902option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01008903 Enable logging of HTTP request, session state and timers
8904 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01008905 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02008906 Arguments :
8907 clf if the "clf" argument is added, then the output format will be
8908 the CLF format instead of HAProxy's default HTTP format. You can
8909 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01008910 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02008911 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008912
8913 By default, the log output format is very poor, as it only contains the
8914 source and destination addresses, and the instance name. By specifying
8915 "option httplog", each log line turns into a much richer format including,
8916 but not limited to, the HTTP request, the connection timers, the session
8917 status, the connections numbers, the captured headers and cookies, the
8918 frontend, backend and server name, and of course the source address and
8919 ports.
8920
PiBa-NLbd556bf2014-12-11 21:31:54 +01008921 Specifying only "option httplog" will automatically clear the 'clf' mode
8922 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02008923
Guillaume de Lafond29f45602017-03-31 19:52:15 +02008924 "option httplog" overrides any previous "log-format" directive.
8925
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008926 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008927
Willy Tarreau55165fe2009-05-10 12:02:55 +02008928
8929option http_proxy
8930no option http_proxy
8931 Enable or disable plain HTTP proxy mode
8932 May be used in sections : defaults | frontend | listen | backend
8933 yes | yes | yes | yes
8934 Arguments : none
8935
8936 It sometimes happens that people need a pure HTTP proxy which understands
8937 basic proxy requests without caching nor any fancy feature. In this case,
8938 it may be worth setting up an HAProxy instance with the "option http_proxy"
8939 set. In this mode, no server is declared, and the connection is forwarded to
8940 the IP address and port found in the URL after the "http://" scheme.
8941
8942 No host address resolution is performed, so this only works when pure IP
8943 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008944 it will probably be used only by experts who know they need exactly it. This
8945 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02008946
8947 If this option has been enabled in a "defaults" section, it can be disabled
8948 in a specific instance by prepending the "no" keyword before it.
8949
8950 Example :
8951 # this backend understands HTTP proxy requests and forwards them directly.
8952 backend direct_forward
8953 option httpclose
8954 option http_proxy
8955
8956 See also : "option httpclose"
8957
Willy Tarreau211ad242009-10-03 21:45:07 +02008958
Jamie Gloudon801a0a32012-08-25 00:18:33 -04008959option independent-streams
8960no option independent-streams
8961 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008962 May be used in sections : defaults | frontend | listen | backend
8963 yes | yes | yes | yes
8964 Arguments : none
8965
8966 By default, when data is sent over a socket, both the write timeout and the
8967 read timeout for that socket are refreshed, because we consider that there is
8968 activity on that socket, and we have no other means of guessing if we should
8969 receive data or not.
8970
Davor Ocelice9ed2812017-12-25 17:49:28 +01008971 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008972 exists a situation where it is desirable to disable it, and only refresh the
8973 read timeout if there are incoming data. This happens on sessions with large
8974 timeouts and low amounts of exchanged data such as telnet session. If the
8975 server suddenly disappears, the output data accumulates in the system's
8976 socket buffers, both timeouts are correctly refreshed, and there is no way
8977 to know the server does not receive them, so we don't timeout. However, when
8978 the underlying protocol always echoes sent data, it would be enough by itself
8979 to detect the issue using the read timeout. Note that this problem does not
8980 happen with more verbose protocols because data won't accumulate long in the
8981 socket buffers.
8982
8983 When this option is set on the frontend, it will disable read timeout updates
8984 on data sent to the client. There probably is little use of this case. When
8985 the option is set on the backend, it will disable read timeout updates on
8986 data sent to the server. Doing so will typically break large HTTP posts from
8987 slow lines, so use it with caution.
8988
Willy Tarreauce887fd2012-05-12 12:50:00 +02008989 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008990
8991
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02008992option ldap-check
8993 Use LDAPv3 health checks for server testing
8994 May be used in sections : defaults | frontend | listen | backend
8995 yes | no | yes | yes
8996 Arguments : none
8997
8998 It is possible to test that the server correctly talks LDAPv3 instead of just
8999 testing that it accepts the TCP connection. When this option is set, an
9000 LDAPv3 anonymous simple bind message is sent to the server, and the response
9001 is analyzed to find an LDAPv3 bind response message.
9002
9003 The server is considered valid only when the LDAP response contains success
9004 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
9005
9006 Logging of bind requests is server dependent see your documentation how to
9007 configure it.
9008
9009 Example :
9010 option ldap-check
9011
9012 See also : "option httpchk"
9013
9014
Simon Horman98637e52014-06-20 12:30:16 +09009015option external-check
9016 Use external processes for server health checks
9017 May be used in sections : defaults | frontend | listen | backend
9018 yes | no | yes | yes
9019
9020 It is possible to test the health of a server using an external command.
9021 This is achieved by running the executable set using "external-check
9022 command".
9023
9024 Requires the "external-check" global to be set.
9025
9026 See also : "external-check", "external-check command", "external-check path"
9027
9028
William Dauchy4711a892022-01-05 22:53:24 +01009029option idle-close-on-response
9030no option idle-close-on-response
9031 Avoid closing idle frontend connections if a soft stop is in progress
9032 May be used in sections : defaults | frontend | listen | backend
9033 yes | yes | yes | no
9034 Arguments : none
9035
9036 By default, idle connections will be closed during a soft stop. In some
9037 environments, a client talking to the proxy may have prepared some idle
9038 connections in order to send requests later. If there is no proper retry on
9039 write errors, this can result in errors while haproxy is reloading. Even
9040 though a proper implementation should retry on connection/write errors, this
9041 option was introduced to support backwards compatibility with haproxy prior
9042 to version 2.4. Indeed before v2.4, haproxy used to wait for a last request
9043 and response to add a "connection: close" header before closing, thus
9044 notifying the client that the connection would not be reusable.
9045
9046 In a real life example, this behavior was seen in AWS using the ALB in front
9047 of a haproxy. The end result was ALB sending 502 during haproxy reloads.
9048
9049 Users are warned that using this option may increase the number of old
9050 processes if connections remain idle for too long. Adjusting the client
9051 timeouts and/or the "hard-stop-after" parameter accordingly might be
9052 needed in case of frequent reloads.
9053
9054 See also: "timeout client", "timeout client-fin", "timeout http-request",
9055 "hard-stop-after"
9056
9057
Willy Tarreau211ad242009-10-03 21:45:07 +02009058option log-health-checks
9059no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02009060 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02009061 May be used in sections : defaults | frontend | listen | backend
9062 yes | no | yes | yes
9063 Arguments : none
9064
Willy Tarreaubef1b322014-05-13 21:01:39 +02009065 By default, failed health check are logged if server is UP and successful
9066 health checks are logged if server is DOWN, so the amount of additional
9067 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02009068
Willy Tarreaubef1b322014-05-13 21:01:39 +02009069 When this option is enabled, any change of the health check status or to
9070 the server's health will be logged, so that it becomes possible to know
9071 that a server was failing occasional checks before crashing, or exactly when
9072 it failed to respond a valid HTTP status, then when the port started to
9073 reject connections, then when the server stopped responding at all.
9074
Davor Ocelice9ed2812017-12-25 17:49:28 +01009075 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02009076 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02009077
Willy Tarreaubef1b322014-05-13 21:01:39 +02009078 See also: "option httpchk", "option ldap-check", "option mysql-check",
9079 "option pgsql-check", "option redis-check", "option smtpchk",
9080 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02009081
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02009082
9083option log-separate-errors
9084no option log-separate-errors
9085 Change log level for non-completely successful connections
9086 May be used in sections : defaults | frontend | listen | backend
9087 yes | yes | yes | no
9088 Arguments : none
9089
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009090 Sometimes looking for errors in logs is not easy. This option makes HAProxy
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02009091 raise the level of logs containing potentially interesting information such
9092 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
9093 level changes from "info" to "err". This makes it possible to log them
9094 separately to a different file with most syslog daemons. Be careful not to
9095 remove them from the original file, otherwise you would lose ordering which
9096 provides very important information.
9097
9098 Using this option, large sites dealing with several thousand connections per
9099 second may log normal traffic to a rotating buffer and only archive smaller
9100 error logs.
9101
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009102 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02009103 logging.
9104
Willy Tarreauc27debf2008-01-06 08:57:02 +01009105
9106option logasap
9107no option logasap
Jerome Magnin95fb57b2020-04-23 19:01:17 +02009108 Enable or disable early logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01009109 May be used in sections : defaults | frontend | listen | backend
9110 yes | yes | yes | no
9111 Arguments : none
9112
Jerome Magnin95fb57b2020-04-23 19:01:17 +02009113 By default, logs are emitted when all the log format variables and sample
9114 fetches used in the definition of the log-format string return a value, or
9115 when the session is terminated. This allows the built in log-format strings
9116 to account for the transfer time, or the number of bytes in log messages.
9117
9118 When handling long lived connections such as large file transfers or RDP,
9119 it may take a while for the request or connection to appear in the logs.
9120 Using "option logasap", the log message is created as soon as the server
9121 connection is established in mode tcp, or as soon as the server sends the
9122 complete headers in mode http. Missing information in the logs will be the
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05009123 total number of bytes which will only indicate the amount of data transferred
Jerome Magnin95fb57b2020-04-23 19:01:17 +02009124 before the message was created and the total time which will not take the
9125 remainder of the connection life or transfer time into account. For the case
9126 of HTTP, it is good practice to capture the Content-Length response header
9127 so that the logs at least indicate how many bytes are expected to be
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05009128 transferred.
Willy Tarreauc27debf2008-01-06 08:57:02 +01009129
Willy Tarreaucc6c8912009-02-22 10:53:55 +01009130 Examples :
9131 listen http_proxy 0.0.0.0:80
9132 mode http
9133 option httplog
9134 option logasap
9135 log 192.168.2.200 local3
9136
9137 >>> Feb 6 12:14:14 localhost \
9138 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
9139 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
9140 "GET /image.iso HTTP/1.0"
9141
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009142 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01009143 logging.
9144
9145
Christopher Faulet62f79fe2020-05-18 18:13:03 +02009146option mysql-check [ user <username> [ { post-41 | pre-41 } ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02009147 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01009148 May be used in sections : defaults | frontend | listen | backend
9149 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02009150 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009151 <username> This is the username which will be used when connecting to MySQL
9152 server.
Christopher Faulet62f79fe2020-05-18 18:13:03 +02009153 post-41 Send post v4.1 client compatible checks (the default)
9154 pre-41 Send pre v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02009155
9156 If you specify a username, the check consists of sending two MySQL packet,
9157 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01009158 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02009159 Error packet. It is a basic but useful test which does not produce error nor
Daniel Black9dc310d2021-07-01 12:09:32 +10009160 aborted connect on the server. However, it requires an unlocked authorised
9161 user without a password. To create a basic limited user in MySQL with
9162 optional resource limits:
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02009163
Daniel Black9dc310d2021-07-01 12:09:32 +10009164 CREATE USER '<username>'@'<ip_of_haproxy|network_of_haproxy/netmask>'
9165 /*!50701 WITH MAX_QUERIES_PER_HOUR 1 MAX_UPDATES_PER_HOUR 0 */
9166 /*M!100201 MAX_STATEMENT_TIME 0.0001 */;
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02009167
9168 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009169 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02009170 Error packet, we don't send anything in this mode. It was reported that it
9171 can generate lockout if check is too frequent and/or if there is not enough
9172 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
9173 value as if a connection is established successfully within fewer than MySQL
9174 "max_connect_errors" attempts after a previous connection was interrupted,
9175 the error count for the host is cleared to zero. If HAProxy's server get
9176 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
9177
9178 Remember that this does not check database presence nor database consistency.
9179 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01009180
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02009181 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01009182
9183 Most often, an incoming MySQL server needs to see the client's IP address for
9184 various purposes, including IP privilege matching and connection logging.
9185 When possible, it is often wise to masquerade the client's IP address when
9186 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02009187 which requires the transparent proxy feature to be compiled in, and the MySQL
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009188 server to route the client via the machine hosting HAProxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01009189
9190 See also: "option httpchk"
9191
9192
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009193option nolinger
9194no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009195 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009196 May be used in sections: defaults | frontend | listen | backend
9197 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009198 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009199
Davor Ocelice9ed2812017-12-25 17:49:28 +01009200 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009201 physically disconnected), the session timeouts triggers and the session is
9202 closed. But it will remain in FIN_WAIT1 state for some time in the system,
9203 using some resources and possibly limiting the ability to establish newer
9204 connections.
9205
9206 When this happens, it is possible to activate "option nolinger" which forces
9207 the system to immediately remove any socket's pending data on close. Thus,
Willy Tarreau4a321032020-10-16 04:55:19 +02009208 a TCP RST is emitted, any pending data are truncated, and the session is
9209 instantly purged from the system's tables. The generally visible effect for
9210 a client is that responses are truncated if the close happens with a last
9211 block of data (e.g. on a redirect or error response). On the server side,
9212 it may help release the source ports immediately when forwarding a client
9213 aborts in tunnels. In both cases, TCP resets are emitted and given that
9214 the session is instantly destroyed, there will be no retransmit. On a lossy
9215 network this can increase problems, especially when there is a firewall on
9216 the lossy side, because the firewall might see and process the reset (hence
9217 purge its session) and block any further traffic for this session,, including
9218 retransmits from the other side. So if the other side doesn't receive it,
9219 it will never receive any RST again, and the firewall might log many blocked
9220 packets.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009221
Willy Tarreau4a321032020-10-16 04:55:19 +02009222 For all these reasons, it is strongly recommended NOT to use this option,
9223 unless absolutely needed as a last resort. In most situations, using the
9224 "client-fin" or "server-fin" timeouts achieves similar results with a more
9225 reliable behavior. On Linux it's also possible to use the "tcp-ut" bind or
9226 server setting.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009227
9228 This option may be used both on frontends and backends, depending on the side
9229 where it is required. Use it on the frontend for clients, and on the backend
Willy Tarreau4a321032020-10-16 04:55:19 +02009230 for servers. While this option is technically supported in "defaults"
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +05009231 sections, it must really not be used there as it risks to accidentally
Willy Tarreau4a321032020-10-16 04:55:19 +02009232 propagate to sections that must no use it and to cause problems there.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009233
9234 If this option has been enabled in a "defaults" section, it can be disabled
9235 in a specific instance by prepending the "no" keyword before it.
9236
Willy Tarreau4a321032020-10-16 04:55:19 +02009237 See also: "timeout client-fin", "timeout server-fin", "tcp-ut" bind or server
9238 keywords.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009239
Willy Tarreau55165fe2009-05-10 12:02:55 +02009240option originalto [ except <network> ] [ header <name> ]
9241 Enable insertion of the X-Original-To header to requests sent to servers
9242 May be used in sections : defaults | frontend | listen | backend
9243 yes | yes | yes | yes
9244 Arguments :
9245 <network> is an optional argument used to disable this option for sources
9246 matching <network>
9247 <name> an optional argument to specify a different "X-Original-To"
9248 header name.
9249
9250 Since HAProxy can work in transparent mode, every request from a client can
9251 be redirected to the proxy and HAProxy itself can proxy every request to a
9252 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
9253 be lost. This is annoying when you want access rules based on destination ip
9254 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
9255 added by HAProxy to all requests sent to the server. This header contains a
9256 value representing the original destination IP address. Since this must be
9257 configured to always use the last occurrence of this header only. Note that
9258 only the last occurrence of the header must be used, since it is really
9259 possible that the client has already brought one.
9260
9261 The keyword "header" may be used to supply a different header name to replace
9262 the default "X-Original-To". This can be useful where you might already
9263 have a "X-Original-To" header from a different application, and you need
9264 preserve it. Also if your backend server doesn't use the "X-Original-To"
9265 header and requires different one.
9266
9267 Sometimes, a same HAProxy instance may be shared between a direct client
9268 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
9269 used to decrypt HTTPS traffic). It is possible to disable the addition of the
Amaury Denoyellef8b42922021-03-04 18:41:14 +01009270 header for a known destination address or network by adding the "except"
9271 keyword followed by the network address. In this case, any destination IP
9272 matching the network will not cause an addition of this header. Most common
9273 uses are with private networks or 127.0.0.1. IPv4 and IPv6 are both
9274 supported.
Willy Tarreau55165fe2009-05-10 12:02:55 +02009275
9276 This option may be specified either in the frontend or in the backend. If at
9277 least one of them uses it, the header will be added. Note that the backend's
9278 setting of the header subargument takes precedence over the frontend's if
9279 both are defined.
9280
Willy Tarreau55165fe2009-05-10 12:02:55 +02009281 Examples :
9282 # Original Destination address
9283 frontend www
9284 mode http
9285 option originalto except 127.0.0.1
9286
9287 # Those servers want the IP Address in X-Client-Dst
9288 backend www
9289 mode http
9290 option originalto header X-Client-Dst
9291
Christopher Faulet315b39c2018-09-21 16:26:19 +02009292 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02009293
9294
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009295option persist
9296no option persist
9297 Enable or disable forced persistence on down servers
9298 May be used in sections: defaults | frontend | listen | backend
9299 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009300 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009301
9302 When an HTTP request reaches a backend with a cookie which references a dead
9303 server, by default it is redispatched to another server. It is possible to
9304 force the request to be sent to the dead server first using "option persist"
9305 if absolutely needed. A common use case is when servers are under extreme
9306 load and spend their time flapping. In this case, the users would still be
9307 directed to the server they opened the session on, in the hope they would be
9308 correctly served. It is recommended to use "option redispatch" in conjunction
9309 with this option so that in the event it would not be possible to connect to
9310 the server at all (server definitely dead), the client would finally be
9311 redirected to another valid server.
9312
9313 If this option has been enabled in a "defaults" section, it can be disabled
9314 in a specific instance by prepending the "no" keyword before it.
9315
Willy Tarreau4de91492010-01-22 19:10:05 +01009316 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009317
9318
Willy Tarreau0c122822013-12-15 18:49:01 +01009319option pgsql-check [ user <username> ]
9320 Use PostgreSQL health checks for server testing
9321 May be used in sections : defaults | frontend | listen | backend
9322 yes | no | yes | yes
9323 Arguments :
9324 <username> This is the username which will be used when connecting to
9325 PostgreSQL server.
9326
9327 The check sends a PostgreSQL StartupMessage and waits for either
9328 Authentication request or ErrorResponse message. It is a basic but useful
9329 test which does not produce error nor aborted connect on the server.
9330 This check is identical with the "mysql-check".
9331
9332 See also: "option httpchk"
9333
9334
Willy Tarreau9420b122013-12-15 18:58:25 +01009335option prefer-last-server
9336no option prefer-last-server
9337 Allow multiple load balanced requests to remain on the same server
9338 May be used in sections: defaults | frontend | listen | backend
9339 yes | no | yes | yes
9340 Arguments : none
9341
9342 When the load balancing algorithm in use is not deterministic, and a previous
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009343 request was sent to a server to which HAProxy still holds a connection, it is
Willy Tarreau9420b122013-12-15 18:58:25 +01009344 sometimes desirable that subsequent requests on a same session go to the same
9345 server as much as possible. Note that this is different from persistence, as
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009346 we only indicate a preference which HAProxy tries to apply without any form
Willy Tarreau9420b122013-12-15 18:58:25 +01009347 of warranty. The real use is for keep-alive connections sent to servers. When
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009348 this option is used, HAProxy will try to reuse the same connection that is
Willy Tarreau9420b122013-12-15 18:58:25 +01009349 attached to the server instead of rebalancing to another server, causing a
9350 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01009351 not make much sense to use this in combination with hashing algorithms. Note,
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009352 HAProxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02009353 to a proxy which sends a 407 (authentication required), when the load
9354 balancing algorithm is not deterministic. This is mandatory for use with the
9355 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01009356 troubleshooting some faulty applications. Option prefer-last-server might be
9357 desirable in these environments as well, to avoid redistributing the traffic
9358 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01009359
9360 If this option has been enabled in a "defaults" section, it can be disabled
9361 in a specific instance by prepending the "no" keyword before it.
9362
9363 See also: "option http-keep-alive"
9364
9365
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009366option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07009367option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009368no option redispatch
9369 Enable or disable session redistribution in case of connection failure
9370 May be used in sections: defaults | frontend | listen | backend
9371 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07009372 Arguments :
9373 <interval> The optional integer value that controls how often redispatches
9374 occur when retrying connections. Positive value P indicates a
9375 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01009376 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07009377 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009378 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07009379 positive value of 1 would indicate a redispatch on every retry,
9380 and a positive value of 3 would indicate a redispatch on every
9381 third retry. You can disable redispatches with a value of 0.
9382
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009383
9384 In HTTP mode, if a server designated by a cookie is down, clients may
9385 definitely stick to it because they cannot flush the cookie, so they will not
9386 be able to access the service anymore.
9387
Willy Tarreau59884a62019-01-02 14:48:31 +01009388 Specifying "option redispatch" will allow the proxy to break cookie or
9389 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009390
Olivier Carrère6e6f59b2020-04-15 11:30:18 +02009391 Active servers are selected from a subset of the list of available
9392 servers. Active servers that are not down or in maintenance (i.e., whose
9393 health is not checked or that have been checked as "up"), are selected in the
9394 following order:
9395
9396 1. Any active, non-backup server, if any, or,
9397
9398 2. If the "allbackups" option is not set, the first backup server in the
9399 list, or
9400
9401 3. If the "allbackups" option is set, any backup server.
9402
9403 When a retry occurs, HAProxy tries to select another server than the last
9404 one. The new server is selected from the current list of servers.
9405
9406 Sometimes, if the list is updated between retries (e.g., if numerous retries
9407 occur and last longer than the time needed to check that a server is down,
9408 remove it from the list and fall back on the list of backup servers),
9409 connections may be redirected to a backup server, though.
9410
Joseph Lynch726ab712015-05-11 23:25:34 -07009411 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009412 connection failures. Of course, it requires having "retries" set to a nonzero
9413 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01009414
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009415 If this option has been enabled in a "defaults" section, it can be disabled
9416 in a specific instance by prepending the "no" keyword before it.
9417
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02009418 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009419
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009420
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02009421option redis-check
9422 Use redis health checks for server testing
9423 May be used in sections : defaults | frontend | listen | backend
9424 yes | no | yes | yes
9425 Arguments : none
9426
9427 It is possible to test that the server correctly talks REDIS protocol instead
9428 of just testing that it accepts the TCP connection. When this option is set,
9429 a PING redis command is sent to the server, and the response is analyzed to
9430 find the "+PONG" response message.
9431
9432 Example :
9433 option redis-check
9434
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009435 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02009436
9437
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009438option smtpchk
9439option smtpchk <hello> <domain>
9440 Use SMTP health checks for server testing
9441 May be used in sections : defaults | frontend | listen | backend
9442 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01009443 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009444 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02009445 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009446 values will be turned into the default command ("HELO").
9447
9448 <domain> is the domain name to present to the server. It may only be
9449 specified (and is mandatory) if the hello command has been
9450 specified. By default, "localhost" is used.
9451
9452 When "option smtpchk" is set, the health checks will consist in TCP
9453 connections followed by an SMTP command. By default, this command is
9454 "HELO localhost". The server's return code is analyzed and only return codes
9455 starting with a "2" will be considered as valid. All other responses,
9456 including a lack of response will constitute an error and will indicate a
9457 dead server.
9458
9459 This test is meant to be used with SMTP servers or relays. Depending on the
9460 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01009461 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009462 25 is often easier than adjusting the configuration.
9463
9464 Most often, an incoming SMTP server needs to see the client's IP address for
9465 various purposes, including spam filtering, anti-spoofing and logging. When
9466 possible, it is often wise to masquerade the client's IP address when
9467 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02009468 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009469
9470 Example :
9471 option smtpchk HELO mydomain.org
9472
9473 See also : "option httpchk", "source"
9474
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009475
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02009476option socket-stats
9477no option socket-stats
9478
9479 Enable or disable collecting & providing separate statistics for each socket.
9480 May be used in sections : defaults | frontend | listen | backend
9481 yes | yes | yes | no
9482
9483 Arguments : none
9484
9485
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009486option splice-auto
9487no option splice-auto
9488 Enable or disable automatic kernel acceleration on sockets in both directions
9489 May be used in sections : defaults | frontend | listen | backend
9490 yes | yes | yes | yes
9491 Arguments : none
9492
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009493 When this option is enabled either on a frontend or on a backend, HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009494 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01009495 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009496 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009497 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009498 are not much aggressive in order to limit excessive use of splicing. This
9499 option requires splicing to be enabled at compile time, and may be globally
9500 disabled with the global option "nosplice". Since splice uses pipes, using it
9501 requires that there are enough spare pipes.
9502
9503 Important note: kernel-based TCP splicing is a Linux-specific feature which
9504 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
9505 transfer data between sockets without copying these data to user-space, thus
9506 providing noticeable performance gains and CPU cycles savings. Since many
9507 early implementations are buggy, corrupt data and/or are inefficient, this
9508 feature is not enabled by default, and it should be used with extreme care.
9509 While it is not possible to detect the correctness of an implementation,
9510 2.6.29 is the first version offering a properly working implementation. In
9511 case of doubt, splicing may be globally disabled using the global "nosplice"
9512 keyword.
9513
9514 Example :
9515 option splice-auto
9516
9517 If this option has been enabled in a "defaults" section, it can be disabled
9518 in a specific instance by prepending the "no" keyword before it.
9519
9520 See also : "option splice-request", "option splice-response", and global
9521 options "nosplice" and "maxpipes"
9522
9523
9524option splice-request
9525no option splice-request
9526 Enable or disable automatic kernel acceleration on sockets for requests
9527 May be used in sections : defaults | frontend | listen | backend
9528 yes | yes | yes | yes
9529 Arguments : none
9530
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009531 When this option is enabled either on a frontend or on a backend, HAProxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009532 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009533 the client to the server. It might still use the recv/send scheme if there
9534 are no spare pipes left. This option requires splicing to be enabled at
9535 compile time, and may be globally disabled with the global option "nosplice".
9536 Since splice uses pipes, using it requires that there are enough spare pipes.
9537
9538 Important note: see "option splice-auto" for usage limitations.
9539
9540 Example :
9541 option splice-request
9542
9543 If this option has been enabled in a "defaults" section, it can be disabled
9544 in a specific instance by prepending the "no" keyword before it.
9545
9546 See also : "option splice-auto", "option splice-response", and global options
9547 "nosplice" and "maxpipes"
9548
9549
9550option splice-response
9551no option splice-response
9552 Enable or disable automatic kernel acceleration on sockets for responses
9553 May be used in sections : defaults | frontend | listen | backend
9554 yes | yes | yes | yes
9555 Arguments : none
9556
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009557 When this option is enabled either on a frontend or on a backend, HAProxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009558 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009559 the server to the client. It might still use the recv/send scheme if there
9560 are no spare pipes left. This option requires splicing to be enabled at
9561 compile time, and may be globally disabled with the global option "nosplice".
9562 Since splice uses pipes, using it requires that there are enough spare pipes.
9563
9564 Important note: see "option splice-auto" for usage limitations.
9565
9566 Example :
9567 option splice-response
9568
9569 If this option has been enabled in a "defaults" section, it can be disabled
9570 in a specific instance by prepending the "no" keyword before it.
9571
9572 See also : "option splice-auto", "option splice-request", and global options
9573 "nosplice" and "maxpipes"
9574
9575
Christopher Fauletba7bc162016-11-07 21:07:38 +01009576option spop-check
9577 Use SPOP health checks for server testing
9578 May be used in sections : defaults | frontend | listen | backend
9579 no | no | no | yes
9580 Arguments : none
9581
9582 It is possible to test that the server correctly talks SPOP protocol instead
9583 of just testing that it accepts the TCP connection. When this option is set,
9584 a HELLO handshake is performed between HAProxy and the server, and the
9585 response is analyzed to check no error is reported.
9586
9587 Example :
9588 option spop-check
9589
9590 See also : "option httpchk"
9591
9592
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009593option srvtcpka
9594no option srvtcpka
9595 Enable or disable the sending of TCP keepalive packets on the server side
9596 May be used in sections : defaults | frontend | listen | backend
9597 yes | no | yes | yes
9598 Arguments : none
9599
9600 When there is a firewall or any session-aware component between a client and
9601 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01009602 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009603 components decides to expire a session which has remained idle for too long.
9604
9605 Enabling socket-level TCP keep-alives makes the system regularly send packets
9606 to the other end of the connection, leaving it active. The delay between
9607 keep-alive probes is controlled by the system only and depends both on the
9608 operating system and its tuning parameters.
9609
9610 It is important to understand that keep-alive packets are neither emitted nor
9611 received at the application level. It is only the network stacks which sees
9612 them. For this reason, even if one side of the proxy already uses keep-alives
9613 to maintain its connection alive, those keep-alive packets will not be
9614 forwarded to the other side of the proxy.
9615
9616 Please note that this has nothing to do with HTTP keep-alive.
9617
9618 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
9619 server side of a connection, which should help when session expirations are
9620 noticed between HAProxy and a server.
9621
9622 If this option has been enabled in a "defaults" section, it can be disabled
9623 in a specific instance by prepending the "no" keyword before it.
9624
9625 See also : "option clitcpka", "option tcpka"
9626
9627
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009628option ssl-hello-chk
9629 Use SSLv3 client hello health checks for server testing
9630 May be used in sections : defaults | frontend | listen | backend
9631 yes | no | yes | yes
9632 Arguments : none
9633
9634 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
9635 possible to test that the server correctly talks SSL instead of just testing
9636 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
9637 SSLv3 client hello messages are sent once the connection is established to
9638 the server, and the response is analyzed to find an SSL server hello message.
9639 The server is considered valid only when the response contains this server
9640 hello message.
9641
9642 All servers tested till there correctly reply to SSLv3 client hello messages,
9643 and most servers tested do not even log the requests containing only hello
9644 messages, which is appreciable.
9645
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009646 Note that this check works even when SSL support was not built into HAProxy
Willy Tarreau763a95b2012-10-04 23:15:39 +02009647 because it forges the SSL message. When SSL support is available, it is best
9648 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009649
Willy Tarreau763a95b2012-10-04 23:15:39 +02009650 See also: "option httpchk", "check-ssl"
9651
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009652
Willy Tarreaued179852013-12-16 01:07:00 +01009653option tcp-check
9654 Perform health checks using tcp-check send/expect sequences
9655 May be used in sections: defaults | frontend | listen | backend
9656 yes | no | yes | yes
9657
9658 This health check method is intended to be combined with "tcp-check" command
9659 lists in order to support send/expect types of health check sequences.
9660
9661 TCP checks currently support 4 modes of operations :
9662 - no "tcp-check" directive : the health check only consists in a connection
9663 attempt, which remains the default mode.
9664
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009665 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01009666 used to send a string along with a connection opening. With some
9667 protocols, it helps sending a "QUIT" message for example that prevents
9668 the server from logging a connection error for each health check. The
9669 check result will still be based on the ability to open the connection
9670 only.
9671
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009672 - "tcp-check expect" only is mentioned : this is used to test a banner.
Daniel Corbett9f0843f2021-05-08 10:50:37 -04009673 The connection is opened and HAProxy waits for the server to present some
Willy Tarreaued179852013-12-16 01:07:00 +01009674 contents which must validate some rules. The check result will be based
9675 on the matching between the contents and the rules. This is suited for
9676 POP, IMAP, SMTP, FTP, SSH, TELNET.
9677
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009678 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009679 used to test a hello-type protocol. HAProxy sends a message, the server
9680 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009681 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01009682 suited for protocols which require a binding or a request/response model.
9683 LDAP, MySQL, Redis and SSL are example of such protocols, though they
9684 already all have their dedicated checks with a deeper understanding of
9685 the respective protocols.
9686 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01009687 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01009688
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009689 A fifth mode can be used to insert comments in different steps of the script.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009690
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009691 For each tcp-check rule you create, you can add a "comment" directive,
9692 followed by a string. This string will be reported in the log and stderr in
9693 debug mode. It is useful to make user-friendly error reporting. The
9694 "comment" is of course optional.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009695
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009696 During the execution of a health check, a variable scope is made available to
9697 store data samples, using the "tcp-check set-var" operation. Freeing those
9698 variable is possible using "tcp-check unset-var".
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +01009699
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009700
Willy Tarreaued179852013-12-16 01:07:00 +01009701 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01009702 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01009703 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009704 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01009705
Davor Ocelice9ed2812017-12-25 17:49:28 +01009706 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01009707 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009708 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01009709
9710 # look for the redis master server after ensuring it speaks well
9711 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009712 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01009713 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009714 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01009715 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009716 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009717 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01009718 tcp-check send info\ replication\r\n
9719 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009720 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01009721 tcp-check send QUIT\r\n
9722 tcp-check expect string +OK
9723
Davor Ocelice9ed2812017-12-25 17:49:28 +01009724 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01009725 (send many headers before analyzing)
9726 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009727 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01009728 tcp-check send HEAD\ /\ HTTP/1.1\r\n
9729 tcp-check send Host:\ www.mydomain.com\r\n
9730 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
9731 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009732 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01009733
9734
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009735 See also : "tcp-check connect", "tcp-check expect" and "tcp-check send".
Willy Tarreaued179852013-12-16 01:07:00 +01009736
9737
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009738option tcp-smart-accept
9739no option tcp-smart-accept
9740 Enable or disable the saving of one ACK packet during the accept sequence
9741 May be used in sections : defaults | frontend | listen | backend
9742 yes | yes | yes | no
9743 Arguments : none
9744
9745 When an HTTP connection request comes in, the system acknowledges it on
9746 behalf of HAProxy, then the client immediately sends its request, and the
9747 system acknowledges it too while it is notifying HAProxy about the new
9748 connection. HAProxy then reads the request and responds. This means that we
9749 have one TCP ACK sent by the system for nothing, because the request could
9750 very well be acknowledged by HAProxy when it sends its response.
9751
9752 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
9753 sending this useless ACK on platforms which support it (currently at least
9754 Linux). It must not cause any problem, because the system will send it anyway
9755 after 40 ms if the response takes more time than expected to come.
9756
9757 During complex network debugging sessions, it may be desirable to disable
9758 this optimization because delayed ACKs can make troubleshooting more complex
9759 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01009760 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009761
9762 It is also possible to force it for non-HTTP proxies by simply specifying
9763 "option tcp-smart-accept". For instance, it can make sense with some services
9764 such as SMTP where the server speaks first.
9765
9766 It is recommended to avoid forcing this option in a defaults section. In case
9767 of doubt, consider setting it back to automatic values by prepending the
9768 "default" keyword before it, or disabling it using the "no" keyword.
9769
Willy Tarreaud88edf22009-06-14 15:48:17 +02009770 See also : "option tcp-smart-connect"
9771
9772
9773option tcp-smart-connect
9774no option tcp-smart-connect
9775 Enable or disable the saving of one ACK packet during the connect sequence
9776 May be used in sections : defaults | frontend | listen | backend
9777 yes | no | yes | yes
9778 Arguments : none
9779
9780 On certain systems (at least Linux), HAProxy can ask the kernel not to
9781 immediately send an empty ACK upon a connection request, but to directly
9782 send the buffer request instead. This saves one packet on the network and
9783 thus boosts performance. It can also be useful for some servers, because they
9784 immediately get the request along with the incoming connection.
9785
9786 This feature is enabled when "option tcp-smart-connect" is set in a backend.
9787 It is not enabled by default because it makes network troubleshooting more
9788 complex.
9789
9790 It only makes sense to enable it with protocols where the client speaks first
9791 such as HTTP. In other situations, if there is no data to send in place of
9792 the ACK, a normal ACK is sent.
9793
9794 If this option has been enabled in a "defaults" section, it can be disabled
9795 in a specific instance by prepending the "no" keyword before it.
9796
9797 See also : "option tcp-smart-accept"
9798
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009799
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009800option tcpka
9801 Enable or disable the sending of TCP keepalive packets on both sides
9802 May be used in sections : defaults | frontend | listen | backend
9803 yes | yes | yes | yes
9804 Arguments : none
9805
9806 When there is a firewall or any session-aware component between a client and
9807 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01009808 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009809 components decides to expire a session which has remained idle for too long.
9810
9811 Enabling socket-level TCP keep-alives makes the system regularly send packets
9812 to the other end of the connection, leaving it active. The delay between
9813 keep-alive probes is controlled by the system only and depends both on the
9814 operating system and its tuning parameters.
9815
9816 It is important to understand that keep-alive packets are neither emitted nor
9817 received at the application level. It is only the network stacks which sees
9818 them. For this reason, even if one side of the proxy already uses keep-alives
9819 to maintain its connection alive, those keep-alive packets will not be
9820 forwarded to the other side of the proxy.
9821
9822 Please note that this has nothing to do with HTTP keep-alive.
9823
9824 Using option "tcpka" enables the emission of TCP keep-alive probes on both
9825 the client and server sides of a connection. Note that this is meaningful
9826 only in "defaults" or "listen" sections. If this option is used in a
9827 frontend, only the client side will get keep-alives, and if this option is
9828 used in a backend, only the server side will get keep-alives. For this
9829 reason, it is strongly recommended to explicitly use "option clitcpka" and
9830 "option srvtcpka" when the configuration is split between frontends and
9831 backends.
9832
9833 See also : "option clitcpka", "option srvtcpka"
9834
Willy Tarreau844e3c52008-01-11 16:28:18 +01009835
9836option tcplog
9837 Enable advanced logging of TCP connections with session state and timers
9838 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01009839 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01009840 Arguments : none
9841
9842 By default, the log output format is very poor, as it only contains the
9843 source and destination addresses, and the instance name. By specifying
9844 "option tcplog", each log line turns into a much richer format including, but
9845 not limited to, the connection timers, the session status, the connections
9846 numbers, the frontend, backend and server name, and of course the source
9847 address and ports. This option is useful for pure TCP proxies in order to
9848 find which of the client or server disconnects or times out. For normal HTTP
9849 proxies, it's better to use "option httplog" which is even more complete.
9850
Guillaume de Lafond29f45602017-03-31 19:52:15 +02009851 "option tcplog" overrides any previous "log-format" directive.
9852
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009853 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009854
9855
Willy Tarreau844e3c52008-01-11 16:28:18 +01009856option transparent
9857no option transparent
9858 Enable client-side transparent proxying
9859 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01009860 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01009861 Arguments : none
9862
9863 This option was introduced in order to provide layer 7 persistence to layer 3
9864 load balancers. The idea is to use the OS's ability to redirect an incoming
9865 connection for a remote address to a local process (here HAProxy), and let
9866 this process know what address was initially requested. When this option is
9867 used, sessions without cookies will be forwarded to the original destination
9868 IP address of the incoming request (which should match that of another
9869 equipment), while requests with cookies will still be forwarded to the
9870 appropriate server.
9871
9872 Note that contrary to a common belief, this option does NOT make HAProxy
9873 present the client's IP to the server when establishing the connection.
9874
Willy Tarreaua1146052011-03-01 09:51:54 +01009875 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009876 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009877
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009878
Simon Horman98637e52014-06-20 12:30:16 +09009879external-check command <command>
9880 Executable to run when performing an external-check
9881 May be used in sections : defaults | frontend | listen | backend
9882 yes | no | yes | yes
9883
9884 Arguments :
9885 <command> is the external command to run
9886
Simon Horman98637e52014-06-20 12:30:16 +09009887 The arguments passed to the to the command are:
9888
Cyril Bonté777be862014-12-02 21:21:35 +01009889 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09009890
Cyril Bonté777be862014-12-02 21:21:35 +01009891 The <proxy_address> and <proxy_port> are derived from the first listener
9892 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
9893 listener the proxy_address will be the path of the socket and the
9894 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
9895 possible to determine a listener, and both <proxy_address> and <proxy_port>
9896 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09009897
Cyril Bonté72cda2a2014-12-27 22:28:39 +01009898 Some values are also provided through environment variables.
9899
9900 Environment variables :
9901 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
9902 applicable, for example in a "backend" section).
9903
9904 HAPROXY_PROXY_ID The backend id.
9905
9906 HAPROXY_PROXY_NAME The backend name.
9907
9908 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
9909 applicable, for example in a "backend" section or
9910 for a UNIX socket).
9911
9912 HAPROXY_SERVER_ADDR The server address.
9913
9914 HAPROXY_SERVER_CURCONN The current number of connections on the server.
9915
9916 HAPROXY_SERVER_ID The server id.
9917
9918 HAPROXY_SERVER_MAXCONN The server max connections.
9919
9920 HAPROXY_SERVER_NAME The server name.
9921
9922 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
9923 socket).
9924
9925 PATH The PATH environment variable used when executing
9926 the command may be set using "external-check path".
9927
William Lallemand4d03e432019-06-14 15:35:37 +02009928 See also "2.3. Environment variables" for other variables.
9929
Simon Horman98637e52014-06-20 12:30:16 +09009930 If the command executed and exits with a zero status then the check is
9931 considered to have passed, otherwise the check is considered to have
9932 failed.
9933
9934 Example :
9935 external-check command /bin/true
9936
9937 See also : "external-check", "option external-check", "external-check path"
9938
9939
9940external-check path <path>
9941 The value of the PATH environment variable used when running an external-check
9942 May be used in sections : defaults | frontend | listen | backend
9943 yes | no | yes | yes
9944
9945 Arguments :
9946 <path> is the path used when executing external command to run
9947
9948 The default path is "".
9949
9950 Example :
9951 external-check path "/usr/bin:/bin"
9952
9953 See also : "external-check", "option external-check",
9954 "external-check command"
9955
9956
Emeric Brun647caf12009-06-30 17:57:00 +02009957persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02009958persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02009959 Enable RDP cookie-based persistence
9960 May be used in sections : defaults | frontend | listen | backend
9961 yes | no | yes | yes
9962 Arguments :
9963 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02009964 default cookie name "msts" will be used. There currently is no
9965 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02009966
9967 This statement enables persistence based on an RDP cookie. The RDP cookie
9968 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01009969 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02009970 and if an RDP cookie is found, it is decoded. If it matches a known server
9971 which is still UP (or if "option persist" is set), then the connection is
9972 forwarded to this server.
9973
9974 Note that this only makes sense in a TCP backend, but for this to work, the
9975 frontend must have waited long enough to ensure that an RDP cookie is present
9976 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009977 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02009978 a single "listen" section.
9979
Willy Tarreau61e28f22010-05-16 22:31:05 +02009980 Also, it is important to understand that the terminal server will emit this
9981 RDP cookie only if it is configured for "token redirection mode", which means
9982 that the "IP address redirection" option is disabled.
9983
Emeric Brun647caf12009-06-30 17:57:00 +02009984 Example :
9985 listen tse-farm
9986 bind :3389
9987 # wait up to 5s for an RDP cookie in the request
9988 tcp-request inspect-delay 5s
9989 tcp-request content accept if RDP_COOKIE
9990 # apply RDP cookie persistence
9991 persist rdp-cookie
9992 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02009993 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02009994 balance rdp-cookie
9995 server srv1 1.1.1.1:3389
9996 server srv2 1.1.1.2:3389
9997
Christian Ruppertfd3184c2022-02-20 22:54:01 +01009998 See also : "balance rdp-cookie", "tcp-request" and the "req.rdp_cookie" ACL.
Emeric Brun647caf12009-06-30 17:57:00 +02009999
10000
Willy Tarreau3a7d2072009-03-05 23:48:25 +010010001rate-limit sessions <rate>
10002 Set a limit on the number of new sessions accepted per second on a frontend
10003 May be used in sections : defaults | frontend | listen | backend
10004 yes | yes | yes | no
10005 Arguments :
10006 <rate> The <rate> parameter is an integer designating the maximum number
10007 of new sessions per second to accept on the frontend.
10008
10009 When the frontend reaches the specified number of new sessions per second, it
10010 stops accepting new connections until the rate drops below the limit again.
10011 During this time, the pending sessions will be kept in the socket's backlog
Daniel Corbett9f0843f2021-05-08 10:50:37 -040010012 (in system buffers) and HAProxy will not even be aware that sessions are
Willy Tarreau3a7d2072009-03-05 23:48:25 +010010013 pending. When applying very low limit on a highly loaded service, it may make
10014 sense to increase the socket's backlog using the "backlog" keyword.
10015
10016 This feature is particularly efficient at blocking connection-based attacks
10017 or service abuse on fragile servers. Since the session rate is measured every
10018 millisecond, it is extremely accurate. Also, the limit applies immediately,
10019 no delay is needed at all to detect the threshold.
10020
10021 Example : limit the connection rate on SMTP to 10 per second max
10022 listen smtp
10023 mode tcp
10024 bind :25
10025 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +020010026 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +010010027
Willy Tarreaua17c2d92011-07-25 08:16:20 +020010028 Note : when the maximum rate is reached, the frontend's status is not changed
10029 but its sockets appear as "WAITING" in the statistics if the
10030 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +010010031
10032 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
10033
10034
Willy Tarreau2e1dca82012-09-12 08:43:15 +020010035redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
10036redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
10037redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +020010038 Return an HTTP redirection if/unless a condition is matched
10039 May be used in sections : defaults | frontend | listen | backend
10040 no | yes | yes | yes
10041
10042 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +010010043 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +020010044
Willy Tarreau0140f252008-11-19 21:07:09 +010010045 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +020010046 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +010010047 the HTTP "Location" header. When used in an "http-request" rule,
10048 <loc> value follows the log-format rules and can include some
10049 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +020010050
10051 <pfx> With "redirect prefix", the "Location" header is built from the
10052 concatenation of <pfx> and the complete URI path, including the
10053 query string, unless the "drop-query" option is specified (see
10054 below). As a special case, if <pfx> equals exactly "/", then
10055 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +010010056 redirect to the same URL (for instance, to insert a cookie). When
10057 used in an "http-request" rule, <pfx> value follows the log-format
10058 rules and can include some dynamic values (see Custom Log Format
10059 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +020010060
10061 <sch> With "redirect scheme", then the "Location" header is built by
10062 concatenating <sch> with "://" then the first occurrence of the
10063 "Host" header, and then the URI path, including the query string
10064 unless the "drop-query" option is specified (see below). If no
10065 path is found or if the path is "*", then "/" is used instead. If
10066 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010067 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +020010068 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +010010069 HTTPS. When used in an "http-request" rule, <sch> value follows
10070 the log-format rules and can include some dynamic values (see
10071 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +010010072
10073 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +010010074 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
10075 with 302 used by default if no code is specified. 301 means
10076 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +020010077 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +010010078 cache the redirection. 303 is equivalent to 302 except that the
10079 browser will fetch the location with a GET method. 307 is just
10080 like 302 but makes it clear that the same method must be reused.
10081 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +010010082
10083 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010084 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +010010085
10086 - "drop-query"
10087 When this keyword is used in a prefix-based redirection, then the
10088 location will be set without any possible query-string, which is useful
10089 for directing users to a non-secure page for instance. It has no effect
10090 with a location-type redirect.
10091
Willy Tarreau81e3b4f2010-01-10 00:42:19 +010010092 - "append-slash"
10093 This keyword may be used in conjunction with "drop-query" to redirect
10094 users who use a URL not ending with a '/' to the same one with the '/'.
10095 It can be useful to ensure that search engines will only see one URL.
10096 For this, a return code 301 is preferred.
10097
Willy Tarreau0140f252008-11-19 21:07:09 +010010098 - "set-cookie NAME[=value]"
10099 A "Set-Cookie" header will be added with NAME (and optionally "=value")
10100 to the response. This is sometimes used to indicate that a user has
10101 been seen, for instance to protect against some types of DoS. No other
10102 cookie option is added, so the cookie will be a session cookie. Note
10103 that for a browser, a sole cookie name without an equal sign is
10104 different from a cookie with an equal sign.
10105
10106 - "clear-cookie NAME[=]"
10107 A "Set-Cookie" header will be added with NAME (and optionally "="), but
10108 with the "Max-Age" attribute set to zero. This will tell the browser to
10109 delete this cookie. It is useful for instance on logout pages. It is
10110 important to note that clearing the cookie "NAME" will not remove a
10111 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
10112 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +020010113
10114 Example: move the login URL only to HTTPS.
10115 acl clear dst_port 80
10116 acl secure dst_port 8080
10117 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +010010118 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +010010119 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +010010120 acl cookie_set hdr_sub(cookie) SEEN=1
10121
10122 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +010010123 redirect prefix https://mysite.com if login_page !secure
10124 redirect prefix http://mysite.com drop-query if login_page !uid_given
10125 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +010010126 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +020010127
Willy Tarreau81e3b4f2010-01-10 00:42:19 +010010128 Example: send redirects for request for articles without a '/'.
10129 acl missing_slash path_reg ^/article/[^/]*$
10130 redirect code 301 prefix / drop-query append-slash if missing_slash
10131
Daniel Corbett9f0843f2021-05-08 10:50:37 -040010132 Example: redirect all HTTP traffic to HTTPS when SSL is handled by HAProxy.
David BERARDe7153042012-11-03 00:11:31 +010010133 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +020010134
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +010010135 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +020010136 http-request redirect code 301 location \
10137 http://www.%[hdr(host)]%[capture.req.uri] \
10138 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +010010139
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010140 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +020010141
Willy Tarreau303c0352008-01-17 19:01:39 +010010142
Willy Tarreaue5c5ce92008-06-20 17:27:19 +020010143retries <value>
10144 Set the number of retries to perform on a server after a connection failure
10145 May be used in sections: defaults | frontend | listen | backend
10146 yes | no | yes | yes
10147 Arguments :
10148 <value> is the number of times a connection attempt should be retried on
10149 a server when a connection either is refused or times out. The
10150 default value is 3.
10151
10152 It is important to understand that this value applies to the number of
10153 connection attempts, not full requests. When a connection has effectively
10154 been established to a server, there will be no more retry.
10155
10156 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -070010157 a turn-around timer of min("timeout connect", one second) is applied before
10158 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +020010159
10160 When "option redispatch" is set, the last retry may be performed on another
10161 server even if a cookie references a different server.
10162
10163 See also : "option redispatch"
10164
10165
Lukas Tribusbb474ff2021-12-08 11:33:01 +010010166retry-on [space-delimited list of keywords]
Jerome Magnin5ce3c142020-05-13 20:09:57 +020010167 Specify when to attempt to automatically retry a failed request.
10168 This setting is only valid when "mode" is set to http and is silently ignored
10169 otherwise.
Olivier Houcharda254a372019-04-05 15:30:12 +020010170 May be used in sections: defaults | frontend | listen | backend
10171 yes | no | yes | yes
10172 Arguments :
Lukas Tribusbb474ff2021-12-08 11:33:01 +010010173 <keywords> is a space-delimited list of keywords or HTTP status codes, each
10174 representing a type of failure event on which an attempt to
10175 retry the request is desired. Please read the notes at the
10176 bottom before changing this setting. The following keywords are
10177 supported :
Olivier Houcharda254a372019-04-05 15:30:12 +020010178
10179 none never retry
10180
10181 conn-failure retry when the connection or the SSL handshake failed
10182 and the request could not be sent. This is the default.
10183
10184 empty-response retry when the server connection was closed after part
10185 of the request was sent, and nothing was received from
10186 the server. This type of failure may be caused by the
10187 request timeout on the server side, poor network
10188 condition, or a server crash or restart while
10189 processing the request.
10190
Olivier Houcharde3249a92019-05-03 23:01:47 +020010191 junk-response retry when the server returned something not looking
10192 like a complete HTTP response. This includes partial
10193 responses headers as well as non-HTTP contents. It
10194 usually is a bad idea to retry on such events, which
10195 may be caused a configuration issue (wrong server port)
10196 or by the request being harmful to the server (buffer
10197 overflow attack for example).
10198
Olivier Houcharda254a372019-04-05 15:30:12 +020010199 response-timeout the server timeout stroke while waiting for the server
10200 to respond to the request. This may be caused by poor
10201 network condition, the reuse of an idle connection
10202 which has expired on the path, or by the request being
10203 extremely expensive to process. It generally is a bad
10204 idea to retry on such events on servers dealing with
10205 heavy database processing (full scans, etc) as it may
10206 amplify denial of service attacks.
10207
Olivier Houchard865d8392019-05-03 22:46:27 +020010208 0rtt-rejected retry requests which were sent over early data and were
10209 rejected by the server. These requests are generally
10210 considered to be safe to retry.
10211
Julien Pivotto2de240a2020-11-12 11:14:05 +010010212 <status> any HTTP status code among "401" (Unauthorized), "403"
10213 (Forbidden), "404" (Not Found), "408" (Request Timeout),
10214 "425" (Too Early), "500" (Server Error), "501" (Not
10215 Implemented), "502" (Bad Gateway), "503" (Service
10216 Unavailable), "504" (Gateway Timeout).
Olivier Houcharda254a372019-04-05 15:30:12 +020010217
Olivier Houchardddf0e032019-05-10 18:05:40 +020010218 all-retryable-errors
10219 retry request for any error that are considered
10220 retryable. This currently activates "conn-failure",
10221 "empty-response", "junk-response", "response-timeout",
10222 "0rtt-rejected", "500", "502", "503", and "504".
10223
Olivier Houcharda254a372019-04-05 15:30:12 +020010224 Using this directive replaces any previous settings with the new ones; it is
10225 not cumulative.
10226
10227 Please note that using anything other than "none" and "conn-failure" requires
10228 to allocate a buffer and copy the whole request into it, so it has memory and
10229 performance impacts. Requests not fitting in a single buffer will never be
10230 retried (see the global tune.bufsize setting).
10231
10232 You have to make sure the application has a replay protection mechanism built
10233 in such as a unique transaction IDs passed in requests, or that replaying the
10234 same request has no consequence, or it is very dangerous to use any retry-on
10235 value beside "conn-failure" and "none". Static file servers and caches are
10236 generally considered safe against any type of retry. Using a status code can
10237 be useful to quickly leave a server showing an abnormal behavior (out of
10238 memory, file system issues, etc), but in this case it may be a good idea to
10239 immediately redispatch the connection to another server (please see "option
10240 redispatch" for this). Last, it is important to understand that most causes
10241 of failures are the requests themselves and that retrying a request causing a
10242 server to misbehave will often make the situation even worse for this server,
10243 or for the whole service in case of redispatch.
10244
10245 Unless you know exactly how the application deals with replayed requests, you
10246 should not use this directive.
10247
10248 The default is "conn-failure".
10249
Lukas Tribusbb474ff2021-12-08 11:33:01 +010010250 Example:
10251 retry-on 503 504
10252
Olivier Houcharda254a372019-04-05 15:30:12 +020010253 See also: "retries", "option redispatch", "tune.bufsize"
10254
David du Colombier486df472011-03-17 10:40:26 +010010255server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010256 Declare a server in a backend
10257 May be used in sections : defaults | frontend | listen | backend
10258 no | no | yes | yes
10259 Arguments :
10260 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +010010261 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -050010262 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010263
David du Colombier486df472011-03-17 10:40:26 +010010264 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
10265 resolvable hostname is supported, but this name will be resolved
10266 during start-up. Address "0.0.0.0" or "*" has a special meaning.
10267 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +020010268 address as the one from the client connection. This is useful in
10269 transparent proxy architectures where the client's connection is
Daniel Corbett9f0843f2021-05-08 10:50:37 -040010270 intercepted and HAProxy must forward to the original destination
Willy Tarreaud669a4f2010-07-13 14:49:50 +020010271 address. This is more or less what the "transparent" keyword does
10272 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +010010273 to report statistics. Optionally, an address family prefix may be
10274 used before the address to force the family regardless of the
10275 address format, which can be useful to specify a path to a unix
10276 socket with no slash ('/'). Currently supported prefixes are :
10277 - 'ipv4@' -> address is always IPv4
10278 - 'ipv6@' -> address is always IPv6
10279 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +020010280 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +020010281 - 'sockpair@' -> address is the FD of a connected unix
10282 socket or of a socketpair. During a connection, the
10283 backend creates a pair of connected sockets, and passes
10284 one of them over the FD. The bind part will use the
10285 received socket as the client FD. Should be used
10286 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +020010287 You may want to reference some environment variables in the
10288 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +010010289 variables. The "init-addr" setting can be used to modify the way
10290 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010291
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010292 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010293 be sent to this port. If unset, the same port the client
10294 connected to will be used. The port may also be prefixed by a "+"
10295 or a "-". In this case, the server's port will be determined by
10296 adding this value to the client's port.
10297
10298 <param*> is a list of parameters for this server. The "server" keywords
10299 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010300 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010301
10302 Examples :
10303 server first 10.1.1.1:1080 cookie first check inter 1000
10304 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +010010305 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +020010306 server backup "${SRV_BACKUP}:1080" backup
10307 server www1_dc1 "${LAN_DC1}.101:80"
10308 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010309
Willy Tarreau55dcaf62015-09-27 15:03:15 +020010310 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
10311 sun_path length is used for the address length. Some other programs
10312 such as socat use the string length only by default. Pass the option
10313 ",unix-tightsocklen=0" to any abstract socket definition in socat to
10314 make it compatible with HAProxy's.
10315
Mark Lamourinec2247f02012-01-04 13:02:01 -050010316 See also: "default-server", "http-send-name-header" and section 5 about
10317 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010318
Christopher Faulet583b6de2021-02-12 09:27:10 +010010319server-state-file-name [ { use-backend-name | <file> } ]
Baptiste Assmann01c6cc32015-08-23 11:45:29 +020010320 Set the server state file to read, load and apply to servers available in
Christopher Faulet583b6de2021-02-12 09:27:10 +010010321 this backend.
10322 May be used in sections: defaults | frontend | listen | backend
10323 no | no | yes | yes
10324
10325 It only applies when the directive "load-server-state-from-file" is set to
10326 "local". When <file> is not provided, if "use-backend-name" is used or if
10327 this directive is not set, then backend name is used. If <file> starts with a
10328 slash '/', then it is considered as an absolute path. Otherwise, <file> is
10329 concatenated to the global directive "server-state-base".
Baptiste Assmann01c6cc32015-08-23 11:45:29 +020010330
10331 Example: the minimal configuration below would make HAProxy look for the
10332 state server file '/etc/haproxy/states/bk':
10333
10334 global
10335 server-state-file-base /etc/haproxy/states
10336
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010010337 backend bk
Baptiste Assmann01c6cc32015-08-23 11:45:29 +020010338 load-server-state-from-file
10339
Christopher Faulet583b6de2021-02-12 09:27:10 +010010340 See also: "server-state-base", "load-server-state-from-file", and
Baptiste Assmann01c6cc32015-08-23 11:45:29 +020010341 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010342
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +020010343server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
10344 Set a template to initialize servers with shared parameters.
10345 The names of these servers are built from <prefix> and <num | range> parameters.
10346 May be used in sections : defaults | frontend | listen | backend
10347 no | no | yes | yes
10348
10349 Arguments:
10350 <prefix> A prefix for the server names to be built.
10351
10352 <num | range>
10353 If <num> is provided, this template initializes <num> servers
10354 with 1 up to <num> as server name suffixes. A range of numbers
10355 <num_low>-<num_high> may also be used to use <num_low> up to
10356 <num_high> as server name suffixes.
10357
10358 <fqdn> A FQDN for all the servers this template initializes.
10359
10360 <port> Same meaning as "server" <port> argument (see "server" keyword).
10361
10362 <params*>
10363 Remaining server parameters among all those supported by "server"
10364 keyword.
10365
10366 Examples:
10367 # Initializes 3 servers with srv1, srv2 and srv3 as names,
10368 # google.com as FQDN, and health-check enabled.
10369 server-template srv 1-3 google.com:80 check
10370
10371 # or
10372 server-template srv 3 google.com:80 check
10373
10374 # would be equivalent to:
10375 server srv1 google.com:80 check
10376 server srv2 google.com:80 check
10377 server srv3 google.com:80 check
10378
10379
10380
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010381source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020010382source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +010010383source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010384 Set the source address for outgoing connections
10385 May be used in sections : defaults | frontend | listen | backend
10386 yes | no | yes | yes
10387 Arguments :
10388 <addr> is the IPv4 address HAProxy will bind to before connecting to a
10389 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +010010390
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010391 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +010010392 the most appropriate address to reach its destination. Optionally
10393 an address family prefix may be used before the address to force
10394 the family regardless of the address format, which can be useful
10395 to specify a path to a unix socket with no slash ('/'). Currently
10396 supported prefixes are :
10397 - 'ipv4@' -> address is always IPv4
10398 - 'ipv6@' -> address is always IPv6
10399 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +020010400 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +020010401 You may want to reference some environment variables in the
10402 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010403
10404 <port> is an optional port. It is normally not needed but may be useful
10405 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020010406 the system will select a free port. Note that port ranges are not
10407 supported in the backend. If you want to force port ranges, you
10408 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010409
10410 <addr2> is the IP address to present to the server when connections are
10411 forwarded in full transparent proxy mode. This is currently only
10412 supported on some patched Linux kernels. When this address is
10413 specified, clients connecting to the server will be presented
10414 with this address, while health checks will still use the address
10415 <addr>.
10416
10417 <port2> is the optional port to present to the server when connections
10418 are forwarded in full transparent proxy mode (see <addr2> above).
10419 The default value of zero means the system will select a free
10420 port.
10421
Willy Tarreaubce70882009-09-07 11:51:47 +020010422 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
10423 This is the name of a comma-separated header list which can
10424 contain multiple IP addresses. By default, the last occurrence is
10425 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +010010426 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +020010427 by previous proxy, typically Stunnel. In order to use another
10428 occurrence from the last one, please see the <occ> parameter
10429 below. When the header (or occurrence) is not found, no binding
10430 is performed so that the proxy's default IP address is used. Also
10431 keep in mind that the header name is case insensitive, as for any
10432 HTTP header.
10433
10434 <occ> is the occurrence number of a value to be used in a multi-value
10435 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010436 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +020010437 address. Positive values indicate a position from the first
10438 occurrence, 1 being the first one. Negative values indicate
10439 positions relative to the last one, -1 being the last one. This
10440 is helpful for situations where an X-Forwarded-For header is set
10441 at the entry point of an infrastructure and must be used several
10442 proxy layers away. When this value is not specified, -1 is
10443 assumed. Passing a zero here disables the feature.
10444
Willy Tarreaud53f96b2009-02-04 18:46:54 +010010445 <name> is an optional interface name to which to bind to for outgoing
10446 traffic. On systems supporting this features (currently, only
10447 Linux), this allows one to bind all traffic to the server to
10448 this interface even if it is not the one the system would select
10449 based on routing tables. This should be used with extreme care.
10450 Note that using this option requires root privileges.
10451
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010452 The "source" keyword is useful in complex environments where a specific
10453 address only is allowed to connect to the servers. It may be needed when a
10454 private address must be used through a public gateway for instance, and it is
10455 known that the system cannot determine the adequate source address by itself.
10456
10457 An extension which is available on certain patched Linux kernels may be used
10458 through the "usesrc" optional keyword. It makes it possible to connect to the
10459 servers with an IP address which does not belong to the system itself. This
10460 is called "full transparent proxy mode". For this to work, the destination
10461 servers have to route their traffic back to this address through the machine
10462 running HAProxy, and IP forwarding must generally be enabled on this machine.
10463
10464 In this "full transparent proxy" mode, it is possible to force a specific IP
10465 address to be presented to the servers. This is not much used in fact. A more
10466 common use is to tell HAProxy to present the client's IP address. For this,
10467 there are two methods :
10468
10469 - present the client's IP and port addresses. This is the most transparent
10470 mode, but it can cause problems when IP connection tracking is enabled on
10471 the machine, because a same connection may be seen twice with different
10472 states. However, this solution presents the huge advantage of not
10473 limiting the system to the 64k outgoing address+port couples, because all
10474 of the client ranges may be used.
10475
10476 - present only the client's IP address and select a spare port. This
10477 solution is still quite elegant but slightly less transparent (downstream
10478 firewalls logs will not match upstream's). It also presents the downside
10479 of limiting the number of concurrent connections to the usual 64k ports.
10480 However, since the upstream and downstream ports are different, local IP
10481 connection tracking on the machine will not be upset by the reuse of the
10482 same session.
10483
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010484 This option sets the default source for all servers in the backend. It may
10485 also be specified in a "defaults" section. Finer source address specification
10486 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010487 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010488
Baptiste Assmann91bd3372015-07-17 21:59:42 +020010489 In order to work, "usesrc" requires root privileges.
10490
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010491 Examples :
10492 backend private
10493 # Connect to the servers using our 192.168.1.200 source address
10494 source 192.168.1.200
10495
10496 backend transparent_ssl1
10497 # Connect to the SSL farm from the client's source address
10498 source 192.168.1.200 usesrc clientip
10499
10500 backend transparent_ssl2
10501 # Connect to the SSL farm from the client's source address and port
10502 # not recommended if IP conntrack is present on the local machine.
10503 source 192.168.1.200 usesrc client
10504
10505 backend transparent_ssl3
10506 # Connect to the SSL farm from the client's source address. It
10507 # is more conntrack-friendly.
10508 source 192.168.1.200 usesrc clientip
10509
10510 backend transparent_smtp
10511 # Connect to the SMTP farm from the client's source address/port
10512 # with Tproxy version 4.
10513 source 0.0.0.0 usesrc clientip
10514
Willy Tarreaubce70882009-09-07 11:51:47 +020010515 backend transparent_http
10516 # Connect to the servers using the client's IP as seen by previous
10517 # proxy.
10518 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
10519
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010520 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010521 the Linux kernel on www.balabit.com, the "bind" keyword.
10522
Willy Tarreau844e3c52008-01-11 16:28:18 +010010523
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010524srvtcpka-cnt <count>
10525 Sets the maximum number of keepalive probes TCP should send before dropping
10526 the connection on the server side.
10527 May be used in sections : defaults | frontend | listen | backend
10528 yes | no | yes | yes
10529 Arguments :
10530 <count> is the maximum number of keepalive probes.
10531
10532 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
10533 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +020010534 The availability of this setting depends on the operating system. It is
10535 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010536
10537 See also : "option srvtcpka", "srvtcpka-idle", "srvtcpka-intvl".
10538
10539
10540srvtcpka-idle <timeout>
10541 Sets the time the connection needs to remain idle before TCP starts sending
10542 keepalive probes, if enabled the sending of TCP keepalive packets on the
10543 server side.
10544 May be used in sections : defaults | frontend | listen | backend
10545 yes | no | yes | yes
10546 Arguments :
10547 <timeout> is the time the connection needs to remain idle before TCP starts
10548 sending keepalive probes. It is specified in seconds by default,
10549 but can be in any other unit if the number is suffixed by the
10550 unit, as explained at the top of this document.
10551
10552 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
10553 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +020010554 The availability of this setting depends on the operating system. It is
10555 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010556
10557 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-intvl".
10558
10559
10560srvtcpka-intvl <timeout>
10561 Sets the time between individual keepalive probes on the server side.
10562 May be used in sections : defaults | frontend | listen | backend
10563 yes | no | yes | yes
10564 Arguments :
10565 <timeout> is the time between individual keepalive probes. It is specified
10566 in seconds by default, but can be in any other unit if the number
10567 is suffixed by the unit, as explained at the top of this
10568 document.
10569
10570 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
10571 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +020010572 The availability of this setting depends on the operating system. It is
10573 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010574
10575 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-idle".
10576
10577
Cyril Bonté66c327d2010-10-12 00:14:37 +020010578stats admin { if | unless } <cond>
10579 Enable statistics admin level if/unless a condition is matched
10580 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010581 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +020010582
10583 This statement enables the statistics admin level if/unless a condition is
10584 matched.
10585
10586 The admin level allows to enable/disable servers from the web interface. By
10587 default, statistics page is read-only for security reasons.
10588
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010589 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10590 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010591 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010592
Cyril Bonté23b39d92011-02-10 22:54:44 +010010593 Currently, the POST request is limited to the buffer size minus the reserved
10594 buffer space, which means that if the list of servers is too long, the
10595 request won't be processed. It is recommended to alter few servers at a
10596 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +020010597
10598 Example :
10599 # statistics admin level only for localhost
10600 backend stats_localhost
10601 stats enable
10602 stats admin if LOCALHOST
10603
10604 Example :
10605 # statistics admin level always enabled because of the authentication
10606 backend stats_auth
10607 stats enable
10608 stats auth admin:AdMiN123
10609 stats admin if TRUE
10610
10611 Example :
10612 # statistics admin level depends on the authenticated user
10613 userlist stats-auth
10614 group admin users admin
10615 user admin insecure-password AdMiN123
10616 group readonly users haproxy
10617 user haproxy insecure-password haproxy
10618
10619 backend stats_auth
10620 stats enable
10621 acl AUTH http_auth(stats-auth)
10622 acl AUTH_ADMIN http_auth_group(stats-auth) admin
10623 stats http-request auth unless AUTH
10624 stats admin if AUTH_ADMIN
10625
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010626 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
10627 "bind-process", section 3.4 about userlists and section 7 about
10628 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +020010629
10630
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010631stats auth <user>:<passwd>
10632 Enable statistics with authentication and grant access to an account
10633 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010634 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010635 Arguments :
10636 <user> is a user name to grant access to
10637
10638 <passwd> is the cleartext password associated to this user
10639
10640 This statement enables statistics with default settings, and restricts access
10641 to declared users only. It may be repeated as many times as necessary to
10642 allow as many users as desired. When a user tries to access the statistics
10643 without a valid account, a "401 Forbidden" response will be returned so that
10644 the browser asks the user to provide a valid user and password. The real
10645 which will be returned to the browser is configurable using "stats realm".
10646
10647 Since the authentication method is HTTP Basic Authentication, the passwords
10648 circulate in cleartext on the network. Thus, it was decided that the
10649 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020010650 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010651
10652 It is also possible to reduce the scope of the proxies which appear in the
10653 report using "stats scope".
10654
10655 Though this statement alone is enough to enable statistics reporting, it is
10656 recommended to set all other settings in order to avoid relying on default
10657 unobvious parameters.
10658
10659 Example :
10660 # public access (limited to this backend only)
10661 backend public_www
10662 server srv1 192.168.0.1:80
10663 stats enable
10664 stats hide-version
10665 stats scope .
10666 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010667 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010668 stats auth admin1:AdMiN123
10669 stats auth admin2:AdMiN321
10670
10671 # internal monitoring access (unlimited)
10672 backend private_monitoring
10673 stats enable
10674 stats uri /admin?stats
10675 stats refresh 5s
10676
10677 See also : "stats enable", "stats realm", "stats scope", "stats uri"
10678
10679
10680stats enable
10681 Enable statistics reporting with default settings
10682 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010683 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010684 Arguments : none
10685
10686 This statement enables statistics reporting with default settings defined
10687 at build time. Unless stated otherwise, these settings are used :
10688 - stats uri : /haproxy?stats
10689 - stats realm : "HAProxy Statistics"
10690 - stats auth : no authentication
10691 - stats scope : no restriction
10692
10693 Though this statement alone is enough to enable statistics reporting, it is
10694 recommended to set all other settings in order to avoid relying on default
10695 unobvious parameters.
10696
10697 Example :
10698 # public access (limited to this backend only)
10699 backend public_www
10700 server srv1 192.168.0.1:80
10701 stats enable
10702 stats hide-version
10703 stats scope .
10704 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010705 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010706 stats auth admin1:AdMiN123
10707 stats auth admin2:AdMiN321
10708
10709 # internal monitoring access (unlimited)
10710 backend private_monitoring
10711 stats enable
10712 stats uri /admin?stats
10713 stats refresh 5s
10714
10715 See also : "stats auth", "stats realm", "stats uri"
10716
10717
Willy Tarreaud63335a2010-02-26 12:56:52 +010010718stats hide-version
10719 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010720 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010721 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010722 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010723
Willy Tarreaud63335a2010-02-26 12:56:52 +010010724 By default, the stats page reports some useful status information along with
10725 the statistics. Among them is HAProxy's version. However, it is generally
10726 considered dangerous to report precise version to anyone, as it can help them
10727 target known weaknesses with specific attacks. The "stats hide-version"
10728 statement removes the version from the statistics report. This is recommended
10729 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010730
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +020010731 Though this statement alone is enough to enable statistics reporting, it is
10732 recommended to set all other settings in order to avoid relying on default
10733 unobvious parameters.
10734
Willy Tarreaud63335a2010-02-26 12:56:52 +010010735 Example :
10736 # public access (limited to this backend only)
10737 backend public_www
10738 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +020010739 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +010010740 stats hide-version
10741 stats scope .
10742 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010743 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +010010744 stats auth admin1:AdMiN123
10745 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010746
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010747 # internal monitoring access (unlimited)
10748 backend private_monitoring
10749 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +010010750 stats uri /admin?stats
10751 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +010010752
Willy Tarreaud63335a2010-02-26 12:56:52 +010010753 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010754
Willy Tarreau983e01e2010-01-11 18:42:06 +010010755
Cyril Bonté2be1b3f2010-09-30 23:46:30 +020010756stats http-request { allow | deny | auth [realm <realm>] }
10757 [ { if | unless } <condition> ]
10758 Access control for statistics
10759
10760 May be used in sections: defaults | frontend | listen | backend
10761 no | no | yes | yes
10762
10763 As "http-request", these set of options allow to fine control access to
10764 statistics. Each option may be followed by if/unless and acl.
10765 First option with matched condition (or option without condition) is final.
10766 For "deny" a 403 error will be returned, for "allow" normal processing is
10767 performed, for "auth" a 401/407 error code is returned so the client
10768 should be asked to enter a username and password.
10769
10770 There is no fixed limit to the number of http-request statements per
10771 instance.
10772
10773 See also : "http-request", section 3.4 about userlists and section 7
10774 about ACL usage.
10775
10776
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010777stats realm <realm>
10778 Enable statistics and set authentication realm
10779 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010780 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010781 Arguments :
10782 <realm> is the name of the HTTP Basic Authentication realm reported to
10783 the browser. The browser uses it to display it in the pop-up
10784 inviting the user to enter a valid username and password.
10785
10786 The realm is read as a single word, so any spaces in it should be escaped
10787 using a backslash ('\').
10788
10789 This statement is useful only in conjunction with "stats auth" since it is
10790 only related to authentication.
10791
10792 Though this statement alone is enough to enable statistics reporting, it is
10793 recommended to set all other settings in order to avoid relying on default
10794 unobvious parameters.
10795
10796 Example :
10797 # public access (limited to this backend only)
10798 backend public_www
10799 server srv1 192.168.0.1:80
10800 stats enable
10801 stats hide-version
10802 stats scope .
10803 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010804 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010805 stats auth admin1:AdMiN123
10806 stats auth admin2:AdMiN321
10807
10808 # internal monitoring access (unlimited)
10809 backend private_monitoring
10810 stats enable
10811 stats uri /admin?stats
10812 stats refresh 5s
10813
10814 See also : "stats auth", "stats enable", "stats uri"
10815
10816
10817stats refresh <delay>
10818 Enable statistics with automatic refresh
10819 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010820 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010821 Arguments :
10822 <delay> is the suggested refresh delay, specified in seconds, which will
10823 be returned to the browser consulting the report page. While the
10824 browser is free to apply any delay, it will generally respect it
10825 and refresh the page this every seconds. The refresh interval may
10826 be specified in any other non-default time unit, by suffixing the
10827 unit after the value, as explained at the top of this document.
10828
10829 This statement is useful on monitoring displays with a permanent page
10830 reporting the load balancer's activity. When set, the HTML report page will
10831 include a link "refresh"/"stop refresh" so that the user can select whether
Jackie Tapia749f74c2020-07-22 18:59:40 -050010832 they want automatic refresh of the page or not.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010833
10834 Though this statement alone is enough to enable statistics reporting, it is
10835 recommended to set all other settings in order to avoid relying on default
10836 unobvious parameters.
10837
10838 Example :
10839 # public access (limited to this backend only)
10840 backend public_www
10841 server srv1 192.168.0.1:80
10842 stats enable
10843 stats hide-version
10844 stats scope .
10845 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010846 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010847 stats auth admin1:AdMiN123
10848 stats auth admin2:AdMiN321
10849
10850 # internal monitoring access (unlimited)
10851 backend private_monitoring
10852 stats enable
10853 stats uri /admin?stats
10854 stats refresh 5s
10855
10856 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10857
10858
10859stats scope { <name> | "." }
10860 Enable statistics and limit access scope
10861 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010862 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010863 Arguments :
10864 <name> is the name of a listen, frontend or backend section to be
10865 reported. The special name "." (a single dot) designates the
10866 section in which the statement appears.
10867
10868 When this statement is specified, only the sections enumerated with this
10869 statement will appear in the report. All other ones will be hidden. This
10870 statement may appear as many times as needed if multiple sections need to be
10871 reported. Please note that the name checking is performed as simple string
10872 comparisons, and that it is never checked that a give section name really
10873 exists.
10874
10875 Though this statement alone is enough to enable statistics reporting, it is
10876 recommended to set all other settings in order to avoid relying on default
10877 unobvious parameters.
10878
10879 Example :
10880 # public access (limited to this backend only)
10881 backend public_www
10882 server srv1 192.168.0.1:80
10883 stats enable
10884 stats hide-version
10885 stats scope .
10886 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010887 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010888 stats auth admin1:AdMiN123
10889 stats auth admin2:AdMiN321
10890
10891 # internal monitoring access (unlimited)
10892 backend private_monitoring
10893 stats enable
10894 stats uri /admin?stats
10895 stats refresh 5s
10896
10897 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10898
Willy Tarreaud63335a2010-02-26 12:56:52 +010010899
Willy Tarreauc9705a12010-07-27 20:05:50 +020010900stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +010010901 Enable reporting of a description on the statistics page.
10902 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010903 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010904
Willy Tarreauc9705a12010-07-27 20:05:50 +020010905 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +010010906 description from global section is automatically used instead.
10907
10908 This statement is useful for users that offer shared services to their
10909 customers, where node or description should be different for each customer.
10910
10911 Though this statement alone is enough to enable statistics reporting, it is
10912 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010913 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010914
10915 Example :
10916 # internal monitoring access (unlimited)
10917 backend private_monitoring
10918 stats enable
10919 stats show-desc Master node for Europe, Asia, Africa
10920 stats uri /admin?stats
10921 stats refresh 5s
10922
10923 See also: "show-node", "stats enable", "stats uri" and "description" in
10924 global section.
10925
10926
10927stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +020010928 Enable reporting additional information on the statistics page
10929 May be used in sections : defaults | frontend | listen | backend
10930 yes | yes | yes | yes
10931 Arguments : none
10932
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010933 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +010010934 - cap: capabilities (proxy)
10935 - mode: one of tcp, http or health (proxy)
10936 - id: SNMP ID (proxy, socket, server)
10937 - IP (socket, server)
10938 - cookie (backend, server)
10939
10940 Though this statement alone is enough to enable statistics reporting, it is
10941 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010942 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010943
10944 See also: "stats enable", "stats uri".
10945
10946
Amaury Denoyelle0b70a8a2020-10-05 11:49:45 +020010947stats show-modules
10948 Enable display of extra statistics module on the statistics page
10949 May be used in sections : defaults | frontend | listen | backend
10950 yes | yes | yes | yes
10951 Arguments : none
10952
10953 New columns are added at the end of the line containing the extra statistics
10954 values as a tooltip.
10955
10956 Though this statement alone is enough to enable statistics reporting, it is
10957 recommended to set all other settings in order to avoid relying on default
10958 unobvious parameters. Default behavior is not to show this information.
10959
10960 See also: "stats enable", "stats uri".
10961
10962
Willy Tarreaud63335a2010-02-26 12:56:52 +010010963stats show-node [ <name> ]
10964 Enable reporting of a host name on the statistics page.
10965 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010966 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010967 Arguments:
10968 <name> is an optional name to be reported. If unspecified, the
10969 node name from global section is automatically used instead.
10970
10971 This statement is useful for users that offer shared services to their
10972 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +010010973 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010974
10975 Though this statement alone is enough to enable statistics reporting, it is
10976 recommended to set all other settings in order to avoid relying on default
10977 unobvious parameters.
10978
10979 Example:
10980 # internal monitoring access (unlimited)
10981 backend private_monitoring
10982 stats enable
10983 stats show-node Europe-1
10984 stats uri /admin?stats
10985 stats refresh 5s
10986
10987 See also: "show-desc", "stats enable", "stats uri", and "node" in global
10988 section.
10989
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010990
10991stats uri <prefix>
10992 Enable statistics and define the URI prefix to access them
10993 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010994 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010995 Arguments :
10996 <prefix> is the prefix of any URI which will be redirected to stats. This
10997 prefix may contain a question mark ('?') to indicate part of a
10998 query string.
10999
11000 The statistics URI is intercepted on the relayed traffic, so it appears as a
11001 page within the normal application. It is strongly advised to ensure that the
11002 selected URI will never appear in the application, otherwise it will never be
11003 possible to reach it in the application.
11004
Daniel Corbett9f0843f2021-05-08 10:50:37 -040011005 The default URI compiled in HAProxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011006 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010011007 It is generally a good idea to include a question mark in the URI so that
11008 intermediate proxies refrain from caching the results. Also, since any string
11009 beginning with the prefix will be accepted as a stats request, the question
11010 mark helps ensuring that no valid URI will begin with the same words.
11011
11012 It is sometimes very convenient to use "/" as the URI prefix, and put that
11013 statement in a "listen" instance of its own. That makes it easy to dedicate
11014 an address or a port to statistics only.
11015
11016 Though this statement alone is enough to enable statistics reporting, it is
11017 recommended to set all other settings in order to avoid relying on default
11018 unobvious parameters.
11019
11020 Example :
11021 # public access (limited to this backend only)
11022 backend public_www
11023 server srv1 192.168.0.1:80
11024 stats enable
11025 stats hide-version
11026 stats scope .
11027 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010011028 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010011029 stats auth admin1:AdMiN123
11030 stats auth admin2:AdMiN321
11031
11032 # internal monitoring access (unlimited)
11033 backend private_monitoring
11034 stats enable
11035 stats uri /admin?stats
11036 stats refresh 5s
11037
11038 See also : "stats auth", "stats enable", "stats realm"
11039
11040
Willy Tarreaud63335a2010-02-26 12:56:52 +010011041stick match <pattern> [table <table>] [{if | unless} <cond>]
11042 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +010011043 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +010011044 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011045
11046 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011047 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011048 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010011049 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011050 stickiness table. This rule is mandatory.
11051
11052 <table> is an optional stickiness table name. If unspecified, the same
11053 backend's table is used. A stickiness table is declared using
11054 the "stick-table" statement.
11055
11056 <cond> is an optional matching condition. It makes it possible to match
11057 on a certain criterion only when other conditions are met (or
11058 not met). For instance, it could be used to match on a source IP
11059 address except when a request passes through a known proxy, in
11060 which case we'd match on a header containing that IP address.
11061
11062 Some protocols or applications require complex stickiness rules and cannot
11063 always simply rely on cookies nor hashing. The "stick match" statement
11064 describes a rule to extract the stickiness criterion from an incoming request
11065 or connection. See section 7 for a complete list of possible patterns and
11066 transformation rules.
11067
11068 The table has to be declared using the "stick-table" statement. It must be of
11069 a type compatible with the pattern. By default it is the one which is present
11070 in the same backend. It is possible to share a table with other backends by
11071 referencing it using the "table" keyword. If another table is referenced,
11072 the server's ID inside the backends are used. By default, all server IDs
11073 start at 1 in each backend, so the server ordering is enough. But in case of
11074 doubt, it is highly recommended to force server IDs using their "id" setting.
11075
11076 It is possible to restrict the conditions where a "stick match" statement
11077 will apply, using "if" or "unless" followed by a condition. See section 7 for
11078 ACL based conditions.
11079
11080 There is no limit on the number of "stick match" statements. The first that
11081 applies and matches will cause the request to be directed to the same server
11082 as was used for the request which created the entry. That way, multiple
11083 matches can be used as fallbacks.
11084
11085 The stick rules are checked after the persistence cookies, so they will not
11086 affect stickiness if a cookie has already been used to select a server. That
11087 way, it becomes very easy to insert cookies and match on IP addresses in
11088 order to maintain stickiness between HTTP and HTTPS.
11089
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011090 Note : Consider not using this feature in multi-process mode (nbproc > 1)
11091 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011092 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011093
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011094 Example :
11095 # forward SMTP users to the same server they just used for POP in the
11096 # last 30 minutes
11097 backend pop
11098 mode tcp
11099 balance roundrobin
11100 stick store-request src
11101 stick-table type ip size 200k expire 30m
11102 server s1 192.168.1.1:110
11103 server s2 192.168.1.1:110
11104
11105 backend smtp
11106 mode tcp
11107 balance roundrobin
11108 stick match src table pop
11109 server s1 192.168.1.1:25
11110 server s2 192.168.1.1:25
11111
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011112 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020011113 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011114
11115
11116stick on <pattern> [table <table>] [{if | unless} <condition>]
11117 Define a request pattern to associate a user to a server
11118 May be used in sections : defaults | frontend | listen | backend
11119 no | no | yes | yes
11120
11121 Note : This form is exactly equivalent to "stick match" followed by
11122 "stick store-request", all with the same arguments. Please refer
11123 to both keywords for details. It is only provided as a convenience
11124 for writing more maintainable configurations.
11125
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011126 Note : Consider not using this feature in multi-process mode (nbproc > 1)
11127 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011128 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011129
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011130 Examples :
11131 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +010011132 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011133
11134 # ...is strictly equivalent to this one :
11135 stick match src table pop if !localhost
11136 stick store-request src table pop if !localhost
11137
11138
11139 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
11140 # well as HTTP without cookie. Share the same table between both accesses.
11141 backend http
11142 mode http
11143 balance roundrobin
11144 stick on src table https
11145 cookie SRV insert indirect nocache
11146 server s1 192.168.1.1:80 cookie s1
11147 server s2 192.168.1.1:80 cookie s2
11148
11149 backend https
11150 mode tcp
11151 balance roundrobin
11152 stick-table type ip size 200k expire 30m
11153 stick on src
11154 server s1 192.168.1.1:443
11155 server s2 192.168.1.1:443
11156
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011157 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011158
11159
11160stick store-request <pattern> [table <table>] [{if | unless} <condition>]
11161 Define a request pattern used to create an entry in a stickiness table
11162 May be used in sections : defaults | frontend | listen | backend
11163 no | no | yes | yes
11164
11165 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011166 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011167 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010011168 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011169 server is selected.
11170
11171 <table> is an optional stickiness table name. If unspecified, the same
11172 backend's table is used. A stickiness table is declared using
11173 the "stick-table" statement.
11174
11175 <cond> is an optional storage condition. It makes it possible to store
11176 certain criteria only when some conditions are met (or not met).
11177 For instance, it could be used to store the source IP address
11178 except when the request passes through a known proxy, in which
11179 case we'd store a converted form of a header containing that IP
11180 address.
11181
11182 Some protocols or applications require complex stickiness rules and cannot
11183 always simply rely on cookies nor hashing. The "stick store-request" statement
11184 describes a rule to decide what to extract from the request and when to do
11185 it, in order to store it into a stickiness table for further requests to
11186 match it using the "stick match" statement. Obviously the extracted part must
11187 make sense and have a chance to be matched in a further request. Storing a
11188 client's IP address for instance often makes sense. Storing an ID found in a
11189 URL parameter also makes sense. Storing a source port will almost never make
11190 any sense because it will be randomly matched. See section 7 for a complete
11191 list of possible patterns and transformation rules.
11192
11193 The table has to be declared using the "stick-table" statement. It must be of
11194 a type compatible with the pattern. By default it is the one which is present
11195 in the same backend. It is possible to share a table with other backends by
11196 referencing it using the "table" keyword. If another table is referenced,
11197 the server's ID inside the backends are used. By default, all server IDs
11198 start at 1 in each backend, so the server ordering is enough. But in case of
11199 doubt, it is highly recommended to force server IDs using their "id" setting.
11200
11201 It is possible to restrict the conditions where a "stick store-request"
11202 statement will apply, using "if" or "unless" followed by a condition. This
11203 condition will be evaluated while parsing the request, so any criteria can be
11204 used. See section 7 for ACL based conditions.
11205
11206 There is no limit on the number of "stick store-request" statements, but
11207 there is a limit of 8 simultaneous stores per request or response. This
11208 makes it possible to store up to 8 criteria, all extracted from either the
11209 request or the response, regardless of the number of rules. Only the 8 first
11210 ones which match will be kept. Using this, it is possible to feed multiple
11211 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010011212 another protocol or access method. Using multiple store-request rules with
11213 the same table is possible and may be used to find the best criterion to rely
11214 on, by arranging the rules by decreasing preference order. Only the first
11215 extracted criterion for a given table will be stored. All subsequent store-
11216 request rules referencing the same table will be skipped and their ACLs will
11217 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011218
11219 The "store-request" rules are evaluated once the server connection has been
11220 established, so that the table will contain the real server that processed
11221 the request.
11222
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011223 Note : Consider not using this feature in multi-process mode (nbproc > 1)
11224 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011225 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011226
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011227 Example :
11228 # forward SMTP users to the same server they just used for POP in the
11229 # last 30 minutes
11230 backend pop
11231 mode tcp
11232 balance roundrobin
11233 stick store-request src
11234 stick-table type ip size 200k expire 30m
11235 server s1 192.168.1.1:110
11236 server s2 192.168.1.1:110
11237
11238 backend smtp
11239 mode tcp
11240 balance roundrobin
11241 stick match src table pop
11242 server s1 192.168.1.1:25
11243 server s2 192.168.1.1:25
11244
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011245 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020011246 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011247
11248
Emeric Brun7c6b82e2010-09-24 16:34:28 +020011249stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Thayne McCombs92149f92020-11-20 01:28:26 -070011250 size <size> [expire <expire>] [nopurge] [peers <peersect>] [srvkey <srvkey>]
Emeric Brunf099e792010-09-27 12:05:28 +020011251 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +080011252 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011253 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +020011254 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011255
11256 Arguments :
11257 ip a table declared with "type ip" will only store IPv4 addresses.
11258 This form is very compact (about 50 bytes per entry) and allows
11259 very fast entry lookup and stores with almost no overhead. This
11260 is mainly used to store client source IP addresses.
11261
David du Colombier9a6d3c92011-03-17 10:40:24 +010011262 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
11263 This form is very compact (about 60 bytes per entry) and allows
11264 very fast entry lookup and stores with almost no overhead. This
11265 is mainly used to store client source IP addresses.
11266
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011267 integer a table declared with "type integer" will store 32bit integers
11268 which can represent a client identifier found in a request for
11269 instance.
11270
11271 string a table declared with "type string" will store substrings of up
11272 to <len> characters. If the string provided by the pattern
11273 extractor is larger than <len>, it will be truncated before
11274 being stored. During matching, at most <len> characters will be
11275 compared between the string in the table and the extracted
11276 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +020011277 to 32 characters.
11278
11279 binary a table declared with "type binary" will store binary blocks
11280 of <len> bytes. If the block provided by the pattern
11281 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +020011282 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +020011283 is shorter than <len>, it will be padded by 0. When not
11284 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011285
11286 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +020011287 "string" type table (See type "string" above). Or the number
11288 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011289 changing this parameter as memory usage will proportionally
11290 increase.
11291
11292 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +010011293 value directly impacts memory usage. Count approximately
11294 50 bytes per entry, plus the size of a string if any. The size
11295 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011296
11297 [nopurge] indicates that we refuse to purge older entries when the table
Daniel Corbett9f0843f2021-05-08 10:50:37 -040011298 is full. When not specified and the table is full when HAProxy
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011299 wants to store an entry in it, it will flush a few of the oldest
11300 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +010011301 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011302 be desirable to refuse new entries instead of purging the older
11303 ones. That may be the case when the amount of data to store is
11304 far above the hardware limits and we prefer not to offer access
11305 to new clients than to reject the ones already connected. When
11306 using this parameter, be sure to properly set the "expire"
11307 parameter (see below).
11308
Emeric Brunf099e792010-09-27 12:05:28 +020011309 <peersect> is the name of the peers section to use for replication. Entries
11310 which associate keys to server IDs are kept synchronized with
11311 the remote peers declared in this section. All entries are also
11312 automatically learned from the local peer (old process) during a
11313 soft restart.
11314
Willy Tarreau1abc6732015-05-01 19:21:02 +020011315 NOTE : each peers section may be referenced only by tables
11316 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011317
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011318 <expire> defines the maximum duration of an entry in the table since it
11319 was last created, refreshed or matched. The expiration delay is
11320 defined using the standard time format, similarly as the various
11321 timeouts. The maximum duration is slightly above 24 days. See
Willy Tarreau4b103022021-02-12 17:59:10 +010011322 section 2.5 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +020011323 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011324 be removed once full. Be sure not to use the "nopurge" parameter
11325 if not expiration delay is specified.
11326
Thayne McCombs92149f92020-11-20 01:28:26 -070011327 <srvkey> specifies how each server is identified for the purposes of the
11328 stick table. The valid values are "name" and "addr". If "name" is
11329 given, then <name> argument for the server (may be generated by
11330 a template). If "addr" is given, then the server is identified
11331 by its current network address, including the port. "addr" is
11332 especially useful if you are using service discovery to generate
11333 the addresses for servers with peered stick-tables and want
11334 to consistently use the same host across peers for a stickiness
11335 token.
11336
Willy Tarreau08d5f982010-06-06 13:34:54 +020011337 <data_type> is used to store additional information in the stick-table. This
11338 may be used by ACLs in order to control various criteria related
11339 to the activity of the client matching the stick-table. For each
11340 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +020011341 that the additional data can fit. Several data types may be
11342 stored with an entry. Multiple data types may be specified after
11343 the "store" keyword, as a comma-separated list. Alternatively,
11344 it is possible to repeat the "store" keyword followed by one or
11345 several data types. Except for the "server_id" type which is
11346 automatically detected and enabled, all data types must be
11347 explicitly declared to be stored. If an ACL references a data
11348 type which is not stored, the ACL will simply not match. Some
11349 data types require an argument which must be passed just after
11350 the type between parenthesis. See below for the supported data
11351 types and their arguments.
11352
11353 The data types that can be stored with an entry are the following :
11354 - server_id : this is an integer which holds the numeric ID of the server a
11355 request was assigned to. It is used by the "stick match", "stick store",
11356 and "stick on" rules. It is automatically enabled when referenced.
11357
11358 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
11359 integer which may be used for anything. Most of the time it will be used
11360 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +010011361 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +020011362
Willy Tarreauba2ffd12013-05-29 15:54:14 +020011363 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
11364 over a period. It is a positive 32-bit integer integer which may be used
11365 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +010011366 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +020011367 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +010011368 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +020011369
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011370 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
11371 integer which may be used for anything. Most of the time it will be used
11372 to put a special tag on some entries, for instance to note that a
11373 specific behavior was detected and must be known for future matches.
11374
11375 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
11376 over a period. It is a positive 32-bit integer integer which may be used
11377 for anything. Just like <gpc1>, it counts events, but instead of keeping
11378 a cumulative number, it maintains the rate at which the counter is
11379 incremented. Most of the time it will be used to measure the frequency of
11380 occurrence of certain events (e.g. requests to a specific URL).
11381
Emeric Bruna5d15312021-07-01 18:34:48 +020011382 - gpt0 : first General Purpose Tag. It is a positive 32-bit integer
11383 integer which may be used for anything. Most of the time it will be used
11384 to put a special tag on some entries, for instance to note that a
11385 specific behavior was detected and must be known for future matches
11386
Willy Tarreauc9705a12010-07-27 20:05:50 +020011387 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
11388 the absolute number of connections received from clients which matched
11389 this entry. It does not mean the connections were accepted, just that
11390 they were received.
11391
11392 - conn_cur : Current Connections. It is a positive 32-bit integer which
11393 stores the concurrent connection counts for the entry. It is incremented
11394 once an incoming connection matches the entry, and decremented once the
11395 connection leaves. That way it is possible to know at any time the exact
11396 number of concurrent connections for an entry.
11397
11398 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11399 integer parameter <period> which indicates in milliseconds the length
11400 of the period over which the average is measured. It reports the average
11401 incoming connection rate over that period, in connections per period. The
11402 result is an integer which can be matched using ACLs.
11403
11404 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
11405 the absolute number of sessions received from clients which matched this
11406 entry. A session is a connection that was accepted by the layer 4 rules.
11407
11408 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11409 integer parameter <period> which indicates in milliseconds the length
11410 of the period over which the average is measured. It reports the average
11411 incoming session rate over that period, in sessions per period. The
11412 result is an integer which can be matched using ACLs.
11413
11414 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
11415 counts the absolute number of HTTP requests received from clients which
11416 matched this entry. It does not matter whether they are valid requests or
11417 not. Note that this is different from sessions when keep-alive is used on
11418 the client side.
11419
11420 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11421 integer parameter <period> which indicates in milliseconds the length
11422 of the period over which the average is measured. It reports the average
11423 HTTP request rate over that period, in requests per period. The result is
11424 an integer which can be matched using ACLs. It does not matter whether
11425 they are valid requests or not. Note that this is different from sessions
11426 when keep-alive is used on the client side.
11427
11428 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
11429 counts the absolute number of HTTP requests errors induced by clients
11430 which matched this entry. Errors are counted on invalid and truncated
11431 requests, as well as on denied or tarpitted requests, and on failed
11432 authentications. If the server responds with 4xx, then the request is
11433 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +010011434 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +020011435
11436 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11437 integer parameter <period> which indicates in milliseconds the length
11438 of the period over which the average is measured. It reports the average
11439 HTTP request error rate over that period, in requests per period (see
11440 http_err_cnt above for what is accounted as an error). The result is an
11441 integer which can be matched using ACLs.
11442
Willy Tarreau826f3ab2021-02-10 12:07:15 +010011443 - http_fail_cnt : HTTP Failure Count. It is a positive 32-bit integer which
11444 counts the absolute number of HTTP response failures induced by servers
11445 which matched this entry. Errors are counted on invalid and truncated
11446 responses, as well as any 5xx response other than 501 or 505. It aims at
11447 being used combined with path or URI to detect service failures.
11448
11449 - http_fail_rate(<period>) : frequency counter (takes 12 bytes). It takes
11450 an integer parameter <period> which indicates in milliseconds the length
11451 of the period over which the average is measured. It reports the average
11452 HTTP response failure rate over that period, in requests per period (see
11453 http_fail_cnt above for what is accounted as a failure). The result is an
11454 integer which can be matched using ACLs.
11455
Willy Tarreauc9705a12010-07-27 20:05:50 +020011456 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010011457 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +020011458 which matched this entry. Headers are included in the count. This may be
11459 used to limit abuse of upload features on photo or video servers.
11460
11461 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11462 integer parameter <period> which indicates in milliseconds the length
11463 of the period over which the average is measured. It reports the average
11464 incoming bytes rate over that period, in bytes per period. It may be used
11465 to detect users which upload too much and too fast. Warning: with large
11466 uploads, it is possible that the amount of uploaded data will be counted
11467 once upon termination, thus causing spikes in the average transfer speed
11468 instead of having a smooth one. This may partially be smoothed with
11469 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
11470 recommended for better fairness.
11471
11472 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010011473 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +020011474 matched this entry. Headers are included in the count. This may be used
11475 to limit abuse of bots sucking the whole site.
11476
11477 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
11478 an integer parameter <period> which indicates in milliseconds the length
11479 of the period over which the average is measured. It reports the average
11480 outgoing bytes rate over that period, in bytes per period. It may be used
11481 to detect users which download too much and too fast. Warning: with large
11482 transfers, it is possible that the amount of transferred data will be
11483 counted once upon termination, thus causing spikes in the average
11484 transfer speed instead of having a smooth one. This may partially be
11485 smoothed with "option contstats" though this is not perfect yet. Use of
11486 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +020011487
Willy Tarreauc00cdc22010-06-06 16:48:26 +020011488 There is only one stick-table per proxy. At the moment of writing this doc,
11489 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011490 to be required, simply create a dummy backend with a stick-table in it and
11491 reference it.
11492
11493 It is important to understand that stickiness based on learning information
11494 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +010011495 lost upon restart unless peers are properly configured to transfer such
11496 information upon restart (recommended). In general it can be good as a
11497 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011498
Willy Tarreauc9705a12010-07-27 20:05:50 +020011499 Last, memory requirements may be important when storing many data types.
11500 Indeed, storing all indicators above at once in each entry requires 116 bytes
11501 per entry, or 116 MB for a 1-million entries table. This is definitely not
11502 something that can be ignored.
11503
11504 Example:
11505 # Keep track of counters of up to 1 million IP addresses over 5 minutes
11506 # and store a general purpose counter and the average connection rate
11507 # computed over a sliding window of 30 seconds.
11508 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
11509
Willy Tarreau4b103022021-02-12 17:59:10 +010011510 See also : "stick match", "stick on", "stick store-request", section 2.5
David du Colombiera13d1b92011-03-17 10:40:22 +010011511 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011512
11513
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011514stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +010011515 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011516 May be used in sections : defaults | frontend | listen | backend
11517 no | no | yes | yes
11518
11519 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011520 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011521 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +010011522 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011523 server is selected.
11524
11525 <table> is an optional stickiness table name. If unspecified, the same
11526 backend's table is used. A stickiness table is declared using
11527 the "stick-table" statement.
11528
11529 <cond> is an optional storage condition. It makes it possible to store
11530 certain criteria only when some conditions are met (or not met).
11531 For instance, it could be used to store the SSL session ID only
11532 when the response is a SSL server hello.
11533
11534 Some protocols or applications require complex stickiness rules and cannot
11535 always simply rely on cookies nor hashing. The "stick store-response"
11536 statement describes a rule to decide what to extract from the response and
11537 when to do it, in order to store it into a stickiness table for further
11538 requests to match it using the "stick match" statement. Obviously the
11539 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011540 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011541 See section 7 for a complete list of possible patterns and transformation
11542 rules.
11543
11544 The table has to be declared using the "stick-table" statement. It must be of
11545 a type compatible with the pattern. By default it is the one which is present
11546 in the same backend. It is possible to share a table with other backends by
11547 referencing it using the "table" keyword. If another table is referenced,
11548 the server's ID inside the backends are used. By default, all server IDs
11549 start at 1 in each backend, so the server ordering is enough. But in case of
11550 doubt, it is highly recommended to force server IDs using their "id" setting.
11551
11552 It is possible to restrict the conditions where a "stick store-response"
11553 statement will apply, using "if" or "unless" followed by a condition. This
11554 condition will be evaluated while parsing the response, so any criteria can
11555 be used. See section 7 for ACL based conditions.
11556
11557 There is no limit on the number of "stick store-response" statements, but
11558 there is a limit of 8 simultaneous stores per request or response. This
11559 makes it possible to store up to 8 criteria, all extracted from either the
11560 request or the response, regardless of the number of rules. Only the 8 first
11561 ones which match will be kept. Using this, it is possible to feed multiple
11562 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010011563 another protocol or access method. Using multiple store-response rules with
11564 the same table is possible and may be used to find the best criterion to rely
11565 on, by arranging the rules by decreasing preference order. Only the first
11566 extracted criterion for a given table will be stored. All subsequent store-
11567 response rules referencing the same table will be skipped and their ACLs will
11568 not be evaluated. However, even if a store-request rule references a table, a
11569 store-response rule may also use the same table. This means that each table
11570 may learn exactly one element from the request and one element from the
11571 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011572
11573 The table will contain the real server that processed the request.
11574
11575 Example :
11576 # Learn SSL session ID from both request and response and create affinity.
11577 backend https
11578 mode tcp
11579 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +020011580 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011581 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011582
Christian Ruppertfd3184c2022-02-20 22:54:01 +010011583 acl clienthello req.ssl_hello_type 1
11584 acl serverhello rep.ssl_hello_type 2
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011585
11586 # use tcp content accepts to detects ssl client and server hello.
11587 tcp-request inspect-delay 5s
11588 tcp-request content accept if clienthello
11589
11590 # no timeout on response inspect delay by default.
11591 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011592
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011593 # SSL session ID (SSLID) may be present on a client or server hello.
11594 # Its length is coded on 1 byte at offset 43 and its value starts
11595 # at offset 44.
11596
11597 # Match and learn on request if client hello.
Christian Ruppertfd3184c2022-02-20 22:54:01 +010011598 stick on req.payload_lv(43,1) if clienthello
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011599
11600 # Learn on response if server hello.
Christian Ruppertfd3184c2022-02-20 22:54:01 +010011601 stick store-response resp.payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +020011602
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011603 server s1 192.168.1.1:443
11604 server s2 192.168.1.1:443
11605
11606 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
11607 extraction.
11608
11609
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011610tcp-check comment <string>
11611 Defines a comment for the following the tcp-check rule, reported in logs if
11612 it fails.
11613 May be used in sections : defaults | frontend | listen | backend
11614 yes | no | yes | yes
11615
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011616 Arguments :
11617 <string> is the comment message to add in logs if the following tcp-check
11618 rule fails.
11619
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011620 It only works for connect, send and expect rules. It is useful to make
11621 user-friendly error reporting.
11622
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011623 See also : "option tcp-check", "tcp-check connect", "tcp-check send" and
11624 "tcp-check expect".
11625
11626
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011627tcp-check connect [default] [port <expr>] [addr <ip>] [send-proxy] [via-socks4]
11628 [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +020011629 [proto <name>] [comment <msg>]
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011630 Opens a new connection
11631 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011632 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011633
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011634 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011635 comment <msg> defines a message to report if the rule evaluation fails.
11636
Christopher Faulet4dce5922020-03-30 13:54:42 +020011637 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -040011638 checks. The server options are used only if not redefined.
Christopher Faulet4dce5922020-03-30 13:54:42 +020011639
Christopher Fauletb7d30092020-03-30 15:19:03 +020011640 port <expr> if not set, check port or server port is used.
Christopher Faulet5c288742020-03-31 08:15:58 +020011641 It tells HAProxy where to open the connection to.
11642 <port> must be a valid TCP port source integer, from 1 to
Christopher Fauletb7d30092020-03-30 15:19:03 +020011643 65535 or an sample-fetch expression.
Christopher Faulet5c288742020-03-31 08:15:58 +020011644
11645 addr <ip> defines the IP address to do the health check.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011646
11647 send-proxy send a PROXY protocol string
11648
Christopher Faulet085426a2020-03-30 13:07:02 +020011649 via-socks4 enables outgoing health checks using upstream socks4 proxy.
11650
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011651 ssl opens a ciphered connection
11652
Christopher Faulet79b31d42020-03-30 13:00:05 +020011653 sni <sni> specifies the SNI to use to do health checks over SSL.
11654
Christopher Faulet98572322020-03-30 13:16:44 +020011655 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
11656 list consists in a comma-delimited list of protocol names,
11657 for instance: "http/1.1,http/1.0" (without quotes).
11658 If it is not set, the server ALPN is used.
11659
Christopher Fauletedc6ed92020-04-23 16:27:59 +020011660 proto <name> forces the multiplexer's protocol to use for this connection.
11661 It must be a TCP mux protocol and it must be usable on the
11662 backend side. The list of available protocols is reported in
11663 haproxy -vv.
11664
Christopher Faulet5c288742020-03-31 08:15:58 +020011665 linger cleanly close the connection instead of using a single RST.
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010011666
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011667 When an application lies on more than a single TCP port or when HAProxy
11668 load-balance many services in a single backend, it makes sense to probe all
11669 the services individually before considering a server as operational.
11670
11671 When there are no TCP port configured on the server line neither server port
11672 directive, then the 'tcp-check connect port <port>' must be the first step
11673 of the sequence.
11674
11675 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
11676 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
11677 do.
11678
11679 When a connect must start the ruleset, if may still be preceded by set-var,
11680 unset-var or comment rules.
11681
11682 Examples :
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011683 # check HTTP and HTTPs services on a server.
11684 # first open port 80 thanks to server line port directive, then
11685 # tcp-check opens port 443, ciphered and run a request on it:
11686 option tcp-check
11687 tcp-check connect
11688 tcp-check send GET\ /\ HTTP/1.0\r\n
11689 tcp-check send Host:\ haproxy.1wt.eu\r\n
11690 tcp-check send \r\n
11691 tcp-check expect rstring (2..|3..)
11692 tcp-check connect port 443 ssl
11693 tcp-check send GET\ /\ HTTP/1.0\r\n
11694 tcp-check send Host:\ haproxy.1wt.eu\r\n
11695 tcp-check send \r\n
11696 tcp-check expect rstring (2..|3..)
11697 server www 10.0.0.1 check port 80
11698
11699 # check both POP and IMAP from a single server:
11700 option tcp-check
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010011701 tcp-check connect port 110 linger
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011702 tcp-check expect string +OK\ POP3\ ready
11703 tcp-check connect port 143
11704 tcp-check expect string *\ OK\ IMAP4\ ready
11705 server mail 10.0.0.1 check
11706
11707 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
11708
11709
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011710tcp-check expect [min-recv <int>] [comment <msg>]
Christopher Fauletec07e382020-04-07 14:56:26 +020011711 [ok-status <st>] [error-status <st>] [tout-status <st>]
Christopher Faulet98cc57c2020-04-01 20:52:31 +020011712 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011713 [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +010011714 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011715 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011716 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011717
11718 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011719 comment <msg> defines a message to report if the rule evaluation fails.
11720
Gaetan Rivet1afd8262020-02-07 15:37:17 +010011721 min-recv is optional and can define the minimum amount of data required to
11722 evaluate the current expect rule. If the number of received bytes
11723 is under this limit, the check will wait for more data. This
11724 option can be used to resolve some ambiguous matching rules or to
11725 avoid executing costly regex matches on content known to be still
11726 incomplete. If an exact string (string or binary) is used, the
11727 minimum between the string length and this parameter is used.
11728 This parameter is ignored if it is set to -1. If the expect rule
11729 does not match, the check will wait for more data. If set to 0,
11730 the evaluation result is always conclusive.
11731
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011732 <match> is a keyword indicating how to look for a specific pattern in the
Gaetan Rivetefab6c62020-02-07 15:37:17 +010011733 response. The keyword may be one of "string", "rstring", "binary" or
11734 "rbinary".
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011735 The keyword may be preceded by an exclamation mark ("!") to negate
11736 the match. Spaces are allowed between the exclamation mark and the
11737 keyword. See below for more details on the supported keywords.
11738
Christopher Fauletec07e382020-04-07 14:56:26 +020011739 ok-status <st> is optional and can be used to set the check status if
11740 the expect rule is successfully evaluated and if it is
11741 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011742 "L6OK" and "L4OK" are supported :
11743 - L7OK : check passed on layer 7
Christopher Faulet83662b52020-11-20 17:47:47 +010011744 - L7OKC : check conditionally passed on layer 7, set
11745 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011746 - L6OK : check passed on layer 6
11747 - L4OK : check passed on layer 4
Christopher Fauletec07e382020-04-07 14:56:26 +020011748 By default "L7OK" is used.
11749
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011750 error-status <st> is optional and can be used to set the check status if
11751 an error occurred during the expect rule evaluation.
Christopher Faulet83662b52020-11-20 17:47:47 +010011752 "L7OKC", "L7RSP", "L7STS", "L6RSP" and "L4CON" are
11753 supported :
11754 - L7OKC : check conditionally passed on layer 7, set
11755 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011756 - L7RSP : layer 7 invalid response - protocol error
11757 - L7STS : layer 7 response error, for example HTTP 5xx
11758 - L6RSP : layer 6 invalid response - protocol error
11759 - L4CON : layer 1-4 connection problem
11760 By default "L7RSP" is used.
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011761
Christopher Fauletec07e382020-04-07 14:56:26 +020011762 tout-status <st> is optional and can be used to set the check status if
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011763 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011764 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
11765 - L7TOUT : layer 7 (HTTP/SMTP) timeout
11766 - L6TOUT : layer 6 (SSL) timeout
11767 - L4TOUT : layer 1-4 timeout
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011768 By default "L7TOUT" is used.
11769
Christopher Fauletbe52b4d2020-04-01 16:30:22 +020011770 on-success <fmt> is optional and can be used to customize the
11771 informational message reported in logs if the expect
11772 rule is successfully evaluated and if it is the last rule
11773 in the tcp-check ruleset. <fmt> is a log-format string.
11774
11775 on-error <fmt> is optional and can be used to customize the
11776 informational message reported in logs if an error
11777 occurred during the expect rule evaluation. <fmt> is a
11778 log-format string.
11779
Christopher Faulet98cc57c2020-04-01 20:52:31 +020011780 status-code <expr> is optional and can be used to set the check status code
11781 reported in logs, on success or on error. <expr> is a
11782 standard HAProxy expression formed by a sample-fetch
11783 followed by some converters.
11784
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011785 <pattern> is the pattern to look for. It may be a string or a regular
11786 expression. If the pattern contains spaces, they must be escaped
11787 with the usual backslash ('\').
11788 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +010011789 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011790 two digits will represent a byte. The hexadecimal digits may be
11791 used upper or lower case.
11792
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011793 The available matches are intentionally similar to their http-check cousins :
11794
11795 string <string> : test the exact string matches in the response buffer.
11796 A health check response will be considered valid if the
11797 response's buffer contains this exact string. If the
11798 "string" keyword is prefixed with "!", then the response
11799 will be considered invalid if the body contains this
11800 string. This can be used to look for a mandatory pattern
11801 in a protocol response, or to detect a failure when a
11802 specific error appears in a protocol banner.
11803
11804 rstring <regex> : test a regular expression on the response buffer.
11805 A health check response will be considered valid if the
11806 response's buffer matches this expression. If the
11807 "rstring" keyword is prefixed with "!", then the response
11808 will be considered invalid if the body matches the
11809 expression.
11810
Christopher Fauletaaab0832020-05-05 15:54:22 +020011811 string-lf <fmt> : test a log-format string match in the response's buffer.
11812 A health check response will be considered valid if the
11813 response's buffer contains the string resulting of the
11814 evaluation of <fmt>, which follows the log-format rules.
11815 If prefixed with "!", then the response will be
11816 considered invalid if the buffer contains the string.
11817
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011818 binary <hexstring> : test the exact string in its hexadecimal form matches
11819 in the response buffer. A health check response will
11820 be considered valid if the response's buffer contains
11821 this exact hexadecimal string.
11822 Purpose is to match data on binary protocols.
11823
Gaetan Rivetefab6c62020-02-07 15:37:17 +010011824 rbinary <regex> : test a regular expression on the response buffer, like
11825 "rstring". However, the response buffer is transformed
11826 into its hexadecimal form, including NUL-bytes. This
11827 allows using all regex engines to match any binary
11828 content. The hexadecimal transformation takes twice the
11829 size of the original response. As such, the expected
11830 pattern should work on at-most half the response buffer
11831 size.
11832
Christopher Fauletaaab0832020-05-05 15:54:22 +020011833 binary-lf <hexfmt> : test a log-format string in its hexadecimal form
11834 match in the response's buffer. A health check response
11835 will be considered valid if the response's buffer
11836 contains the hexadecimal string resulting of the
11837 evaluation of <fmt>, which follows the log-format
11838 rules. If prefixed with "!", then the response will be
11839 considered invalid if the buffer contains the
11840 hexadecimal string. The hexadecimal string is converted
11841 in a binary string before matching the response's
11842 buffer.
11843
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011844 It is important to note that the responses will be limited to a certain size
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011845 defined by the global "tune.bufsize" option, which defaults to 16384 bytes.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011846 Thus, too large responses may not contain the mandatory pattern when using
11847 "string", "rstring" or binary. If a large response is absolutely required, it
11848 is possible to change the default max size by setting the global variable.
11849 However, it is worth keeping in mind that parsing very large responses can
11850 waste some CPU cycles, especially when regular expressions are used, and that
11851 it is always better to focus the checks on smaller resources. Also, in its
11852 current state, the check will not find any string nor regex past a null
11853 character in the response. Similarly it is not possible to request matching
11854 the null character.
11855
11856 Examples :
11857 # perform a POP check
11858 option tcp-check
11859 tcp-check expect string +OK\ POP3\ ready
11860
11861 # perform an IMAP check
11862 option tcp-check
11863 tcp-check expect string *\ OK\ IMAP4\ ready
11864
11865 # look for the redis master server
11866 option tcp-check
11867 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +020011868 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011869 tcp-check send info\ replication\r\n
11870 tcp-check expect string role:master
11871 tcp-check send QUIT\r\n
11872 tcp-check expect string +OK
11873
11874
11875 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011876 "tcp-check send-binary", "http-check expect", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011877
11878
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011879tcp-check send <data> [comment <msg>]
11880tcp-check send-lf <fmt> [comment <msg>]
11881 Specify a string or a log-format string to be sent as a question during a
11882 generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011883 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011884 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011885
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011886 Arguments :
11887 comment <msg> defines a message to report if the rule evaluation fails.
11888
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011889 <data> is the string that will be sent during a generic health
11890 check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020011891
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011892 <fmt> is the log-format string that will be sent, once evaluated,
11893 during a generic health check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011894
11895 Examples :
11896 # look for the redis master server
11897 option tcp-check
11898 tcp-check send info\ replication\r\n
11899 tcp-check expect string role:master
11900
11901 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011902 "tcp-check send-binary", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011903
11904
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011905tcp-check send-binary <hexstring> [comment <msg>]
11906tcp-check send-binary-lf <hexfmt> [comment <msg>]
11907 Specify an hex digits string or an hex digits log-format string to be sent as
11908 a binary question during a raw tcp health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011909 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011910 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011911
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011912 Arguments :
11913 comment <msg> defines a message to report if the rule evaluation fails.
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011914
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011915 <hexstring> is the hexadecimal string that will be send, once converted
11916 to binary, during a generic health check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020011917
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011918 <hexfmt> is the hexadecimal log-format string that will be send, once
11919 evaluated and converted to binary, during a generic health
11920 check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011921
11922 Examples :
11923 # redis check in binary
11924 option tcp-check
11925 tcp-check send-binary 50494e470d0a # PING\r\n
11926 tcp-check expect binary 2b504F4e47 # +PONG
11927
11928
11929 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011930 "tcp-check send", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011931
11932
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011933tcp-check set-var(<var-name>) <expr>
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011934 This operation sets the content of a variable. The variable is declared inline.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011935 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011936 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011937
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011938 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011939 <var-name> The name of the variable starts with an indication about its
11940 scope. The scopes allowed for tcp-check are:
11941 "proc" : the variable is shared with the whole process.
11942 "sess" : the variable is shared with the tcp-check session.
11943 "check": the variable is declared for the lifetime of the tcp-check.
11944 This prefix is followed by a name. The separator is a '.'.
11945 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11946 and '-'.
11947
11948 <expr> Is a sample-fetch expression potentially followed by converters.
11949
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011950 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011951 tcp-check set-var(check.port) int(1234)
11952
11953
11954tcp-check unset-var(<var-name>)
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011955 Free a reference to a variable within its scope.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011956 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011957 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011958
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011959 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011960 <var-name> The name of the variable starts with an indication about its
11961 scope. The scopes allowed for tcp-check are:
11962 "proc" : the variable is shared with the whole process.
11963 "sess" : the variable is shared with the tcp-check session.
11964 "check": the variable is declared for the lifetime of the tcp-check.
11965 This prefix is followed by a name. The separator is a '.'.
11966 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11967 and '-'.
11968
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011969 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011970 tcp-check unset-var(check.port)
11971
11972
Willy Tarreaue9656522010-08-17 15:40:09 +020011973tcp-request connection <action> [{if | unless} <condition>]
11974 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +020011975 May be used in sections : defaults | frontend | listen | backend
11976 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +020011977 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011978 <action> defines the action to perform if the condition applies. See
11979 below.
Willy Tarreau1a687942010-05-23 22:40:30 +020011980
Willy Tarreaue9656522010-08-17 15:40:09 +020011981 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011982
11983 Immediately after acceptance of a new incoming connection, it is possible to
11984 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +020011985 or dropped or have its counters tracked. Those conditions cannot make use of
11986 any data contents because the connection has not been read from yet, and the
11987 buffers are not yet allocated. This is used to selectively and very quickly
11988 accept or drop connections from various sources with a very low overhead. If
11989 some contents need to be inspected in order to take the decision, the
11990 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011991
Willy Tarreaue9656522010-08-17 15:40:09 +020011992 The "tcp-request connection" rules are evaluated in their exact declaration
11993 order. If no rule matches or if there is no rule, the default action is to
11994 accept the incoming connection. There is no specific limit to the number of
11995 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011996
Willy Tarreaua9083d02015-05-08 15:27:59 +020011997 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +020011998 - accept :
11999 accepts the connection if the condition is true (when used with "if")
12000 or false (when used with "unless"). The first such rule executed ends
12001 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012002
Willy Tarreaue9656522010-08-17 15:40:09 +020012003 - reject :
12004 rejects the connection if the condition is true (when used with "if")
12005 or false (when used with "unless"). The first such rule executed ends
12006 the rules evaluation. Rejected connections do not even become a
12007 session, which is why they are accounted separately for in the stats,
12008 as "denied connections". They are not considered for the session
12009 rate-limit and are not logged either. The reason is that these rules
12010 should only be used to filter extremely high connection rates such as
12011 the ones encountered during a massive DDoS attack. Under these extreme
12012 conditions, the simple action of logging each event would make the
12013 system collapse and would considerably lower the filtering capacity. If
12014 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +020012015 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012016
Willy Tarreau4f0d9192013-06-11 20:40:55 +020012017 - expect-proxy layer4 :
12018 configures the client-facing connection to receive a PROXY protocol
12019 header before any byte is read from the socket. This is equivalent to
12020 having the "accept-proxy" keyword on the "bind" line, except that using
12021 the TCP rule allows the PROXY protocol to be accepted only for certain
12022 IP address ranges using an ACL. This is convenient when multiple layers
12023 of load balancers are passed through by traffic coming from public
12024 hosts.
12025
Bertrand Jacquin90759682016-06-06 15:35:39 +010012026 - expect-netscaler-cip layer4 :
12027 configures the client-facing connection to receive a NetScaler Client
12028 IP insertion protocol header before any byte is read from the socket.
12029 This is equivalent to having the "accept-netscaler-cip" keyword on the
12030 "bind" line, except that using the TCP rule allows the PROXY protocol
12031 to be accepted only for certain IP address ranges using an ACL. This
12032 is convenient when multiple layers of load balancers are passed
12033 through by traffic coming from public hosts.
12034
Willy Tarreau18bf01e2014-06-13 16:18:52 +020012035 - capture <sample> len <length> :
12036 This only applies to "tcp-request content" rules. It captures sample
12037 expression <sample> from the request buffer, and converts it to a
12038 string of at most <len> characters. The resulting string is stored into
12039 the next request "capture" slot, so it will possibly appear next to
12040 some captured HTTP headers. It will then automatically appear in the
12041 logs, and it will be possible to extract it using sample fetch rules to
12042 feed it into headers or anything. The length should be limited given
12043 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +020012044 session life. Please check section 7.3 (Fetching samples) and "capture
12045 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +020012046
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012047 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +020012048 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020012049 rules do not stop evaluation and do not change default action. The
12050 number of counters that may be simultaneously tracked by the same
12051 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -050012052 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Matteo Contrini1857b8c2020-10-16 17:35:54 +020012053 and (MAX_SESS_STKCTR-1). The first "track-sc0" rule executed enables
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020012054 tracking of the counters of the specified table as the first set. The
12055 first "track-sc1" rule executed enables tracking of the counters of the
12056 specified table as the second set. The first "track-sc2" rule executed
12057 enables tracking of the counters of the specified table as the third
12058 set. It is a recommended practice to use the first set of counters for
12059 the per-frontend counters and the second set for the per-backend ones.
12060 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012061
Willy Tarreaue9656522010-08-17 15:40:09 +020012062 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020012063 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +020012064 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +010012065 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012066 and used to select which table entry to update the counters.
12067 Note that "tcp-request connection" cannot use content-based
12068 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012069
Willy Tarreaue9656522010-08-17 15:40:09 +020012070 <table> is an optional table to be used instead of the default one,
12071 which is the stick-table declared in the current proxy. All
12072 the counters for the matches and updates for the key will
12073 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012074
Willy Tarreaue9656522010-08-17 15:40:09 +020012075 Once a "track-sc*" rule is executed, the key is looked up in the table
12076 and if it is not found, an entry is allocated for it. Then a pointer to
12077 that entry is kept during all the session's life, and this entry's
12078 counters are updated as often as possible, every time the session's
12079 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012080 Counters are only updated for events that happen after the tracking has
12081 been started. For example, connection counters will not be updated when
12082 tracking layer 7 information, since the connection event happens before
12083 layer7 information is extracted.
12084
Willy Tarreaue9656522010-08-17 15:40:09 +020012085 If the entry tracks concurrent connection counters, one connection is
12086 counted for as long as the entry is tracked, and the entry will not
12087 expire during that time. Tracking counters also provides a performance
12088 advantage over just checking the keys, because only one table lookup is
12089 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012090
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020012091 - sc-inc-gpc0(<sc-id>):
12092 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
12093 counter designated by <sc-id>. If an error occurs, this action silently
12094 fails and the actions evaluation continues.
12095
Frédéric Lécaille6778b272018-01-29 15:22:53 +010012096 - sc-inc-gpc1(<sc-id>):
12097 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
12098 counter designated by <sc-id>. If an error occurs, this action silently
12099 fails and the actions evaluation continues.
12100
Cédric Dufour0d7712d2019-11-06 18:38:53 +010012101 - sc-set-gpt0(<sc-id>) { <int> | <expr> }:
12102 This action sets the 32-bit unsigned GPT0 tag according to the sticky
12103 counter designated by <sc-id> and the value of <int>/<expr>. The
12104 expected result is a boolean. If an error occurs, this action silently
12105 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020012106
William Lallemand2e785f22016-05-25 01:48:42 +020012107 - set-src <expr> :
12108 Is used to set the source IP address to the value of specified
12109 expression. Useful if you want to mask source IP for privacy.
12110 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020012111 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +020012112
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020012113 Arguments:
12114 <expr> Is a standard HAProxy expression formed by a sample-fetch
12115 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +020012116
12117 Example:
William Lallemand2e785f22016-05-25 01:48:42 +020012118 tcp-request connection set-src src,ipmask(24)
12119
Willy Tarreau0c630532016-10-21 17:52:58 +020012120 When possible, set-src preserves the original source port as long as the
12121 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +020012122
William Lallemand44be6402016-05-25 01:51:35 +020012123 - set-src-port <expr> :
12124 Is used to set the source port address to the value of specified
12125 expression.
12126
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020012127 Arguments:
12128 <expr> Is a standard HAProxy expression formed by a sample-fetch
12129 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +020012130
12131 Example:
William Lallemand44be6402016-05-25 01:51:35 +020012132 tcp-request connection set-src-port int(4000)
12133
Willy Tarreau0c630532016-10-21 17:52:58 +020012134 When possible, set-src-port preserves the original source address as long
12135 as the address family supports a port, otherwise it forces the source
12136 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +020012137
William Lallemand13e9b0c2016-05-25 02:34:07 +020012138 - set-dst <expr> :
12139 Is used to set the destination IP address to the value of specified
12140 expression. Useful if you want to mask IP for privacy in log.
12141 If you want to provide an IP from a HTTP header use "http-request
12142 set-dst". If you want to connect to the new address/port, use
12143 '0.0.0.0:0' as a server address in the backend.
12144
12145 <expr> Is a standard HAProxy expression formed by a sample-fetch
12146 followed by some converters.
12147
12148 Example:
12149
12150 tcp-request connection set-dst dst,ipmask(24)
12151 tcp-request connection set-dst ipv4(10.0.0.1)
12152
Willy Tarreau0c630532016-10-21 17:52:58 +020012153 When possible, set-dst preserves the original destination port as long as
12154 the address family allows it, otherwise the destination port is set to 0.
12155
William Lallemand13e9b0c2016-05-25 02:34:07 +020012156 - set-dst-port <expr> :
12157 Is used to set the destination port address to the value of specified
12158 expression. If you want to connect to the new address/port, use
12159 '0.0.0.0:0' as a server address in the backend.
12160
12161
12162 <expr> Is a standard HAProxy expression formed by a sample-fetch
12163 followed by some converters.
12164
12165 Example:
12166
12167 tcp-request connection set-dst-port int(4000)
12168
Willy Tarreau0c630532016-10-21 17:52:58 +020012169 When possible, set-dst-port preserves the original destination address as
12170 long as the address family supports a port, otherwise it forces the
12171 destination address to IPv4 "0.0.0.0" before rewriting the port.
12172
Willy Tarreau2d392c22015-08-24 01:43:45 +020012173 - "silent-drop" :
12174 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012175 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020012176 to prevent the client from being notified. The effect it then that the
12177 client still sees an established connection while there's none on
12178 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
12179 except that it doesn't use any local resource at all on the machine
12180 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010012181 slow down stronger attackers. It is important to understand the impact
12182 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020012183 client and HAProxy (firewalls, proxies, load balancers) will also keep
12184 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012185 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020012186 TCP_REPAIR socket option is used to block the emission of a TCP
12187 reset. On other systems, the socket's TTL is reduced to 1 so that the
12188 TCP reset doesn't pass the first router, though it's still delivered to
12189 local networks. Do not use it unless you fully understand how it works.
12190
Willy Tarreaue9656522010-08-17 15:40:09 +020012191 Note that the "if/unless" condition is optional. If no condition is set on
12192 the action, it is simply performed unconditionally. That can be useful for
12193 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012194
Willy Tarreaue9656522010-08-17 15:40:09 +020012195 Example: accept all connections from white-listed hosts, reject too fast
12196 connection without counting them, and track accepted connections.
12197 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012198
Willy Tarreaue9656522010-08-17 15:40:09 +020012199 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012200 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012201 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012202
Willy Tarreaue9656522010-08-17 15:40:09 +020012203 Example: accept all connections from white-listed hosts, count all other
12204 connections and reject too fast ones. This results in abusive ones
12205 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012206
Willy Tarreaue9656522010-08-17 15:40:09 +020012207 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012208 tcp-request connection track-sc0 src
12209 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012210
Willy Tarreau4f0d9192013-06-11 20:40:55 +020012211 Example: enable the PROXY protocol for traffic coming from all known proxies.
12212
12213 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
12214
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012215 See section 7 about ACL usage.
12216
Willy Tarreau4f614292016-10-21 17:49:36 +020012217 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012218
12219
Willy Tarreaue9656522010-08-17 15:40:09 +020012220tcp-request content <action> [{if | unless} <condition>]
12221 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +020012222 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020012223 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +020012224 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020012225 <action> defines the action to perform if the condition applies. See
12226 below.
Willy Tarreau62644772008-07-16 18:36:06 +020012227
Willy Tarreaue9656522010-08-17 15:40:09 +020012228 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +020012229
Davor Ocelice9ed2812017-12-25 17:49:28 +010012230 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +020012231 called "TCP content inspection". During this stage, ACL-based rules are
12232 evaluated every time the request contents are updated, until either an
Christopher Fauletae863c62021-03-15 12:03:44 +010012233 "accept", a "reject" or a "switch-mode" rule matches, or the TCP request
12234 inspection delay expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +020012235
Willy Tarreaue9656522010-08-17 15:40:09 +020012236 The first difference between these rules and "tcp-request connection" rules
12237 is that "tcp-request content" rules can make use of contents to take a
12238 decision. Most often, these decisions will consider a protocol recognition or
12239 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +010012240 both frontends and backends. In case of HTTP keep-alive with the client, all
Daniel Corbett9f0843f2021-05-08 10:50:37 -040012241 tcp-request content rules are evaluated again, so HAProxy keeps a record of
Willy Tarreauf3338342014-01-28 21:40:28 +010012242 what sticky counters were assigned by a "tcp-request connection" versus a
12243 "tcp-request content" rule, and flushes all the content-related ones after
12244 processing an HTTP request, so that they may be evaluated again by the rules
12245 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012246 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +010012247 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +020012248
Willy Tarreaue9656522010-08-17 15:40:09 +020012249 Content-based rules are evaluated in their exact declaration order. If no
12250 rule matches or if there is no rule, the default action is to accept the
12251 contents. There is no specific limit to the number of rules which may be
12252 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +020012253
Thierry FOURNIER236657b2015-08-19 08:25:14 +020012254 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +020012255 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +010012256 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +020012257 - reject : the request is rejected and the connection is closed
12258 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -040012259 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012260 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020012261 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010012262 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010012263 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020012264 - set-dst <expr>
12265 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012266 - set-var(<var-name>) <expr>
Christopher Fauletae863c62021-03-15 12:03:44 +010012267 - switch-mode http [ proto <name> ]
Christopher Faulet85d79c92016-11-09 16:54:56 +010012268 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +020012269 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +010012270 - send-spoe-group <engine-name> <group-name>
Christopher Faulet579d83b2019-11-22 15:34:17 +010012271 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +020012272
Willy Tarreaue9656522010-08-17 15:40:09 +020012273 They have the same meaning as their counter-parts in "tcp-request connection"
12274 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +010012275 For "do-resolve" action, please check the "http-request do-resolve"
12276 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +020012277
Willy Tarreauf3338342014-01-28 21:40:28 +010012278 While there is nothing mandatory about it, it is recommended to use the
12279 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
12280 content" rules in the frontend, and track-sc2 for "tcp-request content"
12281 rules in the backend, because that makes the configuration more readable
12282 and easier to troubleshoot, but this is just a guideline and all counters
12283 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +020012284
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012285 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +020012286 the action, it is simply performed unconditionally. That can be useful for
12287 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +020012288
Christopher Faulet2079a4a2020-10-02 11:48:57 +020012289 Note also that it is recommended to use a "tcp-request session" rule to track
12290 information that does *not* depend on Layer 7 contents, especially for HTTP
12291 frontends. Some HTTP processing are performed at the session level and may
12292 lead to an early rejection of the requests. Thus, the tracking at the content
12293 level may be disturbed in such case. A warning is emitted during startup to
12294 prevent, as far as possible, such unreliable usage.
12295
Willy Tarreaue9656522010-08-17 15:40:09 +020012296 It is perfectly possible to match layer 7 contents with "tcp-request content"
Christopher Faulet7ea509e2020-10-02 11:38:46 +020012297 rules from a TCP proxy, since HTTP-specific ACL matches are able to
12298 preliminarily parse the contents of a buffer before extracting the required
12299 data. If the buffered contents do not parse as a valid HTTP message, then the
12300 ACL does not match. The parser which is involved there is exactly the same
12301 as for all other HTTP processing, so there is no risk of parsing something
12302 differently. In an HTTP frontend or an HTTP backend, it is guaranteed that
12303 HTTP contents will always be immediately present when the rule is evaluated
12304 first because the HTTP parsing is performed in the early stages of the
12305 connection processing, at the session level. But for such proxies, using
12306 "http-request" rules is much more natural and recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020012307
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012308 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020012309 are present when the rule is processed. The rule processing engine is able to
12310 wait until the inspect delay expires when the data to be tracked is not yet
12311 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012312
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020012313 The "set-dst" and "set-dst-port" are used to set respectively the destination
12314 IP and port. More information on how to use it at "http-request set-dst".
12315
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012316 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +020012317 declared inline. For "tcp-request session" rules, only session-level
12318 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012319
Daniel Schneller0b547052016-03-21 20:46:57 +010012320 <var-name> The name of the variable starts with an indication about
12321 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012322 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012323 "sess" : the variable is shared with the whole session
12324 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012325 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010012326 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012327 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010012328 "res" : the variable is shared only during response
12329 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012330 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010012331 The name may only contain characters 'a-z', 'A-Z', '0-9',
12332 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012333
12334 <expr> Is a standard HAProxy expression formed by a sample-fetch
12335 followed by some converters.
12336
Daniel Corbett9f0843f2021-05-08 10:50:37 -040012337 The "switch-mode" is used to perform a connection upgrade. Only HTTP
Christopher Fauletae863c62021-03-15 12:03:44 +010012338 upgrades are supported for now. The protocol may optionally be
12339 specified. This action is only available for a proxy with the frontend
12340 capability. The connection upgrade is immediately performed, following
12341 "tcp-request content" rules are not evaluated. This upgrade method should be
12342 preferred to the implicit one consisting to rely on the backend mode. When
12343 used, it is possible to set HTTP directives in a frontend without any
Ilya Shipitsin3df59892021-05-10 12:50:00 +050012344 warning. These directives will be conditionally evaluated if the HTTP upgrade
Christopher Fauletae863c62021-03-15 12:03:44 +010012345 is performed. However, an HTTP backend must still be selected. It remains
12346 unsupported to route an HTTP connection (upgraded or not) to a TCP server.
12347
Christopher Faulet4d37e532021-03-26 14:44:00 +010012348 See section 4 about Proxies for more details on HTTP upgrades.
12349
Christopher Faulet85d79c92016-11-09 16:54:56 +010012350 The "unset-var" is used to unset a variable. See above for details about
12351 <var-name>.
12352
Patrick Hemmer268a7072018-05-11 12:52:31 -040012353 The "set-priority-class" is used to set the queue priority class of the
12354 current request. The value must be a sample expression which converts to an
12355 integer in the range -2047..2047. Results outside this range will be
12356 truncated. The priority class determines the order in which queued requests
12357 are processed. Lower values have higher priority.
12358
12359 The "set-priority-offset" is used to set the queue priority timestamp offset
12360 of the current request. The value must be a sample expression which converts
12361 to an integer in the range -524287..524287. Results outside this range will be
12362 truncated. When a request is queued, it is ordered first by the priority
12363 class, then by the current timestamp adjusted by the given offset in
12364 milliseconds. Lower values have higher priority.
12365 Note that the resulting timestamp is is only tracked with enough precision for
12366 524,287ms (8m44s287ms). If the request is queued long enough to where the
12367 adjusted timestamp exceeds this value, it will be misidentified as highest
12368 priority. Thus it is important to set "timeout queue" to a value, where when
12369 combined with the offset, does not exceed this limit.
12370
Christopher Faulet76c09ef2017-09-21 11:03:52 +020012371 The "send-spoe-group" is used to trigger sending of a group of SPOE
12372 messages. To do so, the SPOE engine used to send messages must be defined, as
12373 well as the SPOE group to send. Of course, the SPOE engine must refer to an
12374 existing SPOE filter. If not engine name is provided on the SPOE filter line,
12375 the SPOE agent name must be used.
12376
12377 <engine-name> The SPOE engine name.
12378
12379 <group-name> The SPOE group name as specified in the engine configuration.
12380
Christopher Faulet579d83b2019-11-22 15:34:17 +010012381 The "use-service" is used to executes a TCP service which will reply to the
12382 request and stop the evaluation of the rules. This service may choose to
12383 reply by sending any valid response or it may immediately close the
12384 connection without sending anything. Outside natives services, it is possible
12385 to write your own services in Lua. No further "tcp-request" rules are
12386 evaluated.
12387
12388 Example:
12389 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
12390
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012391 Example:
12392
12393 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +010012394 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012395
Willy Tarreau62644772008-07-16 18:36:06 +020012396 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +020012397 # Accept HTTP requests containing a Host header saying "example.com"
Christopher Fauletae863c62021-03-15 12:03:44 +010012398 # and reject everything else. (Only works for HTTP/1 connections)
Willy Tarreaue9656522010-08-17 15:40:09 +020012399 acl is_host_com hdr(Host) -i example.com
12400 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +020012401 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +020012402 tcp-request content reject
12403
Christopher Fauletae863c62021-03-15 12:03:44 +010012404 # Accept HTTP requests containing a Host header saying "example.com"
12405 # and reject everything else. (works for HTTP/1 and HTTP/2 connections)
12406 acl is_host_com hdr(Host) -i example.com
12407 tcp-request inspect-delay 5s
12408 tcp-request switch-mode http if HTTP
12409 tcp-request reject # non-HTTP traffic is implicit here
12410 ...
12411 http-request reject unless is_host_com
12412
Willy Tarreaue9656522010-08-17 15:40:09 +020012413 Example:
Willy Tarreau62644772008-07-16 18:36:06 +020012414 # reject SMTP connection if client speaks first
12415 tcp-request inspect-delay 30s
Christian Ruppertfd3184c2022-02-20 22:54:01 +010012416 acl content_present req.len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012417 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +020012418
12419 # Forward HTTPS connection only if client speaks
12420 tcp-request inspect-delay 30s
Christian Ruppertfd3184c2022-02-20 22:54:01 +010012421 acl content_present req.len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012422 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +020012423 tcp-request content reject
12424
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012425 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012426 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012427 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020012428 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012429 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
12430 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012431
12432 Example:
12433 # track request counts per "base" (concatenation of Host+URL)
12434 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020012435 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012436
Willy Tarreaue9656522010-08-17 15:40:09 +020012437 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012438 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +020012439
12440 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +010012441 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +020012442 # protecting all our sites
12443 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012444 tcp-request connection track-sc0 src
12445 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +020012446 ...
12447 use_backend http_dynamic if { path_end .php }
12448
12449 backend http_dynamic
12450 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012451 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +020012452 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012453 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012454 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012455 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +020012456 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +020012457
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012458 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +020012459
Jarno Huuskonen95b012b2017-04-06 13:59:14 +030012460 See also : "tcp-request connection", "tcp-request session",
12461 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +020012462
12463
12464tcp-request inspect-delay <timeout>
12465 Set the maximum allowed time to wait for data during content inspection
12466 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020012467 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +020012468 Arguments :
12469 <timeout> is the timeout value specified in milliseconds by default, but
12470 can be in any other unit if the number is suffixed by the unit,
12471 as explained at the top of this document.
12472
Daniel Corbett9f0843f2021-05-08 10:50:37 -040012473 People using HAProxy primarily as a TCP relay are often worried about the
Willy Tarreau62644772008-07-16 18:36:06 +020012474 risk of passing any type of protocol to a server without any analysis. In
12475 order to be able to analyze the request contents, we must first withhold
12476 the data then analyze them. This statement simply enables withholding of
12477 data for at most the specified amount of time.
12478
Willy Tarreaufb356202010-08-03 14:02:05 +020012479 TCP content inspection applies very early when a connection reaches a
12480 frontend, then very early when the connection is forwarded to a backend. This
12481 means that a connection may experience a first delay in the frontend and a
12482 second delay in the backend if both have tcp-request rules.
12483
Daniel Corbett9f0843f2021-05-08 10:50:37 -040012484 Note that when performing content inspection, HAProxy will evaluate the whole
Willy Tarreau62644772008-07-16 18:36:06 +020012485 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012486 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +020012487 a last check is performed upon expiration, this time considering that the
Daniel Corbett9f0843f2021-05-08 10:50:37 -040012488 contents are definitive. If no delay is set, HAProxy will not wait at all
Willy Tarreaud869b242009-03-15 14:43:58 +010012489 and will immediately apply a verdict based on the available information.
12490 Obviously this is unlikely to be very useful and might even be racy, so such
12491 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020012492
12493 As soon as a rule matches, the request is released and continues as usual. If
12494 the timeout is reached and no rule matches, the default policy will be to let
12495 it pass through unaffected.
12496
12497 For most protocols, it is enough to set it to a few seconds, as most clients
12498 send the full request immediately upon connection. Add 3 or more seconds to
12499 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +010012500 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +010012501 before the server (e.g. SMTP), or to wait for a client to talk before passing
12502 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +020012503 least the inspection delay, otherwise it will expire first. If the client
12504 closes the connection or if the buffer is full, the delay immediately expires
12505 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +020012506
Willy Tarreau55165fe2009-05-10 12:02:55 +020012507 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020012508 "timeout client".
12509
12510
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012511tcp-response content <action> [{if | unless} <condition>]
12512 Perform an action on a session response depending on a layer 4-7 condition
12513 May be used in sections : defaults | frontend | listen | backend
12514 no | no | yes | yes
12515 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020012516 <action> defines the action to perform if the condition applies. See
12517 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012518
12519 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
12520
Davor Ocelice9ed2812017-12-25 17:49:28 +010012521 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012522 called "TCP content inspection". During this stage, ACL-based rules are
12523 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020012524 "accept", "close" or a "reject" rule matches, or a TCP response inspection
12525 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012526
12527 Most often, these decisions will consider a protocol recognition or validity.
12528
12529 Content-based rules are evaluated in their exact declaration order. If no
12530 rule matches or if there is no rule, the default action is to accept the
12531 contents. There is no specific limit to the number of rules which may be
12532 inserted.
12533
Thierry FOURNIER236657b2015-08-19 08:25:14 +020012534 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012535 - accept :
12536 accepts the response if the condition is true (when used with "if")
12537 or false (when used with "unless"). The first such rule executed ends
12538 the rules evaluation.
12539
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020012540 - close :
12541 immediately closes the connection with the server if the condition is
12542 true (when used with "if"), or false (when used with "unless"). The
12543 first such rule executed ends the rules evaluation. The main purpose of
12544 this action is to force a connection to be finished between a client
12545 and a server after an exchange when the application protocol expects
12546 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012547 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020012548 protocols.
12549
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012550 - reject :
12551 rejects the response if the condition is true (when used with "if")
12552 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012553 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012554
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012555 - set-var(<var-name>) <expr>
Willy Tarreau9de54ba2021-09-02 20:51:21 +020012556 Sets a variable from an expression.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012557
Christopher Faulet85d79c92016-11-09 16:54:56 +010012558 - unset-var(<var-name>)
12559 Unsets a variable.
12560
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020012561 - sc-inc-gpc0(<sc-id>):
12562 This action increments the GPC0 counter according to the sticky
12563 counter designated by <sc-id>. If an error occurs, this action fails
12564 silently and the actions evaluation continues.
12565
Frédéric Lécaille6778b272018-01-29 15:22:53 +010012566 - sc-inc-gpc1(<sc-id>):
12567 This action increments the GPC1 counter according to the sticky
12568 counter designated by <sc-id>. If an error occurs, this action fails
12569 silently and the actions evaluation continues.
12570
Cédric Dufour0d7712d2019-11-06 18:38:53 +010012571 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
12572 This action sets the 32-bit unsigned GPT0 tag according to the sticky
12573 counter designated by <sc-id> and the value of <int>/<expr>. The
12574 expected result is a boolean. If an error occurs, this action silently
12575 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020012576
Willy Tarreau2d392c22015-08-24 01:43:45 +020012577 - "silent-drop" :
12578 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012579 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020012580 to prevent the client from being notified. The effect it then that the
12581 client still sees an established connection while there's none on
12582 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
12583 except that it doesn't use any local resource at all on the machine
12584 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010012585 slow down stronger attackers. It is important to understand the impact
12586 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020012587 client and HAProxy (firewalls, proxies, load balancers) will also keep
12588 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012589 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020012590 TCP_REPAIR socket option is used to block the emission of a TCP
12591 reset. On other systems, the socket's TTL is reduced to 1 so that the
12592 TCP reset doesn't pass the first router, though it's still delivered to
12593 local networks. Do not use it unless you fully understand how it works.
12594
Christopher Faulet76c09ef2017-09-21 11:03:52 +020012595 - send-spoe-group <engine-name> <group-name>
12596 Send a group of SPOE messages.
12597
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012598 Note that the "if/unless" condition is optional. If no condition is set on
12599 the action, it is simply performed unconditionally. That can be useful for
12600 for changing the default action to a reject.
12601
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012602 It is perfectly possible to match layer 7 contents with "tcp-response
12603 content" rules, but then it is important to ensure that a full response has
12604 been buffered, otherwise no contents will match. In order to achieve this,
12605 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012606 period.
12607
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012608 The "set-var" is used to set the content of a variable. The variable is
12609 declared inline.
12610
Daniel Schneller0b547052016-03-21 20:46:57 +010012611 <var-name> The name of the variable starts with an indication about
12612 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012613 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012614 "sess" : the variable is shared with the whole session
12615 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012616 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010012617 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012618 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010012619 "res" : the variable is shared only during response
12620 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012621 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010012622 The name may only contain characters 'a-z', 'A-Z', '0-9',
12623 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012624
12625 <expr> Is a standard HAProxy expression formed by a sample-fetch
12626 followed by some converters.
12627
Christopher Faulet85d79c92016-11-09 16:54:56 +010012628 The "unset-var" is used to unset a variable. See above for details about
12629 <var-name>.
12630
Christopher Faulet76c09ef2017-09-21 11:03:52 +020012631 The "send-spoe-group" is used to trigger sending of a group of SPOE
12632 messages. To do so, the SPOE engine used to send messages must be defined, as
12633 well as the SPOE group to send. Of course, the SPOE engine must refer to an
12634 existing SPOE filter. If not engine name is provided on the SPOE filter line,
12635 the SPOE agent name must be used.
12636
12637 <engine-name> The SPOE engine name.
12638
12639 <group-name> The SPOE group name as specified in the engine configuration.
12640
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012641 See section 7 about ACL usage.
12642
12643 See also : "tcp-request content", "tcp-response inspect-delay"
12644
12645
Willy Tarreau4f614292016-10-21 17:49:36 +020012646tcp-request session <action> [{if | unless} <condition>]
12647 Perform an action on a validated session depending on a layer 5 condition
12648 May be used in sections : defaults | frontend | listen | backend
12649 no | yes | yes | no
12650 Arguments :
12651 <action> defines the action to perform if the condition applies. See
12652 below.
12653
12654 <condition> is a standard layer5-only ACL-based condition (see section 7).
12655
Davor Ocelice9ed2812017-12-25 17:49:28 +010012656 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020012657 it is possible to evaluate some conditions to decide whether this session
12658 must be accepted or dropped or have its counters tracked. Those conditions
12659 cannot make use of any data contents because no buffers are allocated yet and
12660 the processing cannot wait at this stage. The main use case it to copy some
12661 early information into variables (since variables are accessible in the
12662 session), or to keep track of some information collected after the handshake,
12663 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010012664 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020012665 extracted information can thus be copied to a variable or tracked using
12666 "track-sc" rules. Of course it is also possible to decide to accept/reject as
12667 with other rulesets. Most operations performed here could also be performed
12668 in "tcp-request content" rules, except that in HTTP these rules are evaluated
12669 for each new request, and that might not always be acceptable. For example a
12670 rule might increment a counter on each evaluation. It would also be possible
12671 that a country is resolved by geolocation from the source IP address,
12672 assigned to a session-wide variable, then the source address rewritten from
12673 an HTTP header for all requests. If some contents need to be inspected in
12674 order to take the decision, the "tcp-request content" statements must be used
12675 instead.
12676
12677 The "tcp-request session" rules are evaluated in their exact declaration
12678 order. If no rule matches or if there is no rule, the default action is to
12679 accept the incoming session. There is no specific limit to the number of
12680 rules which may be inserted.
12681
12682 Several types of actions are supported :
12683 - accept : the request is accepted
12684 - reject : the request is rejected and the connection is closed
12685 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
12686 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010012687 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010012688 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Christopher Faulet57759f32021-06-23 12:19:25 +020012689 - set-dst <expr>
12690 - set-dst-port <expr>
12691 - set-src <expr>
12692 - set-src-port <expr>
Willy Tarreau4f614292016-10-21 17:49:36 +020012693 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010012694 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020012695 - silent-drop
12696
12697 These actions have the same meaning as their respective counter-parts in
12698 "tcp-request connection" and "tcp-request content", so please refer to these
12699 sections for a complete description.
12700
12701 Note that the "if/unless" condition is optional. If no condition is set on
12702 the action, it is simply performed unconditionally. That can be useful for
12703 "track-sc*" actions as well as for changing the default action to a reject.
12704
12705 Example: track the original source address by default, or the one advertised
12706 in the PROXY protocol header for connection coming from the local
12707 proxies. The first connection-level rule enables receipt of the
12708 PROXY protocol for these ones, the second rule tracks whatever
12709 address we decide to keep after optional decoding.
12710
12711 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
12712 tcp-request session track-sc0 src
12713
12714 Example: accept all sessions from white-listed hosts, reject too fast
12715 sessions without counting them, and track accepted sessions.
12716 This results in session rate being capped from abusive sources.
12717
12718 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
12719 tcp-request session reject if { src_sess_rate gt 10 }
12720 tcp-request session track-sc0 src
12721
12722 Example: accept all sessions from white-listed hosts, count all other
12723 sessions and reject too fast ones. This results in abusive ones
12724 being blocked as long as they don't slow down.
12725
12726 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
12727 tcp-request session track-sc0 src
12728 tcp-request session reject if { sc0_sess_rate gt 10 }
12729
12730 See section 7 about ACL usage.
12731
12732 See also : "tcp-request connection", "tcp-request content", "stick-table"
12733
12734
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012735tcp-response inspect-delay <timeout>
12736 Set the maximum allowed time to wait for a response during content inspection
12737 May be used in sections : defaults | frontend | listen | backend
12738 no | no | yes | yes
12739 Arguments :
12740 <timeout> is the timeout value specified in milliseconds by default, but
12741 can be in any other unit if the number is suffixed by the unit,
12742 as explained at the top of this document.
12743
12744 See also : "tcp-response content", "tcp-request inspect-delay".
12745
12746
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012747timeout check <timeout>
12748 Set additional check timeout, but only after a connection has been already
12749 established.
12750
12751 May be used in sections: defaults | frontend | listen | backend
12752 yes | no | yes | yes
12753 Arguments:
12754 <timeout> is the timeout value specified in milliseconds by default, but
12755 can be in any other unit if the number is suffixed by the unit,
12756 as explained at the top of this document.
12757
Daniel Corbett9f0843f2021-05-08 10:50:37 -040012758 If set, HAProxy uses min("timeout connect", "inter") as a connect timeout
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012759 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010012760 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012761 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010012762 (Please also note that there is no valid reason to have such long connect
12763 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
12764 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012765
Daniel Corbett9f0843f2021-05-08 10:50:37 -040012766 If "timeout check" is not set HAProxy uses "inter" for complete check
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012767 timeout (connect + read) exactly like all <1.3.15 version.
12768
12769 In most cases check request is much simpler and faster to handle than normal
12770 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010012771 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012772
12773 This parameter is specific to backends, but can be specified once for all in
12774 "defaults" sections. This is in fact one of the easiest solutions not to
12775 forget about it.
12776
Willy Tarreau41a340d2008-01-22 12:25:31 +010012777 See also: "timeout connect", "timeout queue", "timeout server",
12778 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012779
12780
Willy Tarreau0ba27502007-12-24 16:55:16 +010012781timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010012782 Set the maximum inactivity time on the client side.
12783 May be used in sections : defaults | frontend | listen | backend
12784 yes | yes | yes | no
12785 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012786 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010012787 can be in any other unit if the number is suffixed by the unit,
12788 as explained at the top of this document.
12789
12790 The inactivity timeout applies when the client is expected to acknowledge or
12791 send data. In HTTP mode, this timeout is particularly important to consider
12792 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010012793 response while it is reading data sent by the server. That said, for the
12794 first phase, it is preferable to set the "timeout http-request" to better
12795 protect HAProxy from Slowloris like attacks. The value is specified in
12796 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010012797 suffixed by the unit, as specified at the top of this document. In TCP mode
12798 (and to a lesser extent, in HTTP mode), it is highly recommended that the
12799 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012800 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010012801 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010012802 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
12803 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020012804 which overrides "timeout client" and "timeout server" for tunnels, as well as
12805 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012806
12807 This parameter is specific to frontends, but can be specified once for all in
12808 "defaults" sections. This is in fact one of the easiest solutions not to
12809 forget about it. An unspecified timeout results in an infinite timeout, which
12810 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012811 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010012812 the system if the system's timeouts are not configured either.
12813
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012814 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012815
Willy Tarreau0ba27502007-12-24 16:55:16 +010012816
Willy Tarreau05cdd962014-05-10 14:30:07 +020012817timeout client-fin <timeout>
12818 Set the inactivity timeout on the client side for half-closed connections.
12819 May be used in sections : defaults | frontend | listen | backend
12820 yes | yes | yes | no
12821 Arguments :
12822 <timeout> is the timeout value specified in milliseconds by default, but
12823 can be in any other unit if the number is suffixed by the unit,
12824 as explained at the top of this document.
12825
12826 The inactivity timeout applies when the client is expected to acknowledge or
12827 send data while one direction is already shut down. This timeout is different
12828 from "timeout client" in that it only applies to connections which are closed
12829 in one direction. This is particularly useful to avoid keeping connections in
12830 FIN_WAIT state for too long when clients do not disconnect cleanly. This
12831 problem is particularly common long connections such as RDP or WebSocket.
12832 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010012833 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
12834 frame was sent, often indicating an expectation that the connection quickly
12835 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020012836
12837 This parameter is specific to frontends, but can be specified once for all in
12838 "defaults" sections. By default it is not set, so half-closed connections
12839 will use the other timeouts (timeout.client or timeout.tunnel).
12840
12841 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
12842
12843
Willy Tarreau0ba27502007-12-24 16:55:16 +010012844timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010012845 Set the maximum time to wait for a connection attempt to a server to succeed.
12846 May be used in sections : defaults | frontend | listen | backend
12847 yes | no | yes | yes
12848 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012849 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010012850 can be in any other unit if the number is suffixed by the unit,
12851 as explained at the top of this document.
12852
Daniel Corbett9f0843f2021-05-08 10:50:37 -040012853 If the server is located on the same LAN as HAProxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012854 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010012855 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010012856 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012857 connect timeout also presets both queue and tarpit timeouts to the same value
12858 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012859
12860 This parameter is specific to backends, but can be specified once for all in
12861 "defaults" sections. This is in fact one of the easiest solutions not to
12862 forget about it. An unspecified timeout results in an infinite timeout, which
12863 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012864 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010012865 the system if the system's timeouts are not configured either.
12866
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012867 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012868
Willy Tarreau0ba27502007-12-24 16:55:16 +010012869
Willy Tarreaub16a5742010-01-10 14:46:16 +010012870timeout http-keep-alive <timeout>
12871 Set the maximum allowed time to wait for a new HTTP request to appear
12872 May be used in sections : defaults | frontend | listen | backend
12873 yes | yes | yes | yes
12874 Arguments :
12875 <timeout> is the timeout value specified in milliseconds by default, but
12876 can be in any other unit if the number is suffixed by the unit,
12877 as explained at the top of this document.
12878
12879 By default, the time to wait for a new request in case of keep-alive is set
12880 by "timeout http-request". However this is not always convenient because some
12881 people want very short keep-alive timeouts in order to release connections
12882 faster, and others prefer to have larger ones but still have short timeouts
12883 once the request has started to present itself.
12884
12885 The "http-keep-alive" timeout covers these needs. It will define how long to
12886 wait for a new HTTP request to start coming after a response was sent. Once
12887 the first byte of request has been seen, the "http-request" timeout is used
12888 to wait for the complete request to come. Note that empty lines prior to a
12889 new request do not refresh the timeout and are not counted as a new request.
12890
12891 There is also another difference between the two timeouts : when a connection
12892 expires during timeout http-keep-alive, no error is returned, the connection
12893 just closes. If the connection expires in "http-request" while waiting for a
12894 connection to complete, a HTTP 408 error is returned.
12895
12896 In general it is optimal to set this value to a few tens to hundreds of
12897 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010012898 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010012899 1 millisecond) it will probably only accept pipelined requests but not the
12900 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020012901 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010012902
12903 If this parameter is not set, the "http-request" timeout applies, and if both
12904 are not set, "timeout client" still applies at the lower level. It should be
12905 set in the frontend to take effect, unless the frontend is in TCP mode, in
12906 which case the HTTP backend's timeout will be used.
12907
12908 See also : "timeout http-request", "timeout client".
12909
12910
Willy Tarreau036fae02008-01-06 13:24:40 +010012911timeout http-request <timeout>
12912 Set the maximum allowed time to wait for a complete HTTP request
12913 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012914 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010012915 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012916 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010012917 can be in any other unit if the number is suffixed by the unit,
12918 as explained at the top of this document.
12919
12920 In order to offer DoS protection, it may be required to lower the maximum
12921 accepted time to receive a complete HTTP request without affecting the client
12922 timeout. This helps protecting against established connections on which
12923 nothing is sent. The client timeout cannot offer a good protection against
12924 this abuse because it is an inactivity timeout, which means that if the
12925 attacker sends one character every now and then, the timeout will not
12926 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020012927 types, the request will be aborted if it does not complete in time. When the
12928 timeout expires, an HTTP 408 response is sent to the client to inform it
12929 about the problem, and the connection is closed. The logs will report
12930 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012931 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020012932 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
12933 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010012934
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012935 By default, this timeout only applies to the header part of the request,
12936 and not to any data. As soon as the empty line is received, this timeout is
12937 not used anymore. When combined with "option http-buffer-request", this
12938 timeout also applies to the body of the request..
12939 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010012940 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010012941
12942 Generally it is enough to set it to a few seconds, as most clients send the
12943 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010012944 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010012945 generally work on local networks as long as there are no packet losses. This
12946 will prevent people from sending bare HTTP requests using telnet.
12947
12948 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012949 chunk of the incoming request. It should be set in the frontend to take
12950 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
12951 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010012952
Willy Tarreau0f228a02015-05-01 15:37:53 +020012953 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012954 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010012955
Willy Tarreau844e3c52008-01-11 16:28:18 +010012956
12957timeout queue <timeout>
12958 Set the maximum time to wait in the queue for a connection slot to be free
12959 May be used in sections : defaults | frontend | listen | backend
12960 yes | no | yes | yes
12961 Arguments :
12962 <timeout> is the timeout value specified in milliseconds by default, but
12963 can be in any other unit if the number is suffixed by the unit,
12964 as explained at the top of this document.
12965
12966 When a server's maxconn is reached, connections are left pending in a queue
12967 which may be server-specific or global to the backend. In order not to wait
12968 indefinitely, a timeout is applied to requests pending in the queue. If the
12969 timeout is reached, it is considered that the request will almost never be
12970 served, so it is dropped and a 503 error is returned to the client.
12971
12972 The "timeout queue" statement allows to fix the maximum time for a request to
12973 be left pending in a queue. If unspecified, the same value as the backend's
12974 connection timeout ("timeout connect") is used, for backwards compatibility
12975 with older versions with no "timeout queue" parameter.
12976
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012977 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012978
12979
12980timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010012981 Set the maximum inactivity time on the server side.
12982 May be used in sections : defaults | frontend | listen | backend
12983 yes | no | yes | yes
12984 Arguments :
12985 <timeout> is the timeout value specified in milliseconds by default, but
12986 can be in any other unit if the number is suffixed by the unit,
12987 as explained at the top of this document.
12988
12989 The inactivity timeout applies when the server is expected to acknowledge or
12990 send data. In HTTP mode, this timeout is particularly important to consider
12991 during the first phase of the server's response, when it has to send the
12992 headers, as it directly represents the server's processing time for the
12993 request. To find out what value to put there, it's often good to start with
12994 what would be considered as unacceptable response times, then check the logs
12995 to observe the response time distribution, and adjust the value accordingly.
12996
12997 The value is specified in milliseconds by default, but can be in any other
12998 unit if the number is suffixed by the unit, as specified at the top of this
12999 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
13000 recommended that the client timeout remains equal to the server timeout in
13001 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010013002 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010013003 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010013004 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
13005 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020013006 "timeout tunnel", which overrides "timeout client" and "timeout server" for
13007 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013008
13009 This parameter is specific to backends, but can be specified once for all in
13010 "defaults" sections. This is in fact one of the easiest solutions not to
13011 forget about it. An unspecified timeout results in an infinite timeout, which
13012 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050013013 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010013014 the system if the system's timeouts are not configured either.
13015
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020013016 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010013017
Willy Tarreau05cdd962014-05-10 14:30:07 +020013018
13019timeout server-fin <timeout>
13020 Set the inactivity timeout on the server side for half-closed connections.
13021 May be used in sections : defaults | frontend | listen | backend
13022 yes | no | yes | yes
13023 Arguments :
13024 <timeout> is the timeout value specified in milliseconds by default, but
13025 can be in any other unit if the number is suffixed by the unit,
13026 as explained at the top of this document.
13027
13028 The inactivity timeout applies when the server is expected to acknowledge or
13029 send data while one direction is already shut down. This timeout is different
13030 from "timeout server" in that it only applies to connections which are closed
13031 in one direction. This is particularly useful to avoid keeping connections in
13032 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
13033 This problem is particularly common long connections such as RDP or WebSocket.
13034 Note that this timeout can override "timeout tunnel" when a connection shuts
13035 down in one direction. This setting was provided for completeness, but in most
13036 situations, it should not be needed.
13037
13038 This parameter is specific to backends, but can be specified once for all in
13039 "defaults" sections. By default it is not set, so half-closed connections
13040 will use the other timeouts (timeout.server or timeout.tunnel).
13041
13042 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
13043
Willy Tarreau844e3c52008-01-11 16:28:18 +010013044
13045timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010013046 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010013047 May be used in sections : defaults | frontend | listen | backend
13048 yes | yes | yes | yes
13049 Arguments :
13050 <timeout> is the tarpit duration specified in milliseconds by default, but
13051 can be in any other unit if the number is suffixed by the unit,
13052 as explained at the top of this document.
13053
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020013054 When a connection is tarpitted using "http-request tarpit", it is maintained
13055 open with no activity for a certain amount of time, then closed. "timeout
13056 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013057
13058 The value is specified in milliseconds by default, but can be in any other
13059 unit if the number is suffixed by the unit, as specified at the top of this
13060 document. If unspecified, the same value as the backend's connection timeout
13061 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010013062 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013063
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020013064 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010013065
13066
Willy Tarreauce887fd2012-05-12 12:50:00 +020013067timeout tunnel <timeout>
13068 Set the maximum inactivity time on the client and server side for tunnels.
13069 May be used in sections : defaults | frontend | listen | backend
13070 yes | no | yes | yes
13071 Arguments :
13072 <timeout> is the timeout value specified in milliseconds by default, but
13073 can be in any other unit if the number is suffixed by the unit,
13074 as explained at the top of this document.
13075
Jamie Gloudonaaa21002012-08-25 00:18:33 -040013076 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020013077 between a client and a server, and the connection remains inactive in both
13078 directions. This timeout supersedes both the client and server timeouts once
13079 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010013080 analyzer remains attached to either connection (e.g. tcp content rules are
13081 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020013082 when switching to the WebSocket protocol, or forwarding a CONNECT request
13083 to a proxy), or after the first response when no keepalive/close option is
13084 specified.
13085
Willy Tarreau05cdd962014-05-10 14:30:07 +020013086 Since this timeout is usually used in conjunction with long-lived connections,
13087 it usually is a good idea to also set "timeout client-fin" to handle the
13088 situation where a client suddenly disappears from the net and does not
13089 acknowledge a close, or sends a shutdown and does not acknowledge pending
13090 data anymore. This can happen in lossy networks where firewalls are present,
13091 and is detected by the presence of large amounts of sessions in a FIN_WAIT
13092 state.
13093
Willy Tarreauce887fd2012-05-12 12:50:00 +020013094 The value is specified in milliseconds by default, but can be in any other
13095 unit if the number is suffixed by the unit, as specified at the top of this
13096 document. Whatever the expected normal idle time, it is a good practice to
13097 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010013098 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020013099
13100 This parameter is specific to backends, but can be specified once for all in
13101 "defaults" sections. This is in fact one of the easiest solutions not to
13102 forget about it.
13103
13104 Example :
13105 defaults http
13106 option http-server-close
13107 timeout connect 5s
13108 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020013109 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020013110 timeout server 30s
13111 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
13112
Willy Tarreau05cdd962014-05-10 14:30:07 +020013113 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020013114
13115
Willy Tarreau844e3c52008-01-11 16:28:18 +010013116transparent (deprecated)
13117 Enable client-side transparent proxying
13118 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010013119 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010013120 Arguments : none
13121
13122 This keyword was introduced in order to provide layer 7 persistence to layer
13123 3 load balancers. The idea is to use the OS's ability to redirect an incoming
13124 connection for a remote address to a local process (here HAProxy), and let
13125 this process know what address was initially requested. When this option is
13126 used, sessions without cookies will be forwarded to the original destination
13127 IP address of the incoming request (which should match that of another
13128 equipment), while requests with cookies will still be forwarded to the
13129 appropriate server.
13130
13131 The "transparent" keyword is deprecated, use "option transparent" instead.
13132
13133 Note that contrary to a common belief, this option does NOT make HAProxy
13134 present the client's IP to the server when establishing the connection.
13135
Willy Tarreau844e3c52008-01-11 16:28:18 +010013136 See also: "option transparent"
13137
William Lallemanda73203e2012-03-12 12:48:57 +010013138unique-id-format <string>
13139 Generate a unique ID for each request.
13140 May be used in sections : defaults | frontend | listen | backend
13141 yes | yes | yes | no
13142 Arguments :
13143 <string> is a log-format string.
13144
Cyril Bonté108cf6e2012-04-21 23:30:29 +020013145 This keyword creates a ID for each request using the custom log format. A
13146 unique ID is useful to trace a request passing through many components of
13147 a complex infrastructure. The newly created ID may also be logged using the
13148 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010013149
Cyril Bonté108cf6e2012-04-21 23:30:29 +020013150 The format should be composed from elements that are guaranteed to be
Daniel Corbett9f0843f2021-05-08 10:50:37 -040013151 unique when combined together. For instance, if multiple HAProxy instances
Cyril Bonté108cf6e2012-04-21 23:30:29 +020013152 are involved, it might be important to include the node name. It is often
13153 needed to log the incoming connection's source and destination addresses
13154 and ports. Note that since multiple requests may be performed over the same
13155 connection, including a request counter may help differentiate them.
13156 Similarly, a timestamp may protect against a rollover of the counter.
13157 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010013158
Cyril Bonté108cf6e2012-04-21 23:30:29 +020013159 It is recommended to use hexadecimal notation for many fields since it
13160 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010013161
Cyril Bonté108cf6e2012-04-21 23:30:29 +020013162 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010013163
Julien Vehentf21be322014-03-07 08:27:34 -050013164 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010013165
13166 will generate:
13167
13168 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
13169
13170 See also: "unique-id-header"
13171
13172unique-id-header <name>
13173 Add a unique ID header in the HTTP request.
13174 May be used in sections : defaults | frontend | listen | backend
13175 yes | yes | yes | no
13176 Arguments :
13177 <name> is the name of the header.
13178
Cyril Bonté108cf6e2012-04-21 23:30:29 +020013179 Add a unique-id header in the HTTP request sent to the server, using the
13180 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010013181
Cyril Bonté108cf6e2012-04-21 23:30:29 +020013182 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010013183
Julien Vehentf21be322014-03-07 08:27:34 -050013184 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010013185 unique-id-header X-Unique-ID
13186
13187 will generate:
13188
13189 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
13190
13191 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010013192
Willy Tarreauf51658d2014-04-23 01:21:56 +020013193use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020013194 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013195 May be used in sections : defaults | frontend | listen | backend
13196 no | yes | yes | no
13197 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010013198 <backend> is the name of a valid backend or "listen" section, or a
13199 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013200
Willy Tarreauf51658d2014-04-23 01:21:56 +020013201 <condition> is a condition composed of ACLs, as described in section 7. If
13202 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013203
13204 When doing content-switching, connections arrive on a frontend and are then
13205 dispatched to various backends depending on a number of conditions. The
13206 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020013207 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010013208 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020013209 source address validation) or combined with a "tcp-request" rule to wait for
13210 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010013211
13212 There may be as many "use_backend" rules as desired. All of these rules are
13213 evaluated in their declaration order, and the first one which matches will
13214 assign the backend.
13215
13216 In the first form, the backend will be used if the condition is met. In the
13217 second form, the backend will be used if the condition is not met. If no
13218 condition is valid, the backend defined with "default_backend" will be used.
13219 If no default backend is defined, either the servers in the same section are
13220 used (in case of a "listen" section) or, in case of a frontend, no server is
13221 used and a 503 service unavailable response is returned.
13222
Willy Tarreau51aecc72009-07-12 09:47:04 +020013223 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013224 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020013225 and backend processing will immediately follow, or the backend will wait for
13226 a complete HTTP request to get in. This feature is useful when a frontend
13227 must decode several protocols on a unique port, one of them being HTTP.
13228
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010013229 When <backend> is a simple name, it is resolved at configuration time, and an
13230 error is reported if the specified backend does not exist. If <backend> is
13231 a log-format string instead, no check may be done at configuration time, so
13232 the backend name is resolved dynamically at run time. If the resulting
13233 backend name does not correspond to any valid backend, no other rule is
13234 evaluated, and the default_backend directive is applied instead. Note that
13235 when using dynamic backend names, it is highly recommended to use a prefix
13236 that no other backend uses in order to ensure that an unauthorized backend
13237 cannot be forced from the request.
13238
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013239 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010013240 used to detect the association between frontends and backends to compute the
13241 backend's "fullconn" setting. This cannot be done for dynamic names.
13242
13243 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
13244 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010013245
Christopher Fauletb30b3102019-09-12 23:03:09 +020013246use-fcgi-app <name>
13247 Defines the FastCGI application to use for the backend.
13248 May be used in sections : defaults | frontend | listen | backend
13249 no | no | yes | yes
13250 Arguments :
13251 <name> is the name of the FastCGI application to use.
13252
13253 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010013254
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013255use-server <server> if <condition>
13256use-server <server> unless <condition>
13257 Only use a specific server if/unless an ACL-based condition is matched.
13258 May be used in sections : defaults | frontend | listen | backend
13259 no | no | yes | yes
13260 Arguments :
Jerome Magnin824186b2020-03-29 09:37:12 +020013261 <server> is the name of a valid server in the same backend section
13262 or a "log-format" string resolving to a server name.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013263
13264 <condition> is a condition composed of ACLs, as described in section 7.
13265
13266 By default, connections which arrive to a backend are load-balanced across
13267 the available servers according to the configured algorithm, unless a
13268 persistence mechanism such as a cookie is used and found in the request.
13269
13270 Sometimes it is desirable to forward a particular request to a specific
13271 server without having to declare a dedicated backend for this server. This
13272 can be achieved using the "use-server" rules. These rules are evaluated after
13273 the "redirect" rules and before evaluating cookies, and they have precedence
13274 on them. There may be as many "use-server" rules as desired. All of these
13275 rules are evaluated in their declaration order, and the first one which
13276 matches will assign the server.
13277
13278 If a rule designates a server which is down, and "option persist" is not used
13279 and no force-persist rule was validated, it is ignored and evaluation goes on
13280 with the next rules until one matches.
13281
13282 In the first form, the server will be used if the condition is met. In the
13283 second form, the server will be used if the condition is not met. If no
13284 condition is valid, the processing continues and the server will be assigned
13285 according to other persistence mechanisms.
13286
13287 Note that even if a rule is matched, cookie processing is still performed but
13288 does not assign the server. This allows prefixed cookies to have their prefix
13289 stripped.
13290
13291 The "use-server" statement works both in HTTP and TCP mode. This makes it
13292 suitable for use with content-based inspection. For instance, a server could
Lukas Tribusa267b5d2020-07-19 00:25:06 +020013293 be selected in a farm according to the TLS SNI field when using protocols with
Alex18c88772021-06-05 13:23:08 +020013294 implicit TLS (also see "req.ssl_sni"). And if these servers have their weight
Lukas Tribusa267b5d2020-07-19 00:25:06 +020013295 set to zero, they will not be used for other traffic.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013296
13297 Example :
13298 # intercept incoming TLS requests based on the SNI field
Alex18c88772021-06-05 13:23:08 +020013299 use-server www if { req.ssl_sni -i www.example.com }
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013300 server www 192.168.0.1:443 weight 0
Alex18c88772021-06-05 13:23:08 +020013301 use-server mail if { req.ssl_sni -i mail.example.com }
Lukas Tribusa267b5d2020-07-19 00:25:06 +020013302 server mail 192.168.0.1:465 weight 0
Alex18c88772021-06-05 13:23:08 +020013303 use-server imap if { req.ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000013304 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013305 # all the rest is forwarded to this server
13306 server default 192.168.0.2:443 check
13307
Jerome Magnin824186b2020-03-29 09:37:12 +020013308 When <server> is a simple name, it is checked against existing servers in the
13309 configuration and an error is reported if the specified server does not exist.
13310 If it is a log-format, no check is performed when parsing the configuration,
13311 and if we can't resolve a valid server name at runtime but the use-server rule
Ilya Shipitsin11057a32020-06-21 21:18:27 +050013312 was conditioned by an ACL returning true, no other use-server rule is applied
Jerome Magnin824186b2020-03-29 09:37:12 +020013313 and we fall back to load balancing.
13314
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013315 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013316
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013317
Davor Ocelice9ed2812017-12-25 17:49:28 +0100133185. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013319--------------------------
13320
13321The "bind", "server" and "default-server" keywords support a number of settings
13322depending on some build options and on the system HAProxy was built on. These
13323settings generally each consist in one word sometimes followed by a value,
13324written on the same line as the "bind" or "server" line. All these options are
13325described in this section.
13326
13327
133285.1. Bind options
13329-----------------
13330
13331The "bind" keyword supports a certain number of settings which are all passed
13332as arguments on the same line. The order in which those arguments appear makes
13333no importance, provided that they appear after the bind address. All of these
13334parameters are optional. Some of them consist in a single words (booleans),
13335while other ones expect a value after them. In this case, the value must be
13336provided immediately after the setting name.
13337
13338The currently supported settings are the following ones.
13339
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010013340accept-netscaler-cip <magic number>
13341 Enforces the use of the NetScaler Client IP insertion protocol over any
13342 connection accepted by any of the TCP sockets declared on the same line. The
13343 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
13344 the incoming connection to be used everywhere an address is used, with the
13345 only exception of "tcp-request connection" rules which will only see the
13346 real connection address. Logs will reflect the addresses indicated in the
13347 protocol, unless it is violated, in which case the real address will still
13348 be used. This keyword combined with support from external components can be
13349 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010013350 mechanism which is not always reliable and not even always usable. See also
13351 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
13352 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010013353
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013354accept-proxy
13355 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020013356 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
13357 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013358 3/4 addresses of the incoming connection to be used everywhere an address is
13359 used, with the only exception of "tcp-request connection" rules which will
13360 only see the real connection address. Logs will reflect the addresses
13361 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010013362 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013363 components can be used as an efficient and reliable alternative to the
13364 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020013365 usable. See also "tcp-request connection expect-proxy" for a finer-grained
13366 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013367
Olivier Houchardc2aae742017-09-22 18:26:28 +020013368allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010013369 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010013370 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050013371 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010013372 that are idempotent. You can use the "wait-for-handshake" action for any
13373 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020013374
Willy Tarreauab861d32013-04-02 02:30:41 +020013375alpn <protocols>
13376 This enables the TLS ALPN extension and advertises the specified protocol
13377 list as supported on top of ALPN. The protocol list consists in a comma-
13378 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050013379 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020013380 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010013381 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
13382 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
13383 now obsolete NPN extension. At the time of writing this, most browsers still
13384 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
13385 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
13386 are expected to be supported, both versions can be advertised, in order of
13387 preference, like below :
13388
13389 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020013390
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013391backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010013392 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013393 backlog is used instead, which generally defaults to the maxconn value.
13394
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010013395curves <curves>
13396 This setting is only available when support for OpenSSL was built in. It sets
13397 the string describing the list of elliptic curves algorithms ("curve suite")
13398 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
13399 string is a colon-delimited list of curve name.
13400 Example: "X25519:P-256" (without quote)
13401 When "curves" is set, "ecdhe" parameter is ignored.
13402
Emeric Brun7fb34422012-09-28 15:26:15 +020013403ecdhe <named curve>
13404 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010013405 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
13406 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020013407
Emeric Brunfd33a262012-10-11 16:28:27 +020013408ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020013409 This setting is only available when support for OpenSSL was built in. It
13410 designates a PEM file from which to load CA certificates used to verify
13411 client's certificate.
13412
Emeric Brunb6dc9342012-09-28 17:55:37 +020013413ca-ignore-err [all|<errorID>,...]
13414 This setting is only available when support for OpenSSL was built in.
13415 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
13416 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
13417 error is ignored.
13418
Christopher Faulet31af49d2015-06-09 17:29:50 +020013419ca-sign-file <cafile>
13420 This setting is only available when support for OpenSSL was built in. It
13421 designates a PEM file containing both the CA certificate and the CA private
13422 key used to create and sign server's certificates. This is a mandatory
13423 setting when the dynamic generation of certificates is enabled. See
13424 'generate-certificates' for details.
13425
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000013426ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020013427 This setting is only available when support for OpenSSL was built in. It is
13428 the CA private key passphrase. This setting is optional and used only when
13429 the dynamic generation of certificates is enabled. See
13430 'generate-certificates' for details.
13431
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010013432ca-verify-file <cafile>
13433 This setting designates a PEM file from which to load CA certificates used to
13434 verify client's certificate. It designates CA certificates which must not be
13435 included in CA names sent in server hello message. Typically, "ca-file" must
13436 be defined with intermediate certificates, and "ca-verify-file" with
13437 certificates to ending the chain, like root CA.
13438
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013439ciphers <ciphers>
13440 This setting is only available when support for OpenSSL was built in. It sets
13441 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000013442 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013443 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013444 information and recommendations see e.g.
13445 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
13446 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
13447 cipher configuration, please check the "ciphersuites" keyword.
13448
13449ciphersuites <ciphersuites>
13450 This setting is only available when support for OpenSSL was built in and
13451 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
13452 the list of cipher algorithms ("cipher suite") that are negotiated during the
13453 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013454 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
13455 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013456
Emeric Brunfd33a262012-10-11 16:28:27 +020013457crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020013458 This setting is only available when support for OpenSSL was built in. It
13459 designates a PEM file from which to load certificate revocation list used
Remi Tricot-Le Breton02bd6842021-05-04 12:22:34 +020013460 to verify client's certificate. You need to provide a certificate revocation
13461 list for every certificate of your certificate authority chain.
Emeric Brun1a073b42012-09-28 17:07:34 +020013462
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013463crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000013464 This setting is only available when support for OpenSSL was built in. It
13465 designates a PEM file containing both the required certificates and any
13466 associated private keys. This file can be built by concatenating multiple
13467 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
13468 requires an intermediate certificate, this can also be concatenated into this
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +010013469 file. Intermediate certificate can also be shared in a directory via
13470 "issuers-chain-path" directive.
Alex Davies0fbf0162013-03-02 16:04:50 +000013471
William Lallemand4c5adbf2020-02-24 14:23:22 +010013472 If the file does not contain a private key, HAProxy will try to load
13473 the key at the same path suffixed by a ".key".
13474
Alex Davies0fbf0162013-03-02 16:04:50 +000013475 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
13476 are loaded.
13477
13478 If a directory name is used instead of a PEM file, then all files found in
William Lallemand3f25ae32020-02-24 16:30:12 +010013479 that directory will be loaded in alphabetic order unless their name ends
13480 with '.key', '.issuer', '.ocsp' or '.sctl' (reserved extensions). This
13481 directive may be specified multiple times in order to load certificates from
13482 multiple files or directories. The certificates will be presented to clients
13483 who provide a valid TLS Server Name Indication field matching one of their
13484 CN or alt subjects. Wildcards are supported, where a wildcard character '*'
13485 is used instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010013486 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000013487
13488 If no SNI is provided by the client or if the SSL library does not support
13489 TLS extensions, or if the client provides an SNI hostname which does not
13490 match any certificate, then the first loaded certificate will be presented.
13491 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010013492 recommended to load the default one first as a file or to ensure that it will
13493 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000013494
Emeric Brune032bfa2012-09-28 13:01:45 +020013495 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013496
Davor Ocelice9ed2812017-12-25 17:49:28 +010013497 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000013498 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010013499 choose a web server that the CA believes requires an intermediate CA (for
13500 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000013501 others, e.g. nginx, result in a wrong bundle that will not work for some
13502 clients).
13503
Daniel Corbett9f0843f2021-05-08 10:50:37 -040013504 For each PEM file, HAProxy checks for the presence of file at the same path
Emeric Brun4147b2e2014-06-16 18:36:30 +020013505 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
13506 Status Request extension (also known as "OCSP stapling") is automatically
13507 enabled. The content of this file is optional. If not empty, it must contain
13508 a valid OCSP Response in DER format. In order to be valid an OCSP Response
13509 must comply with the following rules: it has to indicate a good status,
13510 it has to be a single response for the certificate of the PEM file, and it
13511 has to be valid at the moment of addition. If these rules are not respected
13512 the OCSP Response is ignored and a warning is emitted. In order to identify
13513 which certificate an OCSP Response applies to, the issuer's certificate is
13514 necessary. If the issuer's certificate is not found in the PEM file, it will
13515 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
13516 if it exists otherwise it will fail with an error.
13517
Daniel Corbett9f0843f2021-05-08 10:50:37 -040013518 For each PEM file, HAProxy also checks for the presence of file at the same
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010013519 path suffixed by ".sctl". If such file is found, support for Certificate
13520 Transparency (RFC6962) TLS extension is enabled. The file must contain a
13521 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
13522 to check basic syntax, but no signatures are verified.
13523
yanbzhu6c25e9e2016-01-05 12:52:02 -050013524 There are cases where it is desirable to support multiple key types, e.g. RSA
13525 and ECDSA in the cipher suites offered to the clients. This allows clients
13526 that support EC certificates to be able to use EC ciphers, while
13527 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050013528
William Lallemandf9ff3ec2020-10-02 17:57:44 +020013529 To achieve this, OpenSSL 1.1.1 is required, you can configure this behavior
13530 by providing one crt entry per certificate type, or by configuring a "cert
13531 bundle" like it was required before HAProxy 1.8. See "ssl-load-extra-files".
yanbzhud19630c2015-12-14 15:10:25 -050013532
Emeric Brunb6dc9342012-09-28 17:55:37 +020013533crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000013534 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010013535 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013536 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000013537 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020013538
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013539crt-list <file>
13540 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013541 designates a list of PEM file with an optional ssl configuration and a SNI
13542 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013543
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013544 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
13545
William Lallemand5d036392020-06-30 16:11:36 +020013546 sslbindconf supports "allow-0rtt", "alpn", "ca-file", "ca-verify-file",
13547 "ciphers", "ciphersuites", "crl-file", "curves", "ecdhe", "no-ca-names",
13548 "npn", "verify" configuration. With BoringSSL and Openssl >= 1.1.1
13549 "ssl-min-ver" and "ssl-max-ver" are also supported. It overrides the
13550 configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013551
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020013552 Wildcards are supported in the SNI filter. Negative filter are also supported,
Joao Moraise51fab02020-11-21 07:42:20 -030013553 useful in combination with a wildcard filter to exclude a particular SNI, or
13554 after the first certificate to exclude a pattern from its CN or Subject Alt
13555 Name (SAN). The certificates will be presented to clients who provide a valid
13556 TLS Server Name Indication field matching one of the SNI filters. If no SNI
13557 filter is specified, the CN and SAN are used. This directive may be specified
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020013558 multiple times. See the "crt" option for more information. The default
13559 certificate is still needed to meet OpenSSL expectations. If it is not used,
13560 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013561
William Lallemandf9ff3ec2020-10-02 17:57:44 +020013562 Multi-cert bundling (see "ssl-load-extra-files") is supported with crt-list,
13563 as long as only the base name is given in the crt-list. SNI filter will do
13564 the same work on all bundled certificates.
yanbzhud19630c2015-12-14 15:10:25 -050013565
William Lallemand7c26ed72020-06-03 17:34:48 +020013566 Empty lines as well as lines beginning with a hash ('#') will be ignored.
13567
Joao Moraisaa8fcc42020-11-24 08:24:30 -030013568 The first declared certificate of a bind line is used as the default
Daniel Corbett9f0843f2021-05-08 10:50:37 -040013569 certificate, either from crt or crt-list option, which HAProxy should use in
Joao Moraisaa8fcc42020-11-24 08:24:30 -030013570 the TLS handshake if no other certificate matches. This certificate will also
13571 be used if the provided SNI matches its CN or SAN, even if a matching SNI
13572 filter is found on any crt-list. The SNI filter !* can be used after the first
13573 declared certificate to not include its CN and SAN in the SNI tree, so it will
13574 never match except if no other certificate matches. This way the first
13575 declared certificate act as a fallback.
Joao Moraise51fab02020-11-21 07:42:20 -030013576
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013577 crt-list file example:
Joao Moraise51fab02020-11-21 07:42:20 -030013578 cert1.pem !*
William Lallemand7c26ed72020-06-03 17:34:48 +020013579 # comment
Emmanuel Hocdet05942112017-02-20 16:11:50 +010013580 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013581 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010013582 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013583
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013584defer-accept
13585 Is an optional keyword which is supported only on certain Linux kernels. It
13586 states that a connection will only be accepted once some data arrive on it,
13587 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010013588 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013589 performance by ensuring that most of the request is already available when
13590 the connection is accepted. On the other hand, it will not be able to detect
13591 connections which don't talk. It is important to note that this option is
13592 broken in all kernels up to 2.6.31, as the connection is never accepted until
13593 the client talks. This can cause issues with front firewalls which would see
13594 an established connection while the proxy will only see it in SYN_RECV. This
13595 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
13596
William Lallemandf6975e92017-05-26 17:42:10 +020013597expose-fd listeners
13598 This option is only usable with the stats socket. It gives your stats socket
13599 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020013600 During a reload with the master-worker mode, the process is automatically
13601 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013602 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020013603
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013604force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013605 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013606 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013607 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013608 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013609
13610force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013611 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013612 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013613 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013614
13615force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013616 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013617 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013618 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013619
13620force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013621 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013622 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013623 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013624
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013625force-tlsv13
13626 This option enforces use of TLSv1.3 only on SSL connections instantiated from
13627 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013628 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013629
Christopher Faulet31af49d2015-06-09 17:29:50 +020013630generate-certificates
13631 This setting is only available when support for OpenSSL was built in. It
13632 enables the dynamic SSL certificates generation. A CA certificate and its
13633 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
13634 a transparent forward proxy, SSL requests generate errors because of a common
13635 name mismatch on the certificate presented to the client. With this option
13636 enabled, HAProxy will try to forge a certificate using the SNI hostname
13637 indicated by the client. This is done only if no certificate matches the SNI
13638 hostname (see 'crt-list'). If an error occurs, the default certificate is
13639 used, else the 'strict-sni' option is set.
13640 It can also be used when HAProxy is configured as a reverse proxy to ease the
13641 deployment of an architecture with many backends.
13642
13643 Creating a SSL certificate is an expensive operation, so a LRU cache is used
13644 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010013645 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020013646 certificate is used many times.
13647
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013648gid <gid>
13649 Sets the group of the UNIX sockets to the designated system gid. It can also
13650 be set by default in the global section's "unix-bind" statement. Note that
13651 some platforms simply ignore this. This setting is equivalent to the "group"
13652 setting except that the group ID is used instead of its name. This setting is
13653 ignored by non UNIX sockets.
13654
13655group <group>
13656 Sets the group of the UNIX sockets to the designated system group. It can
13657 also be set by default in the global section's "unix-bind" statement. Note
13658 that some platforms simply ignore this. This setting is equivalent to the
13659 "gid" setting except that the group name is used instead of its gid. This
13660 setting is ignored by non UNIX sockets.
13661
13662id <id>
13663 Fixes the socket ID. By default, socket IDs are automatically assigned, but
13664 sometimes it is more convenient to fix them to ease monitoring. This value
13665 must be strictly positive and unique within the listener/frontend. This
13666 option can only be used when defining only a single socket.
13667
13668interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010013669 Restricts the socket to a specific interface. When specified, only packets
13670 received from that particular interface are processed by the socket. This is
13671 currently only supported on Linux. The interface must be a primary system
13672 interface, not an aliased interface. It is also possible to bind multiple
13673 frontends to the same address if they are bound to different interfaces. Note
13674 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010013675 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
13676 uses the same interface as inbound traffic, and its associated routing table,
13677 even if there are explicit routes through different interfaces configured.
13678 This can prove useful to address asymmetric routing issues when the same
13679 client IP addresses need to be able to reach frontends hosted on different
13680 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013681
Willy Tarreauabb175f2012-09-24 12:43:26 +020013682level <level>
13683 This setting is used with the stats sockets only to restrict the nature of
13684 the commands that can be issued on the socket. It is ignored by other
13685 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013686 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020013687 read, and no change is allowed. It would make sense on systems where it
13688 is not easy to restrict access to the socket.
13689 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010013690 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020013691 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010013692 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020013693 all counters).
13694
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020013695severity-output <format>
13696 This setting is used with the stats sockets only to configure severity
13697 level output prepended to informational feedback messages. Severity
13698 level of messages can range between 0 and 7, conforming to syslog
13699 rfc5424. Valid and successful socket commands requesting data
13700 (i.e. "show map", "get acl foo" etc.) will never have a severity level
13701 prepended. It is ignored by other sockets. <format> can be one of :
13702 - "none" (default) no severity level is prepended to feedback messages.
13703 - "number" severity level is prepended as a number.
13704 - "string" severity level is prepended as a string following the
13705 rfc5424 convention.
13706
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013707maxconn <maxconn>
13708 Limits the sockets to this number of concurrent connections. Extraneous
13709 connections will remain in the system's backlog until a connection is
13710 released. If unspecified, the limit will be the same as the frontend's
13711 maxconn. Note that in case of port ranges or multiple addresses, the same
13712 value will be applied to each socket. This setting enables different
13713 limitations on expensive sockets, for instance SSL entries which may easily
13714 eat all memory.
13715
13716mode <mode>
13717 Sets the octal mode used to define access permissions on the UNIX socket. It
13718 can also be set by default in the global section's "unix-bind" statement.
13719 Note that some platforms simply ignore this. This setting is ignored by non
13720 UNIX sockets.
13721
13722mss <maxseg>
13723 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
13724 connections. This can be used to force a lower MSS for certain specific
13725 ports, for instance for connections passing through a VPN. Note that this
13726 relies on a kernel feature which is theoretically supported under Linux but
13727 was buggy in all versions prior to 2.6.28. It may or may not work on other
13728 operating systems. It may also not change the advertised value but change the
13729 effective size of outgoing segments. The commonly advertised value for TCPv4
13730 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
13731 positive, it will be used as the advertised MSS. If it is negative, it will
13732 indicate by how much to reduce the incoming connection's advertised MSS for
13733 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
13734
13735name <name>
13736 Sets an optional name for these sockets, which will be reported on the stats
13737 page.
13738
Willy Tarreaud72f0f32015-10-13 14:50:22 +020013739namespace <name>
13740 On Linux, it is possible to specify which network namespace a socket will
13741 belong to. This directive makes it possible to explicitly bind a listener to
13742 a namespace different from the default one. Please refer to your operating
13743 system's documentation to find more details about network namespaces.
13744
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013745nice <nice>
13746 Sets the 'niceness' of connections initiated from the socket. Value must be
13747 in the range -1024..1024 inclusive, and defaults to zero. Positive values
13748 means that such connections are more friendly to others and easily offer
13749 their place in the scheduler. On the opposite, negative values mean that
13750 connections want to run with a higher priority than others. The difference
13751 only happens under high loads when the system is close to saturation.
13752 Negative values are appropriate for low-latency or administration services,
13753 and high values are generally recommended for CPU intensive tasks such as SSL
13754 processing or bulk transfers which are less sensible to latency. For example,
13755 it may make sense to use a positive value for an SMTP socket and a negative
13756 one for an RDP socket.
13757
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020013758no-ca-names
13759 This setting is only available when support for OpenSSL was built in. It
13760 prevents from send CA names in server hello message when ca-file is used.
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010013761 Use "ca-verify-file" instead of "ca-file" with "no-ca-names".
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020013762
Emeric Brun9b3009b2012-10-05 11:55:06 +020013763no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013764 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013765 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013766 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013767 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013768 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
13769 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013770
Emeric Brun90ad8722012-10-02 14:00:59 +020013771no-tls-tickets
13772 This setting is only available when support for OpenSSL was built in. It
13773 disables the stateless session resumption (RFC 5077 TLS Ticket
13774 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013775 session resumption is more expensive in CPU usage. This option is also
13776 available on global statement "ssl-default-bind-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010013777 The TLS ticket mechanism is only used up to TLS 1.2.
13778 Forward Secrecy is compromised with TLS tickets, unless ticket keys
13779 are periodically rotated (via reload or by using "tls-ticket-keys").
Emeric Brun90ad8722012-10-02 14:00:59 +020013780
Emeric Brun9b3009b2012-10-05 11:55:06 +020013781no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013782 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013783 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013784 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013785 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013786 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13787 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013788
Emeric Brun9b3009b2012-10-05 11:55:06 +020013789no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020013790 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013791 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013792 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013793 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013794 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13795 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020013796
Emeric Brun9b3009b2012-10-05 11:55:06 +020013797no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020013798 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013799 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013800 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013801 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013802 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13803 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020013804
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013805no-tlsv13
13806 This setting is only available when support for OpenSSL was built in. It
13807 disables support for TLSv1.3 on any sockets instantiated from the listener
13808 when SSL is supported. Note that SSLv2 is forced disabled in the code and
13809 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013810 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13811 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013812
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020013813npn <protocols>
13814 This enables the NPN TLS extension and advertises the specified protocol list
13815 as supported on top of NPN. The protocol list consists in a comma-delimited
13816 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050013817 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020013818 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010013819 replaced with the ALPN extension (see the "alpn" keyword), though this one is
13820 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
13821 version of OpenSSL, NPN might still be used as most clients still support it
13822 at the time of writing this. It is possible to enable both NPN and ALPN
13823 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020013824
Lukas Tribus53ae85c2017-05-04 15:45:40 +000013825prefer-client-ciphers
13826 Use the client's preference when selecting the cipher suite, by default
13827 the server's preference is enforced. This option is also available on
13828 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020013829 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
13830 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
13831 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000013832
Christopher Fauletc644fa92017-11-23 22:44:11 +010013833process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010013834 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010013835 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010013836 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010013837 between the two is applied. If in the end the listener is not allowed to run
13838 on any remaining process, a warning is emitted, and the listener will either
13839 run on the first process of the listener if a single process was specified,
13840 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010013841 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010013842 connections for this listener, for the the process set. If multiple processes
13843 and threads are configured, a warning is emitted, as it either results from a
13844 configuration error or a misunderstanding of these models. For the unlikely
13845 case where several ranges are needed, this directive may be repeated.
13846 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010013847
13848 all | odd | even | number[-[number]]
13849
13850 Ranges can be partially defined. The higher bound can be omitted. In such
13851 case, it is replaced by the corresponding maximum value. The main purpose of
13852 this directive is to be used with the stats sockets and have one different
13853 socket per process. The second purpose is to have multiple bind lines sharing
13854 the same IP:port but not the same process in a listener, so that the system
13855 can distribute the incoming connections into multiple queues and allow a
13856 smoother inter-process load balancing. Currently Linux 3.9 and above is known
13857 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020013858
Christopher Fauleta717b992018-04-10 14:43:00 +020013859proto <name>
13860 Forces the multiplexer's protocol to use for the incoming connections. It
13861 must be compatible with the mode of the frontend (TCP or HTTP). It must also
13862 be usable on the frontend side. The list of available protocols is reported
Christopher Faulet982e17d2021-03-26 14:44:18 +010013863 in haproxy -vv. The protocols properties are reported : the mode (TCP/HTTP),
13864 the side (FE/BE), the mux name and its flags.
13865
13866 Some protocols report errors on aborts (flag=CLEAN_ABRT). Some others are
13867 subject to the head-of-line blocking on server side (flag=HOL_RISK). Finally
13868 some protocols don't support upgrades (flag=NO_UPG). The HTX compatibility is
13869 also reported (flag=HTX).
13870
13871 Here are the protocols that may be used as argument to a "proto" directive on
13872 a bind line :
13873
13874 h2 : mode=HTTP side=FE|BE mux=H2 flags=HTX|CLEAN_ABRT|HOL_RISK|NO_UPG
13875 h1 : mode=HTTP side=FE|BE mux=H1 flags=HTX|NO_UPG
13876 none : mode=TCP side=FE|BE mux=PASS flags=NO_UPG
13877
Daniel Corbett67a82712020-07-06 23:01:19 -040013878 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauleta717b992018-04-10 14:43:00 +020013879 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080013880 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020013881 h2" on the bind line.
13882
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013883ssl
13884 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013885 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013886 certificate is necessary (see "crt" above). All contents in the buffers will
13887 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020013888 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
13889 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013890
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013891ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
13892 This option enforces use of <version> or lower on SSL connections instantiated
William Lallemand50df1cb2020-06-02 10:52:24 +020013893 from this listener. Using this setting without "ssl-min-ver" can be
13894 ambiguous because the default ssl-min-ver value could change in future HAProxy
13895 versions. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013896 "ssl-default-bind-options". See also "ssl-min-ver".
13897
13898ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
William Lallemand50df1cb2020-06-02 10:52:24 +020013899 This option enforces use of <version> or upper on SSL connections
13900 instantiated from this listener. The default value is "TLSv1.2". This option
13901 is also available on global statement "ssl-default-bind-options".
13902 See also "ssl-max-ver".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013903
Emmanuel Hocdet65623372013-01-24 17:17:15 +010013904strict-sni
13905 This setting is only available when support for OpenSSL was built in. The
13906 SSL/TLS negotiation is allow only if the client provided an SNI which match
13907 a certificate. The default certificate is not used.
13908 See the "crt" option for more information.
13909
Willy Tarreau2af207a2015-02-04 00:45:58 +010013910tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010013911 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010013912 listening socket. This option is available on Linux since version 2.6.37. It
Daniel Corbett9f0843f2021-05-08 10:50:37 -040013913 allows HAProxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010013914 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010013915 useful on long-lived connections experiencing long idle periods such as
13916 remote terminals or database connection pools, where the client and server
13917 timeouts must remain high to allow a long period of idle, but where it is
13918 important to detect that the client has disappeared in order to release all
13919 resources associated with its connection (and the server's session). The
13920 argument is a delay expressed in milliseconds by default. This only works
13921 for regular TCP connections, and is ignored for other protocols.
13922
Willy Tarreau1c862c52012-10-05 16:21:00 +020013923tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010013924 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020013925 enables TCP Fast Open on the listening socket, which means that clients which
13926 support this feature will be able to send a request and receive a response
13927 during the 3-way handshake starting from second connection, thus saving one
13928 round-trip after the first connection. This only makes sense with protocols
13929 that use high connection rates and where each round trip matters. This can
13930 possibly cause issues with many firewalls which do not accept data on SYN
13931 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020013932 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
13933 need to build HAProxy with USE_TFO=1 if your libc doesn't define
13934 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020013935
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010013936tls-ticket-keys <keyfile>
13937 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010013938 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
13939 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
13940 The first key determines the key length used for next keys: you can't mix
13941 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
13942 build option (default 3) and at least as many keys need to be present in
13943 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
13944 penultimate one for encryption. This enables easy key rotation by just
13945 appending new key to the file and reloading the process. Keys must be
13946 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
13947 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010013948 storage such as hard drives (hint: use tmpfs and don't swap those files).
13949 Lifetime hint can be changed using tune.ssl.timeout.
13950
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013951transparent
13952 Is an optional keyword which is supported only on certain Linux kernels. It
13953 indicates that the addresses will be bound even if they do not belong to the
13954 local machine, and that packets targeting any of these addresses will be
13955 intercepted just as if the addresses were locally configured. This normally
13956 requires that IP forwarding is enabled. Caution! do not use this with the
13957 default address '*', as it would redirect any traffic for the specified port.
13958 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
13959 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
13960 kernel version. Some distribution kernels include backports of the feature,
13961 so check for support with your vendor.
13962
Willy Tarreau77e3af92012-11-24 15:07:23 +010013963v4v6
13964 Is an optional keyword which is supported only on most recent systems
13965 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
13966 and IPv6 when it uses the default address. Doing so is sometimes necessary
13967 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013968 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010013969
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013970v6only
13971 Is an optional keyword which is supported only on most recent systems
13972 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
13973 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010013974 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
13975 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013976
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013977uid <uid>
13978 Sets the owner of the UNIX sockets to the designated system uid. It can also
13979 be set by default in the global section's "unix-bind" statement. Note that
13980 some platforms simply ignore this. This setting is equivalent to the "user"
13981 setting except that the user numeric ID is used instead of its name. This
13982 setting is ignored by non UNIX sockets.
13983
13984user <user>
13985 Sets the owner of the UNIX sockets to the designated system user. It can also
13986 be set by default in the global section's "unix-bind" statement. Note that
13987 some platforms simply ignore this. This setting is equivalent to the "uid"
13988 setting except that the user name is used instead of its uid. This setting is
13989 ignored by non UNIX sockets.
13990
Emeric Brun1a073b42012-09-28 17:07:34 +020013991verify [none|optional|required]
13992 This setting is only available when support for OpenSSL was built in. If set
13993 to 'none', client certificate is not requested. This is the default. In other
13994 cases, a client certificate is requested. If the client does not provide a
13995 certificate after the request and if 'verify' is set to 'required', then the
13996 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020013997 certificate provided by the client is always verified using CAs from
13998 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
13999 is aborted, regardless of the 'verify' option, unless the error code exactly
14000 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020014001
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200140025.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010014003------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020014004
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010014005The "server" and "default-server" keywords support a certain number of settings
14006which are all passed as arguments on the server line. The order in which those
14007arguments appear does not count, and they are all optional. Some of those
14008settings are single words (booleans) while others expect one or several values
14009after them. In this case, the values must immediately follow the setting name.
14010Except default-server, all those settings must be specified after the server's
14011address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020014012
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014013 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010014014 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020014015
Frédéric Lécailled2376272017-03-21 18:52:12 +010014016Note that all these settings are supported both by "server" and "default-server"
14017keywords, except "id" which is only supported by "server".
14018
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014019The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010014020
Willy Tarreauceb4ac92012-04-28 00:41:46 +020014021addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014022 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010014023 to send health-checks or to probe the agent-check. On some servers, it may be
14024 desirable to dedicate an IP address to specific component able to perform
14025 complex tests which are more suitable to health-checks than the application.
14026 This parameter is ignored if the "check" parameter is not set. See also the
14027 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014028
Simon Hormand60d6912013-11-25 10:46:36 +090014029agent-check
14030 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010014031 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010014032 to the port set by the "agent-port" parameter and reading an ASCII string
14033 terminated by the first '\r' or '\n' met. The string is made of a series of
14034 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090014035
Willy Tarreau81f5d942013-12-09 20:51:51 +010014036 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090014037 Values in this format will set the weight proportional to the initial
Daniel Corbett9f0843f2021-05-08 10:50:37 -040014038 weight of a server as configured when HAProxy starts. Note that a zero
Willy Tarreauc5af3a62014-10-07 15:27:33 +020014039 weight is reported on the stats page as "DRAIN" since it has the same
14040 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090014041
Davor Ocelice9ed2812017-12-25 17:49:28 +010014042 - The string "maxconn:" followed by an integer (no space between). Values
14043 in this format will set the maxconn of a server. The maximum number of
14044 connections advertised needs to be multiplied by the number of load
14045 balancers and different backends that use this health check to get the
14046 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020014047
Willy Tarreau81f5d942013-12-09 20:51:51 +010014048 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014049 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090014050
Willy Tarreau81f5d942013-12-09 20:51:51 +010014051 - The word "drain". This will turn the server's administrative state to the
14052 DRAIN mode, thus it will not accept any new connections other than those
14053 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090014054
Willy Tarreau81f5d942013-12-09 20:51:51 +010014055 - The word "maint". This will turn the server's administrative state to the
14056 MAINT mode, thus it will not accept any new connections at all, and health
14057 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090014058
William Dauchyf8e795c2020-09-26 13:35:51 +020014059 - The words "down", "fail", or "stopped", optionally followed by a
Willy Tarreau81f5d942013-12-09 20:51:51 +010014060 description string after a sharp ('#'). All of these mark the server's
14061 operating state as DOWN, but since the word itself is reported on the stats
14062 page, the difference allows an administrator to know if the situation was
14063 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010014064 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010014065 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090014066
Willy Tarreau81f5d942013-12-09 20:51:51 +010014067 - The word "up" sets back the server's operating state as UP if health checks
14068 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090014069
Willy Tarreau81f5d942013-12-09 20:51:51 +010014070 Parameters which are not advertised by the agent are not changed. For
14071 example, an agent might be designed to monitor CPU usage and only report a
14072 relative weight and never interact with the operating status. Similarly, an
14073 agent could be designed as an end-user interface with 3 radio buttons
14074 allowing an administrator to change only the administrative state. However,
14075 it is important to consider that only the agent may revert its own actions,
14076 so if a server is set to DRAIN mode or to DOWN state using the agent, the
14077 agent must implement the other equivalent actions to bring the service into
14078 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090014079
Simon Horman2f1f9552013-11-25 10:46:37 +090014080 Failure to connect to the agent is not considered an error as connectivity
14081 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010014082 parameter. Warning though, it is not a good idea to stop an agent after it
14083 reports "down", since only an agent reporting "up" will be able to turn the
14084 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010014085 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090014086
Willy Tarreau81f5d942013-12-09 20:51:51 +010014087 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010014088 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090014089
James Brown55f9ff12015-10-21 18:19:05 -070014090agent-send <string>
Daniel Corbett9f0843f2021-05-08 10:50:37 -040014091 If this option is specified, HAProxy will send the given string (verbatim)
James Brown55f9ff12015-10-21 18:19:05 -070014092 to the agent server upon connection. You could, for example, encode
14093 the backend name into this string, which would enable your agent to send
14094 different responses based on the backend. Make sure to include a '\n' if
14095 you want to terminate your request with a newline.
14096
Simon Hormand60d6912013-11-25 10:46:36 +090014097agent-inter <delay>
14098 The "agent-inter" parameter sets the interval between two agent checks
14099 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
14100
14101 Just as with every other time-based parameter, it may be entered in any
14102 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
14103 parameter also serves as a timeout for agent checks "timeout check" is
14104 not set. In order to reduce "resonance" effects when multiple servers are
14105 hosted on the same hardware, the agent and health checks of all servers
14106 are started with a small time offset between them. It is also possible to
14107 add some random noise in the agent and health checks interval using the
14108 global "spread-checks" keyword. This makes sense for instance when a lot
14109 of backends use the same servers.
14110
14111 See also the "agent-check" and "agent-port" parameters.
14112
Misiek768d8602017-01-09 09:52:43 +010014113agent-addr <addr>
14114 The "agent-addr" parameter sets address for agent check.
14115
14116 You can offload agent-check to another target, so you can make single place
Daniel Corbett9f0843f2021-05-08 10:50:37 -040014117 managing status and weights of servers defined in HAProxy in case you can't
Misiek768d8602017-01-09 09:52:43 +010014118 make self-aware and self-managing services. You can specify both IP or
14119 hostname, it will be resolved.
14120
Simon Hormand60d6912013-11-25 10:46:36 +090014121agent-port <port>
14122 The "agent-port" parameter sets the TCP port used for agent checks.
14123
14124 See also the "agent-check" and "agent-inter" parameters.
14125
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020014126allow-0rtt
14127 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020014128 Note that early data will be sent only if the client used early data, or
14129 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020014130
Olivier Houchardc7566002018-11-20 23:33:50 +010014131alpn <protocols>
14132 This enables the TLS ALPN extension and advertises the specified protocol
14133 list as supported on top of ALPN. The protocol list consists in a comma-
14134 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050014135 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010014136 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
14137 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
14138 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
14139 now obsolete NPN extension.
14140 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
14141 be advertised, in order of preference, like below :
14142
14143 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
14144
Amaury Denoyelle69352ec2021-10-18 14:40:29 +020014145 See also "ws" to use an alternative ALPN for websocket streams.
14146
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014147backup
14148 When "backup" is present on a server line, the server is only used in load
14149 balancing when all other non-backup servers are unavailable. Requests coming
14150 with a persistence cookie referencing the server will always be served
14151 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010014152 the "allbackups" option is set in the backend. See also the "no-backup" and
14153 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014154
Emeric Brunef42d922012-10-11 16:11:36 +020014155ca-file <cafile>
14156 This setting is only available when support for OpenSSL was built in. It
14157 designates a PEM file from which to load CA certificates used to verify
14158 server's certificate.
14159
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014160check
Jerome Magnin90702bc2020-04-26 14:23:04 +020014161 This option enables health checks on a server:
14162 - when not set, no health checking is performed, and the server is always
14163 considered available.
14164 - when set and no other check method is configured, the server is considered
14165 available when a connection can be established at the highest configured
14166 transport layer. This means TCP by default, or SSL/TLS when "ssl" or
14167 "check-ssl" are set, both possibly combined with connection prefixes such
14168 as a PROXY protocol header when "send-proxy" or "check-send-proxy" are
14169 set.
14170 - when set and an application-level health check is defined, the
14171 application-level exchanges are performed on top of the configured
14172 transport layer and the server is considered available if all of the
14173 exchanges succeed.
14174
14175 By default, health checks are performed on the same address and port as
14176 configured on the server, using the same encapsulation parameters (SSL/TLS,
14177 proxy-protocol header, etc... ). It is possible to change the destination
14178 address using "addr" and the port using "port". When done, it is assumed the
14179 server isn't checked on the service port, and configured encapsulation
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +050014180 parameters are not reused. One must explicitly set "check-send-proxy" to send
Jerome Magnin90702bc2020-04-26 14:23:04 +020014181 connection headers, "check-ssl" to use SSL/TLS.
14182
14183 When "sni" or "alpn" are set on the server line, their value is not used for
14184 health checks and one must use "check-sni" or "check-alpn".
14185
14186 The default source address for health check traffic is the same as the one
14187 defined in the backend. It can be changed with the "source" keyword.
14188
14189 The interval between checks can be set using the "inter" keyword, and the
14190 "rise" and "fall" keywords can be used to define how many successful or
14191 failed health checks are required to flag a server available or not
14192 available.
14193
14194 Optional application-level health checks can be configured with "option
14195 httpchk", "option mysql-check" "option smtpchk", "option pgsql-check",
14196 "option ldap-check", or "option redis-check".
14197
14198 Example:
14199 # simple tcp check
14200 backend foo
14201 server s1 192.168.0.1:80 check
14202 # this does a tcp connect + tls handshake
14203 backend foo
14204 server s1 192.168.0.1:443 ssl check
14205 # simple tcp check is enough for check success
14206 backend foo
14207 option tcp-check
14208 tcp-check connect
14209 server s1 192.168.0.1:443 ssl check
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014210
Willy Tarreau6c16adc2012-10-05 00:04:16 +020014211check-send-proxy
14212 This option forces emission of a PROXY protocol line with outgoing health
14213 checks, regardless of whether the server uses send-proxy or not for the
14214 normal traffic. By default, the PROXY protocol is enabled for health checks
14215 if it is already enabled for normal traffic and if no "port" nor "addr"
14216 directive is present. However, if such a directive is present, the
14217 "check-send-proxy" option needs to be used to force the use of the
14218 protocol. See also the "send-proxy" option for more information.
14219
Olivier Houchard92150142018-12-21 19:47:01 +010014220check-alpn <protocols>
14221 Defines which protocols to advertise with ALPN. The protocol list consists in
14222 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
14223 (without quotes). If it is not set, the server ALPN is used.
14224
Christopher Fauletedc6ed92020-04-23 16:27:59 +020014225check-proto <name>
14226 Forces the multiplexer's protocol to use for the server's health-check
14227 connections. It must be compatible with the health-check type (TCP or
14228 HTTP). It must also be usable on the backend side. The list of available
Christopher Faulet982e17d2021-03-26 14:44:18 +010014229 protocols is reported in haproxy -vv. The protocols properties are
14230 reported : the mode (TCP/HTTP), the side (FE/BE), the mux name and its flags.
14231
14232 Some protocols report errors on aborts (flag=CLEAN_ABRT). Some others are
14233 subject to the head-of-line blocking on server side (flag=HOL_RISK). Finally
14234 some protocols don't support upgrades (flag=NO_UPG). The HTX compatibility is
14235 also reported (flag=HTX).
14236
14237 Here are the protocols that may be used as argument to a "check-proto"
14238 directive on a server line:
14239
14240 h2 : mode=HTTP side=FE|BE mux=H2 flags=HTX|CLEAN_ABRT|HOL_RISK|NO_UPG
14241 fcgi : mode=HTTP side=BE mux=FCGI flags=HTX|HOL_RISK|NO_UPG
14242 h1 : mode=HTTP side=FE|BE mux=H1 flags=HTX|NO_UPG
14243 none : mode=TCP side=FE|BE mux=PASS flags=NO_UPG
14244
Daniel Corbett67a82712020-07-06 23:01:19 -040014245 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauletedc6ed92020-04-23 16:27:59 +020014246 protocol for health-check connections established to this server.
14247 If not defined, the server one will be used, if set.
14248
Jérôme Magninae9bb762018-12-09 16:08:26 +010014249check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020014250 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010014251 over SSL. It is only possible to use a string to set <sni>. If you want to
14252 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020014253
Willy Tarreau763a95b2012-10-04 23:15:39 +020014254check-ssl
14255 This option forces encryption of all health checks over SSL, regardless of
14256 whether the server uses SSL or not for the normal traffic. This is generally
14257 used when an explicit "port" or "addr" directive is specified and SSL health
14258 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014259 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020014260 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
14261 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010014262 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010014263 See the "ssl" option for more information and "no-check-ssl" to disable
14264 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014265
Alexander Liu2a54bb72019-05-22 19:44:48 +080014266check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050014267 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080014268 default, the health checks won't go through socks tunnel even it was enabled
14269 for normal traffic.
14270
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014271ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020014272 This setting is only available when support for OpenSSL was built in. This
14273 option sets the string describing the list of cipher algorithms that is
14274 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000014275 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
14276 information and recommendations see e.g.
14277 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
14278 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
14279 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014280
Dirkjan Bussink415150f2018-09-14 11:14:21 +020014281ciphersuites <ciphersuites>
14282 This setting is only available when support for OpenSSL was built in and
14283 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
14284 describing the list of cipher algorithms that is negotiated during the TLS
14285 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000014286 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
14287 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
14288 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020014289
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014290cookie <value>
14291 The "cookie" parameter sets the cookie value assigned to the server to
14292 <value>. This value will be checked in incoming requests, and the first
14293 operational server possessing the same value will be selected. In return, in
14294 cookie insertion or rewrite modes, this value will be assigned to the cookie
14295 sent to the client. There is nothing wrong in having several servers sharing
14296 the same cookie value, and it is in fact somewhat common between normal and
14297 backup servers. See also the "cookie" keyword in backend section.
14298
Emeric Brunef42d922012-10-11 16:11:36 +020014299crl-file <crlfile>
14300 This setting is only available when support for OpenSSL was built in. It
14301 designates a PEM file from which to load certificate revocation list used
14302 to verify server's certificate.
14303
Emeric Bruna7aa3092012-10-26 12:58:00 +020014304crt <cert>
14305 This setting is only available when support for OpenSSL was built in.
14306 It designates a PEM file from which to load both a certificate and the
14307 associated private key. This file can be built by concatenating both PEM
14308 files into one. This certificate will be sent if the server send a client
14309 certificate request.
14310
Remi Tricot-Le Breton7c980df2021-05-07 15:28:08 +020014311 If the file does not contain a private key, HAProxy will try to load the key
14312 at the same path suffixed by a ".key" (provided the "ssl-load-extra-files"
14313 option is set accordingly).
14314
Willy Tarreau96839092010-03-29 10:02:24 +020014315disabled
14316 The "disabled" keyword starts the server in the "disabled" state. That means
14317 that it is marked down in maintenance mode, and no connection other than the
14318 ones allowed by persist mode will reach it. It is very well suited to setup
14319 new servers, because normal traffic will never reach them, while it is still
14320 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014321 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020014322
Frédéric Lécailled2376272017-03-21 18:52:12 +010014323enabled
14324 This option may be used as 'server' setting to reset any 'disabled'
14325 setting which would have been inherited from 'default-server' directive as
14326 default value.
14327 It may also be used as 'default-server' setting to reset any previous
14328 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020014329
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014330error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010014331 If health observing is enabled, the "error-limit" parameter specifies the
14332 number of consecutive errors that triggers event selected by the "on-error"
14333 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014334
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014335 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014336
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014337fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014338 The "fall" parameter states that a server will be considered as dead after
14339 <count> consecutive unsuccessful health checks. This value defaults to 3 if
14340 unspecified. See also the "check", "inter" and "rise" parameters.
14341
Emeric Brun8694b9a2012-10-05 14:39:07 +020014342force-sslv3
14343 This option enforces use of SSLv3 only when SSL is used to communicate with
14344 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014345 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014346 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020014347
14348force-tlsv10
14349 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014350 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014351 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020014352
14353force-tlsv11
14354 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014355 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014356 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020014357
14358force-tlsv12
14359 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014360 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014361 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020014362
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020014363force-tlsv13
14364 This option enforces use of TLSv1.3 only when SSL is used to communicate with
14365 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014366 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020014367
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014368id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020014369 Set a persistent ID for the server. This ID must be positive and unique for
14370 the proxy. An unused ID will automatically be assigned if unset. The first
14371 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014372
Willy Tarreau6a031d12016-11-07 19:42:35 +010014373init-addr {last | libc | none | <ip>},[...]*
14374 Indicate in what order the server's address should be resolved upon startup
14375 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014376 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010014377 method which succeeds is used. If the end of the list is reached without
14378 finding a working method, an error is thrown. Method "last" suggests to pick
14379 the address which appears in the state file (see "server-state-file"). Method
14380 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
14381 depending on the operating system and build options). Method "none"
14382 specifically indicates that the server should start without any valid IP
14383 address in a down state. It can be useful to ignore some DNS issues upon
14384 startup, waiting for the situation to get fixed later. Finally, an IP address
14385 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014386 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010014387 server used to catch old sessions and present them with a decent error
14388 message for example. When the "first" load balancing algorithm is used, this
14389 IP address could point to a fake server used to trigger the creation of new
14390 instances on the fly. This option defaults to "last,libc" indicating that the
14391 previous address found in the state file (if any) is used first, otherwise
14392 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014393 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010014394
14395 Example:
14396 defaults
14397 # never fail on address resolution
14398 default-server init-addr last,libc,none
14399
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014400inter <delay>
14401fastinter <delay>
14402downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014403 The "inter" parameter sets the interval between two consecutive health checks
14404 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
14405 It is also possible to use "fastinter" and "downinter" to optimize delays
14406 between checks depending on the server state :
14407
Pieter Baauw44fc9df2015-09-17 21:30:46 +020014408 Server state | Interval used
14409 ----------------------------------------+----------------------------------
14410 UP 100% (non-transitional) | "inter"
14411 ----------------------------------------+----------------------------------
14412 Transitionally UP (going down "fall"), | "fastinter" if set,
14413 Transitionally DOWN (going up "rise"), | "inter" otherwise.
14414 or yet unchecked. |
14415 ----------------------------------------+----------------------------------
14416 DOWN 100% (non-transitional) | "downinter" if set,
14417 | "inter" otherwise.
14418 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010014419
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014420 Just as with every other time-based parameter, they can be entered in any
14421 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
14422 serves as a timeout for health checks sent to servers if "timeout check" is
14423 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090014424 hosted on the same hardware, the agent and health checks of all servers
14425 are started with a small time offset between them. It is also possible to
14426 add some random noise in the agent and health checks interval using the
14427 global "spread-checks" keyword. This makes sense for instance when a lot
14428 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014429
Emeric Brun97556472020-05-30 01:42:45 +020014430log-proto <logproto>
14431 The "log-proto" specifies the protocol used to forward event messages to
14432 a server configured in a ring section. Possible values are "legacy"
14433 and "octet-count" corresponding respectively to "Non-transparent-framing"
14434 and "Octet counting" in rfc6587. "legacy" is the default.
14435
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014436maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014437 The "maxconn" parameter specifies the maximal number of concurrent
14438 connections that will be sent to this server. If the number of incoming
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010014439 concurrent connections goes higher than this value, they will be queued,
14440 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014441 save fragile servers from going down under extreme loads. If a "minconn"
14442 parameter is specified, the limit becomes dynamic. The default value is "0"
14443 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
14444 the backend's "fullconn" keyword.
14445
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010014446 In HTTP mode this parameter limits the number of concurrent requests instead
14447 of the number of connections. Multiple requests might be multiplexed over a
14448 single TCP connection to the server. As an example if you specify a maxconn
14449 of 50 you might see between 1 and 50 actual server connections, but no more
14450 than 50 concurrent requests.
14451
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014452maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014453 The "maxqueue" parameter specifies the maximal number of connections which
14454 will wait in the queue for this server. If this limit is reached, next
14455 requests will be redispatched to other servers instead of indefinitely
14456 waiting to be served. This will break persistence but may allow people to
Willy Tarreau8ae8c482020-10-22 17:19:07 +020014457 quickly re-log in when the server they try to connect to is dying. Some load
14458 balancing algorithms such as leastconn take this into account and accept to
14459 add requests into a server's queue up to this value if it is explicitly set
14460 to a value greater than zero, which often allows to better smooth the load
14461 when dealing with single-digit maxconn values. The default value is "0" which
14462 means the queue is unlimited. See also the "maxconn" and "minconn" parameters
14463 and "balance leastconn".
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014464
Willy Tarreau9c538e02019-01-23 10:21:49 +010014465max-reuse <count>
14466 The "max-reuse" argument indicates the HTTP connection processors that they
14467 should not reuse a server connection more than this number of times to send
14468 new requests. Permitted values are -1 (the default), which disables this
14469 limit, or any positive value. Value zero will effectively disable keep-alive.
14470 This is only used to work around certain server bugs which cause them to leak
14471 resources over time. The argument is not necessarily respected by the lower
14472 layers as there might be technical limitations making it impossible to
14473 enforce. At least HTTP/2 connections to servers will respect it.
14474
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014475minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014476 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
14477 limit following the backend's load. The server will always accept at least
14478 <minconn> connections, never more than <maxconn>, and the limit will be on
14479 the ramp between both values when the backend has less than <fullconn>
14480 concurrent connections. This makes it possible to limit the load on the
14481 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010014482 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014483 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014484
Willy Tarreaud72f0f32015-10-13 14:50:22 +020014485namespace <name>
14486 On Linux, it is possible to specify which network namespace a socket will
14487 belong to. This directive makes it possible to explicitly bind a server to
14488 a namespace different from the default one. Please refer to your operating
14489 system's documentation to find more details about network namespaces.
14490
Frédéric Lécailled2376272017-03-21 18:52:12 +010014491no-agent-check
14492 This option may be used as "server" setting to reset any "agent-check"
14493 setting which would have been inherited from "default-server" directive as
14494 default value.
14495 It may also be used as "default-server" setting to reset any previous
14496 "default-server" "agent-check" setting.
14497
14498no-backup
14499 This option may be used as "server" setting to reset any "backup"
14500 setting which would have been inherited from "default-server" directive as
14501 default value.
14502 It may also be used as "default-server" setting to reset any previous
14503 "default-server" "backup" setting.
14504
14505no-check
14506 This option may be used as "server" setting to reset any "check"
14507 setting which would have been inherited from "default-server" directive as
14508 default value.
14509 It may also be used as "default-server" setting to reset any previous
14510 "default-server" "check" setting.
14511
14512no-check-ssl
14513 This option may be used as "server" setting to reset any "check-ssl"
14514 setting which would have been inherited from "default-server" directive as
14515 default value.
14516 It may also be used as "default-server" setting to reset any previous
14517 "default-server" "check-ssl" setting.
14518
Frédéric Lécailled2376272017-03-21 18:52:12 +010014519no-send-proxy
14520 This option may be used as "server" setting to reset any "send-proxy"
14521 setting which would have been inherited from "default-server" directive as
14522 default value.
14523 It may also be used as "default-server" setting to reset any previous
14524 "default-server" "send-proxy" setting.
14525
14526no-send-proxy-v2
14527 This option may be used as "server" setting to reset any "send-proxy-v2"
14528 setting which would have been inherited from "default-server" directive as
14529 default value.
14530 It may also be used as "default-server" setting to reset any previous
14531 "default-server" "send-proxy-v2" setting.
14532
14533no-send-proxy-v2-ssl
14534 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
14535 setting which would have been inherited from "default-server" directive as
14536 default value.
14537 It may also be used as "default-server" setting to reset any previous
14538 "default-server" "send-proxy-v2-ssl" setting.
14539
14540no-send-proxy-v2-ssl-cn
14541 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
14542 setting which would have been inherited from "default-server" directive as
14543 default value.
14544 It may also be used as "default-server" setting to reset any previous
14545 "default-server" "send-proxy-v2-ssl-cn" setting.
14546
14547no-ssl
14548 This option may be used as "server" setting to reset any "ssl"
14549 setting which would have been inherited from "default-server" directive as
14550 default value.
14551 It may also be used as "default-server" setting to reset any previous
14552 "default-server" "ssl" setting.
14553
William Dauchyf6370442020-11-14 19:25:33 +010014554 Note that using `default-server ssl` setting and `no-ssl` on server will
14555 however init SSL connection, so it can be later be enabled through the
14556 runtime API: see `set server` commands in management doc.
14557
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010014558no-ssl-reuse
14559 This option disables SSL session reuse when SSL is used to communicate with
14560 the server. It will force the server to perform a full handshake for every
14561 new connection. It's probably only useful for benchmarking, troubleshooting,
14562 and for paranoid users.
14563
Emeric Brun9b3009b2012-10-05 11:55:06 +020014564no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014565 This option disables support for SSLv3 when SSL is used to communicate with
14566 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014567 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014568
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014569 Supported in default-server: No
14570
Emeric Brunf9c5c472012-10-11 15:28:34 +020014571no-tls-tickets
14572 This setting is only available when support for OpenSSL was built in. It
14573 disables the stateless session resumption (RFC 5077 TLS Ticket
14574 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014575 session resumption is more expensive in CPU usage for servers. This option
14576 is also available on global statement "ssl-default-server-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010014577 The TLS ticket mechanism is only used up to TLS 1.2.
14578 Forward Secrecy is compromised with TLS tickets, unless ticket keys
14579 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010014580 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020014581
Emeric Brun9b3009b2012-10-05 11:55:06 +020014582no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020014583 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020014584 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14585 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014586 often makes sense to disable it when communicating with local servers. This
14587 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014588 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014589
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014590 Supported in default-server: No
14591
Emeric Brun9b3009b2012-10-05 11:55:06 +020014592no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020014593 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020014594 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14595 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014596 often makes sense to disable it when communicating with local servers. This
14597 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014598 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014599
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014600 Supported in default-server: No
14601
Emeric Brun9b3009b2012-10-05 11:55:06 +020014602no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020014603 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014604 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14605 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014606 often makes sense to disable it when communicating with local servers. This
14607 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014608 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020014609
14610 Supported in default-server: No
14611
14612no-tlsv13
14613 This option disables support for TLSv1.3 when SSL is used to communicate with
14614 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14615 using any configuration option. TLSv1 is more expensive than SSLv3 so it
14616 often makes sense to disable it when communicating with local servers. This
14617 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014618 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014619
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014620 Supported in default-server: No
14621
Frédéric Lécailled2376272017-03-21 18:52:12 +010014622no-verifyhost
14623 This option may be used as "server" setting to reset any "verifyhost"
14624 setting which would have been inherited from "default-server" directive as
14625 default value.
14626 It may also be used as "default-server" setting to reset any previous
14627 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014628
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020014629no-tfo
14630 This option may be used as "server" setting to reset any "tfo"
14631 setting which would have been inherited from "default-server" directive as
14632 default value.
14633 It may also be used as "default-server" setting to reset any previous
14634 "default-server" "tfo" setting.
14635
Simon Hormanfa461682011-06-25 09:39:49 +090014636non-stick
14637 Never add connections allocated to this sever to a stick-table.
14638 This may be used in conjunction with backup to ensure that
14639 stick-table persistence is disabled for backup servers.
14640
Olivier Houchardc7566002018-11-20 23:33:50 +010014641npn <protocols>
14642 This enables the NPN TLS extension and advertises the specified protocol list
14643 as supported on top of NPN. The protocol list consists in a comma-delimited
14644 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050014645 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010014646 enabled (check with haproxy -vv). Note that the NPN extension has been
14647 replaced with the ALPN extension (see the "alpn" keyword), though this one is
14648 only available starting with OpenSSL 1.0.2.
14649
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014650observe <mode>
14651 This option enables health adjusting based on observing communication with
14652 the server. By default this functionality is disabled and enabling it also
14653 requires to enable health checks. There are two supported modes: "layer4" and
14654 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
14655 significant. In layer7, which is only allowed for http proxies, responses
14656 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010014657 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014658
14659 See also the "check", "on-error" and "error-limit".
14660
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014661on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014662 Select what should happen when enough consecutive errors are detected.
14663 Currently, four modes are available:
14664 - fastinter: force fastinter
14665 - fail-check: simulate a failed check, also forces fastinter (default)
14666 - sudden-death: simulate a pre-fatal failed health check, one more failed
14667 check will mark a server down, forces fastinter
14668 - mark-down: mark the server immediately down and force fastinter
14669
14670 See also the "check", "observe" and "error-limit".
14671
Simon Hormane0d1bfb2011-06-21 14:34:58 +090014672on-marked-down <action>
14673 Modify what occurs when a server is marked down.
14674 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014675 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
14676 all connections to the server are immediately terminated when the server
14677 goes down. It might be used if the health check detects more complex cases
14678 than a simple connection status, and long timeouts would cause the service
14679 to remain unresponsive for too long a time. For instance, a health check
14680 might detect that a database is stuck and that there's no chance to reuse
14681 existing connections anymore. Connections killed this way are logged with
14682 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090014683
14684 Actions are disabled by default
14685
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014686on-marked-up <action>
14687 Modify what occurs when a server is marked up.
14688 Currently one action is available:
14689 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
14690 done only if the server is not in backup state and if it is not disabled
14691 (it must have an effective weight > 0). This can be used sometimes to force
14692 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014693 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
14694 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014695 with extreme care. Sessions killed because a server comes up are logged
14696 with an 'U' termination code (for "Up").
14697
14698 Actions are disabled by default
14699
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020014700pool-low-conn <max>
14701 Set a low threshold on the number of idling connections for a server, below
14702 which a thread will not try to steal a connection from another thread. This
14703 can be useful to improve CPU usage patterns in scenarios involving many very
14704 fast servers, in order to ensure all threads will keep a few idle connections
14705 all the time instead of letting them accumulate over one thread and migrating
14706 them from thread to thread. Typical values of twice the number of threads
14707 seem to show very good performance already with sub-millisecond response
14708 times. The default is zero, indicating that any idle connection can be used
14709 at any time. It is the recommended setting for normal use. This only applies
14710 to connections that can be shared according to the same principles as those
Willy Tarreau0784db82021-02-19 11:45:22 +010014711 applying to "http-reuse". In case connection sharing between threads would
14712 be disabled via "tune.idle-pool.shared", it can become very important to use
14713 this setting to make sure each thread always has a few connections, or the
14714 connection reuse rate will decrease as thread count increases.
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020014715
Olivier Houchard006e3102018-12-10 18:30:32 +010014716pool-max-conn <max>
14717 Set the maximum number of idling connections for a server. -1 means unlimited
14718 connections, 0 means no idle connections. The default is -1. When idle
14719 connections are enabled, orphaned idle connections which do not belong to any
14720 client session anymore are moved to a dedicated pool so that they remain
14721 usable by future clients. This only applies to connections that can be shared
14722 according to the same principles as those applying to "http-reuse".
14723
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010014724pool-purge-delay <delay>
14725 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010014726 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020014727 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010014728
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014729port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014730 Using the "port" parameter, it becomes possible to use a different port to
William Dauchy4858fb22021-02-03 22:30:09 +010014731 send health-checks or to probe the agent-check. On some servers, it may be
14732 desirable to dedicate a port to a specific component able to perform complex
14733 tests which are more suitable to health-checks than the application. It is
14734 common to run a simple script in inetd for instance. This parameter is
14735 ignored if the "check" parameter is not set. See also the "addr" parameter.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014736
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014737proto <name>
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014738 Forces the multiplexer's protocol to use for the outgoing connections to this
14739 server. It must be compatible with the mode of the backend (TCP or HTTP). It
14740 must also be usable on the backend side. The list of available protocols is
Christopher Faulet982e17d2021-03-26 14:44:18 +010014741 reported in haproxy -vv.The protocols properties are reported : the mode
14742 (TCP/HTTP), the side (FE/BE), the mux name and its flags.
14743
14744 Some protocols report errors on aborts (flag=CLEAN_ABRT). Some others are
14745 subject to the head-of-line blocking on server side (flag=HOL_RISK). Finally
14746 some protocols don't support upgrades (flag=NO_UPG). The HTX compatibility is
14747 also reported (flag=HTX).
14748
14749 Here are the protocols that may be used as argument to a "proto" directive on
14750 a server line :
14751
14752 h2 : mode=HTTP side=FE|BE mux=H2 flags=HTX|CLEAN_ABRT|HOL_RISK|NO_UPG
14753 fcgi : mode=HTTP side=BE mux=FCGI flags=HTX|HOL_RISK|NO_UPG
14754 h1 : mode=HTTP side=FE|BE mux=H1 flags=HTX|NO_UPG
14755 none : mode=TCP side=FE|BE mux=PASS flags=NO_UPG
14756
Daniel Corbett67a82712020-07-06 23:01:19 -040014757 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014758 protocol for all connections established to this server.
14759
Amaury Denoyelle69352ec2021-10-18 14:40:29 +020014760 See also "ws" to use an alternative protocol for websocket streams.
14761
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014762redir <prefix>
14763 The "redir" parameter enables the redirection mode for all GET and HEAD
14764 requests addressing this server. This means that instead of having HAProxy
14765 forward the request to the server, it will send an "HTTP 302" response with
14766 the "Location" header composed of this prefix immediately followed by the
14767 requested URI beginning at the leading '/' of the path component. That means
14768 that no trailing slash should be used after <prefix>. All invalid requests
14769 will be rejected, and all non-GET or HEAD requests will be normally served by
14770 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010014771 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014772 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014773 users to a remote location in case of local disaster. Main use consists in
14774 increasing bandwidth for static servers by having the clients directly
14775 connect to them. Note: never use a relative location here, it would cause a
14776 loop between the client and HAProxy!
14777
14778 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
14779
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014780rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014781 The "rise" parameter states that a server will be considered as operational
14782 after <count> consecutive successful health checks. This value defaults to 2
14783 if unspecified. See also the "check", "inter" and "fall" parameters.
14784
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020014785resolve-opts <option>,<option>,...
14786 Comma separated list of options to apply to DNS resolution linked to this
14787 server.
14788
14789 Available options:
14790
14791 * allow-dup-ip
14792 By default, HAProxy prevents IP address duplication in a backend when DNS
14793 resolution at runtime is in operation.
14794 That said, for some cases, it makes sense that two servers (in the same
14795 backend, being resolved by the same FQDN) have the same IP address.
14796 For such case, simply enable this option.
14797 This is the opposite of prevent-dup-ip.
14798
Daniel Corbettf8716912019-11-17 09:48:56 -050014799 * ignore-weight
14800 Ignore any weight that is set within an SRV record. This is useful when
14801 you would like to control the weights using an alternate method, such as
14802 using an "agent-check" or through the runtime api.
14803
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020014804 * prevent-dup-ip
14805 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
14806 an IP address already set to a server in the same backend and sharing the
14807 same fqdn.
14808 This is the opposite of allow-dup-ip.
14809
14810 Example:
14811 backend b_myapp
14812 default-server init-addr none resolvers dns
14813 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
14814 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
14815
14816 With the option allow-dup-ip set:
14817 * if the nameserver returns a single IP address, then both servers will use
14818 it
14819 * If the nameserver returns 2 IP addresses, then each server will pick up a
14820 different address
14821
14822 Default value: not set
14823
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014824resolve-prefer <family>
14825 When DNS resolution is enabled for a server and multiple IP addresses from
14826 different families are returned, HAProxy will prefer using an IP address
14827 from the family mentioned in the "resolve-prefer" parameter.
14828 Available families: "ipv4" and "ipv6"
14829
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020014830 Default value: ipv6
14831
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014832 Example:
14833
14834 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014835
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014836resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050014837 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014838 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010014839 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010014840 different datacenters. The latency between datacenter is not negligible, so
14841 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014842 configured network, another address is selected.
14843
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014844 Example:
14845
14846 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014847
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014848resolvers <id>
14849 Points to an existing "resolvers" section to resolve current server's
14850 hostname.
14851
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014852 Example:
14853
14854 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014855
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014856 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014857
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010014858send-proxy
14859 The "send-proxy" parameter enforces use of the PROXY protocol over any
14860 connection established to this server. The PROXY protocol informs the other
14861 end about the layer 3/4 addresses of the incoming connection, so that it can
14862 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014863 upper layer protocol. For connections accepted by an "accept-proxy" or
14864 "accept-netscaler-cip" listener, the advertised address will be used. Only
14865 TCPv4 and TCPv6 address families are supported. Other families such as
14866 Unix sockets, will report an UNKNOWN family. Servers using this option can
Daniel Corbett9f0843f2021-05-08 10:50:37 -040014867 fully be chained to another instance of HAProxy listening with an
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014868 "accept-proxy" setting. This setting must not be used if the server isn't
14869 aware of the protocol. When health checks are sent to the server, the PROXY
14870 protocol is automatically used when this option is set, unless there is an
14871 explicit "port" or "addr" directive, in which case an explicit
14872 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014873 See also the "no-send-proxy" option of this section and "accept-proxy" and
14874 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010014875
David Safb76832014-05-08 23:42:08 -040014876send-proxy-v2
14877 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
14878 over any connection established to this server. The PROXY protocol informs
14879 the other end about the layer 3/4 addresses of the incoming connection, so
14880 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020014881 whatever the upper layer protocol. It also send ALPN information if an alpn
14882 have been negotiated. This setting must not be used if the server isn't aware
14883 of this version of the protocol. See also the "no-send-proxy-v2" option of
14884 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014885
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010014886proxy-v2-options <option>[,<option>]*
Tim Duesterhuscf6e0c82020-03-13 12:34:24 +010014887 The "proxy-v2-options" parameter add options to send in PROXY protocol
14888 version 2 when "send-proxy-v2" is used. Options available are:
14889
14890 - ssl : See also "send-proxy-v2-ssl".
14891 - cert-cn : See also "send-proxy-v2-ssl-cn".
14892 - ssl-cipher: Name of the used cipher.
14893 - cert-sig : Signature algorithm of the used certificate.
14894 - cert-key : Key algorithm of the used certificate
14895 - authority : Host name value passed by the client (only SNI from a TLS
14896 connection is supported).
14897 - crc32c : Checksum of the PROXYv2 header.
14898 - unique-id : Send a unique ID generated using the frontend's
14899 "unique-id-format" within the PROXYv2 header.
14900 This unique-id is primarily meant for "mode tcp". It can
14901 lead to unexpected results in "mode http", because the
14902 generated unique ID is also used for the first HTTP request
14903 within a Keep-Alive connection.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010014904
David Safb76832014-05-08 23:42:08 -040014905send-proxy-v2-ssl
14906 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
14907 2 over any connection established to this server. The PROXY protocol informs
14908 the other end about the layer 3/4 addresses of the incoming connection, so
14909 that it can know the client's address or the public address it accessed to,
14910 whatever the upper layer protocol. In addition, the SSL information extension
14911 of the PROXY protocol is added to the PROXY protocol header. This setting
14912 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014913 See also the "no-send-proxy-v2-ssl" option of this section and the
14914 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014915
14916send-proxy-v2-ssl-cn
14917 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
14918 2 over any connection established to this server. The PROXY protocol informs
14919 the other end about the layer 3/4 addresses of the incoming connection, so
14920 that it can know the client's address or the public address it accessed to,
14921 whatever the upper layer protocol. In addition, the SSL information extension
14922 of the PROXY protocol, along along with the Common Name from the subject of
14923 the client certificate (if any), is added to the PROXY protocol header. This
14924 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014925 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
14926 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014927
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014928slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014929 The "slowstart" parameter for a server accepts a value in milliseconds which
14930 indicates after how long a server which has just come back up will run at
14931 full speed. Just as with every other time-based parameter, it can be entered
14932 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
14933 linearly from 0 to 100% during this time. The limitation applies to two
14934 parameters :
14935
14936 - maxconn: the number of connections accepted by the server will grow from 1
14937 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
14938
14939 - weight: when the backend uses a dynamic weighted algorithm, the weight
14940 grows linearly from 1 to 100%. In this case, the weight is updated at every
14941 health-check. For this reason, it is important that the "inter" parameter
14942 is smaller than the "slowstart", in order to maximize the number of steps.
14943
Daniel Corbett9f0843f2021-05-08 10:50:37 -040014944 The slowstart never applies when HAProxy starts, otherwise it would cause
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014945 trouble to running servers. It only applies when a server has been previously
14946 seen as failed.
14947
Willy Tarreau732eac42015-07-09 11:40:25 +020014948sni <expression>
14949 The "sni" parameter evaluates the sample fetch expression, converts it to a
14950 string and uses the result as the host name sent in the SNI TLS extension to
14951 the server. A typical use case is to send the SNI received from the client in
14952 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020014953 expression, though alternatives such as req.hdr(host) can also make sense. If
14954 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014955 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010014956 "verify" directive for more details. If you want to set a SNI for health
14957 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020014958
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014959source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020014960source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014961source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014962 The "source" parameter sets the source address which will be used when
14963 connecting to the server. It follows the exact same parameters and principle
14964 as the backend "source" keyword, except that it only applies to the server
14965 referencing it. Please consult the "source" keyword for details.
14966
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014967 Additionally, the "source" statement on a server line allows one to specify a
14968 source port range by indicating the lower and higher bounds delimited by a
14969 dash ('-'). Some operating systems might require a valid IP address when a
14970 source port range is specified. It is permitted to have the same IP/range for
14971 several servers. Doing so makes it possible to bypass the maximum of 64k
14972 total concurrent connections. The limit will then reach 64k connections per
14973 server.
14974
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000014975 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
14976 specifying the source address without port(s).
14977
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014978ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020014979 This option enables SSL ciphering on outgoing connections to the server. It
14980 is critical to verify server certificates using "verify" when using SSL to
14981 connect to servers, otherwise the communication is prone to trivial man in
14982 the-middle attacks rendering SSL useless. When this option is used, health
14983 checks are automatically sent in SSL too unless there is a "port" or an
14984 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014985 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
14986 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014987
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014988ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
14989 This option enforces use of <version> or lower when SSL is used to communicate
14990 with the server. This option is also available on global statement
14991 "ssl-default-server-options". See also "ssl-min-ver".
14992
14993ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
14994 This option enforces use of <version> or upper when SSL is used to communicate
14995 with the server. This option is also available on global statement
14996 "ssl-default-server-options". See also "ssl-max-ver".
14997
Frédéric Lécailled2376272017-03-21 18:52:12 +010014998ssl-reuse
14999 This option may be used as "server" setting to reset any "no-ssl-reuse"
15000 setting which would have been inherited from "default-server" directive as
15001 default value.
15002 It may also be used as "default-server" setting to reset any previous
15003 "default-server" "no-ssl-reuse" setting.
15004
15005stick
15006 This option may be used as "server" setting to reset any "non-stick"
15007 setting which would have been inherited from "default-server" directive as
15008 default value.
15009 It may also be used as "default-server" setting to reset any previous
15010 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020015011
Alexander Liu2a54bb72019-05-22 19:44:48 +080015012socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050015013 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080015014 server. Using this option won't force the health check to go via socks4 by
15015 default. You will have to use the keyword "check-via-socks4" to enable it.
15016
Willy Tarreau163d4622015-10-13 16:16:41 +020015017tcp-ut <delay>
15018 Sets the TCP User Timeout for all outgoing connections to this server. This
Daniel Corbett9f0843f2021-05-08 10:50:37 -040015019 option is available on Linux since version 2.6.37. It allows HAProxy to
Willy Tarreau163d4622015-10-13 16:16:41 +020015020 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010015021 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020015022 long-lived connections experiencing long idle periods such as remote
15023 terminals or database connection pools, where the client and server timeouts
15024 must remain high to allow a long period of idle, but where it is important to
15025 detect that the server has disappeared in order to release all resources
15026 associated with its connection (and the client's session). One typical use
15027 case is also to force dead server connections to die when health checks are
15028 too slow or during a soft reload since health checks are then disabled. The
15029 argument is a delay expressed in milliseconds by default. This only works for
15030 regular TCP connections, and is ignored for other protocols.
15031
Willy Tarreau034c88c2017-01-23 23:36:45 +010015032tfo
15033 This option enables using TCP fast open when connecting to servers, on
15034 systems that support it (currently only the Linux kernel >= 4.11).
15035 See the "tfo" bind option for more information about TCP fast open.
15036 Please note that when using tfo, you should also use the "conn-failure",
Daniel Corbett9f0843f2021-05-08 10:50:37 -040015037 "empty-response" and "response-timeout" keywords for "retry-on", or HAProxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020015038 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010015039
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015040track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020015041 This option enables ability to set the current state of the server by tracking
15042 another one. It is possible to track a server which itself tracks another
15043 server, provided that at the end of the chain, a server has health checks
15044 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015045 used, it has to be enabled on both proxies.
15046
Frédéric Lécailled2376272017-03-21 18:52:12 +010015047tls-tickets
15048 This option may be used as "server" setting to reset any "no-tls-tickets"
15049 setting which would have been inherited from "default-server" directive as
15050 default value.
Lukas Tribusbdb386d2020-03-10 00:56:09 +010015051 The TLS ticket mechanism is only used up to TLS 1.2.
15052 Forward Secrecy is compromised with TLS tickets, unless ticket keys
15053 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010015054 It may also be used as "default-server" setting to reset any previous
Bjoern Jacke5ab7eb62020-02-13 14:16:16 +010015055 "default-server" "no-tls-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010015056
Emeric Brunef42d922012-10-11 16:11:36 +020015057verify [none|required]
15058 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010015059 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020015060 certificate provided by the server is verified using CAs from 'ca-file' and
15061 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010015062 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020015063 the name passed using the "sni" directive, or if not provided, the static
15064 host name passed using the "verifyhost" directive. When no name is found, the
15065 certificate's names are ignored. For this reason, without SNI it's important
15066 to use "verifyhost". On verification failure the handshake is aborted. It is
15067 critically important to verify server certificates when using SSL to connect
15068 to servers, otherwise the communication is prone to trivial man-in-the-middle
15069 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
15070 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020015071
Evan Broderbe554312013-06-27 00:05:25 -070015072verifyhost <hostname>
15073 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020015074 only takes effect if 'verify required' is also specified. This directive sets
15075 a default static hostname to check the server's certificate against when no
15076 SNI was used to connect to the server. If SNI is not used, this is the only
15077 way to enable hostname verification. This static hostname, when set, will
15078 also be used for health checks (which cannot provide an SNI value). If none
15079 of the hostnames in the certificate match the specified hostname, the
15080 handshake is aborted. The hostnames in the server-provided certificate may
15081 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070015082
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010015083weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015084 The "weight" parameter is used to adjust the server's weight relative to
15085 other servers. All servers will receive a load proportional to their weight
15086 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020015087 load. The default weight is 1, and the maximal value is 256. A value of 0
15088 means the server will not participate in load-balancing but will still accept
15089 persistent connections. If this parameter is used to distribute the load
15090 according to server's capacity, it is recommended to start with values which
15091 can both grow and shrink, for instance between 10 and 100 to leave enough
15092 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015093
Amaury Denoyelle69352ec2021-10-18 14:40:29 +020015094ws { auto | h1 | h2 }
15095 This option allows to configure the protocol used when relaying websocket
15096 streams. This is most notably useful when using an HTTP/2 backend without the
15097 support for H2 websockets through the RFC8441.
15098
15099 The default mode is "auto". This will reuse the same protocol as the main
15100 one. The only difference is when using ALPN. In this case, it can try to
15101 downgrade the ALPN to "http/1.1" only for websocket streams if the configured
15102 server ALPN contains it.
15103
15104 The value "h1" is used to force HTTP/1.1 for websockets streams, through ALPN
15105 if SSL ALPN is activated for the server. Similarly, "h2" can be used to
15106 force HTTP/2.0 websockets. Use this value with care : the server must support
15107 RFC8441 or an error will be reported by haproxy when relaying websockets.
15108
15109 Note that NPN is not taken into account as its usage has been deprecated in
15110 favor of the ALPN extension.
15111
15112 See also "alpn" and "proto".
15113
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015114
Cyril Bonté46175dd2015-07-02 22:45:32 +0200151155.3. Server IP address resolution using DNS
15116-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015117
Baptiste Assmann62b75b42015-09-09 01:11:36 +020015118HAProxy allows using a host name on the server line to retrieve its IP address
15119using name servers. By default, HAProxy resolves the name when parsing the
Thayne McCombscdbcca92021-01-07 21:24:41 -070015120configuration file, at startup and cache the result for the process's life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015121This is not sufficient in some cases, such as in Amazon where a server's IP
15122can change after a reboot or an ELB Virtual IP can change based on current
15123workload.
15124This chapter describes how HAProxy can be configured to process server's name
15125resolution at run time.
15126Whether run time server name resolution has been enable or not, HAProxy will
15127carry on doing the first resolution when parsing the configuration.
15128
15129
Cyril Bonté46175dd2015-07-02 22:45:32 +0200151305.3.1. Global overview
15131----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015132
15133As we've seen in introduction, name resolution in HAProxy occurs at two
15134different steps of the process life:
15135
15136 1. when starting up, HAProxy parses the server line definition and matches a
15137 host name. It uses libc functions to get the host name resolved. This
15138 resolution relies on /etc/resolv.conf file.
15139
Christopher Faulet67957bd2017-09-27 11:00:59 +020015140 2. at run time, HAProxy performs periodically name resolutions for servers
15141 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015142
15143A few other events can trigger a name resolution at run time:
15144 - when a server's health check ends up in a connection timeout: this may be
15145 because the server has a new IP address. So we need to trigger a name
15146 resolution to know this new IP.
15147
Christopher Faulet67957bd2017-09-27 11:00:59 +020015148When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010015149HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020015150SRV label is specified, then the corresponding SRV records will be retrieved
15151from the DNS server, and the provided hostnames will be used. The SRV label
Daniel Corbett9f0843f2021-05-08 10:50:37 -040015152will be checked periodically, and if any server are added or removed, HAProxy
Christopher Faulet67957bd2017-09-27 11:00:59 +020015153will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020015154
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015155A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050015156 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015157 first valid response.
15158
15159 - a resolution is considered as invalid (NX, timeout, refused), when all the
15160 servers return an error.
15161
15162
Cyril Bonté46175dd2015-07-02 22:45:32 +0200151635.3.2. The resolvers section
15164----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015165
15166This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020015167HAProxy. There can be as many as resolvers section as needed. Each section can
15168contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015169
Baptiste Assmann62b75b42015-09-09 01:11:36 +020015170When multiple name servers are configured in a resolvers section, then HAProxy
15171uses the first valid response. In case of invalid responses, only the last one
15172is treated. Purpose is to give the chance to a slow server to deliver a valid
15173answer after a fast faulty or outdated server.
15174
15175When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020015176used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020015177
Christopher Faulet67957bd2017-09-27 11:00:59 +020015178 1. HAProxy retries the same DNS query with a new query type. The A queries are
15179 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
15180 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020015181
Christopher Faulet67957bd2017-09-27 11:00:59 +020015182 2. When the fallback on the query type was done (or not applicable), HAProxy
15183 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020015184
Thierry Fournierfc13f792021-12-15 19:03:52 +010015185 3. HAProxy retries previous steps <resolve_retries> times. If no valid
Christopher Faulet67957bd2017-09-27 11:00:59 +020015186 response is received after that, it stops the DNS resolution and reports
15187 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020015188
Christopher Faulet67957bd2017-09-27 11:00:59 +020015189For example, with 2 name servers configured in a resolvers section, the
15190following scenarios are possible:
15191
15192 - First response is valid and is applied directly, second response is
15193 ignored
15194
15195 - First response is invalid and second one is valid, then second response is
15196 applied
15197
15198 - First response is a NX domain and second one a truncated response, then
15199 HAProxy retries the query with a new type
15200
15201 - First response is a NX domain and second one is a timeout, then HAProxy
15202 retries the query with a new type
15203
15204 - Query timed out for both name servers, then HAProxy retries it with the
15205 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020015206
Daniel Corbett9f0843f2021-05-08 10:50:37 -040015207As a DNS server may not answer all the IPs in one DNS request, HAProxy keeps
Olivier Houcharda8c6db82017-07-06 18:46:47 +020015208a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020015209<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020015210
Baptiste Assmann62b75b42015-09-09 01:11:36 +020015211
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015212resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010015213 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015214
15215A resolvers section accept the following parameters:
15216
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020015217accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010015218 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020015219 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020015220 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
15221 by RFC 6891)
15222
Emeric Brun4c751952021-03-08 16:41:29 +010015223 Note: the maximum allowed value is 65535. Recommended value for UDP is
15224 4096 and it is not recommended to exceed 8192 except if you are sure
15225 that your system and network can handle this (over 65507 makes no sense
15226 since is the maximum UDP payload size). If you are using only TCP
15227 nameservers to handle huge DNS responses, you should put this value
15228 to the max: 65535.
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020015229
Emeric Brunc8f3e452021-04-07 16:04:54 +020015230nameserver <name> <address>[:port] [param*]
15231 Used to configure a nameserver. <name> of the nameserver should ne unique.
15232 By default the <address> is considered of type datagram. This means if an
15233 IPv4 or IPv6 is configured without special address prefixes (paragraph 11.)
15234 the UDP protocol will be used. If an stream protocol address prefix is used,
15235 the nameserver will be considered as a stream server (TCP for instance) and
15236 "server" parameters found in 5.2 paragraph which are relevant for DNS
15237 resolving will be considered. Note: currently, in TCP mode, 4 queries are
15238 pipelined on the same connections. A batch of idle connections are removed
15239 every 5 seconds. "maxconn" can be configured to limit the amount of those
Emeric Brun56fc5d92021-02-12 20:05:45 +010015240 concurrent connections and TLS should also usable if the server supports.
15241
Ben Draut44e609b2018-05-29 15:40:08 -060015242parse-resolv-conf
15243 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
15244 list. Ordered as if each nameserver in /etc/resolv.conf was individually
15245 placed in the resolvers section in place of this directive.
15246
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015247hold <status> <period>
15248 Defines <period> during which the last name resolution should be kept based
15249 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010015250 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020015251 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015252 <period> : interval between two successive name resolution when the last
15253 answer was in <status>. It follows the HAProxy time format.
15254 <period> is in milliseconds by default.
15255
Baptiste Assmann686408b2017-08-18 10:15:42 +020015256 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015257
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015258resolve_retries <nb>
15259 Defines the number <nb> of queries to send to resolve a server name before
15260 giving up.
15261 Default value: 3
15262
Baptiste Assmann62b75b42015-09-09 01:11:36 +020015263 A retry occurs on name server timeout or when the full sequence of DNS query
15264 type failover is over and we need to start up from the default ANY query
15265 type.
15266
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015267timeout <event> <time>
15268 Defines timeouts related to name resolution
15269 <event> : the event on which the <time> timeout period applies to.
15270 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010015271 - resolve : default time to trigger name resolutions when no
15272 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020015273 Default value: 1s
15274 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010015275 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020015276 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015277 <time> : time related to the event. It follows the HAProxy time format.
15278 <time> is expressed in milliseconds.
15279
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020015280 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015281
15282 resolvers mydns
15283 nameserver dns1 10.0.0.1:53
15284 nameserver dns2 10.0.0.2:53
Emeric Brunc8f3e452021-04-07 16:04:54 +020015285 nameserver dns3 tcp@10.0.0.3:53
Ben Draut44e609b2018-05-29 15:40:08 -060015286 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015287 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020015288 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015289 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010015290 hold other 30s
15291 hold refused 30s
15292 hold nx 30s
15293 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015294 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020015295 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020015296
15297
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200152986. Cache
15299---------
15300
15301HAProxy provides a cache, which was designed to perform cache on small objects
15302(favicon, css...). This is a minimalist low-maintenance cache which runs in
15303RAM.
15304
15305The cache is based on a memory which is shared between processes and threads,
15306this memory is split in blocks of 1k.
15307
15308If an object is not used anymore, it can be deleted to store a new object
15309independently of its expiration date. The oldest objects are deleted first
15310when we try to allocate a new one.
15311
15312The cache uses a hash of the host header and the URI as the key.
15313
15314It's possible to view the status of a cache using the Unix socket command
15315"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
15316for more details.
15317
15318When an object is delivered from the cache, the server name in the log is
15319replaced by "<CACHE>".
15320
15321
153226.1. Limitation
15323----------------
15324
15325The cache won't store and won't deliver objects in these cases:
15326
15327- If the response is not a 200
Remi Tricot-Le Breton4f730832020-11-26 15:51:50 +010015328- If the response contains a Vary header and either the process-vary option is
15329 disabled, or a currently unmanaged header is specified in the Vary value (only
15330 accept-encoding and referer are managed for now)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020015331- If the Content-Length + the headers size is greater than "max-object-size"
15332- If the response is not cacheable
Remi Tricot-Le Bretond493bc82020-11-26 15:51:29 +010015333- If the response does not have an explicit expiration time (s-maxage or max-age
15334 Cache-Control directives or Expires header) or a validator (ETag or Last-Modified
15335 headers)
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010015336- If the process-vary option is enabled and there are already max-secondary-entries
15337 entries with the same primary key as the current response
Remi Tricot-Le Breton6ca89162021-01-07 14:50:51 +010015338- If the process-vary option is enabled and the response has an unknown encoding (not
15339 mentioned in https://www.iana.org/assignments/http-parameters/http-parameters.xhtml)
15340 while varying on the accept-encoding client header
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020015341
15342- If the request is not a GET
15343- If the HTTP version of the request is smaller than 1.1
15344- If the request contains an Authorization header
15345
15346
153476.2. Setup
15348-----------
15349
15350To setup a cache, you must define a cache section and use it in a proxy with
15351the corresponding http-request and response actions.
15352
15353
153546.2.1. Cache section
15355---------------------
15356
15357cache <name>
15358 Declare a cache section, allocate a shared cache memory named <name>, the
15359 size of cache is mandatory.
15360
15361total-max-size <megabytes>
15362 Define the size in RAM of the cache in megabytes. This size is split in
15363 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
15364
15365max-object-size <bytes>
15366 Define the maximum size of the objects to be cached. Must not be greater than
15367 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
15368 All objects with sizes larger than "max-object-size" will not be cached.
15369
15370max-age <seconds>
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010015371 Define the maximum expiration duration. The expiration is set as the lowest
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020015372 value between the s-maxage or max-age (in this order) directive in the
15373 Cache-Control response header and this value. The default value is 60
15374 seconds, which means that you can't cache an object more than 60 seconds by
15375 default.
15376
Remi Tricot-Le Bretone6cc5b52020-12-23 18:13:53 +010015377process-vary <on/off>
15378 Enable or disable the processing of the Vary header. When disabled, a response
Remi Tricot-Le Breton754b2422020-11-16 15:56:10 +010015379 containing such a header will never be cached. When enabled, we need to calculate
15380 a preliminary hash for a subset of request headers on all the incoming requests
15381 (which might come with a cpu cost) which will be used to build a secondary key
Remi Tricot-Le Bretone6cc5b52020-12-23 18:13:53 +010015382 for a given request (see RFC 7234#4.1). The default value is off (disabled).
Remi Tricot-Le Breton754b2422020-11-16 15:56:10 +010015383
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010015384max-secondary-entries <number>
15385 Define the maximum number of simultaneous secondary entries with the same primary
15386 key in the cache. This needs the vary support to be enabled. Its default value is 10
15387 and should be passed a strictly positive integer.
15388
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020015389
153906.2.2. Proxy section
15391---------------------
15392
15393http-request cache-use <name> [ { if | unless } <condition> ]
15394 Try to deliver a cached object from the cache <name>. This directive is also
15395 mandatory to store the cache as it calculates the cache hash. If you want to
15396 use a condition for both storage and delivering that's a good idea to put it
15397 after this one.
15398
15399http-response cache-store <name> [ { if | unless } <condition> ]
15400 Store an http-response within the cache. The storage of the response headers
15401 is done at this step, which means you can use others http-response actions
15402 to modify headers before or after the storage of the response. This action
15403 is responsible for the setup of the cache storage filter.
15404
15405
15406Example:
15407
15408 backend bck1
15409 mode http
15410
15411 http-request cache-use foobar
15412 http-response cache-store foobar
15413 server srv1 127.0.0.1:80
15414
15415 cache foobar
15416 total-max-size 4
15417 max-age 240
15418
15419
Willy Tarreau74ca5042013-06-11 23:12:07 +0200154207. Using ACLs and fetching samples
15421----------------------------------
15422
Davor Ocelice9ed2812017-12-25 17:49:28 +010015423HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020015424client or server information, from tables, environmental information etc...
15425The action of extracting such data is called fetching a sample. Once retrieved,
15426these samples may be used for various purposes such as a key to a stick-table,
15427but most common usages consist in matching them against predefined constant
15428data called patterns.
15429
15430
154317.1. ACL basics
15432---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015433
15434The use of Access Control Lists (ACL) provides a flexible solution to perform
15435content switching and generally to take decisions based on content extracted
15436from the request, the response or any environmental status. The principle is
15437simple :
15438
Willy Tarreau74ca5042013-06-11 23:12:07 +020015439 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015440 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020015441 - apply one or multiple pattern matching methods on this sample
15442 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015443
Willy Tarreau74ca5042013-06-11 23:12:07 +020015444The actions generally consist in blocking a request, selecting a backend, or
15445adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015446
15447In order to define a test, the "acl" keyword is used. The syntax is :
15448
Willy Tarreau74ca5042013-06-11 23:12:07 +020015449 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015450
15451This creates a new ACL <aclname> or completes an existing one with new tests.
15452Those tests apply to the portion of request/response specified in <criterion>
15453and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015454an operator which may be specified before the set of values. Optionally some
15455conversion operators may be applied to the sample, and they will be specified
15456as a comma-delimited list of keywords just after the first keyword. The values
15457are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015458
15459ACL names must be formed from upper and lower case letters, digits, '-' (dash),
15460'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
15461which means that "my_acl" and "My_Acl" are two different ACLs.
15462
15463There is no enforced limit to the number of ACLs. The unused ones do not affect
15464performance, they just consume a small amount of memory.
15465
Willy Tarreau74ca5042013-06-11 23:12:07 +020015466The criterion generally is the name of a sample fetch method, or one of its ACL
15467specific declinations. The default test method is implied by the output type of
15468this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015469methods of a same sample fetch method. The sample fetch methods are the only
15470ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015471
15472Sample fetch methods return data which can be of the following types :
15473 - boolean
15474 - integer (signed or unsigned)
15475 - IPv4 or IPv6 address
15476 - string
15477 - data block
15478
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015479Converters transform any of these data into any of these. For example, some
15480converters might convert a string to a lower-case string while other ones
15481would turn a string to an IPv4 address, or apply a netmask to an IP address.
15482The resulting sample is of the type of the last converter applied to the list,
15483which defaults to the type of the sample fetch method.
15484
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015485Each sample or converter returns data of a specific type, specified with its
15486keyword in this documentation. When an ACL is declared using a standard sample
15487fetch method, certain types automatically involved a default matching method
15488which are summarized in the table below :
15489
15490 +---------------------+-----------------+
15491 | Sample or converter | Default |
15492 | output type | matching method |
15493 +---------------------+-----------------+
15494 | boolean | bool |
15495 +---------------------+-----------------+
15496 | integer | int |
15497 +---------------------+-----------------+
15498 | ip | ip |
15499 +---------------------+-----------------+
15500 | string | str |
15501 +---------------------+-----------------+
15502 | binary | none, use "-m" |
15503 +---------------------+-----------------+
15504
15505Note that in order to match a binary samples, it is mandatory to specify a
15506matching method, see below.
15507
Willy Tarreau74ca5042013-06-11 23:12:07 +020015508The ACL engine can match these types against patterns of the following types :
15509 - boolean
15510 - integer or integer range
15511 - IP address / network
15512 - string (exact, substring, suffix, prefix, subdir, domain)
15513 - regular expression
15514 - hex block
15515
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015516The following ACL flags are currently supported :
15517
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015518 -i : ignore case during matching of all subsequent patterns.
15519 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015520 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010015521 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010015522 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010015523 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020015524 -- : force end of flags. Useful when a string looks like one of the flags.
15525
Willy Tarreau74ca5042013-06-11 23:12:07 +020015526The "-f" flag is followed by the name of a file from which all lines will be
15527read as individual values. It is even possible to pass multiple "-f" arguments
15528if the patterns are to be loaded from multiple files. Empty lines as well as
15529lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
15530will be stripped. If it is absolutely necessary to insert a valid pattern
15531beginning with a sharp, just prefix it with a space so that it is not taken for
Daniel Corbett9f0843f2021-05-08 10:50:37 -040015532a comment. Depending on the data type and match method, HAProxy may load the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015533lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
15534exact string matching. In this case, duplicates will automatically be removed.
15535
Thierry FOURNIER9860c412014-01-29 14:23:29 +010015536The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
15537parsed as two column file. The first column contains the patterns used by the
15538ACL, and the second column contain the samples. The sample can be used later by
15539a map. This can be useful in some rare cases where an ACL would just be used to
15540check for the existence of a pattern in a map before a mapping is applied.
15541
Thierry FOURNIER3534d882014-01-20 17:01:44 +010015542The "-u" flag forces the unique id of the ACL. This unique id is used with the
15543socket interface to identify ACL and dynamically change its values. Note that a
15544file is always identified by its name even if an id is set.
15545
Willy Tarreau74ca5042013-06-11 23:12:07 +020015546Also, note that the "-i" flag applies to subsequent entries and not to entries
15547loaded from files preceding it. For instance :
15548
15549 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
15550
15551In this example, each line of "exact-ua.lst" will be exactly matched against
15552the "user-agent" header of the request. Then each line of "generic-ua" will be
15553case-insensitively matched. Then the word "test" will be insensitively matched
15554as well.
15555
15556The "-m" flag is used to select a specific pattern matching method on the input
15557sample. All ACL-specific criteria imply a pattern matching method and generally
15558do not need this flag. However, this flag is useful with generic sample fetch
15559methods to describe how they're going to be matched against the patterns. This
15560is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010015561obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020015562followed by a pattern matching method name, this method is used instead of the
15563default one for the criterion. This makes it possible to match contents in ways
15564that were not initially planned, or with sample fetch methods which return a
15565string. The matching method also affects the way the patterns are parsed.
15566
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010015567The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
15568By default, if the parser cannot parse ip address it considers that the parsed
15569string is maybe a domain name and try dns resolution. The flag "-n" disable this
15570resolution. It is useful for detecting malformed ip lists. Note that if the DNS
Daniel Corbett9f0843f2021-05-08 10:50:37 -040015571server is not reachable, the HAProxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050015572waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010015573flag "-n" disable this behavior. Note also that during the runtime, this
15574function is disabled for the dynamic acl modifications.
15575
Willy Tarreau74ca5042013-06-11 23:12:07 +020015576There are some restrictions however. Not all methods can be used with all
15577sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
15578be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020015579
15580 - "found" : only check if the requested sample could be found in the stream,
15581 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020015582 to pass any pattern to avoid confusion. This matching method is
15583 particularly useful to detect presence of certain contents such
15584 as headers, cookies, etc... even if they are empty and without
15585 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015586
15587 - "bool" : check the value as a boolean. It can only be applied to fetches
15588 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015589 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015590
15591 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020015592 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015593
15594 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020015595 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015596
Davor Ocelice9ed2812017-12-25 17:49:28 +010015597 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020015598 binary sequence. This may be used with binary or string samples.
15599
15600 - "len" : match the sample's length as an integer. This may be used with
15601 binary or string samples.
15602
Willy Tarreau74ca5042013-06-11 23:12:07 +020015603 - "str" : exact match : match the contents against a string. This may be
15604 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015605
Willy Tarreau74ca5042013-06-11 23:12:07 +020015606 - "sub" : substring match : check that the contents contain at least one of
15607 the provided string patterns. This may be used with binary or
15608 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015609
Willy Tarreau74ca5042013-06-11 23:12:07 +020015610 - "reg" : regex match : match the contents against a list of regular
15611 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015612
Willy Tarreau74ca5042013-06-11 23:12:07 +020015613 - "beg" : prefix match : check that the contents begin like the provided
15614 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015615
Willy Tarreau74ca5042013-06-11 23:12:07 +020015616 - "end" : suffix match : check that the contents end like the provided
15617 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015618
Willy Tarreau74ca5042013-06-11 23:12:07 +020015619 - "dir" : subdir match : check that a slash-delimited portion of the
15620 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015621 This may be used with binary or string samples.
15622
Willy Tarreau74ca5042013-06-11 23:12:07 +020015623 - "dom" : domain match : check that a dot-delimited portion of the contents
15624 exactly match one of the provided string patterns. This may be
15625 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015626
15627For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
15628request, it is possible to do :
15629
Christian Ruppertfd3184c2022-02-20 22:54:01 +010015630 acl jsess_present req.cook(JSESSIONID) -m found
Willy Tarreau5adeda12013-03-31 22:13:34 +020015631
15632In order to apply a regular expression on the 500 first bytes of data in the
15633buffer, one would use the following acl :
15634
Christian Ruppertfd3184c2022-02-20 22:54:01 +010015635 acl script_tag req.payload(0,500) -m reg -i <script>
Willy Tarreau5adeda12013-03-31 22:13:34 +020015636
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015637On systems where the regex library is much slower when using "-i", it is
15638possible to convert the sample to lowercase before matching, like this :
15639
Christian Ruppertfd3184c2022-02-20 22:54:01 +010015640 acl script_tag req.payload(0,500),lower -m reg <script>
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015641
Willy Tarreau74ca5042013-06-11 23:12:07 +020015642All ACL-specific criteria imply a default matching method. Most often, these
15643criteria are composed by concatenating the name of the original sample fetch
15644method and the matching method. For example, "hdr_beg" applies the "beg" match
15645to samples retrieved using the "hdr" fetch method. Since all ACL-specific
15646criteria rely on a sample fetch method, it is always possible instead to use
15647the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015648
Willy Tarreau74ca5042013-06-11 23:12:07 +020015649If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015650the matching method is simply applied to the underlying sample fetch method.
15651For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015652
Willy Tarreau74ca5042013-06-11 23:12:07 +020015653 acl short_form hdr_beg(host) www.
15654 acl alternate1 hdr_beg(host) -m beg www.
15655 acl alternate2 hdr_dom(host) -m beg www.
15656 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015657
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015658
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015659The table below summarizes the compatibility matrix between sample or converter
15660types and the pattern types to fetch against. It indicates for each compatible
15661combination the name of the matching method to be used, surrounded with angle
15662brackets ">" and "<" when the method is the default one and will work by
15663default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015664
Willy Tarreau74ca5042013-06-11 23:12:07 +020015665 +-------------------------------------------------+
15666 | Input sample type |
15667 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015668 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015669 +----------------------+---------+---------+---------+---------+---------+
15670 | none (presence only) | found | found | found | found | found |
15671 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015672 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015673 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015674 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015675 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015676 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015677 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015678 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015679 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015680 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015681 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015682 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015683 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015684 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015685 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015686 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015687 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015688 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015689 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015690 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015691 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015692 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015693 +----------------------+---------+---------+---------+---------+---------+
15694 | hex block | | | | bin | bin |
15695 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020015696
15697
Willy Tarreau74ca5042013-06-11 23:12:07 +0200156987.1.1. Matching booleans
15699------------------------
15700
15701In order to match a boolean, no value is needed and all values are ignored.
15702Boolean matching is used by default for all fetch methods of type "boolean".
15703When boolean matching is used, the fetched value is returned as-is, which means
15704that a boolean "true" will always match and a boolean "false" will never match.
15705
15706Boolean matching may also be enforced using "-m bool" on fetch methods which
15707return an integer value. Then, integer value 0 is converted to the boolean
15708"false" and all other values are converted to "true".
15709
Willy Tarreau6a06a402007-07-15 20:15:28 +020015710
Willy Tarreau74ca5042013-06-11 23:12:07 +0200157117.1.2. Matching integers
15712------------------------
15713
15714Integer matching applies by default to integer fetch methods. It can also be
15715enforced on boolean fetches using "-m int". In this case, "false" is converted
15716to the integer 0, and "true" is converted to the integer 1.
15717
15718Integer matching also supports integer ranges and operators. Note that integer
15719matching only applies to positive values. A range is a value expressed with a
15720lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015721
15722For instance, "1024:65535" is a valid range to represent a range of
15723unprivileged ports, and "1024:" would also work. "0:1023" is a valid
15724representation of privileged ports, and ":1023" would also work.
15725
Willy Tarreau62644772008-07-16 18:36:06 +020015726As a special case, some ACL functions support decimal numbers which are in fact
15727two integers separated by a dot. This is used with some version checks for
15728instance. All integer properties apply to those decimal numbers, including
15729ranges and operators.
15730
Willy Tarreau6a06a402007-07-15 20:15:28 +020015731For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010015732operators with ranges does not make much sense and is strongly discouraged.
15733Similarly, it does not make much sense to perform order comparisons with a set
15734of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015735
Willy Tarreau0ba27502007-12-24 16:55:16 +010015736Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020015737
15738 eq : true if the tested value equals at least one value
15739 ge : true if the tested value is greater than or equal to at least one value
15740 gt : true if the tested value is greater than at least one value
15741 le : true if the tested value is less than or equal to at least one value
15742 lt : true if the tested value is less than at least one value
15743
Willy Tarreau0ba27502007-12-24 16:55:16 +010015744For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020015745
Christian Ruppertfd3184c2022-02-20 22:54:01 +010015746 acl negative-length req.hdr_val(content-length) lt 0
Willy Tarreau6a06a402007-07-15 20:15:28 +020015747
Willy Tarreau62644772008-07-16 18:36:06 +020015748This one matches SSL versions between 3.0 and 3.1 (inclusive) :
15749
Christian Ruppertfd3184c2022-02-20 22:54:01 +010015750 acl sslv3 req.ssl_ver 3:3.1
Willy Tarreau62644772008-07-16 18:36:06 +020015751
Willy Tarreau6a06a402007-07-15 20:15:28 +020015752
Willy Tarreau74ca5042013-06-11 23:12:07 +0200157537.1.3. Matching strings
15754-----------------------
15755
15756String matching applies to string or binary fetch methods, and exists in 6
15757different forms :
15758
15759 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015760 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020015761
15762 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015763 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020015764
15765 - prefix match (-m beg) : the patterns are compared with the beginning of
15766 the extracted string, and the ACL matches if any of them matches.
15767
15768 - suffix match (-m end) : the patterns are compared with the end of the
15769 extracted string, and the ACL matches if any of them matches.
15770
Baptiste Assmann33db6002016-03-06 23:32:10 +010015771 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020015772 string, delimited with slashes ("/"), and the ACL matches if any of them
15773 matches.
15774
15775 - domain match (-m dom) : the patterns are looked up inside the extracted
15776 string, delimited with dots ("."), and the ACL matches if any of them
15777 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015778
15779String matching applies to verbatim strings as they are passed, with the
15780exception of the backslash ("\") which makes it possible to escape some
15781characters such as the space. If the "-i" flag is passed before the first
15782string, then the matching will be performed ignoring the case. In order
15783to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010015784before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020015785
Mathias Weiersmuellercb250fc2019-12-02 09:43:40 +010015786Do not use string matches for binary fetches which might contain null bytes
15787(0x00), as the comparison stops at the occurrence of the first null byte.
15788Instead, convert the binary fetch to a hex string with the hex converter first.
15789
15790Example:
15791 # matches if the string <tag> is present in the binary sample
15792 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
15793
Willy Tarreau6a06a402007-07-15 20:15:28 +020015794
Willy Tarreau74ca5042013-06-11 23:12:07 +0200157957.1.4. Matching regular expressions (regexes)
15796---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020015797
15798Just like with string matching, regex matching applies to verbatim strings as
15799they are passed, with the exception of the backslash ("\") which makes it
15800possible to escape some characters such as the space. If the "-i" flag is
15801passed before the first regex, then the matching will be performed ignoring
15802the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010015803the "--" flag before the first string. Same principle applies of course to
15804match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020015805
15806
Willy Tarreau74ca5042013-06-11 23:12:07 +0200158077.1.5. Matching arbitrary data blocks
15808-------------------------------------
15809
15810It is possible to match some extracted samples against a binary block which may
15811not safely be represented as a string. For this, the patterns must be passed as
15812a series of hexadecimal digits in an even number, when the match method is set
15813to binary. Each sequence of two digits will represent a byte. The hexadecimal
15814digits may be used upper or lower case.
15815
15816Example :
15817 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
Christian Ruppertfd3184c2022-02-20 22:54:01 +010015818 acl hello req.payload(0,6) -m bin 48656c6c6f0a
Willy Tarreau74ca5042013-06-11 23:12:07 +020015819
15820
158217.1.6. Matching IPv4 and IPv6 addresses
15822---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020015823
15824IPv4 addresses values can be specified either as plain addresses or with a
15825netmask appended, in which case the IPv4 address matches whenever it is
15826within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010015827host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010015828difficult to read and debug configurations. If hostnames are used, you should
15829at least ensure that they are present in /etc/hosts so that the configuration
15830does not depend on any random DNS match at the moment the configuration is
15831parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015832
Daniel Schnellereba56342016-04-13 00:26:52 +020015833The dotted IPv4 address notation is supported in both regular as well as the
15834abbreviated form with all-0-octets omitted:
15835
15836 +------------------+------------------+------------------+
15837 | Example 1 | Example 2 | Example 3 |
15838 +------------------+------------------+------------------+
15839 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
15840 | 192.168.1 | 10.12 | 127.1 |
15841 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
15842 | 192.168.1/22 | 10.12/8 | 127.1/8 |
15843 +------------------+------------------+------------------+
15844
15845Notice that this is different from RFC 4632 CIDR address notation in which
15846192.168.42/24 would be equivalent to 192.168.42.0/24.
15847
Willy Tarreauceb4ac92012-04-28 00:41:46 +020015848IPv6 may be entered in their usual form, with or without a netmask appended.
15849Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
15850trouble with randomly resolved IP addresses, host names are never allowed in
15851IPv6 patterns.
15852
15853HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
15854following situations :
15855 - tested address is IPv4, pattern address is IPv4, the match applies
15856 in IPv4 using the supplied mask if any.
15857 - tested address is IPv6, pattern address is IPv6, the match applies
15858 in IPv6 using the supplied mask if any.
15859 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
15860 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
15861 ::IPV4 or ::ffff:IPV4, otherwise it fails.
15862 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
15863 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
15864 applied in IPv6 using the supplied IPv6 mask.
15865
Willy Tarreau74ca5042013-06-11 23:12:07 +020015866
158677.2. Using ACLs to form conditions
15868----------------------------------
15869
15870Some actions are only performed upon a valid condition. A condition is a
15871combination of ACLs with operators. 3 operators are supported :
15872
15873 - AND (implicit)
15874 - OR (explicit with the "or" keyword or the "||" operator)
15875 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020015876
Willy Tarreau74ca5042013-06-11 23:12:07 +020015877A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020015878
Willy Tarreau74ca5042013-06-11 23:12:07 +020015879 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020015880
Willy Tarreau74ca5042013-06-11 23:12:07 +020015881Such conditions are generally used after an "if" or "unless" statement,
15882indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020015883
Willy Tarreau74ca5042013-06-11 23:12:07 +020015884For instance, to block HTTP requests to the "*" URL with methods other than
15885"OPTIONS", as well as POST requests without content-length, and GET or HEAD
15886requests with a content-length greater than 0, and finally every request which
15887is not either GET/HEAD/POST/OPTIONS !
15888
Christian Ruppertfd3184c2022-02-20 22:54:01 +010015889 acl missing_cl req.hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015890 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
15891 http-request deny if METH_GET HTTP_CONTENT
15892 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020015893
15894To select a different backend for requests to static contents on the "www" site
15895and to every request on the "img", "video", "download" and "ftp" hosts :
15896
15897 acl url_static path_beg /static /images /img /css
15898 acl url_static path_end .gif .png .jpg .css .js
15899 acl host_www hdr_beg(host) -i www
15900 acl host_static hdr_beg(host) -i img. video. download. ftp.
15901
Davor Ocelice9ed2812017-12-25 17:49:28 +010015902 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020015903 # of host "www". Use backend "www" for the rest.
15904 use_backend static if host_static or host_www url_static
15905 use_backend www if host_www
15906
15907It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
15908expressions that are built on the fly without needing to be declared. They must
15909be enclosed between braces, with a space before and after each brace (because
15910the braces must be seen as independent words). Example :
15911
15912 The following rule :
15913
Christian Ruppertfd3184c2022-02-20 22:54:01 +010015914 acl missing_cl req.hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015915 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020015916
15917 Can also be written that way :
15918
Christian Ruppertfd3184c2022-02-20 22:54:01 +010015919 http-request deny if METH_POST { req.hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020015920
15921It is generally not recommended to use this construct because it's a lot easier
15922to leave errors in the configuration when written that way. However, for very
15923simple rules matching only one source IP address for instance, it can make more
15924sense to use them than to declare ACLs with random names. Another example of
15925good use is the following :
15926
15927 With named ACLs :
15928
15929 acl site_dead nbsrv(dynamic) lt 2
15930 acl site_dead nbsrv(static) lt 2
15931 monitor fail if site_dead
15932
15933 With anonymous ACLs :
15934
15935 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
15936
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015937See section 4.2 for detailed help on the "http-request deny" and "use_backend"
15938keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015939
15940
159417.3. Fetching samples
15942---------------------
15943
15944Historically, sample fetch methods were only used to retrieve data to match
15945against patterns using ACLs. With the arrival of stick-tables, a new class of
15946sample fetch methods was created, most often sharing the same syntax as their
15947ACL counterpart. These sample fetch methods are also known as "fetches". As
15948of now, ACLs and fetches have converged. All ACL fetch methods have been made
15949available as fetch methods, and ACLs may use any sample fetch method as well.
15950
15951This section details all available sample fetch methods and their output type.
15952Some sample fetch methods have deprecated aliases that are used to maintain
15953compatibility with existing configurations. They are then explicitly marked as
15954deprecated and should not be used in new setups.
15955
15956The ACL derivatives are also indicated when available, with their respective
15957matching methods. These ones all have a well defined default pattern matching
15958method, so it is never necessary (though allowed) to pass the "-m" option to
15959indicate how the sample will be matched using ACLs.
15960
15961As indicated in the sample type versus matching compatibility matrix above,
15962when using a generic sample fetch method in an ACL, the "-m" option is
15963mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
15964the same keyword exists as an ACL keyword and as a standard fetch method, the
15965ACL engine will automatically pick the ACL-only one by default.
15966
15967Some of these keywords support one or multiple mandatory arguments, and one or
15968multiple optional arguments. These arguments are strongly typed and are checked
15969when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010015970incorrect argument (e.g. an unresolved backend name). Fetch function arguments
15971are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020015972is optional, it will be indicated below between square brackets ('[ ]'). When
15973all arguments are optional, the parenthesis may be omitted.
15974
15975Thus, the syntax of a standard sample fetch method is one of the following :
15976 - name
15977 - name(arg1)
15978 - name(arg1,arg2)
15979
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015980
159817.3.1. Converters
15982-----------------
15983
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015984Sample fetch methods may be combined with transformations to be applied on top
15985of the fetched sample (also called "converters"). These combinations form what
15986is called "sample expressions" and the result is a "sample". Initially this
15987was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010015988has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015989unique-id-format, add-header, ...).
15990
15991These transformations are enumerated as a series of specific keywords after the
15992sample fetch method. These keywords may equally be appended immediately after
15993the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010015994support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015995
Willy Tarreau97707872015-01-27 15:12:13 +010015996A certain category of converters are bitwise and arithmetic operators which
15997support performing basic operations on integers. Some bitwise operations are
15998supported (and, or, xor, cpl) and some arithmetic operations are supported
15999(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
16000bool) which make it possible to report a match without having to write an ACL.
16001
Willy Tarreau74ca5042013-06-11 23:12:07 +020016002The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010016003
Ben Shillitof25e8e52016-12-02 14:25:37 +00001600451d.single(<prop>[,<prop>*])
16005 Returns values for the properties requested as a string, where values are
16006 separated by the delimiter specified with "51degrees-property-separator".
16007 The device is identified using the User-Agent header passed to the
16008 converter. The function can be passed up to five property names, and if a
16009 property name can't be found, the value "NoData" is returned.
16010
16011 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010016012 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
16013 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000016014 # User-Agent passed to the converter.
16015 frontend http-in
16016 bind *:8081
16017 default_backend servers
16018 http-request set-header X-51D-DeviceTypeMobileTablet \
16019 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
16020
Willy Tarreau97707872015-01-27 15:12:13 +010016021add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016022 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016023 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016024 name. The name of the variable starts with an indication about its scope. The
16025 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016026 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016027 "sess" : the variable is shared with the whole session
16028 "txn" : the variable is shared with the transaction (request and response)
16029 "req" : the variable is shared only during request processing
16030 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016031 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016032 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016033
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010016034aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
16035 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
16036 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
16037 need to be base64 encoded and the returned result is in raw byte format.
16038 If the <aead_tag> validation fails, the converter doesn't return any data.
16039 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
16040 converter requires at least OpenSSL 1.0.1.
16041
16042 Example:
16043 http-response set-header X-Decrypted-Text %[var(txn.enc),\
16044 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
16045
Willy Tarreau97707872015-01-27 15:12:13 +010016046and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016047 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016048 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010016049 numeric value or a variable name. The name of the variable starts with an
16050 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016051 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016052 "sess" : the variable is shared with the whole session
16053 "txn" : the variable is shared with the transaction (request and response)
16054 "req" : the variable is shared only during request processing
16055 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016056 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016057 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016058
Holger Just1bfc24b2017-05-06 00:56:53 +020016059b64dec
16060 Converts (decodes) a base64 encoded input string to its binary
16061 representation. It performs the inverse operation of base64().
Moemen MHEDHBI92f7d432021-04-01 20:53:59 +020016062 For base64url("URL and Filename Safe Alphabet" (RFC 4648)) variant
16063 see "ub64dec".
Holger Just1bfc24b2017-05-06 00:56:53 +020016064
Emeric Brun53d1a982014-04-30 18:21:37 +020016065base64
16066 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010016067 transfer binary content in a way that can be reliably transferred (e.g.
Moemen MHEDHBI92f7d432021-04-01 20:53:59 +020016068 an SSL ID can be copied in a header). For base64url("URL and Filename
16069 Safe Alphabet" (RFC 4648)) variant see "ub64enc".
Emeric Brun53d1a982014-04-30 18:21:37 +020016070
Willy Tarreau97707872015-01-27 15:12:13 +010016071bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016072 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010016073 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016074 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010016075 presence of a flag).
16076
Emeric Brun54c4ac82014-11-03 15:32:43 +010016077bytes(<offset>[,<length>])
16078 Extracts some bytes from an input binary sample. The result is a binary
16079 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010016080 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010016081
Willy Tarreau280f42b2018-02-19 15:34:12 +010016082concat([<start>],[<var>],[<end>])
16083 Concatenates up to 3 fields after the current sample which is then turned to
16084 a string. The first one, <start>, is a constant string, that will be appended
16085 immediately after the existing sample. It may be omitted if not used. The
16086 second one, <var>, is a variable name. The variable will be looked up, its
16087 contents converted to a string, and it will be appended immediately after the
16088 <first> part. If the variable is not found, nothing is appended. It may be
16089 omitted as well. The third field, <end> is a constant string that will be
16090 appended after the variable. It may also be omitted. Together, these elements
16091 allow to concatenate variables with delimiters to an existing set of
16092 variables. This can be used to build new variables made of a succession of
Willy Tarreauef21fac2020-02-14 13:37:20 +010016093 other variables, such as colon-delimited values. If commas or closing
Daniel Corbett67a82712020-07-06 23:01:19 -040016094 parenthesis are needed as delimiters, they must be protected by quotes or
Willy Tarreauef21fac2020-02-14 13:37:20 +010016095 backslashes, themselves protected so that they are not stripped by the first
16096 level parser. See examples below.
Willy Tarreau280f42b2018-02-19 15:34:12 +010016097
16098 Example:
16099 tcp-request session set-var(sess.src) src
16100 tcp-request session set-var(sess.dn) ssl_c_s_dn
16101 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
Willy Tarreauef21fac2020-02-14 13:37:20 +010016102 tcp-request session set-var(txn.ipport) "str(),concat('addr=(',sess.ip),concat(',',sess.port,')')"
Willy Tarreau280f42b2018-02-19 15:34:12 +010016103 http-request set-header x-hap-sig %[var(txn.sig)]
16104
Willy Tarreau97707872015-01-27 15:12:13 +010016105cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016106 Takes the input value of type signed integer, applies a ones-complement
16107 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010016108
Willy Tarreau80599772015-01-20 19:35:24 +010016109crc32([<avalanche>])
16110 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
16111 hash function. Optionally, it is possible to apply a full avalanche hash
16112 function to the output if the optional <avalanche> argument equals 1. This
16113 converter uses the same functions as used by the various hash-based load
16114 balancing algorithms, so it will provide exactly the same results. It is
16115 provided for compatibility with other software which want a CRC32 to be
16116 computed on some input keys, so it follows the most common implementation as
16117 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
16118 but may provide a better or at least less predictable distribution. It must
16119 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010016120 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
16121
16122crc32c([<avalanche>])
16123 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
16124 hash function. Optionally, it is possible to apply a full avalanche hash
16125 function to the output if the optional <avalanche> argument equals 1. This
16126 converter uses the same functions as described in RFC4960, Appendix B [8].
16127 It is provided for compatibility with other software which want a CRC32C to be
16128 computed on some input keys. It is slower than the other algorithms and it must
16129 not be used for security purposes as a 32-bit hash is trivial to break. See
16130 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010016131
Christopher Fauletea159d62020-04-01 16:21:44 +020016132cut_crlf
16133 Cuts the string representation of the input sample on the first carriage
16134 return ('\r') or newline ('\n') character found. Only the string length is
16135 updated.
16136
David Carlier29b3ca32015-09-25 14:09:21 +010016137da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020016138 Asks the DeviceAtlas converter to identify the User Agent string passed on
16139 input, and to emit a string made of the concatenation of the properties
16140 enumerated in argument, delimited by the separator defined by the global
16141 keyword "deviceatlas-property-separator", or by default the pipe character
Daniel Corbett9f0843f2021-05-08 10:50:37 -040016142 ('|'). There's a limit of 12 different properties imposed by the HAProxy
David Carlier4542b102015-06-01 13:54:29 +020016143 configuration language.
16144
16145 Example:
16146 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020016147 bind *:8881
16148 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000016149 http-request set-header X-DeviceAtlas-Data %[req.fhdr(User-Agent),da-csv(primaryHardwareType,osName,osVersion,browserName,browserVersion,browserRenderingEngine)]
David Carlier4542b102015-06-01 13:54:29 +020016150
Willy Tarreau0851fd52019-12-17 10:07:25 +010016151debug([<prefix][,<destination>])
16152 This converter is used as debug tool. It takes a capture of the input sample
16153 and sends it to event sink <destination>, which may designate a ring buffer
16154 such as "buf0", as well as "stdout", or "stderr". Available sinks may be
16155 checked at run time by issuing "show events" on the CLI. When not specified,
16156 the output will be "buf0", which may be consulted via the CLI's "show events"
16157 command. An optional prefix <prefix> may be passed to help distinguish
16158 outputs from multiple expressions. It will then appear before the colon in
16159 the output message. The input sample is passed as-is on the output, so that
16160 it is safe to insert the debug converter anywhere in a chain, even with non-
16161 printable sample types.
16162
16163 Example:
16164 tcp-request connection track-sc0 src,debug(track-sc)
Thierry FOURNIER9687c772015-05-07 15:46:29 +020016165
Patrick Gansterer8e366512020-04-22 16:47:57 +020016166digest(<algorithm>)
16167 Converts a binary input sample to a message digest. The result is a binary
16168 sample. The <algorithm> must be an OpenSSL message digest name (e.g. sha256).
16169
Daniel Corbett9f0843f2021-05-08 10:50:37 -040016170 Please note that this converter is only available when HAProxy has been
Patrick Gansterer8e366512020-04-22 16:47:57 +020016171 compiled with USE_OPENSSL.
16172
Willy Tarreau97707872015-01-27 15:12:13 +010016173div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016174 Divides the input value of type signed integer by <value>, and returns the
16175 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016176 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010016177 variable name. The name of the variable starts with an indication about its
16178 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016179 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016180 "sess" : the variable is shared with the whole session
16181 "txn" : the variable is shared with the transaction (request and response)
16182 "req" : the variable is shared only during request processing
16183 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016184 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016185 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016186
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016187djb2([<avalanche>])
16188 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
16189 hash function. Optionally, it is possible to apply a full avalanche hash
16190 function to the output if the optional <avalanche> argument equals 1. This
16191 converter uses the same functions as used by the various hash-based load
16192 balancing algorithms, so it will provide exactly the same results. It is
16193 mostly intended for debugging, but can be used as a stick-table entry to
16194 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010016195 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
16196 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016197
Willy Tarreau97707872015-01-27 15:12:13 +010016198even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016199 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010016200 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
16201
Marcin Deranek9631a282018-04-16 14:30:46 +020016202field(<index>,<delimiters>[,<count>])
16203 Extracts the substring at the given index counting from the beginning
16204 (positive index) or from the end (negative index) considering given delimiters
16205 from an input string. Indexes start at 1 or -1 and delimiters are a string
16206 formatted list of chars. Optionally you can specify <count> of fields to
16207 extract (default: 1). Value of 0 indicates extraction of all remaining
16208 fields.
16209
16210 Example :
16211 str(f1_f2_f3__f5),field(5,_) # f5
16212 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
16213 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
16214 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
16215 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010016216
Baptiste Assmanne138dda2020-10-22 15:39:03 +020016217fix_is_valid
16218 Parses a binary payload and performs sanity checks regarding FIX (Financial
16219 Information eXchange):
16220
16221 - checks that all tag IDs and values are not empty and the tags IDs are well
16222 numeric
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050016223 - checks the BeginString tag is the first tag with a valid FIX version
Baptiste Assmanne138dda2020-10-22 15:39:03 +020016224 - checks the BodyLength tag is the second one with the right body length
Christopher Fauleted4bef72021-03-18 17:40:56 +010016225 - checks the MsgType tag is the third tag.
Baptiste Assmanne138dda2020-10-22 15:39:03 +020016226 - checks that last tag in the message is the CheckSum tag with a valid
16227 checksum
16228
16229 Due to current HAProxy design, only the first message sent by the client and
16230 the server can be parsed.
16231
16232 This converter returns a boolean, true if the payload contains a valid FIX
16233 message, false if not.
16234
16235 See also the fix_tag_value converter.
16236
16237 Example:
16238 tcp-request inspect-delay 10s
16239 tcp-request content reject unless { req.payload(0,0),fix_is_valid }
16240
16241fix_tag_value(<tag>)
16242 Parses a FIX (Financial Information eXchange) message and extracts the value
16243 from the tag <tag>. <tag> can be a string or an integer pointing to the
16244 desired tag. Any integer value is accepted, but only the following strings
16245 are translated into their integer equivalent: BeginString, BodyLength,
Daniel Corbettbefef702021-03-09 23:00:34 -050016246 MsgType, SenderCompID, TargetCompID, CheckSum. More tag names can be easily
Baptiste Assmanne138dda2020-10-22 15:39:03 +020016247 added.
16248
16249 Due to current HAProxy design, only the first message sent by the client and
16250 the server can be parsed. No message validation is performed by this
16251 converter. It is highly recommended to validate the message first using
16252 fix_is_valid converter.
16253
16254 See also the fix_is_valid converter.
16255
16256 Example:
16257 tcp-request inspect-delay 10s
16258 tcp-request content reject unless { req.payload(0,0),fix_is_valid }
16259 # MsgType tag ID is 35, so both lines below will return the same content
16260 tcp-request content set-var(txn.foo) req.payload(0,0),fix_tag_value(35)
16261 tcp-request content set-var(txn.bar) req.payload(0,0),fix_tag_value(MsgType)
16262
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016263hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010016264 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016265 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010016266 in a way that can be reliably transferred (e.g. an SSL ID can be copied in a
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016267 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010016268
Dragan Dosen3f957b22017-10-24 09:27:34 +020016269hex2i
16270 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050016271 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020016272
Christopher Faulet4ccc12f2020-04-01 09:08:32 +020016273htonl
16274 Converts the input integer value to its 32-bit binary representation in the
16275 network byte order. Because sample fetches own signed 64-bit integer, when
16276 this converter is used, the input integer value is first casted to an
16277 unsigned 32-bit integer.
16278
Tim Duesterhusa3082092021-01-21 17:40:49 +010016279hmac(<algorithm>,<key>)
Patrick Gansterer8e366512020-04-22 16:47:57 +020016280 Converts a binary input sample to a message authentication code with the given
16281 key. The result is a binary sample. The <algorithm> must be one of the
16282 registered OpenSSL message digest names (e.g. sha256). The <key> parameter must
16283 be base64 encoded and can either be a string or a variable.
16284
Daniel Corbett9f0843f2021-05-08 10:50:37 -040016285 Please note that this converter is only available when HAProxy has been
Patrick Gansterer8e366512020-04-22 16:47:57 +020016286 compiled with USE_OPENSSL.
16287
Cyril Bonté6bcd1822019-11-05 23:13:59 +010016288http_date([<offset],[<unit>])
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016289 Converts an integer supposed to contain a date since epoch to a string
16290 representing this date in a format suitable for use in HTTP header fields. If
Damien Claisseae6f1252019-10-30 15:57:28 +000016291 an offset value is specified, then it is added to the date before the
16292 conversion is operated. This is particularly useful to emit Date header fields,
16293 Expires values in responses when combined with a positive offset, or
16294 Last-Modified values when the offset is negative.
16295 If a unit value is specified, then consider the timestamp as either
16296 "s" for seconds (default behavior), "ms" for milliseconds, or "us" for
16297 microseconds since epoch. Offset is assumed to have the same unit as
16298 input timestamp.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016299
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020016300iif(<true>,<false>)
16301 Returns the <true> string if the input value is true. Returns the <false>
16302 string otherwise.
16303
16304 Example:
Tim Duesterhus870713b2020-09-11 17:13:12 +020016305 http-request set-header x-forwarded-proto %[ssl_fc,iif(https,http)]
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020016306
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016307in_table(<table>)
16308 Uses the string representation of the input sample to perform a look up in
16309 the specified table. If the key is not found in the table, a boolean false
16310 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010016311 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016312 or not a source IP address or an Authorization header was already seen).
16313
Tim Duesterhusa3082092021-01-21 17:40:49 +010016314ipmask(<mask4>,[<mask6>])
Tim Duesterhus1478aa72018-01-25 16:24:51 +010016315 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020016316 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010016317 table entries and as such use the same server. The mask4 can be passed in
16318 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
16319 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
16320 If no mask6 is given IPv6 addresses will fail to convert for backwards
16321 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020016322
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016323json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010016324 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016325 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020016326 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016327 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
16328 of errors:
16329 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
16330 bytes, ...)
16331 - invalid range (the decoded value is within a UTF-8 prohibited range),
16332 - code overlong (the value is encoded with more bytes than necessary).
16333
16334 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
16335 character is greater than 0xffff because the JSON string escape specification
16336 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
16337 in 4 variants designated by a combination of two suffix letters : "p" for
16338 "permissive" and "s" for "silently ignore". The behaviors of the decoders
16339 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010016340 - "ascii" : never fails;
16341 - "utf8" : fails on any detected errors;
16342 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016343 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010016344 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016345 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
16346 characters corresponding to the other errors.
16347
16348 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010016349 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016350
16351 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016352 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020016353 capture request header user-agent len 150
16354 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016355
16356 Input request from client 127.0.0.1:
16357 GET / HTTP/1.0
16358 User-Agent: Very "Ugly" UA 1/2
16359
16360 Output log:
16361 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
16362
Alex51c8ad42021-04-15 16:45:15 +020016363json_query(<json_path>,[<output_type>])
16364 The json_query converter supports the JSON types string, boolean and
16365 number. Floating point numbers will be returned as a string. By
16366 specifying the output_type 'int' the value will be converted to an
16367 Integer. If conversion is not possible the json_query converter fails.
16368
16369 <json_path> must be a valid JSON Path string as defined in
16370 https://datatracker.ietf.org/doc/draft-ietf-jsonpath-base/
16371
16372 Example:
16373 # get a integer value from the request body
16374 # "{"integer":4}" => 5
16375 http-request set-var(txn.pay_int) req.body,json_query('$.integer','int'),add(1)
16376
16377 # get a key with '.' in the name
16378 # {"my.key":"myvalue"} => myvalue
16379 http-request set-var(txn.pay_mykey) req.body,json_query('$.my\\.key')
16380
16381 # {"boolean-false":false} => 0
16382 http-request set-var(txn.pay_boolean_false) req.body,json_query('$.boolean-false')
16383
16384 # get the value of the key 'iss' from a JWT Bearer token
16385 http-request set-var(txn.token_payload) req.hdr(Authorization),word(2,.),ub64dec,json_query('$.iss')
16386
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016387language(<value>[,<default>])
16388 Returns the value with the highest q-factor from a list as extracted from the
16389 "accept-language" header using "req.fhdr". Values with no q-factor have a
16390 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
16391 belong to the list of semi-colon delimited <values> will be considered. The
16392 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
16393 given list and a default value is provided, it is returned. Note that language
16394 names may have a variant after a dash ('-'). If this variant is present in the
16395 list, it will be matched, but if it is not, only the base language is checked.
16396 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010016397 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016398 ordering of the values in the request counts, as the first value among
16399 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020016400
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016401 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020016402
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016403 # this configuration switches to the backend matching a
16404 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020016405
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016406 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
16407 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
16408 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
16409 use_backend spanish if es
16410 use_backend french if fr
16411 use_backend english if en
16412 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020016413
Willy Tarreau60a2ee72017-12-15 07:13:48 +010016414length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010016415 Get the length of the string. This can only be placed after a string
16416 sample fetch function or after a transformation keyword returning a string
16417 type. The result is of type integer.
16418
Willy Tarreauffcb2e42014-07-10 16:29:08 +020016419lower
16420 Convert a string sample to lower case. This can only be placed after a string
16421 sample fetch function or after a transformation keyword returning a string
16422 type. The result is of type string.
16423
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016424ltime(<format>[,<offset>])
16425 Converts an integer supposed to contain a date since epoch to a string
16426 representing this date in local time using a format defined by the <format>
16427 string using strftime(3). The purpose is to allow any date format to be used
16428 in logs. An optional <offset> in seconds may be applied to the input date
16429 (positive or negative). See the strftime() man page for the format supported
16430 by your operating system. See also the utime converter.
16431
16432 Example :
16433
16434 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010016435 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016436 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
16437
Christopher Faulet51fc9d12020-04-01 17:24:41 +020016438ltrim(<chars>)
16439 Skips any characters from <chars> from the beginning of the string
16440 representation of the input sample.
16441
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016442map(<map_file>[,<default_value>])
16443map_<match_type>(<map_file>[,<default_value>])
16444map_<match_type>_<output_type>(<map_file>[,<default_value>])
16445 Search the input value from <map_file> using the <match_type> matching method,
16446 and return the associated value converted to the type <output_type>. If the
16447 input value cannot be found in the <map_file>, the converter returns the
16448 <default_value>. If the <default_value> is not set, the converter fails and
16449 acts as if no input value could be fetched. If the <match_type> is not set, it
16450 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
16451 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
16452 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016453
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016454 It is important to avoid overlapping between the keys : IP addresses and
16455 strings are stored in trees, so the first of the finest match will be used.
16456 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016457
Tim Düsterhus4896c442016-11-29 02:15:19 +010016458 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016459 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016460
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016461 input type | match method | output type str | output type int | output type ip
16462 -----------+--------------+-----------------+-----------------+---------------
16463 str | str | map_str | map_str_int | map_str_ip
16464 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020016465 str | beg | map_beg | map_beg_int | map_end_ip
16466 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016467 str | sub | map_sub | map_sub_int | map_sub_ip
16468 -----------+--------------+-----------------+-----------------+---------------
16469 str | dir | map_dir | map_dir_int | map_dir_ip
16470 -----------+--------------+-----------------+-----------------+---------------
16471 str | dom | map_dom | map_dom_int | map_dom_ip
16472 -----------+--------------+-----------------+-----------------+---------------
16473 str | end | map_end | map_end_int | map_end_ip
16474 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080016475 str | reg | map_reg | map_reg_int | map_reg_ip
16476 -----------+--------------+-----------------+-----------------+---------------
16477 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016478 -----------+--------------+-----------------+-----------------+---------------
16479 int | int | map_int | map_int_int | map_int_ip
16480 -----------+--------------+-----------------+-----------------+---------------
16481 ip | ip | map_ip | map_ip_int | map_ip_ip
16482 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016483
Thierry Fournier8feaa662016-02-10 22:55:20 +010016484 The special map called "map_regm" expect matching zone in the regular
16485 expression and modify the output replacing back reference (like "\1") by
16486 the corresponding match text.
16487
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016488 The file contains one key + value per line. Lines which start with '#' are
16489 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
16490 is then the first "word" (series of non-space/tabs characters), and the value
16491 is what follows this series of space/tab till the end of the line excluding
16492 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016493
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016494 Example :
16495
16496 # this is a comment and is ignored
16497 2.22.246.0/23 United Kingdom \n
16498 <-><-----------><--><------------><---->
16499 | | | | `- trailing spaces ignored
16500 | | | `---------- value
16501 | | `-------------------- middle spaces ignored
16502 | `---------------------------- key
16503 `------------------------------------ leading spaces ignored
16504
Willy Tarreau97707872015-01-27 15:12:13 +010016505mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016506 Divides the input value of type signed integer by <value>, and returns the
16507 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016508 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016509 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016510 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016511 "sess" : the variable is shared with the whole session
16512 "txn" : the variable is shared with the transaction (request and response)
16513 "req" : the variable is shared only during request processing
16514 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016515 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016516 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016517
Alexbf1bd5a2021-04-24 13:02:21 +020016518mqtt_field_value(<packettype>,<fieldname_or_property_ID>)
Baptiste Assmanne279ca62020-10-27 18:10:06 +010016519 Returns value of <fieldname> found in input MQTT payload of type
16520 <packettype>.
16521 <packettype> can be either a string (case insensitive matching) or a numeric
16522 value corresponding to the type of packet we're supposed to extract data
16523 from.
16524 Supported string and integers can be found here:
16525 https://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html#_Toc398718021
16526 https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html#_Toc3901022
16527
16528 <fieldname> depends on <packettype> and can be any of the following below.
16529 (note that <fieldname> matching is case insensitive).
16530 <property id> can only be found in MQTT v5.0 streams. check this table:
16531 https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html#_Toc3901029
16532
16533 - CONNECT (or 1): flags, protocol_name, protocol_version, client_identifier,
16534 will_topic, will_payload, username, password, keepalive
16535 OR any property ID as a numeric value (for MQTT v5.0
16536 packets only):
16537 17: Session Expiry Interval
16538 33: Receive Maximum
16539 39: Maximum Packet Size
16540 34: Topic Alias Maximum
16541 25: Request Response Information
16542 23: Request Problem Information
16543 21: Authentication Method
16544 22: Authentication Data
16545 18: Will Delay Interval
16546 1: Payload Format Indicator
16547 2: Message Expiry Interval
16548 3: Content Type
16549 8: Response Topic
16550 9: Correlation Data
16551 Not supported yet:
16552 38: User Property
16553
16554 - CONNACK (or 2): flags, protocol_version, reason_code
16555 OR any property ID as a numeric value (for MQTT v5.0
16556 packets only):
16557 17: Session Expiry Interval
16558 33: Receive Maximum
16559 36: Maximum QoS
16560 37: Retain Available
16561 39: Maximum Packet Size
16562 18: Assigned Client Identifier
16563 34: Topic Alias Maximum
16564 31: Reason String
16565 40; Wildcard Subscription Available
16566 41: Subscription Identifiers Available
16567 42: Shared Subscription Available
16568 19: Server Keep Alive
16569 26: Response Information
16570 28: Server Reference
16571 21: Authentication Method
16572 22: Authentication Data
16573 Not supported yet:
16574 38: User Property
16575
16576 Due to current HAProxy design, only the first message sent by the client and
16577 the server can be parsed. Thus this converter can extract data only from
16578 CONNECT and CONNACK packet types. CONNECT is the first message sent by the
16579 client and CONNACK is the first response sent by the server.
16580
16581 Example:
16582
16583 acl data_in_buffer req.len ge 4
16584 tcp-request content set-var(txn.username) \
16585 req.payload(0,0),mqtt_field_value(connect,protocol_name) \
16586 if data_in_buffer
16587 # do the same as above
16588 tcp-request content set-var(txn.username) \
16589 req.payload(0,0),mqtt_field_value(1,protocol_name) \
16590 if data_in_buffer
16591
16592mqtt_is_valid
16593 Checks that the binary input is a valid MQTT packet. It returns a boolean.
16594
16595 Due to current HAProxy design, only the first message sent by the client and
16596 the server can be parsed. Thus this converter can extract data only from
16597 CONNECT and CONNACK packet types. CONNECT is the first message sent by the
16598 client and CONNACK is the first response sent by the server.
16599
Christopher Fauletc7907732022-03-22 09:41:11 +010016600 Only MQTT 3.1, 3.1.1 and 5.0 are supported.
16601
Baptiste Assmanne279ca62020-10-27 18:10:06 +010016602 Example:
16603
16604 acl data_in_buffer req.len ge 4
Daniel Corbettcc9d9b02021-05-13 10:46:07 -040016605 tcp-request content reject unless { req.payload(0,0),mqtt_is_valid }
Baptiste Assmanne279ca62020-10-27 18:10:06 +010016606
Willy Tarreau97707872015-01-27 15:12:13 +010016607mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016608 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020016609 the product as an signed integer. In case of overflow, the largest possible
16610 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016611 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016612 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016613 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016614 "sess" : the variable is shared with the whole session
16615 "txn" : the variable is shared with the transaction (request and response)
16616 "req" : the variable is shared only during request processing
16617 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016618 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016619 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016620
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010016621nbsrv
16622 Takes an input value of type string, interprets it as a backend name and
16623 returns the number of usable servers in that backend. Can be used in places
16624 where we want to look up a backend from a dynamic name, like a result of a
16625 map lookup.
16626
Willy Tarreau97707872015-01-27 15:12:13 +010016627neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016628 Takes the input value of type signed integer, computes the opposite value,
16629 and returns the remainder as an signed integer. 0 is identity. This operator
16630 is provided for reversed subtracts : in order to subtract the input from a
16631 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010016632
16633not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016634 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010016635 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016636 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010016637 absence of a flag).
16638
16639odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016640 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010016641 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
16642
16643or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016644 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016645 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010016646 numeric value or a variable name. The name of the variable starts with an
16647 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016648 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016649 "sess" : the variable is shared with the whole session
16650 "txn" : the variable is shared with the transaction (request and response)
16651 "req" : the variable is shared only during request processing
16652 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016653 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016654 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016655
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016656protobuf(<field_number>,[<field_type>])
16657 This extracts the protocol buffers message field in raw mode of an input binary
16658 sample representation of a protocol buffer message with <field_number> as field
16659 number (dotted notation) if <field_type> is not present, or as an integer sample
16660 if this field is present (see also "ungrpc" below).
16661 The list of the authorized types is the following one: "int32", "int64", "uint32",
16662 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
16663 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
16664 "float" for the wire type 5. Note that "string" is considered as a length-delimited
16665 type, so it does not require any <field_type> argument to be extracted.
16666 More information may be found here about the protocol buffers message field types:
16667 https://developers.google.com/protocol-buffers/docs/encoding
16668
Willy Tarreauc4dc3502015-01-23 20:39:28 +010016669regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010016670 Applies a regex-based substitution to the input string. It does the same
16671 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
16672 default it will replace in the input string the first occurrence of the
16673 largest part matching the regular expression <regex> with the substitution
16674 string <subst>. It is possible to replace all occurrences instead by adding
16675 the flag "g" in the third argument <flags>. It is also possible to make the
16676 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
16677 string, it is made up from the concatenation of all desired flags. Thus if
16678 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
Willy Tarreauef21fac2020-02-14 13:37:20 +010016679 The first use of this converter is to replace certain characters or sequence
16680 of characters with other ones.
16681
16682 It is highly recommended to enclose the regex part using protected quotes to
16683 improve clarity and never have a closing parenthesis from the regex mixed up
16684 with the parenthesis from the function. Just like in Bourne shell, the first
16685 level of quotes is processed when delimiting word groups on the line, a
16686 second level is usable for argument. It is recommended to use single quotes
16687 outside since these ones do not try to resolve backslashes nor dollar signs.
Willy Tarreau7eda8492015-01-20 19:47:06 +010016688
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010016689 Examples:
Willy Tarreau7eda8492015-01-20 19:47:06 +010016690
16691 # de-duplicate "/" in header "x-path".
16692 # input: x-path: /////a///b/c/xzxyz/
16693 # output: x-path: /a/b/c/xzxyz/
Willy Tarreauef21fac2020-02-14 13:37:20 +010016694 http-request set-header x-path "%[hdr(x-path),regsub('/+','/','g')]"
Willy Tarreau7eda8492015-01-20 19:47:06 +010016695
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010016696 # copy query string to x-query and drop all leading '?', ';' and '&'
16697 http-request set-header x-query "%[query,regsub([?;&]*,'')]"
16698
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010016699 # capture groups and backreferences
16700 # both lines do the same.
Willy Tarreau465dc7d2020-10-08 18:05:56 +020016701 http-request redirect location %[url,'regsub("(foo|bar)([0-9]+)?","\2\1",i)']
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010016702 http-request redirect location %[url,regsub(\"(foo|bar)([0-9]+)?\",\"\2\1\",i)]
16703
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020016704capture-req(<id>)
16705 Capture the string entry in the request slot <id> and returns the entry as
16706 is. If the slot doesn't exist, the capture fails silently.
16707
16708 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020016709 "http-response capture", "capture.req.hdr" and
16710 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020016711
16712capture-res(<id>)
16713 Capture the string entry in the response slot <id> and returns the entry as
16714 is. If the slot doesn't exist, the capture fails silently.
16715
16716 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020016717 "http-response capture", "capture.req.hdr" and
16718 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020016719
Christopher Faulet568415a2020-04-01 17:24:47 +020016720rtrim(<chars>)
16721 Skips any characters from <chars> from the end of the string representation
16722 of the input sample.
16723
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016724sdbm([<avalanche>])
16725 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
16726 hash function. Optionally, it is possible to apply a full avalanche hash
16727 function to the output if the optional <avalanche> argument equals 1. This
16728 converter uses the same functions as used by the various hash-based load
16729 balancing algorithms, so it will provide exactly the same results. It is
16730 mostly intended for debugging, but can be used as a stick-table entry to
16731 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010016732 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
16733 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016734
Tim Duesterhusf38175c2020-06-09 11:48:42 +020016735secure_memcmp(<var>)
16736 Compares the contents of <var> with the input value. Both values are treated
16737 as a binary string. Returns a boolean indicating whether both binary strings
16738 match.
16739
16740 If both binary strings have the same length then the comparison will be
16741 performed in constant time.
16742
Daniel Corbett9f0843f2021-05-08 10:50:37 -040016743 Please note that this converter is only available when HAProxy has been
Tim Duesterhusf38175c2020-06-09 11:48:42 +020016744 compiled with USE_OPENSSL.
16745
16746 Example :
16747
16748 http-request set-var(txn.token) hdr(token)
16749 # Check whether the token sent by the client matches the secret token
16750 # value, without leaking the contents using a timing attack.
16751 acl token_given str(my_secret_token),secure_memcmp(txn.token)
16752
Tim Duesterhusef4e45c2021-01-21 17:40:50 +010016753set-var(<var>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010016754 Sets a variable with the input content and returns the content on the output
16755 as-is. The variable keeps the value and the associated input type. The name of
16756 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016757 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016758 "sess" : the variable is shared with the whole session
16759 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016760 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016761 "req" : the variable is shared only during request processing,
16762 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016763 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016764 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016765
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020016766sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020016767 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020016768 sample with length of 20 bytes.
16769
Tim Duesterhusd4376302019-06-17 12:41:44 +020016770sha2([<bits>])
16771 Converts a binary input sample to a digest in the SHA-2 family. The result
16772 is a binary sample with length of <bits>/8 bytes.
16773
16774 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
16775 SHA-<bits>. The default value is 256.
16776
Daniel Corbett9f0843f2021-05-08 10:50:37 -040016777 Please note that this converter is only available when HAProxy has been
Tim Duesterhusd4376302019-06-17 12:41:44 +020016778 compiled with USE_OPENSSL.
16779
Nenad Merdanovic177adc92019-08-27 01:58:13 +020016780srv_queue
16781 Takes an input value of type string, either a server name or <backend>/<server>
16782 format and returns the number of queued sessions on that server. Can be used
16783 in places where we want to look up queued sessions from a dynamic name, like a
16784 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
16785 persistence or direct a request elsewhere.
16786
Tim Duesterhusca097c12018-04-27 21:18:45 +020016787strcmp(<var>)
16788 Compares the contents of <var> with the input value of type string. Returns
16789 the result as a signed integer compatible with strcmp(3): 0 if both strings
16790 are identical. A value less than 0 if the left string is lexicographically
16791 smaller than the right string or if the left string is shorter. A value greater
16792 than 0 otherwise (right string greater than left string or the right string is
16793 shorter).
16794
Tim Duesterhusf38175c2020-06-09 11:48:42 +020016795 See also the secure_memcmp converter if you need to compare two binary
16796 strings in constant time.
16797
Tim Duesterhusca097c12018-04-27 21:18:45 +020016798 Example :
16799
16800 http-request set-var(txn.host) hdr(host)
16801 # Check whether the client is attempting domain fronting.
16802 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
16803
16804
Willy Tarreau97707872015-01-27 15:12:13 +010016805sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016806 Subtracts <value> from the input value of type signed integer, and returns
16807 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016808 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010016809 or a variable name. The name of the variable starts with an indication about
16810 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016811 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016812 "sess" : the variable is shared with the whole session
16813 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016814 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016815 "req" : the variable is shared only during request processing,
16816 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016817 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016818 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016819
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016820table_bytes_in_rate(<table>)
16821 Uses the string representation of the input sample to perform a look up in
16822 the specified table. If the key is not found in the table, integer value zero
16823 is returned. Otherwise the converter returns the average client-to-server
16824 bytes rate associated with the input sample in the designated table, measured
16825 in amount of bytes over the period configured in the table. See also the
16826 sc_bytes_in_rate sample fetch keyword.
16827
16828
16829table_bytes_out_rate(<table>)
16830 Uses the string representation of the input sample to perform a look up in
16831 the specified table. If the key is not found in the table, integer value zero
16832 is returned. Otherwise the converter returns the average server-to-client
16833 bytes rate associated with the input sample in the designated table, measured
16834 in amount of bytes over the period configured in the table. See also the
16835 sc_bytes_out_rate sample fetch keyword.
16836
16837table_conn_cnt(<table>)
16838 Uses the string representation of the input sample to perform a look up in
16839 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016840 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016841 connections associated with the input sample in the designated table. See
16842 also the sc_conn_cnt sample fetch keyword.
16843
16844table_conn_cur(<table>)
16845 Uses the string representation of the input sample to perform a look up in
16846 the specified table. If the key is not found in the table, integer value zero
16847 is returned. Otherwise the converter returns the current amount of concurrent
16848 tracked connections associated with the input sample in the designated table.
16849 See also the sc_conn_cur sample fetch keyword.
16850
16851table_conn_rate(<table>)
16852 Uses the string representation of the input sample to perform a look up in
16853 the specified table. If the key is not found in the table, integer value zero
16854 is returned. Otherwise the converter returns the average incoming connection
16855 rate associated with the input sample in the designated table. See also the
16856 sc_conn_rate sample fetch keyword.
16857
Thierry FOURNIER236657b2015-08-19 08:25:14 +020016858table_gpt0(<table>)
16859 Uses the string representation of the input sample to perform a look up in
16860 the specified table. If the key is not found in the table, boolean value zero
16861 is returned. Otherwise the converter returns the current value of the first
16862 general purpose tag associated with the input sample in the designated table.
16863 See also the sc_get_gpt0 sample fetch keyword.
16864
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016865table_gpc0(<table>)
16866 Uses the string representation of the input sample to perform a look up in
16867 the specified table. If the key is not found in the table, integer value zero
16868 is returned. Otherwise the converter returns the current value of the first
16869 general purpose counter associated with the input sample in the designated
16870 table. See also the sc_get_gpc0 sample fetch keyword.
16871
16872table_gpc0_rate(<table>)
16873 Uses the string representation of the input sample to perform a look up in
16874 the specified table. If the key is not found in the table, integer value zero
16875 is returned. Otherwise the converter returns the frequency which the gpc0
16876 counter was incremented over the configured period in the table, associated
16877 with the input sample in the designated table. See also the sc_get_gpc0_rate
16878 sample fetch keyword.
16879
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016880table_gpc1(<table>)
16881 Uses the string representation of the input sample to perform a look up in
16882 the specified table. If the key is not found in the table, integer value zero
16883 is returned. Otherwise the converter returns the current value of the second
16884 general purpose counter associated with the input sample in the designated
16885 table. See also the sc_get_gpc1 sample fetch keyword.
16886
16887table_gpc1_rate(<table>)
16888 Uses the string representation of the input sample to perform a look up in
16889 the specified table. If the key is not found in the table, integer value zero
16890 is returned. Otherwise the converter returns the frequency which the gpc1
16891 counter was incremented over the configured period in the table, associated
16892 with the input sample in the designated table. See also the sc_get_gpc1_rate
16893 sample fetch keyword.
16894
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016895table_http_err_cnt(<table>)
16896 Uses the string representation of the input sample to perform a look up in
16897 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016898 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016899 errors associated with the input sample in the designated table. See also the
16900 sc_http_err_cnt sample fetch keyword.
16901
16902table_http_err_rate(<table>)
16903 Uses the string representation of the input sample to perform a look up in
16904 the specified table. If the key is not found in the table, integer value zero
16905 is returned. Otherwise the average rate of HTTP errors associated with the
16906 input sample in the designated table, measured in amount of errors over the
16907 period configured in the table. See also the sc_http_err_rate sample fetch
16908 keyword.
16909
Willy Tarreau826f3ab2021-02-10 12:07:15 +010016910table_http_fail_cnt(<table>)
16911 Uses the string representation of the input sample to perform a look up in
16912 the specified table. If the key is not found in the table, integer value zero
16913 is returned. Otherwise the converter returns the cumulative number of HTTP
16914 failures associated with the input sample in the designated table. See also
16915 the sc_http_fail_cnt sample fetch keyword.
16916
16917table_http_fail_rate(<table>)
16918 Uses the string representation of the input sample to perform a look up in
16919 the specified table. If the key is not found in the table, integer value zero
16920 is returned. Otherwise the average rate of HTTP failures associated with the
16921 input sample in the designated table, measured in amount of failures over the
16922 period configured in the table. See also the sc_http_fail_rate sample fetch
16923 keyword.
16924
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016925table_http_req_cnt(<table>)
16926 Uses the string representation of the input sample to perform a look up in
16927 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016928 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016929 requests associated with the input sample in the designated table. See also
16930 the sc_http_req_cnt sample fetch keyword.
16931
16932table_http_req_rate(<table>)
16933 Uses the string representation of the input sample to perform a look up in
16934 the specified table. If the key is not found in the table, integer value zero
16935 is returned. Otherwise the average rate of HTTP requests associated with the
16936 input sample in the designated table, measured in amount of requests over the
16937 period configured in the table. See also the sc_http_req_rate sample fetch
16938 keyword.
16939
16940table_kbytes_in(<table>)
16941 Uses the string representation of the input sample to perform a look up in
16942 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016943 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016944 to-server data associated with the input sample in the designated table,
16945 measured in kilobytes. The test is currently performed on 32-bit integers,
16946 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
16947 keyword.
16948
16949table_kbytes_out(<table>)
16950 Uses the string representation of the input sample to perform a look up in
16951 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016952 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016953 to-client data associated with the input sample in the designated table,
16954 measured in kilobytes. The test is currently performed on 32-bit integers,
16955 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
16956 keyword.
16957
16958table_server_id(<table>)
16959 Uses the string representation of the input sample to perform a look up in
16960 the specified table. If the key is not found in the table, integer value zero
16961 is returned. Otherwise the converter returns the server ID associated with
16962 the input sample in the designated table. A server ID is associated to a
16963 sample by a "stick" rule when a connection to a server succeeds. A server ID
16964 zero means that no server is associated with this key.
16965
16966table_sess_cnt(<table>)
16967 Uses the string representation of the input sample to perform a look up in
16968 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016969 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016970 sessions associated with the input sample in the designated table. Note that
16971 a session here refers to an incoming connection being accepted by the
16972 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
16973 keyword.
16974
16975table_sess_rate(<table>)
16976 Uses the string representation of the input sample to perform a look up in
16977 the specified table. If the key is not found in the table, integer value zero
16978 is returned. Otherwise the converter returns the average incoming session
16979 rate associated with the input sample in the designated table. Note that a
16980 session here refers to an incoming connection being accepted by the
16981 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
16982 keyword.
16983
16984table_trackers(<table>)
16985 Uses the string representation of the input sample to perform a look up in
16986 the specified table. If the key is not found in the table, integer value zero
16987 is returned. Otherwise the converter returns the current amount of concurrent
16988 connections tracking the same key as the input sample in the designated
16989 table. It differs from table_conn_cur in that it does not rely on any stored
16990 information but on the table's reference count (the "use" value which is
16991 returned by "show table" on the CLI). This may sometimes be more suited for
16992 layer7 tracking. It can be used to tell a server how many concurrent
16993 connections there are from a given address for example. See also the
16994 sc_trackers sample fetch keyword.
16995
Moemen MHEDHBI92f7d432021-04-01 20:53:59 +020016996ub64dec
16997 This converter is the base64url variant of b64dec converter. base64url
16998 encoding is the "URL and Filename Safe Alphabet" variant of base64 encoding.
16999 It is also the encoding used in JWT (JSON Web Token) standard.
17000
17001 Example:
17002 # Decoding a JWT payload:
17003 http-request set-var(txn.token_payload) req.hdr(Authorization),word(2,.),ub64dec
17004
17005ub64enc
17006 This converter is the base64url variant of base64 converter.
17007
Willy Tarreauffcb2e42014-07-10 16:29:08 +020017008upper
17009 Convert a string sample to upper case. This can only be placed after a string
17010 sample fetch function or after a transformation keyword returning a string
17011 type. The result is of type string.
17012
Willy Tarreau62ba9ba2020-04-23 17:54:47 +020017013url_dec([<in_form>])
17014 Takes an url-encoded string provided as input and returns the decoded version
17015 as output. The input and the output are of type string. If the <in_form>
17016 argument is set to a non-zero integer value, the input string is assumed to
17017 be part of a form or query string and the '+' character will be turned into a
17018 space (' '). Otherwise this will only happen after a question mark indicating
17019 a query string ('?').
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020017020
William Dauchy888b0ae2021-01-06 23:39:50 +010017021url_enc([<enc_type>])
17022 Takes a string provided as input and returns the encoded version as output.
17023 The input and the output are of type string. By default the type of encoding
17024 is meant for `query` type. There is no other type supported for now but the
17025 optional argument is here for future changes.
17026
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010017027ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010017028 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010017029 sample representation of a gRPC message with <field_number> as field number
17030 (dotted notation) if <field_type> is not present, or as an integer sample if this
17031 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010017032 The list of the authorized types is the following one: "int32", "int64", "uint32",
17033 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
17034 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
17035 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010017036 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010017037 More information may be found here about the protocol buffers message field types:
17038 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010017039
17040 Example:
17041 // with such a protocol buffer .proto file content adapted from
17042 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
17043
17044 message Point {
17045 int32 latitude = 1;
17046 int32 longitude = 2;
17047 }
17048
17049 message PPoint {
17050 Point point = 59;
17051 }
17052
17053 message Rectangle {
17054 // One corner of the rectangle.
17055 PPoint lo = 48;
17056 // The other corner of the rectangle.
17057 PPoint hi = 49;
17058 }
17059
Peter Gervaidf4c9d22020-06-11 18:05:11 +020017060 let's say a body request is made of a "Rectangle" object value (two PPoint
17061 protocol buffers messages), the four protocol buffers fields could be
17062 extracted with these "ungrpc" directives:
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010017063
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010017064 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
17065 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050017066 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010017067 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
17068
Peter Gervaidf4c9d22020-06-11 18:05:11 +020017069 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010017070
Frédéric Lécaille93d33162019-03-06 09:35:59 +010017071 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010017072
Peter Gervaidf4c9d22020-06-11 18:05:11 +020017073 As a gRPC message is always made of a gRPC header followed by protocol buffers
17074 messages, in the previous example the "latitude" of "lo" first PPoint
17075 could be extracted with these equivalent directives:
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010017076
17077 req.body,ungrpc(48.59),protobuf(1,int32)
17078 req.body,ungrpc(48),protobuf(59.1,int32)
17079 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
17080
Peter Gervaidf4c9d22020-06-11 18:05:11 +020017081 Note that the first convert must be "ungrpc", the remaining ones must be
17082 "protobuf" and only the last one may have or not a second argument to
17083 interpret the previous binary sample.
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010017084
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010017085
Tim Duesterhusef4e45c2021-01-21 17:40:50 +010017086unset-var(<var>)
Christopher Faulet85d79c92016-11-09 16:54:56 +010017087 Unsets a variable if the input content is defined. The name of the variable
17088 starts with an indication about its scope. The scopes allowed are:
17089 "proc" : the variable is shared with the whole process
17090 "sess" : the variable is shared with the whole session
17091 "txn" : the variable is shared with the transaction (request and
17092 response),
17093 "req" : the variable is shared only during request processing,
17094 "res" : the variable is shared only during response processing.
17095 This prefix is followed by a name. The separator is a '.'. The name may only
17096 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
17097
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020017098utime(<format>[,<offset>])
17099 Converts an integer supposed to contain a date since epoch to a string
17100 representing this date in UTC time using a format defined by the <format>
17101 string using strftime(3). The purpose is to allow any date format to be used
17102 in logs. An optional <offset> in seconds may be applied to the input date
17103 (positive or negative). See the strftime() man page for the format supported
17104 by your operating system. See also the ltime converter.
17105
17106 Example :
17107
17108 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010017109 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020017110 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
17111
Marcin Deranek9631a282018-04-16 14:30:46 +020017112word(<index>,<delimiters>[,<count>])
17113 Extracts the nth word counting from the beginning (positive index) or from
17114 the end (negative index) considering given delimiters from an input string.
17115 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
Jerome Magnin88209322020-01-28 13:33:44 +010017116 Delimiters at the beginning or end of the input string are ignored.
Marcin Deranek9631a282018-04-16 14:30:46 +020017117 Optionally you can specify <count> of words to extract (default: 1).
17118 Value of 0 indicates extraction of all remaining words.
17119
17120 Example :
17121 str(f1_f2_f3__f5),word(4,_) # f5
17122 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
17123 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
17124 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
17125 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Jerome Magnin88209322020-01-28 13:33:44 +010017126 str(/f1/f2/f3/f4),word(1,/) # f1
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010017127
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020017128wt6([<avalanche>])
17129 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
17130 hash function. Optionally, it is possible to apply a full avalanche hash
17131 function to the output if the optional <avalanche> argument equals 1. This
17132 converter uses the same functions as used by the various hash-based load
17133 balancing algorithms, so it will provide exactly the same results. It is
17134 mostly intended for debugging, but can be used as a stick-table entry to
17135 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010017136 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
17137 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020017138
Willy Tarreau97707872015-01-27 15:12:13 +010017139xor(<value>)
17140 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020017141 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020017142 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010017143 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010017144 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010017145 "sess" : the variable is shared with the whole session
17146 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020017147 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010017148 "req" : the variable is shared only during request processing,
17149 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020017150 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010017151 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010017152
Dragan Dosen04bf0cc2020-12-22 21:44:33 +010017153xxh3([<seed>])
17154 Hashes a binary input sample into a signed 64-bit quantity using the XXH3
17155 64-bit variant of the XXhash hash function. This hash supports a seed which
17156 defaults to zero but a different value maybe passed as the <seed> argument.
17157 This hash is known to be very good and very fast so it can be used to hash
17158 URLs and/or URL parameters for use as stick-table keys to collect statistics
17159 with a low collision rate, though care must be taken as the algorithm is not
17160 considered as cryptographically secure.
17161
Thierry FOURNIER01e09742016-12-26 11:46:11 +010017162xxh32([<seed>])
17163 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
17164 variant of the XXHash hash function. This hash supports a seed which defaults
17165 to zero but a different value maybe passed as the <seed> argument. This hash
17166 is known to be very good and very fast so it can be used to hash URLs and/or
17167 URL parameters for use as stick-table keys to collect statistics with a low
17168 collision rate, though care must be taken as the algorithm is not considered
17169 as cryptographically secure.
17170
17171xxh64([<seed>])
17172 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
17173 variant of the XXHash hash function. This hash supports a seed which defaults
17174 to zero but a different value maybe passed as the <seed> argument. This hash
17175 is known to be very good and very fast so it can be used to hash URLs and/or
17176 URL parameters for use as stick-table keys to collect statistics with a low
17177 collision rate, though care must be taken as the algorithm is not considered
17178 as cryptographically secure.
17179
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010017180
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200171817.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020017182--------------------------------------------
17183
17184A first set of sample fetch methods applies to internal information which does
17185not even relate to any client information. These ones are sometimes used with
17186"monitor-fail" directives to report an internal status to external watchers.
17187The sample fetch methods described in this section are usable anywhere.
17188
17189always_false : boolean
17190 Always returns the boolean "false" value. It may be used with ACLs as a
17191 temporary replacement for another one when adjusting configurations.
17192
17193always_true : boolean
17194 Always returns the boolean "true" value. It may be used with ACLs as a
17195 temporary replacement for another one when adjusting configurations.
17196
17197avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010017198 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020017199 divided by the number of active servers. The current backend is used if no
17200 backend is specified. This is very similar to "queue" except that the size of
17201 the farm is considered, in order to give a more accurate measurement of the
17202 time it may take for a new connection to be processed. The main usage is with
17203 ACL to return a sorry page to new users when it becomes certain they will get
17204 a degraded service, or to pass to the backend servers in a header so that
17205 they decide to work in degraded mode or to disable some functions to speed up
17206 the processing a bit. Note that in the event there would not be any active
17207 server anymore, twice the number of queued connections would be considered as
17208 the measured value. This is a fair estimate, as we expect one server to get
17209 back soon anyway, but we still prefer to send new traffic to another backend
17210 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
17211 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010017212
Willy Tarreau74ca5042013-06-11 23:12:07 +020017213be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020017214 Applies to the number of currently established connections on the backend,
17215 possibly including the connection being evaluated. If no backend name is
17216 specified, the current one is used. But it is also possible to check another
17217 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040017218 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
17219
17220be_conn_free([<backend>]) : integer
17221 Returns an integer value corresponding to the number of available connections
17222 across available servers in the backend. Queue slots are not included. Backup
17223 servers are also not included, unless all other servers are down. If no
17224 backend name is specified, the current one is used. But it is also possible
17225 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040017226 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
17227 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040017228
17229 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
17230 (meaning unlimited), then this fetch clearly does not make sense, in which
17231 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020017232
Willy Tarreau74ca5042013-06-11 23:12:07 +020017233be_sess_rate([<backend>]) : integer
17234 Returns an integer value corresponding to the sessions creation rate on the
17235 backend, in number of new sessions per second. This is used with ACLs to
17236 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010017237 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020017238 online dictionary). It can also be useful to add this element to logs using a
17239 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017240
17241 Example :
17242 # Redirect to an error page if the dictionary is requested too often
17243 backend dynamic
17244 mode http
17245 acl being_scanned be_sess_rate gt 100
17246 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010017247
Davor Ocelice9ed2812017-12-25 17:49:28 +010017248bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020017249 Returns a binary chain. The input is the hexadecimal representation
17250 of the string.
17251
17252bool(<bool>) : bool
17253 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
17254 'false' and '0' are the same. 'true' and '1' are the same.
17255
Willy Tarreau74ca5042013-06-11 23:12:07 +020017256connslots([<backend>]) : integer
17257 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017258 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020017259 connections on all servers and the maximum queue size. This is probably only
17260 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050017261
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080017262 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020017263 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080017264 usage; see "use_backend" keyword) can be redirected to a different backend.
17265
Willy Tarreau55165fe2009-05-10 12:02:55 +020017266 'connslots' = number of available server connection slots, + number of
17267 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080017268
Willy Tarreaua36af912009-10-10 12:02:45 +020017269 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020017270 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020017271 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020017272 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010017273 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020017274 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020017275 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080017276
Willy Tarreau55165fe2009-05-10 12:02:55 +020017277 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
17278 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020017279 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020017280 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080017281
Willy Tarreau70fe9442018-11-22 16:07:39 +010017282cpu_calls : integer
17283 Returns the number of calls to the task processing the stream or current
17284 request since it was allocated. This number is reset for each new request on
17285 the same connections in case of HTTP keep-alive. This value should usually be
17286 low and stable (around 2 calls for a typically simple request) but may become
17287 high if some processing (compression, caching or analysis) is performed. This
17288 is purely for performance monitoring purposes.
17289
17290cpu_ns_avg : integer
17291 Returns the average number of nanoseconds spent in each call to the task
17292 processing the stream or current request. This number is reset for each new
17293 request on the same connections in case of HTTP keep-alive. This value
17294 indicates the overall cost of processing the request or the connection for
17295 each call. There is no good nor bad value but the time spent in a call
17296 automatically causes latency for other processing (see lat_ns_avg below),
17297 and may affect other connection's apparent response time. Certain operations
17298 like compression, complex regex matching or heavy Lua operations may directly
17299 affect this value, and having it in the logs will make it easier to spot the
17300 faulty processing that needs to be fixed to recover decent performance.
17301 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
17302
17303cpu_ns_tot : integer
17304 Returns the total number of nanoseconds spent in each call to the task
17305 processing the stream or current request. This number is reset for each new
17306 request on the same connections in case of HTTP keep-alive. This value
17307 indicates the overall cost of processing the request or the connection for
17308 each call. There is no good nor bad value but the time spent in a call
17309 automatically causes latency for other processing (see lat_ns_avg below),
17310 induces CPU costs on the machine, and may affect other connection's apparent
17311 response time. Certain operations like compression, complex regex matching or
17312 heavy Lua operations may directly affect this value, and having it in the
17313 logs will make it easier to spot the faulty processing that needs to be fixed
17314 to recover decent performance. The value may be artificially high due to a
17315 high cpu_calls count, for example when processing many HTTP chunks, and for
17316 this reason it is often preferred to log cpu_ns_avg instead.
17317
Cyril Bonté6bcd1822019-11-05 23:13:59 +010017318date([<offset>],[<unit>]) : integer
Willy Tarreau6236d3a2013-07-25 14:28:25 +020017319 Returns the current date as the epoch (number of seconds since 01/01/1970).
Damien Claisseae6f1252019-10-30 15:57:28 +000017320
17321 If an offset value is specified, then it is added to the current date before
17322 returning the value. This is particularly useful to compute relative dates,
17323 as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020017324 It is useful combined with the http_date converter.
17325
Damien Claisseae6f1252019-10-30 15:57:28 +000017326 <unit> is facultative, and can be set to "s" for seconds (default behavior),
17327 "ms" for milliseconds or "us" for microseconds.
17328 If unit is set, return value is an integer reflecting either seconds,
17329 milliseconds or microseconds since epoch, plus offset.
17330 It is useful when a time resolution of less than a second is needed.
17331
Willy Tarreau276fae92013-07-25 14:36:01 +020017332 Example :
17333
17334 # set an expires header to now+1 hour in every response
17335 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020017336
Damien Claisseae6f1252019-10-30 15:57:28 +000017337 # set an expires header to now+1 hour in every response, with
17338 # millisecond granularity
17339 http-response set-header Expires %[date(3600000,ms),http_date(0,ms)]
17340
Etienne Carrierea792a0a2018-01-17 13:43:24 +010017341date_us : integer
17342 Return the microseconds part of the date (the "second" part is returned by
17343 date sample). This sample is coherent with the date sample as it is comes
17344 from the same timeval structure.
17345
Willy Tarreaud716f9b2017-10-13 11:03:15 +020017346distcc_body(<token>[,<occ>]) : binary
17347 Parses a distcc message and returns the body associated to occurrence #<occ>
17348 of the token <token>. Occurrences start at 1, and when unspecified, any may
17349 match though in practice only the first one is checked for now. This can be
17350 used to extract file names or arguments in files built using distcc through
Daniel Corbett9f0843f2021-05-08 10:50:37 -040017351 HAProxy. Please refer to distcc's protocol documentation for the complete
Willy Tarreaud716f9b2017-10-13 11:03:15 +020017352 list of supported tokens.
17353
17354distcc_param(<token>[,<occ>]) : integer
17355 Parses a distcc message and returns the parameter associated to occurrence
17356 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
17357 any may match though in practice only the first one is checked for now. This
17358 can be used to extract certain information such as the protocol version, the
Daniel Corbett9f0843f2021-05-08 10:50:37 -040017359 file size or the argument in files built using distcc through HAProxy.
Willy Tarreaud716f9b2017-10-13 11:03:15 +020017360 Another use case consists in waiting for the start of the preprocessed file
17361 contents before connecting to the server to avoid keeping idle connections.
17362 Please refer to distcc's protocol documentation for the complete list of
17363 supported tokens.
17364
17365 Example :
17366 # wait up to 20s for the pre-processed file to be uploaded
17367 tcp-request inspect-delay 20s
17368 tcp-request content accept if { distcc_param(DOTI) -m found }
17369 # send large files to the big farm
17370 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
17371
Willy Tarreau595ec542013-06-12 21:34:28 +020017372env(<name>) : string
17373 Returns a string containing the value of environment variable <name>. As a
17374 reminder, environment variables are per-process and are sampled when the
17375 process starts. This can be useful to pass some information to a next hop
17376 server, or with ACLs to take specific action when the process is started a
17377 certain way.
17378
17379 Examples :
17380 # Pass the Via header to next hop with the local hostname in it
17381 http-request add-header Via 1.1\ %[env(HOSTNAME)]
17382
17383 # reject cookie-less requests when the STOP environment variable is set
Christian Ruppertfd3184c2022-02-20 22:54:01 +010017384 http-request deny if !{ req.cook(SESSIONID) -m found } { env(STOP) -m found }
Willy Tarreau595ec542013-06-12 21:34:28 +020017385
Willy Tarreau74ca5042013-06-11 23:12:07 +020017386fe_conn([<frontend>]) : integer
17387 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010017388 possibly including the connection being evaluated. If no frontend name is
17389 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020017390 frontend. It can be used to return a sorry page before hard-blocking, or to
17391 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010017392 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020017393 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
17394 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020017395
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020017396fe_req_rate([<frontend>]) : integer
17397 Returns an integer value corresponding to the number of HTTP requests per
17398 second sent to a frontend. This number can differ from "fe_sess_rate" in
17399 situations where client-side keep-alive is enabled.
17400
Willy Tarreau74ca5042013-06-11 23:12:07 +020017401fe_sess_rate([<frontend>]) : integer
17402 Returns an integer value corresponding to the sessions creation rate on the
17403 frontend, in number of new sessions per second. This is used with ACLs to
17404 limit the incoming session rate to an acceptable range in order to prevent
17405 abuse of service at the earliest moment, for example when combined with other
17406 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
17407 down below the limit. It can also be useful to add this element to logs using
17408 a log-format directive. See also the "rate-limit sessions" directive for use
17409 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010017410
17411 Example :
17412 # This frontend limits incoming mails to 10/s with a max of 100
17413 # concurrent connections. We accept any connection below 10/s, and
17414 # force excess clients to wait for 100 ms. Since clients are limited to
17415 # 100 max, there cannot be more than 10 incoming mails per second.
17416 frontend mail
17417 bind :25
17418 mode tcp
17419 maxconn 100
17420 acl too_fast fe_sess_rate ge 10
17421 tcp-request inspect-delay 100ms
17422 tcp-request content accept if ! too_fast
17423 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010017424
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010017425hostname : string
17426 Returns the system hostname.
17427
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020017428int(<integer>) : signed integer
17429 Returns a signed integer.
17430
Thierry FOURNIERcc103292015-06-06 19:30:17 +020017431ipv4(<ipv4>) : ipv4
17432 Returns an ipv4.
17433
17434ipv6(<ipv6>) : ipv6
17435 Returns an ipv6.
17436
Willy Tarreau70fe9442018-11-22 16:07:39 +010017437lat_ns_avg : integer
17438 Returns the average number of nanoseconds spent between the moment the task
17439 handling the stream is woken up and the moment it is effectively called. This
17440 number is reset for each new request on the same connections in case of HTTP
17441 keep-alive. This value indicates the overall latency inflicted to the current
17442 request by all other requests being processed in parallel, and is a direct
17443 indicator of perceived performance due to noisy neighbours. In order to keep
17444 the value low, it is possible to reduce the scheduler's run queue depth using
17445 "tune.runqueue-depth", to reduce the number of concurrent events processed at
17446 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020017447 the "nice" option on the "bind" lines or in the frontend, to enable low
17448 latency scheduling using "tune.sched.low-latency", or to look for other heavy
17449 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
17450 processing needs to be adjusted or fixed. Compression of large buffers could
17451 be a culprit, like heavy regex or long lists of regex. Note: this value is
17452 exactly lat_ns_tot divided by cpu_calls.
Willy Tarreau70fe9442018-11-22 16:07:39 +010017453
17454lat_ns_tot : integer
17455 Returns the total number of nanoseconds spent between the moment the task
17456 handling the stream is woken up and the moment it is effectively called. This
17457 number is reset for each new request on the same connections in case of HTTP
17458 keep-alive. This value indicates the overall latency inflicted to the current
17459 request by all other requests being processed in parallel, and is a direct
17460 indicator of perceived performance due to noisy neighbours. In order to keep
17461 the value low, it is possible to reduce the scheduler's run queue depth using
17462 "tune.runqueue-depth", to reduce the number of concurrent events processed at
17463 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020017464 the "nice" option on the "bind" lines or in the frontend, to enable low
17465 latency scheduling using "tune.sched.low-latency", or to look for other heavy
17466 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
17467 processing needs to be adjusted or fixed. Compression of large buffers could
17468 be a culprit, like heavy regex or long lists of regex. Note: while it
Willy Tarreau70fe9442018-11-22 16:07:39 +010017469 may intuitively seem that the total latency adds to a transfer time, it is
17470 almost never true because while a task waits for the CPU, network buffers
17471 continue to fill up and the next call will process more at once. The value
17472 may be artificially high due to a high cpu_calls count, for example when
17473 processing many HTTP chunks, and for this reason it is often preferred to log
17474 lat_ns_avg instead, which is a more relevant performance indicator.
17475
Thierry FOURNIERcc103292015-06-06 19:30:17 +020017476meth(<method>) : method
17477 Returns a method.
17478
Willy Tarreau0f30d262014-11-24 16:02:05 +010017479nbproc : integer
17480 Returns an integer value corresponding to the number of processes that were
17481 started (it equals the global "nbproc" setting). This is useful for logging
17482 and debugging purposes.
17483
Willy Tarreau74ca5042013-06-11 23:12:07 +020017484nbsrv([<backend>]) : integer
17485 Returns an integer value corresponding to the number of usable servers of
17486 either the current backend or the named backend. This is mostly used with
17487 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010017488 switch to an alternate backend when the number of servers is too low to
17489 to handle some load. It is useful to report a failure when combined with
17490 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010017491
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040017492prio_class : integer
17493 Returns the priority class of the current session for http mode or connection
17494 for tcp mode. The value will be that set by the last call to "http-request
17495 set-priority-class" or "tcp-request content set-priority-class".
17496
17497prio_offset : integer
17498 Returns the priority offset of the current session for http mode or
17499 connection for tcp mode. The value will be that set by the last call to
17500 "http-request set-priority-offset" or "tcp-request content
17501 set-priority-offset".
17502
Willy Tarreau0f30d262014-11-24 16:02:05 +010017503proc : integer
17504 Returns an integer value corresponding to the position of the process calling
17505 the function, between 1 and global.nbproc. This is useful for logging and
17506 debugging purposes.
17507
Willy Tarreau74ca5042013-06-11 23:12:07 +020017508queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010017509 Returns the total number of queued connections of the designated backend,
17510 including all the connections in server queues. If no backend name is
17511 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020017512 one. This is useful with ACLs or to pass statistics to backend servers. This
17513 can be used to take actions when queuing goes above a known level, generally
17514 indicating a surge of traffic or a massive slowdown on the servers. One
17515 possible action could be to reject new users but still accept old ones. See
17516 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
17517
Willy Tarreau84310e22014-02-14 11:59:04 +010017518rand([<range>]) : integer
17519 Returns a random integer value within a range of <range> possible values,
17520 starting at zero. If the range is not specified, it defaults to 2^32, which
17521 gives numbers between 0 and 4294967295. It can be useful to pass some values
17522 needed to take some routing decisions for example, or just for debugging
17523 purposes. This random must not be used for security purposes.
17524
Willy Tarreau74ca5042013-06-11 23:12:07 +020017525srv_conn([<backend>/]<server>) : integer
17526 Returns an integer value corresponding to the number of currently established
17527 connections on the designated server, possibly including the connection being
17528 evaluated. If <backend> is omitted, then the server is looked up in the
17529 current backend. It can be used to use a specific farm when one server is
17530 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040017531 connections with it. See also the "fe_conn", "be_conn", "queue", and
17532 "srv_conn_free" fetch methods.
17533
17534srv_conn_free([<backend>/]<server>) : integer
17535 Returns an integer value corresponding to the number of available connections
17536 on the designated server, possibly including the connection being evaluated.
17537 The value does not include queue slots. If <backend> is omitted, then the
17538 server is looked up in the current backend. It can be used to use a specific
17539 farm when one server is full, or to inform the server about our view of the
17540 number of active connections with it. See also the "be_conn_free" and
17541 "srv_conn" fetch methods.
17542
17543 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
17544 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017545
17546srv_is_up([<backend>/]<server>) : boolean
17547 Returns true when the designated server is UP, and false when it is either
17548 DOWN or in maintenance mode. If <backend> is omitted, then the server is
17549 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010017550 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017551 availability). Another possible use which is more of a hack consists in
17552 using dummy servers as boolean variables that can be enabled or disabled from
17553 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
17554
Willy Tarreauff2b7af2017-10-13 11:46:26 +020017555srv_queue([<backend>/]<server>) : integer
17556 Returns an integer value corresponding to the number of connections currently
17557 pending in the designated server's queue. If <backend> is omitted, then the
17558 server is looked up in the current backend. It can sometimes be used together
17559 with the "use-server" directive to force to use a known faster server when it
17560 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
17561 fetch methods.
17562
Willy Tarreau74ca5042013-06-11 23:12:07 +020017563srv_sess_rate([<backend>/]<server>) : integer
17564 Returns an integer corresponding to the sessions creation rate on the
17565 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017566 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020017567 used with ACLs but can make sense with logs too. This is used to switch to an
17568 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010017569 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020017570 overloading servers).
17571
17572 Example :
17573 # Redirect to a separate back
17574 acl srv1_full srv_sess_rate(be1/srv1) gt 50
17575 acl srv2_full srv_sess_rate(be1/srv2) gt 50
17576 use_backend be2 if srv1_full or srv2_full
17577
Alexbf1bd5a2021-04-24 13:02:21 +020017578srv_iweight([<backend>/]<server>) : integer
Christopher Faulet1bea8652020-07-10 16:03:45 +020017579 Returns an integer corresponding to the server's initial weight. If <backend>
17580 is omitted, then the server is looked up in the current backend. See also
17581 "srv_weight" and "srv_uweight".
17582
Alexbf1bd5a2021-04-24 13:02:21 +020017583srv_uweight([<backend>/]<server>) : integer
Christopher Faulet1bea8652020-07-10 16:03:45 +020017584 Returns an integer corresponding to the user visible server's weight. If
17585 <backend> is omitted, then the server is looked up in the current
17586 backend. See also "srv_weight" and "srv_iweight".
17587
Alexbf1bd5a2021-04-24 13:02:21 +020017588srv_weight([<backend>/]<server>) : integer
Christopher Faulet1bea8652020-07-10 16:03:45 +020017589 Returns an integer corresponding to the current (or effective) server's
17590 weight. If <backend> is omitted, then the server is looked up in the current
17591 backend. See also "srv_iweight" and "srv_uweight".
17592
Willy Tarreau0f30d262014-11-24 16:02:05 +010017593stopping : boolean
17594 Returns TRUE if the process calling the function is currently stopping. This
17595 can be useful for logging, or for relaxing certain checks or helping close
17596 certain connections upon graceful shutdown.
17597
Thierry FOURNIERcc103292015-06-06 19:30:17 +020017598str(<string>) : string
17599 Returns a string.
17600
Willy Tarreau74ca5042013-06-11 23:12:07 +020017601table_avl([<table>]) : integer
17602 Returns the total number of available entries in the current proxy's
17603 stick-table or in the designated stick-table. See also table_cnt.
17604
17605table_cnt([<table>]) : integer
17606 Returns the total number of entries currently in use in the current proxy's
17607 stick-table or in the designated stick-table. See also src_conn_cnt and
17608 table_avl for other entry counting methods.
17609
Christopher Faulet34adb2a2017-11-21 21:45:38 +010017610thread : integer
17611 Returns an integer value corresponding to the position of the thread calling
17612 the function, between 0 and (global.nbthread-1). This is useful for logging
17613 and debugging purposes.
17614
Alexandar Lazica429ad32021-06-01 00:27:01 +020017615uuid([<version>]) : string
17616 Returns a UUID following the RFC4122 standard. If the version is not
17617 specified, a UUID version 4 (fully random) is returned.
17618 Currently, only version 4 is supported.
17619
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017620var(<var-name>) : undefined
17621 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010017622 sample fetch fails. The name of the variable starts with an indication
17623 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010017624 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010017625 "sess" : the variable is shared with the whole session
17626 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017627 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010017628 "req" : the variable is shared only during request processing,
17629 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017630 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010017631 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017632
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200176337.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020017634----------------------------------
17635
Daniel Corbett9f0843f2021-05-08 10:50:37 -040017636The layer 4 usually describes just the transport layer which in HAProxy is
Willy Tarreau74ca5042013-06-11 23:12:07 +020017637closest to the connection, where no content is yet made available. The fetch
17638methods described here are usable as low as the "tcp-request connection" rule
17639sets unless they require some future information. Those generally include
17640TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017641the incoming connection. For retrieving a value from a sticky counters, the
17642counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020017643"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
17644used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
17645can be specified as the first integer argument when using the "sc_" prefix.
17646Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
17647table may be specified with the "sc*" form, in which case the currently
17648tracked key will be looked up into this alternate table instead of the table
17649currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017650
Christopher Faulet7d081f02021-04-15 09:38:37 +020017651bc_dst : ip
17652 This is the destination ip address of the connection on the server side,
17653 which is the server address HAProxy connected to. It is of type IP and works
17654 on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 address is mapped to its
17655 IPv6 equivalent, according to RFC 4291.
17656
17657bc_dst_port : integer
17658 Returns an integer value corresponding to the destination TCP port of the
Daniel Corbett9f0843f2021-05-08 10:50:37 -040017659 connection on the server side, which is the port HAProxy connected to.
Christopher Faulet7d081f02021-04-15 09:38:37 +020017660
Jérôme Magnin35e53a62019-01-16 14:38:37 +010017661bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010017662 Returns the backend connection's HTTP major version encoding, which may be 1
17663 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
17664 encoding and not the version present in the request header.
17665
Christopher Faulet7d081f02021-04-15 09:38:37 +020017666bc_src : ip
17667 This is the source ip address of the connection on the server side, which is
Daniel Corbett9f0843f2021-05-08 10:50:37 -040017668 the server address HAProxy connected from. It is of type IP and works on both
Christopher Faulet7d081f02021-04-15 09:38:37 +020017669 IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are mapped to their IPv6
17670 equivalent, according to RFC 4291.
17671
17672bc_src_port : integer
17673 Returns an integer value corresponding to the TCP source port of the
Daniel Corbett9f0843f2021-05-08 10:50:37 -040017674 connection on the server side, which is the port HAProxy connected from.
Christopher Faulet7d081f02021-04-15 09:38:37 +020017675
Willy Tarreau74ca5042013-06-11 23:12:07 +020017676be_id : integer
17677 Returns an integer containing the current backend's id. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020017678 frontends with responses to check which backend processed the request. It can
17679 also be used in a tcp-check or an http-check ruleset.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017680
Marcin Deranekd2471c22016-12-12 14:08:05 +010017681be_name : string
17682 Returns a string containing the current backend's name. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020017683 frontends with responses to check which backend processed the request. It can
17684 also be used in a tcp-check or an http-check ruleset.
Marcin Deranekd2471c22016-12-12 14:08:05 +010017685
Amaury Denoyelled91d7792020-12-10 13:43:56 +010017686be_server_timeout : integer
17687 Returns the configuration value in millisecond for the server timeout of the
17688 current backend. This timeout can be overwritten by a "set-timeout" rule. See
17689 also the "cur_server_timeout".
17690
17691be_tunnel_timeout : integer
17692 Returns the configuration value in millisecond for the tunnel timeout of the
17693 current backend. This timeout can be overwritten by a "set-timeout" rule. See
17694 also the "cur_tunnel_timeout".
17695
Amaury Denoyellef7719a22020-12-10 13:43:58 +010017696cur_server_timeout : integer
17697 Returns the currently applied server timeout in millisecond for the stream.
17698 In the default case, this will be equal to be_server_timeout unless a
17699 "set-timeout" rule has been applied. See also "be_server_timeout".
17700
17701cur_tunnel_timeout : integer
17702 Returns the currently applied tunnel timeout in millisecond for the stream.
17703 In the default case, this will be equal to be_tunnel_timeout unless a
17704 "set-timeout" rule has been applied. See also "be_tunnel_timeout".
17705
Willy Tarreau74ca5042013-06-11 23:12:07 +020017706dst : ip
17707 This is the destination IPv4 address of the connection on the client side,
17708 which is the address the client connected to. It can be useful when running
17709 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
17710 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010017711 RFC 4291. When the incoming connection passed through address translation or
17712 redirection involving connection tracking, the original destination address
17713 before the redirection will be reported. On Linux systems, the source and
17714 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
17715 is set, because a late response may reopen a timed out connection and switch
17716 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017717
17718dst_conn : integer
17719 Returns an integer value corresponding to the number of currently established
17720 connections on the same socket including the one being evaluated. It is
17721 normally used with ACLs but can as well be used to pass the information to
17722 servers in an HTTP header or in logs. It can be used to either return a sorry
17723 page before hard-blocking, or to use a specific backend to drain new requests
17724 when the socket is considered saturated. This offers the ability to assign
17725 different limits to different listening ports or addresses. See also the
17726 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017727
Willy Tarreau16e01562016-08-09 16:46:18 +020017728dst_is_local : boolean
17729 Returns true if the destination address of the incoming connection is local
17730 to the system, or false if the address doesn't exist on the system, meaning
17731 that it was intercepted in transparent mode. It can be useful to apply
17732 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010017733 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020017734 be delivered only on this address, or SSH access could be locally redirected.
17735 Please note that the check involves a few system calls, so it's better to do
17736 it only once per connection.
17737
Willy Tarreau74ca5042013-06-11 23:12:07 +020017738dst_port : integer
17739 Returns an integer value corresponding to the destination TCP port of the
17740 connection on the client side, which is the port the client connected to.
17741 This might be used when running in transparent mode, when assigning dynamic
17742 ports to some clients for a whole application session, to stick all users to
17743 a same server, or to pass the destination port information to a server using
17744 an HTTP header.
17745
Christopher Faulet9f074f62021-10-25 16:18:15 +020017746fc_fackets : integer
17747 Returns the fack counter measured by the kernel for the client
17748 connection. If the server connection is not established, if the connection is
17749 not TCP or if the operating system does not support TCP_INFO, for example
17750 Linux kernels before 2.4, the sample fetch fails.
17751
Willy Tarreau60ca10a2017-08-18 15:26:54 +020017752fc_http_major : integer
17753 Reports the front connection's HTTP major version encoding, which may be 1
17754 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
17755 encoding and not on the version present in the request header.
17756
Christopher Faulet9f074f62021-10-25 16:18:15 +020017757fc_lost : integer
17758 Returns the lost counter measured by the kernel for the client
17759 connection. If the server connection is not established, if the connection is
17760 not TCP or if the operating system does not support TCP_INFO, for example
17761 Linux kernels before 2.4, the sample fetch fails.
17762
Geoff Simmons7185b782019-08-27 18:31:16 +020017763fc_pp_authority : string
17764 Returns the authority TLV sent by the client in the PROXY protocol header,
17765 if any.
17766
Tim Duesterhusd1b15b62020-03-13 12:34:23 +010017767fc_pp_unique_id : string
17768 Returns the unique ID TLV sent by the client in the PROXY protocol header,
17769 if any.
17770
Emeric Brun4f603012017-01-05 15:11:44 +010017771fc_rcvd_proxy : boolean
17772 Returns true if the client initiated the connection with a PROXY protocol
17773 header.
17774
Christopher Faulet9f074f62021-10-25 16:18:15 +020017775fc_reordering : integer
17776 Returns the reordering counter measured by the kernel for the client
17777 connection. If the server connection is not established, if the connection is
17778 not TCP or if the operating system does not support TCP_INFO, for example
17779 Linux kernels before 2.4, the sample fetch fails.
17780
17781fc_retrans : integer
17782 Returns the retransmits counter measured by the kernel for the client
17783 connection. If the server connection is not established, if the connection is
17784 not TCP or if the operating system does not support TCP_INFO, for example
17785 Linux kernels before 2.4, the sample fetch fails.
17786
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020017787fc_rtt(<unit>) : integer
17788 Returns the Round Trip Time (RTT) measured by the kernel for the client
17789 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
17790 can be set to "ms" for milliseconds or "us" for microseconds. If the server
17791 connection is not established, if the connection is not TCP or if the
17792 operating system does not support TCP_INFO, for example Linux kernels before
17793 2.4, the sample fetch fails.
17794
17795fc_rttvar(<unit>) : integer
17796 Returns the Round Trip Time (RTT) variance measured by the kernel for the
17797 client connection. <unit> is facultative, by default the unit is milliseconds.
17798 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
17799 server connection is not established, if the connection is not TCP or if the
17800 operating system does not support TCP_INFO, for example Linux kernels before
17801 2.4, the sample fetch fails.
17802
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017803fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017804 Returns the sacked counter measured by the kernel for the client connection.
17805 If the server connection is not established, if the connection is not TCP or
17806 if the operating system does not support TCP_INFO, for example Linux kernels
17807 before 2.4, the sample fetch fails.
17808
Joe Williams30fcd392016-08-10 07:06:44 -070017809
Christopher Faulet9f074f62021-10-25 16:18:15 +020017810fc_unacked : integer
17811 Returns the unacked counter measured by the kernel for the client connection.
17812 If the server connection is not established, if the connection is not TCP or
17813 if the operating system does not support TCP_INFO, for example Linux kernels
17814 before 2.4, the sample fetch fails.
Joe Williams30fcd392016-08-10 07:06:44 -070017815
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020017816fe_defbe : string
17817 Returns a string containing the frontend's default backend name. It can be
17818 used in frontends to check which backend will handle requests by default.
17819
Willy Tarreau74ca5042013-06-11 23:12:07 +020017820fe_id : integer
17821 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010017822 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020017823 coming via a same frontend to the same server.
17824
Marcin Deranekd2471c22016-12-12 14:08:05 +010017825fe_name : string
17826 Returns a string containing the current frontend's name. It can be used in
17827 backends to check from which frontend it was called, or to stick all users
17828 coming via a same frontend to the same server.
17829
Amaury Denoyelleda184d52020-12-10 13:43:55 +010017830fe_client_timeout : integer
17831 Returns the configuration value in millisecond for the client timeout of the
17832 current frontend.
17833
Cyril Bonté62ba8702014-04-22 23:52:25 +020017834sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017835sc0_bytes_in_rate([<table>]) : integer
17836sc1_bytes_in_rate([<table>]) : integer
17837sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017838 Returns the average client-to-server bytes rate from the currently tracked
17839 counters, measured in amount of bytes over the period configured in the
17840 table. See also src_bytes_in_rate.
17841
Cyril Bonté62ba8702014-04-22 23:52:25 +020017842sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017843sc0_bytes_out_rate([<table>]) : integer
17844sc1_bytes_out_rate([<table>]) : integer
17845sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017846 Returns the average server-to-client bytes rate from the currently tracked
17847 counters, measured in amount of bytes over the period configured in the
17848 table. See also src_bytes_out_rate.
17849
Cyril Bonté62ba8702014-04-22 23:52:25 +020017850sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017851sc0_clr_gpc0([<table>]) : integer
17852sc1_clr_gpc0([<table>]) : integer
17853sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020017854 Clears the first General Purpose Counter associated to the currently tracked
17855 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010017856 stored value is zero, so first invocation will always return zero. This is
17857 typically used as a second ACL in an expression in order to mark a connection
17858 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020017859
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017860 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020017861 # block if 5 consecutive requests continue to come faster than 10 sess
17862 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017863 acl abuse sc0_http_req_rate gt 10
17864 acl kill sc0_inc_gpc0 gt 5
17865 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020017866 tcp-request connection accept if !abuse save
17867 tcp-request connection reject if abuse kill
17868
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017869sc_clr_gpc1(<ctr>[,<table>]) : integer
17870sc0_clr_gpc1([<table>]) : integer
17871sc1_clr_gpc1([<table>]) : integer
17872sc2_clr_gpc1([<table>]) : integer
17873 Clears the second General Purpose Counter associated to the currently tracked
17874 counters, and returns its previous value. Before the first invocation, the
17875 stored value is zero, so first invocation will always return zero. This is
17876 typically used as a second ACL in an expression in order to mark a connection
17877 when a first ACL was verified.
17878
Cyril Bonté62ba8702014-04-22 23:52:25 +020017879sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017880sc0_conn_cnt([<table>]) : integer
17881sc1_conn_cnt([<table>]) : integer
17882sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017883 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017884 counters. See also src_conn_cnt.
17885
Cyril Bonté62ba8702014-04-22 23:52:25 +020017886sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017887sc0_conn_cur([<table>]) : integer
17888sc1_conn_cur([<table>]) : integer
17889sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017890 Returns the current amount of concurrent connections tracking the same
17891 tracked counters. This number is automatically incremented when tracking
17892 begins and decremented when tracking stops. See also src_conn_cur.
17893
Cyril Bonté62ba8702014-04-22 23:52:25 +020017894sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017895sc0_conn_rate([<table>]) : integer
17896sc1_conn_rate([<table>]) : integer
17897sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017898 Returns the average connection rate from the currently tracked counters,
17899 measured in amount of connections over the period configured in the table.
17900 See also src_conn_rate.
17901
Cyril Bonté62ba8702014-04-22 23:52:25 +020017902sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017903sc0_get_gpc0([<table>]) : integer
17904sc1_get_gpc0([<table>]) : integer
17905sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017906 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017907 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017908
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017909sc_get_gpc1(<ctr>[,<table>]) : integer
17910sc0_get_gpc1([<table>]) : integer
17911sc1_get_gpc1([<table>]) : integer
17912sc2_get_gpc1([<table>]) : integer
17913 Returns the value of the second General Purpose Counter associated to the
17914 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
17915
Thierry FOURNIER236657b2015-08-19 08:25:14 +020017916sc_get_gpt0(<ctr>[,<table>]) : integer
17917sc0_get_gpt0([<table>]) : integer
17918sc1_get_gpt0([<table>]) : integer
17919sc2_get_gpt0([<table>]) : integer
17920 Returns the value of the first General Purpose Tag associated to the
17921 currently tracked counters. See also src_get_gpt0.
17922
Cyril Bonté62ba8702014-04-22 23:52:25 +020017923sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017924sc0_gpc0_rate([<table>]) : integer
17925sc1_gpc0_rate([<table>]) : integer
17926sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017927 Returns the average increment rate of the first General Purpose Counter
17928 associated to the currently tracked counters. It reports the frequency
17929 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017930 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
17931 that the "gpc0_rate" counter must be stored in the stick-table for a value to
17932 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020017933
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017934sc_gpc1_rate(<ctr>[,<table>]) : integer
17935sc0_gpc1_rate([<table>]) : integer
17936sc1_gpc1_rate([<table>]) : integer
17937sc2_gpc1_rate([<table>]) : integer
17938 Returns the average increment rate of the second General Purpose Counter
17939 associated to the currently tracked counters. It reports the frequency
17940 which the gpc1 counter was incremented over the configured period. See also
17941 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
17942 that the "gpc1_rate" counter must be stored in the stick-table for a value to
17943 be returned, as "gpc1" only holds the event count.
17944
Cyril Bonté62ba8702014-04-22 23:52:25 +020017945sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017946sc0_http_err_cnt([<table>]) : integer
17947sc1_http_err_cnt([<table>]) : integer
17948sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017949 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017950 counters. This includes the both request errors and 4xx error responses.
17951 See also src_http_err_cnt.
17952
Cyril Bonté62ba8702014-04-22 23:52:25 +020017953sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017954sc0_http_err_rate([<table>]) : integer
17955sc1_http_err_rate([<table>]) : integer
17956sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017957 Returns the average rate of HTTP errors from the currently tracked counters,
17958 measured in amount of errors over the period configured in the table. This
17959 includes the both request errors and 4xx error responses. See also
17960 src_http_err_rate.
17961
Willy Tarreau826f3ab2021-02-10 12:07:15 +010017962sc_http_fail_cnt(<ctr>[,<table>]) : integer
17963sc0_http_fail_cnt([<table>]) : integer
17964sc1_http_fail_cnt([<table>]) : integer
17965sc2_http_fail_cnt([<table>]) : integer
17966 Returns the cumulative number of HTTP response failures from the currently
17967 tracked counters. This includes the both response errors and 5xx status codes
17968 other than 501 and 505. See also src_http_fail_cnt.
17969
17970sc_http_fail_rate(<ctr>[,<table>]) : integer
17971sc0_http_fail_rate([<table>]) : integer
17972sc1_http_fail_rate([<table>]) : integer
17973sc2_http_fail_rate([<table>]) : integer
17974 Returns the average rate of HTTP response failures from the currently tracked
17975 counters, measured in amount of failures over the period configured in the
17976 table. This includes the both response errors and 5xx status codes other than
17977 501 and 505. See also src_http_fail_rate.
17978
Cyril Bonté62ba8702014-04-22 23:52:25 +020017979sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017980sc0_http_req_cnt([<table>]) : integer
17981sc1_http_req_cnt([<table>]) : integer
17982sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017983 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017984 counters. This includes every started request, valid or not. See also
17985 src_http_req_cnt.
17986
Cyril Bonté62ba8702014-04-22 23:52:25 +020017987sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017988sc0_http_req_rate([<table>]) : integer
17989sc1_http_req_rate([<table>]) : integer
17990sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017991 Returns the average rate of HTTP requests from the currently tracked
17992 counters, measured in amount of requests over the period configured in
17993 the table. This includes every started request, valid or not. See also
17994 src_http_req_rate.
17995
Cyril Bonté62ba8702014-04-22 23:52:25 +020017996sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017997sc0_inc_gpc0([<table>]) : integer
17998sc1_inc_gpc0([<table>]) : integer
17999sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020018000 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010018001 tracked counters, and returns its new value. Before the first invocation,
18002 the stored value is zero, so first invocation will increase it to 1 and will
18003 return 1. This is typically used as a second ACL in an expression in order
18004 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020018005
Jarno Huuskonen676f6222017-03-30 09:19:45 +030018006 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020018007 acl abuse sc0_http_req_rate gt 10
18008 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020018009 tcp-request connection reject if abuse kill
18010
Frédéric Lécaille6778b272018-01-29 15:22:53 +010018011sc_inc_gpc1(<ctr>[,<table>]) : integer
18012sc0_inc_gpc1([<table>]) : integer
18013sc1_inc_gpc1([<table>]) : integer
18014sc2_inc_gpc1([<table>]) : integer
18015 Increments the second General Purpose Counter associated to the currently
18016 tracked counters, and returns its new value. Before the first invocation,
18017 the stored value is zero, so first invocation will increase it to 1 and will
18018 return 1. This is typically used as a second ACL in an expression in order
18019 to mark a connection when a first ACL was verified.
18020
Cyril Bonté62ba8702014-04-22 23:52:25 +020018021sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020018022sc0_kbytes_in([<table>]) : integer
18023sc1_kbytes_in([<table>]) : integer
18024sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020018025 Returns the total amount of client-to-server data from the currently tracked
18026 counters, measured in kilobytes. The test is currently performed on 32-bit
18027 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020018028
Cyril Bonté62ba8702014-04-22 23:52:25 +020018029sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020018030sc0_kbytes_out([<table>]) : integer
18031sc1_kbytes_out([<table>]) : integer
18032sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020018033 Returns the total amount of server-to-client data from the currently tracked
18034 counters, measured in kilobytes. The test is currently performed on 32-bit
18035 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020018036
Cyril Bonté62ba8702014-04-22 23:52:25 +020018037sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020018038sc0_sess_cnt([<table>]) : integer
18039sc1_sess_cnt([<table>]) : integer
18040sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010018041 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020018042 into sessions, which means that they were accepted by a "tcp-request
18043 connection" rule, from the currently tracked counters. A backend may count
18044 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040018045 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020018046 with the client. See also src_sess_cnt.
18047
Cyril Bonté62ba8702014-04-22 23:52:25 +020018048sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020018049sc0_sess_rate([<table>]) : integer
18050sc1_sess_rate([<table>]) : integer
18051sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020018052 Returns the average session rate from the currently tracked counters,
18053 measured in amount of sessions over the period configured in the table. A
18054 session is a connection that got past the early "tcp-request connection"
18055 rules. A backend may count more sessions than connections because each
18056 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040018057 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020018058
Cyril Bonté62ba8702014-04-22 23:52:25 +020018059sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020018060sc0_tracked([<table>]) : boolean
18061sc1_tracked([<table>]) : boolean
18062sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020018063 Returns true if the designated session counter is currently being tracked by
18064 the current session. This can be useful when deciding whether or not we want
18065 to set some values in a header passed to the server.
18066
Cyril Bonté62ba8702014-04-22 23:52:25 +020018067sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020018068sc0_trackers([<table>]) : integer
18069sc1_trackers([<table>]) : integer
18070sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010018071 Returns the current amount of concurrent connections tracking the same
18072 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020018073 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010018074 that it does not rely on any stored information but on the table's reference
18075 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020018076 may sometimes be more suited for layer7 tracking. It can be used to tell a
18077 server how many concurrent connections there are from a given address for
18078 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010018079
Willy Tarreau74ca5042013-06-11 23:12:07 +020018080so_id : integer
18081 Returns an integer containing the current listening socket's id. It is useful
18082 in frontends involving many "bind" lines, or to stick all users coming via a
18083 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018084
Jerome Magnineb421b22020-03-27 22:08:40 +010018085so_name : string
18086 Returns a string containing the current listening socket's name, as defined
18087 with name on a "bind" line. It can serve the same purposes as so_id but with
18088 strings instead of integers.
18089
Willy Tarreau74ca5042013-06-11 23:12:07 +020018090src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010018091 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020018092 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
18093 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
18094 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010018095 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
18096 directive is used, it can be the address of a client behind another
18097 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010018098 "tcp-request connection" which sees the real address. When the incoming
18099 connection passed through address translation or redirection involving
18100 connection tracking, the original destination address before the redirection
18101 will be reported. On Linux systems, the source and destination may seldom
18102 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
18103 response may reopen a timed out connection and switch what is believed to be
18104 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018105
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010018106 Example:
18107 # add an HTTP header in requests with the originating address' country
18108 http-request set-header X-Country %[src,map_ip(geoip.lst)]
18109
Willy Tarreau74ca5042013-06-11 23:12:07 +020018110src_bytes_in_rate([<table>]) : integer
18111 Returns the average bytes rate from the incoming connection's source address
18112 in the current proxy's stick-table or in the designated stick-table, measured
18113 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018114 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018115
Willy Tarreau74ca5042013-06-11 23:12:07 +020018116src_bytes_out_rate([<table>]) : integer
18117 Returns the average bytes rate to the incoming connection's source address in
18118 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020018119 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018120 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018121
Willy Tarreau74ca5042013-06-11 23:12:07 +020018122src_clr_gpc0([<table>]) : integer
18123 Clears the first General Purpose Counter associated to the incoming
18124 connection's source address in the current proxy's stick-table or in the
18125 designated stick-table, and returns its previous value. If the address is not
18126 found, an entry is created and 0 is returned. This is typically used as a
18127 second ACL in an expression in order to mark a connection when a first ACL
18128 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020018129
Jarno Huuskonen676f6222017-03-30 09:19:45 +030018130 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020018131 # block if 5 consecutive requests continue to come faster than 10 sess
18132 # per second, and reset the counter as soon as the traffic slows down.
18133 acl abuse src_http_req_rate gt 10
18134 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010018135 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020018136 tcp-request connection accept if !abuse save
18137 tcp-request connection reject if abuse kill
18138
Frédéric Lécaille6778b272018-01-29 15:22:53 +010018139src_clr_gpc1([<table>]) : integer
18140 Clears the second General Purpose Counter associated to the incoming
18141 connection's source address in the current proxy's stick-table or in the
18142 designated stick-table, and returns its previous value. If the address is not
18143 found, an entry is created and 0 is returned. This is typically used as a
18144 second ACL in an expression in order to mark a connection when a first ACL
18145 was verified.
18146
Willy Tarreau74ca5042013-06-11 23:12:07 +020018147src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010018148 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020018149 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020018150 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018151 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018152
Willy Tarreau74ca5042013-06-11 23:12:07 +020018153src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020018154 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020018155 current incoming connection's source address in the current proxy's
18156 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018157 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018158
Willy Tarreau74ca5042013-06-11 23:12:07 +020018159src_conn_rate([<table>]) : integer
18160 Returns the average connection rate from the incoming connection's source
18161 address in the current proxy's stick-table or in the designated stick-table,
18162 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018163 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018164
Willy Tarreau74ca5042013-06-11 23:12:07 +020018165src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020018166 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020018167 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020018168 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018169 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018170
Frédéric Lécaille6778b272018-01-29 15:22:53 +010018171src_get_gpc1([<table>]) : integer
18172 Returns the value of the second General Purpose Counter associated to the
18173 incoming connection's source address in the current proxy's stick-table or in
18174 the designated stick-table. If the address is not found, zero is returned.
18175 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
18176
Thierry FOURNIER236657b2015-08-19 08:25:14 +020018177src_get_gpt0([<table>]) : integer
18178 Returns the value of the first General Purpose Tag associated to the
18179 incoming connection's source address in the current proxy's stick-table or in
18180 the designated stick-table. If the address is not found, zero is returned.
18181 See also sc/sc0/sc1/sc2_get_gpt0.
18182
Willy Tarreau74ca5042013-06-11 23:12:07 +020018183src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020018184 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020018185 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020018186 stick-table or in the designated stick-table. It reports the frequency
18187 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018188 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
18189 that the "gpc0_rate" counter must be stored in the stick-table for a value to
18190 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020018191
Frédéric Lécaille6778b272018-01-29 15:22:53 +010018192src_gpc1_rate([<table>]) : integer
18193 Returns the average increment rate of the second General Purpose Counter
18194 associated to the incoming connection's source address in the current proxy's
18195 stick-table or in the designated stick-table. It reports the frequency
18196 which the gpc1 counter was incremented over the configured period. See also
18197 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
18198 that the "gpc1_rate" counter must be stored in the stick-table for a value to
18199 be returned, as "gpc1" only holds the event count.
18200
Willy Tarreau74ca5042013-06-11 23:12:07 +020018201src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010018202 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020018203 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020018204 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018205 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020018206 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018207
Willy Tarreau74ca5042013-06-11 23:12:07 +020018208src_http_err_rate([<table>]) : integer
18209 Returns the average rate of HTTP errors from the incoming connection's source
18210 address in the current proxy's stick-table or in the designated stick-table,
18211 measured in amount of errors over the period configured in the table. This
18212 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018213 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018214
Willy Tarreau826f3ab2021-02-10 12:07:15 +010018215src_http_fail_cnt([<table>]) : integer
18216 Returns the cumulative number of HTTP response failures triggered by the
18217 incoming connection's source address in the current proxy's stick-table or in
Ilya Shipitsin0de36ad2021-02-20 00:23:36 +050018218 the designated stick-table. This includes the both response errors and 5xx
Willy Tarreau826f3ab2021-02-10 12:07:15 +010018219 status codes other than 501 and 505. See also sc/sc0/sc1/sc2_http_fail_cnt.
18220 If the address is not found, zero is returned.
18221
18222src_http_fail_rate([<table>]) : integer
18223 Returns the average rate of HTTP response failures triggered by the incoming
18224 connection's source address in the current proxy's stick-table or in the
18225 designated stick-table, measured in amount of failures over the period
18226 configured in the table. This includes the both response errors and 5xx
18227 status codes other than 501 and 505. If the address is not found, zero is
18228 returned. See also sc/sc0/sc1/sc2_http_fail_rate.
18229
Willy Tarreau74ca5042013-06-11 23:12:07 +020018230src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010018231 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020018232 source address in the current proxy's stick-table or in the designated stick-
18233 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018234 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018235
Willy Tarreau74ca5042013-06-11 23:12:07 +020018236src_http_req_rate([<table>]) : integer
18237 Returns the average rate of HTTP requests from the incoming connection's
18238 source address in the current proxy's stick-table or in the designated stick-
18239 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020018240 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018241 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018242
Willy Tarreau74ca5042013-06-11 23:12:07 +020018243src_inc_gpc0([<table>]) : integer
18244 Increments the first General Purpose Counter associated to the incoming
18245 connection's source address in the current proxy's stick-table or in the
18246 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020018247 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018248 This is typically used as a second ACL in an expression in order to mark a
18249 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020018250
Jarno Huuskonen676f6222017-03-30 09:19:45 +030018251 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020018252 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010018253 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020018254 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020018255
Frédéric Lécaille6778b272018-01-29 15:22:53 +010018256src_inc_gpc1([<table>]) : integer
18257 Increments the second General Purpose Counter associated to the incoming
18258 connection's source address in the current proxy's stick-table or in the
18259 designated stick-table, and returns its new value. If the address is not
18260 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
18261 This is typically used as a second ACL in an expression in order to mark a
18262 connection when a first ACL was verified.
18263
Willy Tarreau16e01562016-08-09 16:46:18 +020018264src_is_local : boolean
18265 Returns true if the source address of the incoming connection is local to the
18266 system, or false if the address doesn't exist on the system, meaning that it
18267 comes from a remote machine. Note that UNIX addresses are considered local.
18268 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010018269 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020018270 note that the check involves a few system calls, so it's better to do it only
18271 once per connection.
18272
Willy Tarreau74ca5042013-06-11 23:12:07 +020018273src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020018274 Returns the total amount of data received from the incoming connection's
18275 source address in the current proxy's stick-table or in the designated
18276 stick-table, measured in kilobytes. If the address is not found, zero is
18277 returned. The test is currently performed on 32-bit integers, which limits
18278 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018279
Willy Tarreau74ca5042013-06-11 23:12:07 +020018280src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020018281 Returns the total amount of data sent to the incoming connection's source
18282 address in the current proxy's stick-table or in the designated stick-table,
18283 measured in kilobytes. If the address is not found, zero is returned. The
18284 test is currently performed on 32-bit integers, which limits values to 4
18285 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020018286
Willy Tarreau74ca5042013-06-11 23:12:07 +020018287src_port : integer
18288 Returns an integer value corresponding to the TCP source port of the
18289 connection on the client side, which is the port the client connected from.
18290 Usage of this function is very limited as modern protocols do not care much
18291 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010018292
Willy Tarreau74ca5042013-06-11 23:12:07 +020018293src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010018294 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020018295 connection's source IPv4 address in the current proxy's stick-table or in the
18296 designated stick-table, that were transformed into sessions, which means that
18297 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018298 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018299
Willy Tarreau74ca5042013-06-11 23:12:07 +020018300src_sess_rate([<table>]) : integer
18301 Returns the average session rate from the incoming connection's source
18302 address in the current proxy's stick-table or in the designated stick-table,
18303 measured in amount of sessions over the period configured in the table. A
18304 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020018305 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020018306
Willy Tarreau74ca5042013-06-11 23:12:07 +020018307src_updt_conn_cnt([<table>]) : integer
18308 Creates or updates the entry associated to the incoming connection's source
18309 address in the current proxy's stick-table or in the designated stick-table.
18310 This table must be configured to store the "conn_cnt" data type, otherwise
18311 the match will be ignored. The current count is incremented by one, and the
18312 expiration timer refreshed. The updated count is returned, so this match
18313 can't return zero. This was used to reject service abusers based on their
18314 source address. Note: it is recommended to use the more complete "track-sc*"
18315 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020018316
18317 Example :
18318 # This frontend limits incoming SSH connections to 3 per 10 second for
18319 # each source address, and rejects excess connections until a 10 second
18320 # silence is observed. At most 20 addresses are tracked.
18321 listen ssh
18322 bind :22
18323 mode tcp
18324 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020018325 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020018326 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020018327 server local 127.0.0.1:22
18328
Willy Tarreau74ca5042013-06-11 23:12:07 +020018329srv_id : integer
18330 Returns an integer containing the server's id when processing the response.
18331 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020018332 debugging. It can also be used in a tcp-check or an http-check ruleset.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020018333
vkill1dfd1652019-10-30 16:58:14 +080018334srv_name : string
18335 Returns a string containing the server's name when processing the response.
18336 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020018337 debugging. It can also be used in a tcp-check or an http-check ruleset.
vkill1dfd1652019-10-30 16:58:14 +080018338
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200183397.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020018340----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020018341
Daniel Corbett9f0843f2021-05-08 10:50:37 -040018342The layer 5 usually describes just the session layer which in HAProxy is
Willy Tarreau74ca5042013-06-11 23:12:07 +020018343closest to the session once all the connection handshakes are finished, but
18344when no content is yet made available. The fetch methods described here are
18345usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018346future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020018347
Ben Shillitof25e8e52016-12-02 14:25:37 +00001834851d.all(<prop>[,<prop>*]) : string
18349 Returns values for the properties requested as a string, where values are
18350 separated by the delimiter specified with "51degrees-property-separator".
18351 The device is identified using all the important HTTP headers from the
18352 request. The function can be passed up to five property names, and if a
18353 property name can't be found, the value "NoData" is returned.
18354
18355 Example :
18356 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
18357 # containing the three properties requested using all relevant headers from
18358 # the request.
18359 frontend http-in
18360 bind *:8081
18361 default_backend servers
18362 http-request set-header X-51D-DeviceTypeMobileTablet \
18363 %[51d.all(DeviceType,IsMobile,IsTablet)]
18364
Emeric Brun645ae792014-04-30 14:21:06 +020018365ssl_bc : boolean
18366 Returns true when the back connection was made via an SSL/TLS transport
18367 layer and is locally deciphered. This means the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018368 other a server with the "ssl" option. It can be used in a tcp-check or an
18369 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018370
18371ssl_bc_alg_keysize : integer
18372 Returns the symmetric cipher key size supported in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018373 connection was made over an SSL/TLS transport layer. It can be used in a
18374 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018375
Olivier Houchard6b77f492018-11-22 18:18:29 +010018376ssl_bc_alpn : string
18377 This extracts the Application Layer Protocol Negotiation field from an
18378 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020018379 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010018380 server. The SSL library must have been built with support for TLS
18381 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
18382 not advertised unless the "alpn" keyword on the "server" line specifies a
18383 protocol list. Also, nothing forces the server to pick a protocol from this
18384 list, any other one may be requested. The TLS ALPN extension is meant to
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018385 replace the TLS NPN extension. See also "ssl_bc_npn". It can be used in a
18386 tcp-check or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010018387
Emeric Brun645ae792014-04-30 14:21:06 +020018388ssl_bc_cipher : string
18389 Returns the name of the used cipher when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018390 over an SSL/TLS transport layer. It can be used in a tcp-check or an
18391 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018392
Patrick Hemmer65674662019-06-04 08:13:03 -040018393ssl_bc_client_random : binary
18394 Returns the client random of the back connection when the incoming connection
18395 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
18396 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018397 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040018398
Emeric Brun74f7ffa2018-02-19 16:14:12 +010018399ssl_bc_is_resumed : boolean
18400 Returns true when the back connection was made over an SSL/TLS transport
18401 layer and the newly created SSL session was resumed using a cached
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018402 session or a TLS ticket. It can be used in a tcp-check or an http-check
18403 ruleset.
Emeric Brun74f7ffa2018-02-19 16:14:12 +010018404
Olivier Houchard6b77f492018-11-22 18:18:29 +010018405ssl_bc_npn : string
18406 This extracts the Next Protocol Negotiation field from an outgoing connection
18407 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020018408 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010018409 built with support for TLS extensions enabled (check haproxy -vv). Note that
18410 the TLS NPN extension is not advertised unless the "npn" keyword on the
18411 "server" line specifies a protocol list. Also, nothing forces the server to
18412 pick a protocol from this list, any other one may be used. Please note that
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018413 the TLS NPN extension was replaced with ALPN. It can be used in a tcp-check
18414 or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010018415
Emeric Brun645ae792014-04-30 14:21:06 +020018416ssl_bc_protocol : string
18417 Returns the name of the used protocol when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018418 over an SSL/TLS transport layer. It can be used in a tcp-check or an
18419 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018420
Emeric Brunb73a9b02014-04-30 18:49:19 +020018421ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020018422 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020018423 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018424 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64". It
18425 can be used in a tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018426
Patrick Hemmer65674662019-06-04 08:13:03 -040018427ssl_bc_server_random : binary
18428 Returns the server random of the back connection when the incoming connection
18429 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
18430 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018431 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040018432
Emeric Brun645ae792014-04-30 14:21:06 +020018433ssl_bc_session_id : binary
18434 Returns the SSL ID of the back connection when the outgoing connection was
18435 made over an SSL/TLS transport layer. It is useful to log if we want to know
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018436 if session was reused or not. It can be used in a tcp-check or an http-check
18437 ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018438
Patrick Hemmere0275472018-04-28 19:15:51 -040018439ssl_bc_session_key : binary
18440 Returns the SSL session master key of the back connection when the outgoing
18441 connection was made over an SSL/TLS transport layer. It is useful to decrypt
18442 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018443 BoringSSL. It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmere0275472018-04-28 19:15:51 -040018444
Emeric Brun645ae792014-04-30 14:21:06 +020018445ssl_bc_use_keysize : integer
18446 Returns the symmetric cipher key size used in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018447 connection was made over an SSL/TLS transport layer. It can be used in a
18448 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018449
Willy Tarreau74ca5042013-06-11 23:12:07 +020018450ssl_c_ca_err : integer
18451 When the incoming connection was made over an SSL/TLS transport layer,
18452 returns the ID of the first error detected during verification of the client
18453 certificate at depth > 0, or 0 if no error was encountered during this
18454 verification process. Please refer to your SSL library's documentation to
18455 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020018456
Willy Tarreau74ca5042013-06-11 23:12:07 +020018457ssl_c_ca_err_depth : integer
18458 When the incoming connection was made over an SSL/TLS transport layer,
18459 returns the depth in the CA chain of the first error detected during the
18460 verification of the client certificate. If no error is encountered, 0 is
18461 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010018462
Christopher Faulet70d10d12020-11-06 12:10:33 +010018463ssl_c_chain_der : binary
William Dauchya598b502020-08-06 18:11:38 +020018464 Returns the DER formatted chain certificate presented by the client when the
18465 incoming connection was made over an SSL/TLS transport layer. When used for
18466 an ACL, the value(s) to match against can be passed in hexadecimal form. One
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050018467 can parse the result with any lib accepting ASN.1 DER data. It currently
William Dauchya598b502020-08-06 18:11:38 +020018468 does not support resumed sessions.
18469
Christopher Faulet70d10d12020-11-06 12:10:33 +010018470ssl_c_der : binary
18471 Returns the DER formatted certificate presented by the client when the
18472 incoming connection was made over an SSL/TLS transport layer. When used for
18473 an ACL, the value(s) to match against can be passed in hexadecimal form.
18474
Willy Tarreau74ca5042013-06-11 23:12:07 +020018475ssl_c_err : integer
18476 When the incoming connection was made over an SSL/TLS transport layer,
18477 returns the ID of the first error detected during verification at depth 0, or
18478 0 if no error was encountered during this verification process. Please refer
18479 to your SSL library's documentation to find the exhaustive list of error
18480 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020018481
Elliot Otchet71f82972020-01-15 08:12:14 -050018482ssl_c_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018483 When the incoming connection was made over an SSL/TLS transport layer,
18484 returns the full distinguished name of the issuer of the certificate
18485 presented by the client when no <entry> is specified, or the value of the
18486 first given entry found from the beginning of the DN. If a positive/negative
18487 occurrence number is specified as the optional second argument, it returns
18488 the value of the nth given entry value from the beginning/end of the DN.
18489 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
18490 "ssl_c_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018491 The <format> parameter allows you to receive the DN suitable for
18492 consumption by different protocols. Currently supported is rfc2253 for
18493 LDAP v3.
18494 If you'd like to modify the format only you can specify an empty string
18495 and zero for the first two parameters. Example: ssl_c_i_dn(,0,rfc2253)
Willy Tarreau62644772008-07-16 18:36:06 +020018496
Willy Tarreau74ca5042013-06-11 23:12:07 +020018497ssl_c_key_alg : string
18498 Returns the name of the algorithm used to generate the key of the certificate
18499 presented by the client when the incoming connection was made over an SSL/TLS
18500 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020018501
Willy Tarreau74ca5042013-06-11 23:12:07 +020018502ssl_c_notafter : string
18503 Returns the end date presented by the client as a formatted string
18504 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18505 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020018506
Willy Tarreau74ca5042013-06-11 23:12:07 +020018507ssl_c_notbefore : string
18508 Returns the start date presented by the client as a formatted string
18509 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18510 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010018511
Elliot Otchet71f82972020-01-15 08:12:14 -050018512ssl_c_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018513 When the incoming connection was made over an SSL/TLS transport layer,
18514 returns the full distinguished name of the subject of the certificate
18515 presented by the client when no <entry> is specified, or the value of the
18516 first given entry found from the beginning of the DN. If a positive/negative
18517 occurrence number is specified as the optional second argument, it returns
18518 the value of the nth given entry value from the beginning/end of the DN.
18519 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
18520 "ssl_c_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018521 The <format> parameter allows you to receive the DN suitable for
18522 consumption by different protocols. Currently supported is rfc2253 for
18523 LDAP v3.
18524 If you'd like to modify the format only you can specify an empty string
18525 and zero for the first two parameters. Example: ssl_c_s_dn(,0,rfc2253)
Willy Tarreaub6672b52011-12-12 17:23:41 +010018526
Willy Tarreau74ca5042013-06-11 23:12:07 +020018527ssl_c_serial : binary
18528 Returns the serial of the certificate presented by the client when the
18529 incoming connection was made over an SSL/TLS transport layer. When used for
18530 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020018531
Willy Tarreau74ca5042013-06-11 23:12:07 +020018532ssl_c_sha1 : binary
18533 Returns the SHA-1 fingerprint of the certificate presented by the client when
18534 the incoming connection was made over an SSL/TLS transport layer. This can be
18535 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020018536 Note that the output is binary, so if you want to pass that signature to the
18537 server, you need to encode it in hex or base64, such as in the example below:
18538
Jarno Huuskonen676f6222017-03-30 09:19:45 +030018539 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020018540 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020018541
Willy Tarreau74ca5042013-06-11 23:12:07 +020018542ssl_c_sig_alg : string
18543 Returns the name of the algorithm used to sign the certificate presented by
18544 the client when the incoming connection was made over an SSL/TLS transport
18545 layer.
Emeric Brun87855892012-10-17 17:39:35 +020018546
Willy Tarreau74ca5042013-06-11 23:12:07 +020018547ssl_c_used : boolean
18548 Returns true if current SSL session uses a client certificate even if current
18549 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020018550
Willy Tarreau74ca5042013-06-11 23:12:07 +020018551ssl_c_verify : integer
18552 Returns the verify result error ID when the incoming connection was made over
18553 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
18554 refer to your SSL library's documentation for an exhaustive list of error
18555 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020018556
Willy Tarreau74ca5042013-06-11 23:12:07 +020018557ssl_c_version : integer
18558 Returns the version of the certificate presented by the client when the
18559 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020018560
Emeric Brun43e79582014-10-29 19:03:26 +010018561ssl_f_der : binary
18562 Returns the DER formatted certificate presented by the frontend when the
18563 incoming connection was made over an SSL/TLS transport layer. When used for
18564 an ACL, the value(s) to match against can be passed in hexadecimal form.
18565
Elliot Otchet71f82972020-01-15 08:12:14 -050018566ssl_f_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018567 When the incoming connection was made over an SSL/TLS transport layer,
18568 returns the full distinguished name of the issuer of the certificate
18569 presented by the frontend when no <entry> is specified, or the value of the
18570 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020018571 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020018572 the value of the nth given entry value from the beginning/end of the DN.
18573 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
18574 "ssl_f_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018575 The <format> parameter allows you to receive the DN suitable for
18576 consumption by different protocols. Currently supported is rfc2253 for
18577 LDAP v3.
18578 If you'd like to modify the format only you can specify an empty string
18579 and zero for the first two parameters. Example: ssl_f_i_dn(,0,rfc2253)
Emeric Brun87855892012-10-17 17:39:35 +020018580
Willy Tarreau74ca5042013-06-11 23:12:07 +020018581ssl_f_key_alg : string
18582 Returns the name of the algorithm used to generate the key of the certificate
18583 presented by the frontend when the incoming connection was made over an
18584 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020018585
Willy Tarreau74ca5042013-06-11 23:12:07 +020018586ssl_f_notafter : string
18587 Returns the end date presented by the frontend as a formatted string
18588 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18589 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020018590
Willy Tarreau74ca5042013-06-11 23:12:07 +020018591ssl_f_notbefore : string
18592 Returns the start date presented by the frontend as a formatted string
18593 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18594 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020018595
Elliot Otchet71f82972020-01-15 08:12:14 -050018596ssl_f_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018597 When the incoming connection was made over an SSL/TLS transport layer,
18598 returns the full distinguished name of the subject of the certificate
18599 presented by the frontend when no <entry> is specified, or the value of the
18600 first given entry found from the beginning of the DN. If a positive/negative
18601 occurrence number is specified as the optional second argument, it returns
18602 the value of the nth given entry value from the beginning/end of the DN.
18603 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
18604 "ssl_f_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018605 The <format> parameter allows you to receive the DN suitable for
18606 consumption by different protocols. Currently supported is rfc2253 for
18607 LDAP v3.
18608 If you'd like to modify the format only you can specify an empty string
18609 and zero for the first two parameters. Example: ssl_f_s_dn(,0,rfc2253)
Emeric Brunce5ad802012-10-22 14:11:22 +020018610
Willy Tarreau74ca5042013-06-11 23:12:07 +020018611ssl_f_serial : binary
18612 Returns the serial of the certificate presented by the frontend when the
18613 incoming connection was made over an SSL/TLS transport layer. When used for
18614 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020018615
Emeric Brun55f4fa82014-04-30 17:11:25 +020018616ssl_f_sha1 : binary
18617 Returns the SHA-1 fingerprint of the certificate presented by the frontend
18618 when the incoming connection was made over an SSL/TLS transport layer. This
18619 can be used to know which certificate was chosen using SNI.
18620
Willy Tarreau74ca5042013-06-11 23:12:07 +020018621ssl_f_sig_alg : string
18622 Returns the name of the algorithm used to sign the certificate presented by
18623 the frontend when the incoming connection was made over an SSL/TLS transport
18624 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020018625
Willy Tarreau74ca5042013-06-11 23:12:07 +020018626ssl_f_version : integer
18627 Returns the version of the certificate presented by the frontend when the
18628 incoming connection was made over an SSL/TLS transport layer.
18629
18630ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020018631 Returns true when the front connection was made via an SSL/TLS transport
18632 layer and is locally deciphered. This means it has matched a socket declared
18633 with a "bind" line having the "ssl" option.
18634
Willy Tarreau74ca5042013-06-11 23:12:07 +020018635 Example :
18636 # This passes "X-Proto: https" to servers when client connects over SSL
18637 listen http-https
18638 bind :80
18639 bind :443 ssl crt /etc/haproxy.pem
18640 http-request add-header X-Proto https if { ssl_fc }
18641
18642ssl_fc_alg_keysize : integer
18643 Returns the symmetric cipher key size supported in bits when the incoming
18644 connection was made over an SSL/TLS transport layer.
18645
18646ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018647 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020018648 incoming connection made via a TLS transport layer and locally deciphered by
Daniel Corbett9f0843f2021-05-08 10:50:37 -040018649 HAProxy. The result is a string containing the protocol name advertised by
Willy Tarreau74ca5042013-06-11 23:12:07 +020018650 the client. The SSL library must have been built with support for TLS
18651 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
18652 not advertised unless the "alpn" keyword on the "bind" line specifies a
18653 protocol list. Also, nothing forces the client to pick a protocol from this
18654 list, any other one may be requested. The TLS ALPN extension is meant to
18655 replace the TLS NPN extension. See also "ssl_fc_npn".
18656
Willy Tarreau74ca5042013-06-11 23:12:07 +020018657ssl_fc_cipher : string
18658 Returns the name of the used cipher when the incoming connection was made
18659 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020018660
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018661ssl_fc_cipherlist_bin : binary
18662 Returns the binary form of the client hello cipher list. The maximum returned
18663 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010018664 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018665
18666ssl_fc_cipherlist_hex : string
18667 Returns the binary form of the client hello cipher list encoded as
18668 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010018669 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018670
18671ssl_fc_cipherlist_str : string
18672 Returns the decoded text form of the client hello cipher list. The maximum
18673 number of ciphers returned is according with the value of
18674 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010018675 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020018676 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018677
18678ssl_fc_cipherlist_xxh : integer
18679 Returns a xxh64 of the cipher list. This hash can be return only is the value
18680 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010018681 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018682
Patrick Hemmer65674662019-06-04 08:13:03 -040018683ssl_fc_client_random : binary
18684 Returns the client random of the front connection when the incoming connection
18685 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
18686 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
18687
William Lallemand7d42ef52020-07-06 11:41:30 +020018688ssl_fc_client_early_traffic_secret : string
18689 Return the CLIENT_EARLY_TRAFFIC_SECRET as an hexadecimal string for the
18690 front connection when the incoming connection was made over a TLS 1.3
18691 transport layer.
18692 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18693 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18694 activated with "tune.ssl.keylog on" in the global section. See also
18695 "tune.ssl.keylog"
18696
18697ssl_fc_client_handshake_traffic_secret : string
18698 Return the CLIENT_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
18699 front connection when the incoming connection was made over a TLS 1.3
18700 transport layer.
18701 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18702 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18703 activated with "tune.ssl.keylog on" in the global section. See also
18704 "tune.ssl.keylog"
18705
18706ssl_fc_client_traffic_secret_0 : string
18707 Return the CLIENT_TRAFFIC_SECRET_0 as an hexadecimal string for the
18708 front connection when the incoming connection was made over a TLS 1.3
18709 transport layer.
18710 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18711 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18712 activated with "tune.ssl.keylog on" in the global section. See also
18713 "tune.ssl.keylog"
18714
18715ssl_fc_exporter_secret : string
18716 Return the EXPORTER_SECRET as an hexadecimal string for the
18717 front connection when the incoming connection was made over a TLS 1.3
18718 transport layer.
18719 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18720 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18721 activated with "tune.ssl.keylog on" in the global section. See also
18722 "tune.ssl.keylog"
18723
18724ssl_fc_early_exporter_secret : string
18725 Return the EARLY_EXPORTER_SECRET as an hexadecimal string for the
18726 front connection when the incoming connection was made over an TLS 1.3
18727 transport layer.
18728 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18729 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18730 activated with "tune.ssl.keylog on" in the global section. See also
18731 "tune.ssl.keylog"
18732
Willy Tarreau74ca5042013-06-11 23:12:07 +020018733ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020018734 Returns true if a client certificate is present in an incoming connection over
18735 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010018736 Note: on SSL session resumption with Session ID or TLS ticket, client
18737 certificate is not present in the current connection but may be retrieved
18738 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
18739 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020018740
Olivier Houchardccaa7de2017-10-02 11:51:03 +020018741ssl_fc_has_early : boolean
18742 Returns true if early data were sent, and the handshake didn't happen yet. As
18743 it has security implications, it is useful to be able to refuse those, or
18744 wait until the handshake happened.
18745
Willy Tarreau74ca5042013-06-11 23:12:07 +020018746ssl_fc_has_sni : boolean
18747 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020018748 in an incoming connection was made over an SSL/TLS transport layer. Returns
18749 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050018750 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020018751 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020018752
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020018753ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020018754 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010018755 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
18756 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020018757
Willy Tarreau74ca5042013-06-11 23:12:07 +020018758ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018759 This extracts the Next Protocol Negotiation field from an incoming connection
Daniel Corbett9f0843f2021-05-08 10:50:37 -040018760 made via a TLS transport layer and locally deciphered by HAProxy. The result
Willy Tarreau74ca5042013-06-11 23:12:07 +020018761 is a string containing the protocol name advertised by the client. The SSL
18762 library must have been built with support for TLS extensions enabled (check
18763 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
18764 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
18765 forces the client to pick a protocol from this list, any other one may be
18766 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020018767
Willy Tarreau74ca5042013-06-11 23:12:07 +020018768ssl_fc_protocol : string
18769 Returns the name of the used protocol when the incoming connection was made
18770 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020018771
Emeric Brunb73a9b02014-04-30 18:49:19 +020018772ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040018773 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020018774 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
Christopher Fauletc5de4192021-11-09 14:23:36 +010018775 can be encoded to base64 using the converter: "ssl_fc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040018776
William Lallemand7d42ef52020-07-06 11:41:30 +020018777ssl_fc_server_handshake_traffic_secret : string
18778 Return the SERVER_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
18779 front connection when the incoming connection was made over a TLS 1.3
18780 transport layer.
18781 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18782 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18783 activated with "tune.ssl.keylog on" in the global section. See also
18784 "tune.ssl.keylog"
18785
18786ssl_fc_server_traffic_secret_0 : string
18787 Return the SERVER_TRAFFIC_SECRET_0 as an hexadecimal string for the
18788 front connection when the incoming connection was made over an TLS 1.3
18789 transport layer.
18790 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18791 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18792 activated with "tune.ssl.keylog on" in the global section. See also
18793 "tune.ssl.keylog"
18794
Patrick Hemmer65674662019-06-04 08:13:03 -040018795ssl_fc_server_random : binary
18796 Returns the server random of the front connection when the incoming connection
18797 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
18798 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
18799
Willy Tarreau74ca5042013-06-11 23:12:07 +020018800ssl_fc_session_id : binary
18801 Returns the SSL ID of the front connection when the incoming connection was
18802 made over an SSL/TLS transport layer. It is useful to stick a given client to
18803 a server. It is important to note that some browsers refresh their session ID
18804 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020018805
Patrick Hemmere0275472018-04-28 19:15:51 -040018806ssl_fc_session_key : binary
18807 Returns the SSL session master key of the front connection when the incoming
18808 connection was made over an SSL/TLS transport layer. It is useful to decrypt
18809 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
18810 BoringSSL.
18811
18812
Willy Tarreau74ca5042013-06-11 23:12:07 +020018813ssl_fc_sni : string
18814 This extracts the Server Name Indication TLS extension (SNI) field from an
18815 incoming connection made via an SSL/TLS transport layer and locally
Daniel Corbett9f0843f2021-05-08 10:50:37 -040018816 deciphered by HAProxy. The result (when present) typically is a string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018817 matching the HTTPS host name (253 chars or less). The SSL library must have
18818 been built with support for TLS extensions enabled (check haproxy -vv).
18819
Alex18c88772021-06-05 13:23:08 +020018820 This fetch is different from "req.ssl_sni" above in that it applies to the
Daniel Corbett9f0843f2021-05-08 10:50:37 -040018821 connection being deciphered by HAProxy and not to SSL contents being blindly
Willy Tarreau74ca5042013-06-11 23:12:07 +020018822 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050018823 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020018824 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020018825
Willy Tarreau74ca5042013-06-11 23:12:07 +020018826 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020018827 ssl_fc_sni_end : suffix match
18828 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020018829
Willy Tarreau74ca5042013-06-11 23:12:07 +020018830ssl_fc_use_keysize : integer
18831 Returns the symmetric cipher key size used in bits when the incoming
18832 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018833
William Lallemandbfa3e812020-06-25 20:07:18 +020018834ssl_s_der : binary
18835 Returns the DER formatted certificate presented by the server when the
18836 outgoing connection was made over an SSL/TLS transport layer. When used for
18837 an ACL, the value(s) to match against can be passed in hexadecimal form.
18838
William Dauchya598b502020-08-06 18:11:38 +020018839ssl_s_chain_der : binary
18840 Returns the DER formatted chain certificate presented by the server when the
18841 outgoing connection was made over an SSL/TLS transport layer. When used for
18842 an ACL, the value(s) to match against can be passed in hexadecimal form. One
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050018843 can parse the result with any lib accepting ASN.1 DER data. It currently
William Dauchya598b502020-08-06 18:11:38 +020018844 does not support resumed sessions.
18845
William Lallemandbfa3e812020-06-25 20:07:18 +020018846ssl_s_key_alg : string
18847 Returns the name of the algorithm used to generate the key of the certificate
18848 presented by the server when the outgoing connection was made over an
18849 SSL/TLS transport layer.
18850
18851ssl_s_notafter : string
18852 Returns the end date presented by the server as a formatted string
18853 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
18854 transport layer.
18855
18856ssl_s_notbefore : string
18857 Returns the start date presented by the server as a formatted string
18858 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
18859 transport layer.
18860
18861ssl_s_i_dn([<entry>[,<occ>[,<format>]]]) : string
18862 When the outgoing connection was made over an SSL/TLS transport layer,
18863 returns the full distinguished name of the issuer of the certificate
18864 presented by the server when no <entry> is specified, or the value of the
18865 first given entry found from the beginning of the DN. If a positive/negative
18866 occurrence number is specified as the optional second argument, it returns
18867 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020018868 For instance, "ssl_s_i_dn(OU,2)" the second organization unit, and
18869 "ssl_s_i_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020018870 The <format> parameter allows you to receive the DN suitable for
18871 consumption by different protocols. Currently supported is rfc2253 for
18872 LDAP v3.
18873 If you'd like to modify the format only you can specify an empty string
18874 and zero for the first two parameters. Example: ssl_s_i_dn(,0,rfc2253)
18875
18876ssl_s_s_dn([<entry>[,<occ>[,<format>]]]) : string
18877 When the outgoing connection was made over an SSL/TLS transport layer,
18878 returns the full distinguished name of the subject of the certificate
18879 presented by the server when no <entry> is specified, or the value of the
18880 first given entry found from the beginning of the DN. If a positive/negative
18881 occurrence number is specified as the optional second argument, it returns
18882 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020018883 For instance, "ssl_s_s_dn(OU,2)" the second organization unit, and
18884 "ssl_s_s_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020018885 The <format> parameter allows you to receive the DN suitable for
18886 consumption by different protocols. Currently supported is rfc2253 for
18887 LDAP v3.
18888 If you'd like to modify the format only you can specify an empty string
18889 and zero for the first two parameters. Example: ssl_s_s_dn(,0,rfc2253)
18890
18891ssl_s_serial : binary
18892 Returns the serial of the certificate presented by the server when the
18893 outgoing connection was made over an SSL/TLS transport layer. When used for
18894 an ACL, the value(s) to match against can be passed in hexadecimal form.
18895
18896ssl_s_sha1 : binary
18897 Returns the SHA-1 fingerprint of the certificate presented by the server
18898 when the outgoing connection was made over an SSL/TLS transport layer. This
18899 can be used to know which certificate was chosen using SNI.
18900
18901ssl_s_sig_alg : string
18902 Returns the name of the algorithm used to sign the certificate presented by
18903 the server when the outgoing connection was made over an SSL/TLS transport
18904 layer.
18905
18906ssl_s_version : integer
18907 Returns the version of the certificate presented by the server when the
18908 outgoing connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018909
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200189107.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020018911------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018912
Willy Tarreau74ca5042013-06-11 23:12:07 +020018913Fetching samples from buffer contents is a bit different from the previous
18914sample fetches above because the sampled data are ephemeral. These data can
18915only be used when they're available and will be lost when they're forwarded.
18916For this reason, samples fetched from buffer contents during a request cannot
18917be used in a response for example. Even while the data are being fetched, they
18918can change. Sometimes it is necessary to set some delays or combine multiple
18919sample fetch methods to ensure that the expected data are complete and usable,
18920for example through TCP request content inspection. Please see the "tcp-request
18921content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020018922
Christopher Fauleta434a002021-03-25 11:58:51 +010018923Warning : Following sample fetches are ignored if used from HTTP proxies. They
18924 only deal with raw contents found in the buffers. On their side,
Daniel Corbett9f0843f2021-05-08 10:50:37 -040018925 HTTP proxies use structured content. Thus raw representation of
Christopher Fauleta434a002021-03-25 11:58:51 +010018926 these data are meaningless. A warning is emitted if an ACL relies on
18927 one of the following sample fetches. But it is not possible to detect
18928 all invalid usage (for instance inside a log-format string or a
18929 sample expression). So be careful.
18930
Willy Tarreau74ca5042013-06-11 23:12:07 +020018931payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010018932 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018933 "stick on", "stick match"), and for "res.payload" when used in the context of
18934 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010018935
Willy Tarreau74ca5042013-06-11 23:12:07 +020018936payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
18937 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010018938 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020018939 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010018940
Willy Tarreau74ca5042013-06-11 23:12:07 +020018941req.len : integer
18942req_len : integer (deprecated)
18943 Returns an integer value corresponding to the number of bytes present in the
18944 request buffer. This is mostly used in ACL. It is important to understand
18945 that this test does not return false as long as the buffer is changing. This
18946 means that a check with equality to zero will almost always immediately match
18947 at the beginning of the session, while a test for more data will wait for
Daniel Corbett9f0843f2021-05-08 10:50:37 -040018948 that data to come in and return false only when HAProxy is certain that no
Willy Tarreau74ca5042013-06-11 23:12:07 +020018949 more data will come in. This test was designed to be used with TCP request
18950 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018951
Willy Tarreau74ca5042013-06-11 23:12:07 +020018952req.payload(<offset>,<length>) : binary
18953 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020018954 in the request buffer. As a special case, if the <length> argument is zero,
18955 the the whole buffer from <offset> to the end is extracted. This can be used
18956 with ACLs in order to check for the presence of some content in a buffer at
18957 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018958
Christian Ruppertfd3184c2022-02-20 22:54:01 +010018959 ACL derivatives :
18960 req.payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018961
Willy Tarreau74ca5042013-06-11 23:12:07 +020018962req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
18963 This extracts a binary block whose size is specified at <offset1> for <length>
18964 bytes, and which starts at <offset2> if specified or just after the length in
18965 the request buffer. The <offset2> parameter also supports relative offsets if
18966 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018967
Christian Ruppertfd3184c2022-02-20 22:54:01 +010018968 ACL derivatives :
18969 req.payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018970
Willy Tarreau74ca5042013-06-11 23:12:07 +020018971 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018972
Willy Tarreau74ca5042013-06-11 23:12:07 +020018973req.proto_http : boolean
18974req_proto_http : boolean (deprecated)
18975 Returns true when data in the request buffer look like HTTP and correctly
18976 parses as such. It is the same parser as the common HTTP request parser which
18977 is used so there should be no surprises. The test does not match until the
18978 request is complete, failed or timed out. This test may be used to report the
18979 protocol in TCP logs, but the biggest use is to block TCP request analysis
18980 until a complete HTTP request is present in the buffer, for example to track
18981 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018982
Willy Tarreau74ca5042013-06-11 23:12:07 +020018983 Example:
18984 # track request counts per "base" (concatenation of Host+URL)
18985 tcp-request inspect-delay 10s
18986 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020018987 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018988
Willy Tarreau74ca5042013-06-11 23:12:07 +020018989req.rdp_cookie([<name>]) : string
18990rdp_cookie([<name>]) : string (deprecated)
18991 When the request buffer looks like the RDP protocol, extracts the RDP cookie
18992 <name>, or any cookie if unspecified. The parser only checks for the first
18993 cookie, as illustrated in the RDP protocol specification. The cookie name is
18994 case insensitive. Generally the "MSTS" cookie name will be used, as it can
18995 contain the user name of the client connecting to the server if properly
18996 configured on the client. The "MSTSHASH" cookie is often used as well for
18997 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018998
Willy Tarreau74ca5042013-06-11 23:12:07 +020018999 This differs from "balance rdp-cookie" in that any balancing algorithm may be
19000 used and thus the distribution of clients to backend servers is not linked to
19001 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
19002 such as "balance roundrobin" or "balance leastconn" will lead to a more even
19003 distribution of clients to backend servers than the hash used by "balance
19004 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020019005
Willy Tarreau74ca5042013-06-11 23:12:07 +020019006 ACL derivatives :
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019007 req.rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020019008
Willy Tarreau74ca5042013-06-11 23:12:07 +020019009 Example :
19010 listen tse-farm
19011 bind 0.0.0.0:3389
19012 # wait up to 5s for an RDP cookie in the request
19013 tcp-request inspect-delay 5s
19014 tcp-request content accept if RDP_COOKIE
19015 # apply RDP cookie persistence
19016 persist rdp-cookie
19017 # Persist based on the mstshash cookie
19018 # This is only useful makes sense if
19019 # balance rdp-cookie is not used
19020 stick-table type string size 204800
19021 stick on req.rdp_cookie(mstshash)
19022 server srv1 1.1.1.1:3389
19023 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020019024
Willy Tarreau74ca5042013-06-11 23:12:07 +020019025 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019026 "req.rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020019027
Willy Tarreau74ca5042013-06-11 23:12:07 +020019028req.rdp_cookie_cnt([name]) : integer
19029rdp_cookie_cnt([name]) : integer (deprecated)
19030 Tries to parse the request buffer as RDP protocol, then returns an integer
19031 corresponding to the number of RDP cookies found. If an optional cookie name
19032 is passed, only cookies matching this name are considered. This is mostly
19033 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020019034
Willy Tarreau74ca5042013-06-11 23:12:07 +020019035 ACL derivatives :
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019036 req.rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020019037
Alex Zorin4afdd132018-12-30 13:56:28 +110019038req.ssl_alpn : string
19039 Returns a string containing the values of the Application-Layer Protocol
19040 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
19041 ClientHello message. Note that this only applies to raw contents found in the
19042 request buffer and not to the contents deciphered via an SSL data layer, so
19043 this will not work with "bind" lines having the "ssl" option. This is useful
19044 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020019045 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110019046
19047 Examples :
19048 # Wait for a client hello for at most 5 seconds
19049 tcp-request inspect-delay 5s
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019050 tcp-request content accept if { req.ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020019051 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110019052 default_backend bk_default
19053
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020019054req.ssl_ec_ext : boolean
19055 Returns a boolean identifying if client sent the Supported Elliptic Curves
19056 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020019057 message. This can be used to present ECC compatible clients with EC
19058 certificate and to use RSA for all others, on the same IP address. Note that
19059 this only applies to raw contents found in the request buffer and not to
19060 contents deciphered via an SSL data layer, so this will not work with "bind"
19061 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020019062
Willy Tarreau74ca5042013-06-11 23:12:07 +020019063req.ssl_hello_type : integer
19064req_ssl_hello_type : integer (deprecated)
19065 Returns an integer value containing the type of the SSL hello message found
19066 in the request buffer if the buffer contains data that parse as a complete
19067 SSL (v3 or superior) client hello message. Note that this only applies to raw
19068 contents found in the request buffer and not to contents deciphered via an
19069 SSL data layer, so this will not work with "bind" lines having the "ssl"
19070 option. This is mostly used in ACL to detect presence of an SSL hello message
19071 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020019072
Willy Tarreau74ca5042013-06-11 23:12:07 +020019073req.ssl_sni : string
19074req_ssl_sni : string (deprecated)
19075 Returns a string containing the value of the Server Name TLS extension sent
19076 by a client in a TLS stream passing through the request buffer if the buffer
19077 contains data that parse as a complete SSL (v3 or superior) client hello
19078 message. Note that this only applies to raw contents found in the request
19079 buffer and not to contents deciphered via an SSL data layer, so this will not
Lukas Tribusa267b5d2020-07-19 00:25:06 +020019080 work with "bind" lines having the "ssl" option. This will only work for actual
19081 implicit TLS based protocols like HTTPS (443), IMAPS (993), SMTPS (465),
19082 however it will not work for explicit TLS based protocols, like SMTP (25/587)
19083 or IMAP (143). SNI normally contains the name of the host the client tries to
19084 connect to (for recent browsers). SNI is useful for allowing or denying access
19085 to certain hosts when SSL/TLS is used by the client. This test was designed to
19086 be used with TCP request content inspection. If content switching is needed,
19087 it is recommended to first wait for a complete client hello (type 1), like in
19088 the example below. See also "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020019089
Willy Tarreau74ca5042013-06-11 23:12:07 +020019090 ACL derivatives :
Alex18c88772021-06-05 13:23:08 +020019091 req.ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020019092
Willy Tarreau74ca5042013-06-11 23:12:07 +020019093 Examples :
19094 # Wait for a client hello for at most 5 seconds
19095 tcp-request inspect-delay 5s
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019096 tcp-request content accept if { req.ssl_hello_type 1 }
Alex18c88772021-06-05 13:23:08 +020019097 use_backend bk_allow if { req.ssl_sni -f allowed_sites }
Willy Tarreau74ca5042013-06-11 23:12:07 +020019098 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020019099
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053019100req.ssl_st_ext : integer
19101 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
19102 Returns 1 if the client sent SessionTicket TLS Extension
19103 Returns 2 if the client also sent non-zero length TLS SessionTicket
19104 Note that this only applies to raw contents found in the request buffer and
19105 not to contents deciphered via an SSL data layer, so this will not work with
19106 "bind" lines having the "ssl" option. This can for example be used to detect
19107 whether the client sent a SessionTicket or not and stick it accordingly, if
19108 no SessionTicket then stick on SessionID or don't stick as there's no server
19109 side state is there when SessionTickets are in use.
19110
Willy Tarreau74ca5042013-06-11 23:12:07 +020019111req.ssl_ver : integer
19112req_ssl_ver : integer (deprecated)
19113 Returns an integer value containing the version of the SSL/TLS protocol of a
19114 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
19115 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
19116 composed of the major version multiplied by 65536, added to the minor
19117 version. Note that this only applies to raw contents found in the request
19118 buffer and not to contents deciphered via an SSL data layer, so this will not
19119 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010019120 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020019121 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019122
Willy Tarreau74ca5042013-06-11 23:12:07 +020019123 ACL derivatives :
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019124 req.ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010019125
Willy Tarreau47e8eba2013-09-11 23:28:46 +020019126res.len : integer
19127 Returns an integer value corresponding to the number of bytes present in the
19128 response buffer. This is mostly used in ACL. It is important to understand
19129 that this test does not return false as long as the buffer is changing. This
19130 means that a check with equality to zero will almost always immediately match
19131 at the beginning of the session, while a test for more data will wait for
Daniel Corbett9f0843f2021-05-08 10:50:37 -040019132 that data to come in and return false only when HAProxy is certain that no
Willy Tarreau47e8eba2013-09-11 23:28:46 +020019133 more data will come in. This test was designed to be used with TCP response
Christopher Faulete596d182020-05-05 17:46:34 +020019134 content inspection. But it may also be used in tcp-check based expect rules.
Willy Tarreau47e8eba2013-09-11 23:28:46 +020019135
Willy Tarreau74ca5042013-06-11 23:12:07 +020019136res.payload(<offset>,<length>) : binary
19137 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020019138 in the response buffer. As a special case, if the <length> argument is zero,
Christopher Faulete596d182020-05-05 17:46:34 +020019139 the whole buffer from <offset> to the end is extracted. This can be used
Willy Tarreau00f00842013-08-02 11:07:32 +020019140 with ACLs in order to check for the presence of some content in a buffer at
Christopher Faulete596d182020-05-05 17:46:34 +020019141 any location. It may also be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019142
Willy Tarreau74ca5042013-06-11 23:12:07 +020019143res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
19144 This extracts a binary block whose size is specified at <offset1> for <length>
19145 bytes, and which starts at <offset2> if specified or just after the length in
19146 the response buffer. The <offset2> parameter also supports relative offsets
Christopher Faulete596d182020-05-05 17:46:34 +020019147 if prepended with a '+' or '-' sign. It may also be used in tcp-check based
19148 expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019149
Willy Tarreau74ca5042013-06-11 23:12:07 +020019150 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019151
Willy Tarreau971f7b62015-09-29 14:06:59 +020019152res.ssl_hello_type : integer
19153rep_ssl_hello_type : integer (deprecated)
19154 Returns an integer value containing the type of the SSL hello message found
19155 in the response buffer if the buffer contains data that parses as a complete
19156 SSL (v3 or superior) hello message. Note that this only applies to raw
19157 contents found in the response buffer and not to contents deciphered via an
19158 SSL data layer, so this will not work with "server" lines having the "ssl"
19159 option. This is mostly used in ACL to detect presence of an SSL hello message
19160 that is supposed to contain an SSL session ID usable for stickiness.
19161
Willy Tarreau74ca5042013-06-11 23:12:07 +020019162wait_end : boolean
19163 This fetch either returns true when the inspection period is over, or does
19164 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010019165 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020019166 actions, such as a delayed reject for some special addresses. Since it either
19167 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010019168 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020019169 "WAIT_END" is always usable without prior declaration. This test was designed
19170 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019171
Willy Tarreau74ca5042013-06-11 23:12:07 +020019172 Examples :
19173 # delay every incoming request by 2 seconds
19174 tcp-request inspect-delay 2s
19175 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010019176
Willy Tarreau74ca5042013-06-11 23:12:07 +020019177 # don't immediately tell bad guys they are rejected
19178 tcp-request inspect-delay 10s
19179 acl goodguys src 10.0.0.0/24
19180 acl badguys src 10.0.1.0/24
19181 tcp-request content accept if goodguys
19182 tcp-request content reject if badguys WAIT_END
19183 tcp-request content reject
19184
19185
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200191867.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020019187--------------------------------------
19188
19189It is possible to fetch samples from HTTP contents, requests and responses.
19190This application layer is also called layer 7. It is only possible to fetch the
19191data in this section when a full HTTP request or response has been parsed from
19192its respective request or response buffer. This is always the case with all
19193HTTP specific rules and for sections running with "mode http". When using TCP
19194content inspection, it may be necessary to support an inspection delay in order
19195to let the request or response come in first. These fetches may require a bit
19196more CPU resources than the layer 4 ones, but not much since the request and
19197response are indexed.
19198
Christopher Faulet4d37e532021-03-26 14:44:00 +010019199Note : Regarding HTTP processing from the tcp-request content rules, everything
19200 will work as expected from an HTTP proxy. However, from a TCP proxy,
19201 without an HTTP upgrade, it will only work for HTTP/1 content. For
19202 HTTP/2 content, only the preface is visible. Thus, it is only possible
19203 to rely to "req.proto_http", "req.ver" and eventually "method" sample
19204 fetches. All other L7 sample fetches will fail. After an HTTP upgrade,
19205 they will work in the same manner than from an HTTP proxy.
19206
Willy Tarreau74ca5042013-06-11 23:12:07 +020019207base : string
19208 This returns the concatenation of the first Host header and the path part of
19209 the request, which starts at the first slash and ends before the question
19210 mark. It can be useful in virtual hosted environments to detect URL abuses as
19211 well as to improve shared caches efficiency. Using this with a limited size
19212 stick table also allows one to collect statistics about most commonly
19213 requested objects by host/path. With ACLs it can allow simple content
19214 switching rules involving the host and the path at the same time, such as
19215 "www.example.com/favicon.ico". See also "path" and "uri".
19216
19217 ACL derivatives :
19218 base : exact string match
19219 base_beg : prefix match
19220 base_dir : subdir match
19221 base_dom : domain match
19222 base_end : suffix match
19223 base_len : length match
19224 base_reg : regex match
19225 base_sub : substring match
19226
19227base32 : integer
19228 This returns a 32-bit hash of the value returned by the "base" fetch method
19229 above. This is useful to track per-URL activity on high traffic sites without
19230 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020019231 memory. The output type is an unsigned integer. The hash function used is
19232 SDBM with full avalanche on the output. Technically, base32 is exactly equal
19233 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020019234
19235base32+src : binary
19236 This returns the concatenation of the base32 fetch above and the src fetch
19237 below. The resulting type is of type binary, with a size of 8 or 20 bytes
19238 depending on the source address family. This can be used to track per-IP,
19239 per-URL counters.
19240
Yves Lafonb4d37082021-02-11 11:01:28 +010019241baseq : string
19242 This returns the concatenation of the first Host header and the path part of
19243 the request with the query-string, which starts at the first slash. Using this
19244 instead of "base" allows one to properly identify the target resource, for
19245 statistics or caching use cases. See also "path", "pathq" and "base".
19246
William Lallemand65ad6e12014-01-31 15:08:02 +010019247capture.req.hdr(<idx>) : string
19248 This extracts the content of the header captured by the "capture request
19249 header", idx is the position of the capture keyword in the configuration.
19250 The first entry is an index of 0. See also: "capture request header".
19251
19252capture.req.method : string
19253 This extracts the METHOD of an HTTP request. It can be used in both request
19254 and response. Unlike "method", it can be used in both request and response
19255 because it's allocated.
19256
19257capture.req.uri : string
19258 This extracts the request's URI, which starts at the first slash and ends
19259 before the first space in the request (without the host part). Unlike "path"
19260 and "url", it can be used in both request and response because it's
19261 allocated.
19262
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020019263capture.req.ver : string
19264 This extracts the request's HTTP version and returns either "HTTP/1.0" or
19265 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
19266 logs because it relies on a persistent flag.
19267
William Lallemand65ad6e12014-01-31 15:08:02 +010019268capture.res.hdr(<idx>) : string
19269 This extracts the content of the header captured by the "capture response
19270 header", idx is the position of the capture keyword in the configuration.
19271 The first entry is an index of 0.
19272 See also: "capture response header"
19273
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020019274capture.res.ver : string
19275 This extracts the response's HTTP version and returns either "HTTP/1.0" or
19276 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
19277 persistent flag.
19278
Willy Tarreaua5910cc2015-05-02 00:46:08 +020019279req.body : binary
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020019280 This returns the HTTP request's available body as a block of data. It is
19281 recommended to use "option http-buffer-request" to be sure to wait, as much
19282 as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020019283
Thierry FOURNIER9826c772015-05-20 15:50:54 +020019284req.body_param([<name>) : string
19285 This fetch assumes that the body of the POST request is url-encoded. The user
19286 can check if the "content-type" contains the value
19287 "application/x-www-form-urlencoded". This extracts the first occurrence of the
19288 parameter <name> in the body, which ends before '&'. The parameter name is
19289 case-sensitive. If no name is given, any parameter will match, and the first
19290 one will be returned. The result is a string corresponding to the value of the
19291 parameter <name> as presented in the request body (no URL decoding is
19292 performed). Note that the ACL version of this fetch iterates over multiple
19293 parameters and will iteratively report all parameters values if no name is
19294 given.
19295
Willy Tarreaua5910cc2015-05-02 00:46:08 +020019296req.body_len : integer
19297 This returns the length of the HTTP request's available body in bytes. It may
19298 be lower than the advertised length if the body is larger than the buffer. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020019299 is recommended to use "option http-buffer-request" to be sure to wait, as
19300 much as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020019301
19302req.body_size : integer
19303 This returns the advertised length of the HTTP request's body in bytes. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020019304 will represent the advertised Content-Length header, or the size of the
19305 available data in case of chunked encoding.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020019306
Willy Tarreau74ca5042013-06-11 23:12:07 +020019307req.cook([<name>]) : string
19308cook([<name>]) : string (deprecated)
19309 This extracts the last occurrence of the cookie name <name> on a "Cookie"
19310 header line from the request, and returns its value as string. If no name is
19311 specified, the first cookie value is returned. When used with ACLs, all
19312 matching cookies are evaluated. Spaces around the name and the value are
19313 ignored as requested by the Cookie header specification (RFC6265). The cookie
19314 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
19315 well return an empty value if it is present. Use the "found" match to detect
19316 presence. Use the res.cook() variant for response cookies sent by the server.
19317
19318 ACL derivatives :
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019319 req.cook([<name>]) : exact string match
19320 req.cook_beg([<name>]) : prefix match
19321 req.cook_dir([<name>]) : subdir match
19322 req.cook_dom([<name>]) : domain match
19323 req.cook_end([<name>]) : suffix match
19324 req.cook_len([<name>]) : length match
19325 req.cook_reg([<name>]) : regex match
19326 req.cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010019327
Willy Tarreau74ca5042013-06-11 23:12:07 +020019328req.cook_cnt([<name>]) : integer
19329cook_cnt([<name>]) : integer (deprecated)
19330 Returns an integer value representing the number of occurrences of the cookie
19331 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019332
Willy Tarreau74ca5042013-06-11 23:12:07 +020019333req.cook_val([<name>]) : integer
19334cook_val([<name>]) : integer (deprecated)
19335 This extracts the last occurrence of the cookie name <name> on a "Cookie"
19336 header line from the request, and converts its value to an integer which is
19337 returned. If no name is specified, the first cookie value is returned. When
19338 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020019339
Willy Tarreau74ca5042013-06-11 23:12:07 +020019340cookie([<name>]) : string (deprecated)
19341 This extracts the last occurrence of the cookie name <name> on a "Cookie"
19342 header line from the request, or a "Set-Cookie" header from the response, and
19343 returns its value as a string. A typical use is to get multiple clients
19344 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020019345 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020019346 multi-peer synchronization and state keeping across restarts. If no name is
19347 specified, the first cookie value is returned. This fetch should not be used
19348 anymore and should be replaced by req.cook() or res.cook() instead as it
19349 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019350
Willy Tarreau74ca5042013-06-11 23:12:07 +020019351hdr([<name>[,<occ>]]) : string
19352 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
19353 used on responses. Please refer to these respective fetches for more details.
19354 In case of doubt about the fetch direction, please use the explicit ones.
19355 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019356 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019357
Willy Tarreau74ca5042013-06-11 23:12:07 +020019358req.fhdr(<name>[,<occ>]) : string
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019359 This returns the full value of the last occurrence of header <name> in an
19360 HTTP request. It differs from req.hdr() in that any commas present in the
19361 value are returned and are not used as delimiters. This is sometimes useful
19362 with headers such as User-Agent.
19363
19364 When used from an ACL, all occurrences are iterated over until a match is
19365 found.
19366
Willy Tarreau74ca5042013-06-11 23:12:07 +020019367 Optionally, a specific occurrence might be specified as a position number.
19368 Positive values indicate a position from the first occurrence, with 1 being
19369 the first one. Negative values indicate positions relative to the last one,
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019370 with -1 being the last one.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019371
Willy Tarreau74ca5042013-06-11 23:12:07 +020019372req.fhdr_cnt([<name>]) : integer
19373 Returns an integer value representing the number of occurrences of request
19374 header field name <name>, or the total number of header fields if <name> is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019375 not specified. Like req.fhdr() it differs from res.hdr_cnt() by not splitting
19376 headers at commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019377
Willy Tarreau74ca5042013-06-11 23:12:07 +020019378req.hdr([<name>[,<occ>]]) : string
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019379 This returns the last comma-separated value of the header <name> in an HTTP
19380 request. The fetch considers any comma as a delimiter for distinct values.
19381 This is useful if you need to process headers that are defined to be a list
19382 of values, such as Accept, or X-Forwarded-For. If full-line headers are
19383 desired instead, use req.fhdr(). Please carefully check RFC 7231 to know how
19384 certain headers are supposed to be parsed. Also, some of them are case
19385 insensitive (e.g. Connection).
19386
19387 When used from an ACL, all occurrences are iterated over until a match is
19388 found.
19389
Willy Tarreau74ca5042013-06-11 23:12:07 +020019390 Optionally, a specific occurrence might be specified as a position number.
19391 Positive values indicate a position from the first occurrence, with 1 being
19392 the first one. Negative values indicate positions relative to the last one,
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019393 with -1 being the last one.
19394
19395 A typical use is with the X-Forwarded-For header once converted to IP,
19396 associated with an IP stick-table.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019397
Willy Tarreau74ca5042013-06-11 23:12:07 +020019398 ACL derivatives :
19399 hdr([<name>[,<occ>]]) : exact string match
19400 hdr_beg([<name>[,<occ>]]) : prefix match
19401 hdr_dir([<name>[,<occ>]]) : subdir match
19402 hdr_dom([<name>[,<occ>]]) : domain match
19403 hdr_end([<name>[,<occ>]]) : suffix match
19404 hdr_len([<name>[,<occ>]]) : length match
19405 hdr_reg([<name>[,<occ>]]) : regex match
19406 hdr_sub([<name>[,<occ>]]) : substring match
19407
19408req.hdr_cnt([<name>]) : integer
19409hdr_cnt([<header>]) : integer (deprecated)
19410 Returns an integer value representing the number of occurrences of request
19411 header field name <name>, or the total number of header field values if
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019412 <name> is not specified. Like req.hdr() it counts each comma separated
19413 part of the header's value. If counting of full-line headers is desired,
19414 then req.fhdr_cnt() should be used instead.
19415
19416 With ACLs, it can be used to detect presence, absence or abuse of a specific
19417 header, as well as to block request smuggling attacks by rejecting requests
19418 which contain more than one of certain headers.
19419
19420 Refer to req.hdr() for more information on header matching.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019421
19422req.hdr_ip([<name>[,<occ>]]) : ip
19423hdr_ip([<name>[,<occ>]]) : ip (deprecated)
19424 This extracts the last occurrence of header <name> in an HTTP request,
19425 converts it to an IPv4 or IPv6 address and returns this address. When used
19426 with ACLs, all occurrences are checked, and if <name> is omitted, every value
Willy Tarreau7b0e00d2021-03-25 14:12:29 +010019427 of every header is checked. The parser strictly adheres to the format
19428 described in RFC7239, with the extension that IPv4 addresses may optionally
19429 be followed by a colon (':') and a valid decimal port number (0 to 65535),
19430 which will be silently dropped. All other forms will not match and will
19431 cause the address to be ignored.
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019432
19433 The <occ> parameter is processed as with req.hdr().
19434
19435 A typical use is with the X-Forwarded-For and X-Client-IP headers.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019436
19437req.hdr_val([<name>[,<occ>]]) : integer
19438hdr_val([<name>[,<occ>]]) : integer (deprecated)
19439 This extracts the last occurrence of header <name> in an HTTP request, and
19440 converts it to an integer value. When used with ACLs, all occurrences are
19441 checked, and if <name> is omitted, every value of every header is checked.
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019442
19443 The <occ> parameter is processed as with req.hdr().
19444
19445 A typical use is with the X-Forwarded-For header.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019446
Christopher Faulet687a68e2020-11-24 17:13:24 +010019447req.hdrs : string
19448 Returns the current request headers as string including the last empty line
19449 separating headers from the request body. The last empty line can be used to
19450 detect a truncated header block. This sample fetch is useful for some SPOE
19451 headers analyzers and for advanced logging.
19452
19453req.hdrs_bin : binary
19454 Returns the current request headers contained in preparsed binary form. This
19455 is useful for offloading some processing with SPOE. Each string is described
19456 by a length followed by the number of bytes indicated in the length. The
19457 length is represented using the variable integer encoding detailed in the
19458 SPOE documentation. The end of the list is marked by a couple of empty header
19459 names and values (length of 0 for both).
19460
19461 *(<str:header-name><str:header-value>)<empty string><empty string>
Frédéric Lécailleec891192019-02-26 15:02:35 +010019462
Christopher Faulet687a68e2020-11-24 17:13:24 +010019463 int: refer to the SPOE documentation for the encoding
19464 str: <int:length><bytes>
Frédéric Lécailleec891192019-02-26 15:02:35 +010019465
Willy Tarreau74ca5042013-06-11 23:12:07 +020019466http_auth(<userlist>) : boolean
19467 Returns a boolean indicating whether the authentication data received from
19468 the client match a username & password stored in the specified userlist. This
19469 fetch function is not really useful outside of ACLs. Currently only http
19470 basic auth is supported.
19471
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010019472http_auth_group(<userlist>) : string
19473 Returns a string corresponding to the user name found in the authentication
19474 data received from the client if both the user name and password are valid
19475 according to the specified userlist. The main purpose is to use it in ACLs
19476 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019477 This fetch function is not really useful outside of ACLs. Currently only http
19478 basic auth is supported.
19479
19480 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010019481 http_auth_group(<userlist>) : group ...
19482 Returns true when the user extracted from the request and whose password is
19483 valid according to the specified userlist belongs to at least one of the
19484 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019485
Christopher Fauleta4063562019-08-02 11:51:37 +020019486http_auth_pass : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010019487 Returns the user's password found in the authentication data received from
19488 the client, as supplied in the Authorization header. Not checks are
19489 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020019490
19491http_auth_type : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010019492 Returns the authentication method found in the authentication data received from
19493 the client, as supplied in the Authorization header. Not checks are
19494 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020019495
19496http_auth_user : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010019497 Returns the user name found in the authentication data received from the
19498 client, as supplied in the Authorization header. Not checks are performed by
19499 this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020019500
Willy Tarreau74ca5042013-06-11 23:12:07 +020019501http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020019502 Returns true when the request being processed is the first one of the
19503 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020019504 from some requests when a request is not the first one, or to help grouping
19505 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020019506
Willy Tarreau74ca5042013-06-11 23:12:07 +020019507method : integer + string
19508 Returns an integer value corresponding to the method in the HTTP request. For
19509 example, "GET" equals 1 (check sources to establish the matching). Value 9
19510 means "other method" and may be converted to a string extracted from the
19511 stream. This should not be used directly as a sample, this is only meant to
19512 be used from ACLs, which transparently convert methods from patterns to these
19513 integer + string values. Some predefined ACL already check for most common
19514 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019515
Willy Tarreau74ca5042013-06-11 23:12:07 +020019516 ACL derivatives :
19517 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020019518
Willy Tarreau74ca5042013-06-11 23:12:07 +020019519 Example :
19520 # only accept GET and HEAD requests
19521 acl valid_method method GET HEAD
19522 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020019523
Willy Tarreau74ca5042013-06-11 23:12:07 +020019524path : string
19525 This extracts the request's URL path, which starts at the first slash and
19526 ends before the question mark (without the host part). A typical use is with
19527 prefetch-capable caches, and with portals which need to aggregate multiple
19528 information from databases and keep them in caches. Note that with outgoing
19529 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010019530 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020019531 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019532
Willy Tarreau74ca5042013-06-11 23:12:07 +020019533 ACL derivatives :
19534 path : exact string match
19535 path_beg : prefix match
19536 path_dir : subdir match
19537 path_dom : domain match
19538 path_end : suffix match
19539 path_len : length match
19540 path_reg : regex match
19541 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020019542
Christopher Faulete720c322020-09-02 17:25:18 +020019543pathq : string
19544 This extracts the request's URL path with the query-string, which starts at
19545 the first slash. This sample fetch is pretty handy to always retrieve a
19546 relative URI, excluding the scheme and the authority part, if any. Indeed,
19547 while it is the common representation for an HTTP/1.1 request target, in
19548 HTTP/2, an absolute URI is often used. This sample fetch will return the same
19549 result in both cases.
19550
Willy Tarreau49ad95c2015-01-19 15:06:26 +010019551query : string
19552 This extracts the request's query string, which starts after the first
19553 question mark. If no question mark is present, this fetch returns nothing. If
19554 a question mark is present but nothing follows, it returns an empty string.
19555 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010019556 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010019557 which stops before the question mark.
19558
Willy Tarreaueb27ec72015-02-20 13:55:29 +010019559req.hdr_names([<delim>]) : string
19560 This builds a string made from the concatenation of all header names as they
19561 appear in the request when the rule is evaluated. The default delimiter is
19562 the comma (',') but it may be overridden as an optional argument <delim>. In
19563 this case, only the first character of <delim> is considered.
19564
Willy Tarreau74ca5042013-06-11 23:12:07 +020019565req.ver : string
19566req_ver : string (deprecated)
19567 Returns the version string from the HTTP request, for example "1.1". This can
19568 be useful for logs, but is mostly there for ACL. Some predefined ACL already
19569 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019570
Willy Tarreau74ca5042013-06-11 23:12:07 +020019571 ACL derivatives :
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019572 req.ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020019573
Christopher Faulete596d182020-05-05 17:46:34 +020019574res.body : binary
19575 This returns the HTTP response's available body as a block of data. Unlike
19576 the request side, there is no directive to wait for the response's body. This
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019577 sample fetch is really useful (and usable) in the health-check context.
19578
19579 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019580
19581res.body_len : integer
19582 This returns the length of the HTTP response available body in bytes. Unlike
19583 the request side, there is no directive to wait for the response's body. This
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019584 sample fetch is really useful (and usable) in the health-check context.
19585
19586 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019587
19588res.body_size : integer
19589 This returns the advertised length of the HTTP response body in bytes. It
19590 will represent the advertised Content-Length header, or the size of the
19591 available data in case of chunked encoding. Unlike the request side, there is
19592 no directive to wait for the response body. This sample fetch is really
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019593 useful (and usable) in the health-check context.
19594
19595 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019596
Remi Tricot-Le Bretonbf971212020-10-27 11:55:57 +010019597res.cache_hit : boolean
19598 Returns the boolean "true" value if the response has been built out of an
19599 HTTP cache entry, otherwise returns boolean "false".
19600
19601res.cache_name : string
19602 Returns a string containing the name of the HTTP cache that was used to
19603 build the HTTP response if res.cache_hit is true, otherwise returns an
19604 empty string.
19605
Willy Tarreau74ca5042013-06-11 23:12:07 +020019606res.comp : boolean
19607 Returns the boolean "true" value if the response has been compressed by
19608 HAProxy, otherwise returns boolean "false". This may be used to add
19609 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019610
Willy Tarreau74ca5042013-06-11 23:12:07 +020019611res.comp_algo : string
19612 Returns a string containing the name of the algorithm used if the response
19613 was compressed by HAProxy, for example : "deflate". This may be used to add
19614 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019615
Willy Tarreau74ca5042013-06-11 23:12:07 +020019616res.cook([<name>]) : string
19617scook([<name>]) : string (deprecated)
19618 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
19619 header line from the response, and returns its value as string. If no name is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019620 specified, the first cookie value is returned.
19621
19622 It may be used in tcp-check based expect rules.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020019623
Willy Tarreau74ca5042013-06-11 23:12:07 +020019624 ACL derivatives :
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019625 res.scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020019626
Willy Tarreau74ca5042013-06-11 23:12:07 +020019627res.cook_cnt([<name>]) : integer
19628scook_cnt([<name>]) : integer (deprecated)
19629 Returns an integer value representing the number of occurrences of the cookie
19630 <name> in the response, or all cookies if <name> is not specified. This is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019631 mostly useful when combined with ACLs to detect suspicious responses.
19632
19633 It may be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019634
Willy Tarreau74ca5042013-06-11 23:12:07 +020019635res.cook_val([<name>]) : integer
19636scook_val([<name>]) : integer (deprecated)
19637 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
19638 header line from the response, and converts its value to an integer which is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019639 returned. If no name is specified, the first cookie value is returned.
19640
19641 It may be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019642
Willy Tarreau74ca5042013-06-11 23:12:07 +020019643res.fhdr([<name>[,<occ>]]) : string
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019644 This fetch works like the req.fhdr() fetch with the difference that it acts
19645 on the headers within an HTTP response.
19646
19647 Like req.fhdr() the res.fhdr() fetch returns full values. If the header is
19648 defined to be a list you should use res.hdr().
19649
19650 This fetch is sometimes useful with headers such as Date or Expires.
19651
19652 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019653
Willy Tarreau74ca5042013-06-11 23:12:07 +020019654res.fhdr_cnt([<name>]) : integer
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019655 This fetch works like the req.fhdr_cnt() fetch with the difference that it
19656 acts on the headers within an HTTP response.
19657
19658 Like req.fhdr_cnt() the res.fhdr_cnt() fetch acts on full values. If the
19659 header is defined to be a list you should use res.hdr_cnt().
19660
19661 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019662
Willy Tarreau74ca5042013-06-11 23:12:07 +020019663res.hdr([<name>[,<occ>]]) : string
19664shdr([<name>[,<occ>]]) : string (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019665 This fetch works like the req.hdr() fetch with the difference that it acts
19666 on the headers within an HTTP response.
19667
Ilya Shipitsinacf84592021-02-06 22:29:08 +050019668 Like req.hdr() the res.hdr() fetch considers the comma to be a delimiter. If
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019669 this is not desired res.fhdr() should be used.
19670
19671 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019672
Willy Tarreau74ca5042013-06-11 23:12:07 +020019673 ACL derivatives :
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019674 res.hdr([<name>[,<occ>]]) : exact string match
19675 res.hdr_beg([<name>[,<occ>]]) : prefix match
19676 res.hdr_dir([<name>[,<occ>]]) : subdir match
19677 res.hdr_dom([<name>[,<occ>]]) : domain match
19678 res.hdr_end([<name>[,<occ>]]) : suffix match
19679 res.hdr_len([<name>[,<occ>]]) : length match
19680 res.hdr_reg([<name>[,<occ>]]) : regex match
19681 res.hdr_sub([<name>[,<occ>]]) : substring match
Willy Tarreau74ca5042013-06-11 23:12:07 +020019682
19683res.hdr_cnt([<name>]) : integer
19684shdr_cnt([<name>]) : integer (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019685 This fetch works like the req.hdr_cnt() fetch with the difference that it
19686 acts on the headers within an HTTP response.
19687
19688 Like req.hdr_cnt() the res.hdr_cnt() fetch considers the comma to be a
Ilya Shipitsinacf84592021-02-06 22:29:08 +050019689 delimiter. If this is not desired res.fhdr_cnt() should be used.
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019690
19691 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019692
Willy Tarreau74ca5042013-06-11 23:12:07 +020019693res.hdr_ip([<name>[,<occ>]]) : ip
19694shdr_ip([<name>[,<occ>]]) : ip (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019695 This fetch works like the req.hdr_ip() fetch with the difference that it
19696 acts on the headers within an HTTP response.
19697
19698 This can be useful to learn some data into a stick table.
19699
19700 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019701
Willy Tarreaueb27ec72015-02-20 13:55:29 +010019702res.hdr_names([<delim>]) : string
19703 This builds a string made from the concatenation of all header names as they
19704 appear in the response when the rule is evaluated. The default delimiter is
19705 the comma (',') but it may be overridden as an optional argument <delim>. In
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019706 this case, only the first character of <delim> is considered.
19707
19708 It may be used in tcp-check based expect rules.
Willy Tarreaueb27ec72015-02-20 13:55:29 +010019709
Willy Tarreau74ca5042013-06-11 23:12:07 +020019710res.hdr_val([<name>[,<occ>]]) : integer
19711shdr_val([<name>[,<occ>]]) : integer (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019712 This fetch works like the req.hdr_val() fetch with the difference that it
19713 acts on the headers within an HTTP response.
19714
19715 This can be useful to learn some data into a stick table.
19716
19717 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019718
19719res.hdrs : string
19720 Returns the current response headers as string including the last empty line
19721 separating headers from the request body. The last empty line can be used to
19722 detect a truncated header block. This sample fetch is useful for some SPOE
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019723 headers analyzers and for advanced logging.
19724
19725 It may also be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019726
19727res.hdrs_bin : binary
19728 Returns the current response headers contained in preparsed binary form. This
19729 is useful for offloading some processing with SPOE. It may be used in
19730 tcp-check based expect rules. Each string is described by a length followed
19731 by the number of bytes indicated in the length. The length is represented
19732 using the variable integer encoding detailed in the SPOE documentation. The
19733 end of the list is marked by a couple of empty header names and values
19734 (length of 0 for both).
19735
19736 *(<str:header-name><str:header-value>)<empty string><empty string>
19737
19738 int: refer to the SPOE documentation for the encoding
19739 str: <int:length><bytes>
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010019740
Willy Tarreau74ca5042013-06-11 23:12:07 +020019741res.ver : string
19742resp_ver : string (deprecated)
19743 Returns the version string from the HTTP response, for example "1.1". This
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019744 can be useful for logs, but is mostly there for ACL.
19745
19746 It may be used in tcp-check based expect rules.
Willy Tarreau0e698542011-09-16 08:32:32 +020019747
Willy Tarreau74ca5042013-06-11 23:12:07 +020019748 ACL derivatives :
Christian Ruppertfd3184c2022-02-20 22:54:01 +010019749 resp.ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010019750
Willy Tarreau74ca5042013-06-11 23:12:07 +020019751set-cookie([<name>]) : string (deprecated)
19752 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
19753 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020019754 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020019755 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010019756
Willy Tarreau74ca5042013-06-11 23:12:07 +020019757 This fetch function is deprecated and has been superseded by the "res.cook"
19758 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010019759
Willy Tarreau74ca5042013-06-11 23:12:07 +020019760status : integer
19761 Returns an integer containing the HTTP status code in the HTTP response, for
19762 example, 302. It is mostly used within ACLs and integer ranges, for example,
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019763 to remove any Location header if the response is not a 3xx.
19764
19765 It may be used in tcp-check based expect rules.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019766
Thierry Fournier0e00dca2016-04-07 15:47:40 +020019767unique-id : string
19768 Returns the unique-id attached to the request. The directive
19769 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
19770 fails. Note that the unique-id is usually used with HTTP requests, however this
19771 sample fetch can be used with other protocols. Obviously, if it is used with
19772 other protocols than HTTP, the unique-id-format directive must not contain
19773 HTTP parts. See: unique-id-format and unique-id-header
19774
Willy Tarreau74ca5042013-06-11 23:12:07 +020019775url : string
19776 This extracts the request's URL as presented in the request. A typical use is
19777 with prefetch-capable caches, and with portals which need to aggregate
19778 multiple information from databases and keep them in caches. With ACLs, using
19779 "path" is preferred over using "url", because clients may send a full URL as
19780 is normally done with proxies. The only real use is to match "*" which does
19781 not match in "path", and for which there is already a predefined ACL. See
19782 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019783
Willy Tarreau74ca5042013-06-11 23:12:07 +020019784 ACL derivatives :
19785 url : exact string match
19786 url_beg : prefix match
19787 url_dir : subdir match
19788 url_dom : domain match
19789 url_end : suffix match
19790 url_len : length match
19791 url_reg : regex match
19792 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019793
Willy Tarreau74ca5042013-06-11 23:12:07 +020019794url_ip : ip
19795 This extracts the IP address from the request's URL when the host part is
19796 presented as an IP address. Its use is very limited. For instance, a
19797 monitoring system might use this field as an alternative for the source IP in
19798 order to test what path a given source address would follow, or to force an
19799 entry in a table for a given source address. With ACLs it can be used to
19800 restrict access to certain systems through a proxy, for example when combined
19801 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019802
Willy Tarreau74ca5042013-06-11 23:12:07 +020019803url_port : integer
19804 This extracts the port part from the request's URL. Note that if the port is
19805 not specified in the request, port 80 is assumed. With ACLs it can be used to
19806 restrict access to certain systems through a proxy, for example when combined
19807 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019808
Willy Tarreau1ede1da2015-05-07 16:06:18 +020019809urlp([<name>[,<delim>]]) : string
19810url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020019811 This extracts the first occurrence of the parameter <name> in the query
19812 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020019813 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
19814 any parameter will match, and the first one will be returned. The result is
19815 a string corresponding to the value of the parameter <name> as presented in
19816 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020019817 stickiness based on a client ID, to extract an application cookie passed as a
19818 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020019819 this fetch iterates over multiple parameters and will iteratively report all
19820 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019821
Willy Tarreau74ca5042013-06-11 23:12:07 +020019822 ACL derivatives :
19823 urlp(<name>[,<delim>]) : exact string match
19824 urlp_beg(<name>[,<delim>]) : prefix match
19825 urlp_dir(<name>[,<delim>]) : subdir match
19826 urlp_dom(<name>[,<delim>]) : domain match
19827 urlp_end(<name>[,<delim>]) : suffix match
19828 urlp_len(<name>[,<delim>]) : length match
19829 urlp_reg(<name>[,<delim>]) : regex match
19830 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019831
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019832
Willy Tarreau74ca5042013-06-11 23:12:07 +020019833 Example :
19834 # match http://example.com/foo?PHPSESSIONID=some_id
19835 stick on urlp(PHPSESSIONID)
19836 # match http://example.com/foo;JSESSIONID=some_id
19837 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019838
Jarno Huuskonen676f6222017-03-30 09:19:45 +030019839urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020019840 See "urlp" above. This one extracts the URL parameter <name> in the request
19841 and converts it to an integer value. This can be used for session stickiness
19842 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020019843
Dragan Dosen0070cd52016-06-16 12:19:49 +020019844url32 : integer
19845 This returns a 32-bit hash of the value obtained by concatenating the first
19846 Host header and the whole URL including parameters (not only the path part of
19847 the request, as in the "base32" fetch above). This is useful to track per-URL
19848 activity. A shorter hash is stored, saving a lot of memory. The output type
19849 is an unsigned integer.
19850
19851url32+src : binary
19852 This returns the concatenation of the "url32" fetch and the "src" fetch. The
19853 resulting type is of type binary, with a size of 8 or 20 bytes depending on
19854 the source address family. This can be used to track per-IP, per-URL counters.
19855
Christopher Faulet16032ab2020-04-30 11:30:00 +020019856
Christopher Faulete596d182020-05-05 17:46:34 +0200198577.3.7. Fetching samples for developers
Christopher Fauletd47941d2020-01-08 14:40:19 +010019858---------------------------------------
19859
19860This set of sample fetch methods is reserved to developers and must never be
19861used on a production environment, except on developer demand, for debugging
19862purposes. Moreover, no special care will be taken on backwards compatibility.
19863There is no warranty the following sample fetches will never change, be renamed
19864or simply removed. So be really careful if you should use one of them. To avoid
19865any ambiguity, these sample fetches are placed in the dedicated scope "internal",
19866for instance "internal.strm.is_htx".
19867
19868internal.htx.data : integer
19869 Returns the size in bytes used by data in the HTX message associated to a
19870 channel. The channel is chosen depending on the sample direction.
19871
19872internal.htx.free : integer
19873 Returns the free space (size - used) in bytes in the HTX message associated
19874 to a channel. The channel is chosen depending on the sample direction.
19875
19876internal.htx.free_data : integer
19877 Returns the free space for the data in bytes in the HTX message associated to
19878 a channel. The channel is chosen depending on the sample direction.
19879
19880internal.htx.has_eom : boolean
Christopher Fauletd1ac2b92020-12-02 19:12:22 +010019881 Returns true if the HTX message associated to a channel contains the
19882 end-of-message flag (EOM). Otherwise, it returns false. The channel is chosen
19883 depending on the sample direction.
Christopher Fauletd47941d2020-01-08 14:40:19 +010019884
19885internal.htx.nbblks : integer
19886 Returns the number of blocks present in the HTX message associated to a
19887 channel. The channel is chosen depending on the sample direction.
19888
19889internal.htx.size : integer
19890 Returns the total size in bytes of the HTX message associated to a
19891 channel. The channel is chosen depending on the sample direction.
19892
19893internal.htx.used : integer
19894 Returns the total size used in bytes (data + metadata) in the HTX message
19895 associated to a channel. The channel is chosen depending on the sample
19896 direction.
19897
19898internal.htx_blk.size(<idx>) : integer
19899 Returns the size of the block at the position <idx> in the HTX message
19900 associated to a channel or 0 if it does not exist. The channel is chosen
19901 depending on the sample direction. <idx> may be any positive integer or one
19902 of the special value :
19903 * head : The oldest inserted block
19904 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019905 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019906
19907internal.htx_blk.type(<idx>) : string
19908 Returns the type of the block at the position <idx> in the HTX message
19909 associated to a channel or "HTX_BLK_UNUSED" if it does not exist. The channel
19910 is chosen depending on the sample direction. <idx> may be any positive
19911 integer or one of the special value :
19912 * head : The oldest inserted block
19913 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019914 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019915
19916internal.htx_blk.data(<idx>) : binary
19917 Returns the value of the DATA block at the position <idx> in the HTX message
19918 associated to a channel or an empty string if it does not exist or if it is
19919 not a DATA block. The channel is chosen depending on the sample direction.
19920 <idx> may be any positive integer or one of the special value :
19921
19922 * head : The oldest inserted block
19923 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019924 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019925
19926internal.htx_blk.hdrname(<idx>) : string
19927 Returns the header name of the HEADER block at the position <idx> in the HTX
19928 message associated to a channel or an empty string if it does not exist or if
19929 it is not an HEADER block. The channel is chosen depending on the sample
19930 direction. <idx> may be any positive integer or one of the special value :
19931
19932 * head : The oldest inserted block
19933 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019934 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019935
19936internal.htx_blk.hdrval(<idx>) : string
19937 Returns the header value of the HEADER block at the position <idx> in the HTX
19938 message associated to a channel or an empty string if it does not exist or if
19939 it is not an HEADER block. The channel is chosen depending on the sample
19940 direction. <idx> may be any positive integer or one of the special value :
19941
19942 * head : The oldest inserted block
19943 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019944 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019945
19946internal.htx_blk.start_line(<idx>) : string
19947 Returns the value of the REQ_SL or RES_SL block at the position <idx> in the
19948 HTX message associated to a channel or an empty string if it does not exist
19949 or if it is not a SL block. The channel is chosen depending on the sample
19950 direction. <idx> may be any positive integer or one of the special value :
19951
19952 * head : The oldest inserted block
19953 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019954 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019955
19956internal.strm.is_htx : boolean
19957 Returns true if the current stream is an HTX stream. It means the data in the
19958 channels buffers are stored using the internal HTX representation. Otherwise,
19959 it returns false.
19960
19961
Willy Tarreau74ca5042013-06-11 23:12:07 +0200199627.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019963---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010019964
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019965Some predefined ACLs are hard-coded so that they do not have to be declared in
19966every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020019967order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010019968
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019969ACL name Equivalent to Usage
Christopher Faulet779184e2021-04-01 17:24:04 +020019970---------------+----------------------------------+------------------------------------------------------
19971FALSE always_false never match
19972HTTP req.proto_http match if request protocol is valid HTTP
19973HTTP_1.0 req.ver 1.0 match if HTTP request version is 1.0
19974HTTP_1.1 req.ver 1.1 match if HTTP request version is 1.1
Christopher Faulet8043e832021-03-26 16:00:54 +010019975HTTP_2.0 req.ver 2.0 match if HTTP request version is 2.0
Christopher Faulet779184e2021-04-01 17:24:04 +020019976HTTP_CONTENT req.hdr_val(content-length) gt 0 match an existing content-length in the HTTP request
19977HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
19978HTTP_URL_SLASH url_beg / match URL beginning with "/"
19979HTTP_URL_STAR url * match URL equal to "*"
19980LOCALHOST src 127.0.0.1/8 match connection from local host
19981METH_CONNECT method CONNECT match HTTP CONNECT method
19982METH_DELETE method DELETE match HTTP DELETE method
19983METH_GET method GET HEAD match HTTP GET or HEAD method
19984METH_HEAD method HEAD match HTTP HEAD method
19985METH_OPTIONS method OPTIONS match HTTP OPTIONS method
19986METH_POST method POST match HTTP POST method
19987METH_PUT method PUT match HTTP PUT method
19988METH_TRACE method TRACE match HTTP TRACE method
19989RDP_COOKIE req.rdp_cookie_cnt gt 0 match presence of an RDP cookie in the request buffer
19990REQ_CONTENT req.len gt 0 match data in the request buffer
19991TRUE always_true always match
19992WAIT_END wait_end wait for end of content analysis
19993---------------+----------------------------------+------------------------------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010019994
Willy Tarreaub937b7e2010-01-12 15:27:54 +010019995
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200199968. Logging
19997----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010019998
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019999One of HAProxy's strong points certainly lies is its precise logs. It probably
20000provides the finest level of information available for such a product, which is
20001very important for troubleshooting complex environments. Standard information
20002provided in logs include client ports, TCP/HTTP state timers, precise session
20003state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020004to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020005headers.
20006
20007In order to improve administrators reactivity, it offers a great transparency
20008about encountered problems, both internal and external, and it is possible to
20009send logs to different sources at the same time with different level filters :
20010
20011 - global process-level logs (system errors, start/stop, etc..)
20012 - per-instance system and internal errors (lack of resource, bugs, ...)
20013 - per-instance external troubles (servers up/down, max connections)
20014 - per-instance activity (client connections), either at the establishment or
20015 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010020016 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060020017 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020018
20019The ability to distribute different levels of logs to different log servers
20020allow several production teams to interact and to fix their problems as soon
20021as possible. For example, the system team might monitor system-wide errors,
20022while the application team might be monitoring the up/down for their servers in
20023real time, and the security team might analyze the activity logs with one hour
20024delay.
20025
20026
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200200278.1. Log levels
20028---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020029
Simon Hormandf791f52011-05-29 15:01:10 +090020030TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020031source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090020032HTTP request, HTTP return code, number of bytes transmitted, conditions
20033in which the session ended, and even exchanged cookies values. For example
20034track a particular user's problems. All messages may be sent to up to two
20035syslog servers. Check the "log" keyword in section 4.2 for more information
20036about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020037
20038
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200200398.2. Log formats
20040----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020041
William Lallemand48940402012-01-30 16:47:22 +010020042HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090020043and will be detailed in the following sections. A few of them may vary
20044slightly with the configuration, due to indicators specific to certain
20045options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020046
20047 - the default format, which is very basic and very rarely used. It only
20048 provides very basic information about the incoming connection at the moment
20049 it is accepted : source IP:port, destination IP:port, and frontend-name.
20050 This mode will eventually disappear so it will not be described to great
20051 extents.
20052
20053 - the TCP format, which is more advanced. This format is enabled when "option
20054 tcplog" is set on the frontend. HAProxy will then usually wait for the
20055 connection to terminate before logging. This format provides much richer
20056 information, such as timers, connection counts, queue size, etc... This
20057 format is recommended for pure TCP proxies.
20058
20059 - the HTTP format, which is the most advanced for HTTP proxying. This format
20060 is enabled when "option httplog" is set on the frontend. It provides the
20061 same information as the TCP format with some HTTP-specific fields such as
20062 the request, the status code, and captures of headers and cookies. This
20063 format is recommended for HTTP proxies.
20064
Emeric Brun3a058f32009-06-30 18:26:00 +020020065 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
20066 fields arranged in the same order as the CLF format. In this mode, all
20067 timers, captures, flags, etc... appear one per field after the end of the
20068 common fields, in the same order they appear in the standard HTTP format.
20069
William Lallemand48940402012-01-30 16:47:22 +010020070 - the custom log format, allows you to make your own log line.
20071
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020072Next sections will go deeper into details for each of these formats. Format
20073specification will be performed on a "field" basis. Unless stated otherwise, a
20074field is a portion of text delimited by any number of spaces. Since syslog
20075servers are susceptible of inserting fields at the beginning of a line, it is
20076always assumed that the first field is the one containing the process name and
20077identifier.
20078
20079Note : Since log lines may be quite long, the log examples in sections below
20080 might be broken into multiple lines. The example log lines will be
20081 prefixed with 3 closing angle brackets ('>>>') and each time a log is
20082 broken into multiple lines, each non-final line will end with a
20083 backslash ('\') and the next line will start indented by two characters.
20084
20085
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200200868.2.1. Default log format
20087-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020088
20089This format is used when no specific option is set. The log is emitted as soon
20090as the connection is accepted. One should note that this currently is the only
20091format which logs the request's destination IP and ports.
20092
20093 Example :
20094 listen www
20095 mode http
20096 log global
20097 server srv1 127.0.0.1:8000
20098
20099 >>> Feb 6 12:12:09 localhost \
20100 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
20101 (www/HTTP)
20102
20103 Field Format Extract from the example above
20104 1 process_name '[' pid ']:' haproxy[14385]:
20105 2 'Connect from' Connect from
20106 3 source_ip ':' source_port 10.0.1.2:33312
20107 4 'to' to
20108 5 destination_ip ':' destination_port 10.0.3.31:8012
20109 6 '(' frontend_name '/' mode ')' (www/HTTP)
20110
20111Detailed fields description :
20112 - "source_ip" is the IP address of the client which initiated the connection.
20113 - "source_port" is the TCP port of the client which initiated the connection.
20114 - "destination_ip" is the IP address the client connected to.
20115 - "destination_port" is the TCP port the client connected to.
20116 - "frontend_name" is the name of the frontend (or listener) which received
20117 and processed the connection.
20118 - "mode is the mode the frontend is operating (TCP or HTTP).
20119
Willy Tarreauceb24bc2010-11-09 12:46:41 +010020120In case of a UNIX socket, the source and destination addresses are marked as
20121"unix:" and the ports reflect the internal ID of the socket which accepted the
20122connection (the same ID as reported in the stats).
20123
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020124It is advised not to use this deprecated format for newer installations as it
20125will eventually disappear.
20126
20127
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200201288.2.2. TCP log format
20129---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020130
20131The TCP format is used when "option tcplog" is specified in the frontend, and
20132is the recommended format for pure TCP proxies. It provides a lot of precious
20133information for troubleshooting. Since this format includes timers and byte
20134counts, the log is normally emitted at the end of the session. It can be
20135emitted earlier if "option logasap" is specified, which makes sense in most
20136environments with long sessions such as remote terminals. Sessions which match
20137the "monitor" rules are never logged. It is also possible not to emit logs for
20138sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020020139specifying "option dontlognull" in the frontend. Successful connections will
20140not be logged if "option dontlog-normal" is specified in the frontend. A few
20141fields may slightly vary depending on some configuration options, those are
20142marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020143
20144 Example :
20145 frontend fnt
20146 mode tcp
20147 option tcplog
20148 log global
20149 default_backend bck
20150
20151 backend bck
20152 server srv1 127.0.0.1:8000
20153
20154 >>> Feb 6 12:12:56 localhost \
20155 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
20156 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
20157
20158 Field Format Extract from the example above
20159 1 process_name '[' pid ']:' haproxy[14387]:
20160 2 client_ip ':' client_port 10.0.1.2:33313
20161 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
20162 4 frontend_name fnt
20163 5 backend_name '/' server_name bck/srv1
20164 6 Tw '/' Tc '/' Tt* 0/0/5007
20165 7 bytes_read* 212
20166 8 termination_state --
20167 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
20168 10 srv_queue '/' backend_queue 0/0
20169
20170Detailed fields description :
20171 - "client_ip" is the IP address of the client which initiated the TCP
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020172 connection to HAProxy. If the connection was accepted on a UNIX socket
Willy Tarreauceb24bc2010-11-09 12:46:41 +010020173 instead, the IP address would be replaced with the word "unix". Note that
20174 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010020175 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010020176 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010020177 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020178
20179 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010020180 If the connection was accepted on a UNIX socket instead, the port would be
20181 replaced with the ID of the accepting socket, which is also reported in the
20182 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020183
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020184 - "accept_date" is the exact date when the connection was received by HAProxy
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020185 (which might be very slightly different from the date observed on the
20186 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020020187 the same date which may appear in any upstream firewall's log. When used in
20188 HTTP mode, the accept_date field will be reset to the first moment the
20189 connection is ready to receive a new request (end of previous response for
20190 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020191
20192 - "frontend_name" is the name of the frontend (or listener) which received
20193 and processed the connection.
20194
20195 - "backend_name" is the name of the backend (or listener) which was selected
20196 to manage the connection to the server. This will be the same as the
20197 frontend if no switching rule has been applied, which is common for TCP
20198 applications.
20199
20200 - "server_name" is the name of the last server to which the connection was
20201 sent, which might differ from the first one if there were connection errors
20202 and a redispatch occurred. Note that this server belongs to the backend
20203 which processed the request. If the connection was aborted before reaching
20204 a server, "<NOSRV>" is indicated instead of a server name.
20205
20206 - "Tw" is the total time in milliseconds spent waiting in the various queues.
20207 It can be "-1" if the connection was aborted before reaching the queue.
20208 See "Timers" below for more details.
20209
20210 - "Tc" is the total time in milliseconds spent waiting for the connection to
20211 establish to the final server, including retries. It can be "-1" if the
20212 connection was aborted before a connection could be established. See
20213 "Timers" below for more details.
20214
20215 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020216 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020217 "option logasap" was specified, then the time counting stops at the moment
20218 the log is emitted. In this case, a '+' sign is prepended before the value,
20219 indicating that the final one will be larger. See "Timers" below for more
20220 details.
20221
20222 - "bytes_read" is the total number of bytes transmitted from the server to
20223 the client when the log is emitted. If "option logasap" is specified, the
20224 this value will be prefixed with a '+' sign indicating that the final one
20225 may be larger. Please note that this value is a 64-bit counter, so log
20226 analysis tools must be able to handle it without overflowing.
20227
20228 - "termination_state" is the condition the session was in when the session
20229 ended. This indicates the session state, which side caused the end of
20230 session to happen, and for what reason (timeout, error, ...). The normal
20231 flags should be "--", indicating the session was closed by either end with
20232 no data remaining in buffers. See below "Session state at disconnection"
20233 for more details.
20234
20235 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020236 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020237 limits have been reached. For instance, if actconn is close to 512 when
20238 multiple connection errors occur, chances are high that the system limits
20239 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020240 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020241
20242 - "feconn" is the total number of concurrent connections on the frontend when
20243 the session was logged. It is useful to estimate the amount of resource
20244 required to sustain high loads, and to detect when the frontend's "maxconn"
20245 has been reached. Most often when this value increases by huge jumps, it is
20246 because there is congestion on the backend servers, but sometimes it can be
20247 caused by a denial of service attack.
20248
20249 - "beconn" is the total number of concurrent connections handled by the
20250 backend when the session was logged. It includes the total number of
20251 concurrent connections active on servers as well as the number of
20252 connections pending in queues. It is useful to estimate the amount of
20253 additional servers needed to support high loads for a given application.
20254 Most often when this value increases by huge jumps, it is because there is
20255 congestion on the backend servers, but sometimes it can be caused by a
20256 denial of service attack.
20257
20258 - "srv_conn" is the total number of concurrent connections still active on
20259 the server when the session was logged. It can never exceed the server's
20260 configured "maxconn" parameter. If this value is very often close or equal
20261 to the server's "maxconn", it means that traffic regulation is involved a
20262 lot, meaning that either the server's maxconn value is too low, or that
20263 there aren't enough servers to process the load with an optimal response
20264 time. When only one of the server's "srv_conn" is high, it usually means
20265 that this server has some trouble causing the connections to take longer to
20266 be processed than on other servers.
20267
20268 - "retries" is the number of connection retries experienced by this session
20269 when trying to connect to the server. It must normally be zero, unless a
20270 server is being stopped at the same moment the connection was attempted.
20271 Frequent retries generally indicate either a network problem between
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020272 HAProxy and the server, or a misconfigured system backlog on the server
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020273 preventing new connections from being queued. This field may optionally be
20274 prefixed with a '+' sign, indicating that the session has experienced a
20275 redispatch after the maximal retry count has been reached on the initial
20276 server. In this case, the server name appearing in the log is the one the
20277 connection was redispatched to, and not the first one, though both may
20278 sometimes be the same in case of hashing for instance. So as a general rule
20279 of thumb, when a '+' is present in front of the retry count, this count
20280 should not be attributed to the logged server.
20281
20282 - "srv_queue" is the total number of requests which were processed before
20283 this one in the server queue. It is zero when the request has not gone
20284 through the server queue. It makes it possible to estimate the approximate
20285 server's response time by dividing the time spent in queue by the number of
20286 requests in the queue. It is worth noting that if a session experiences a
20287 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010020288 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020289 backend queue unless a redispatch occurs.
20290
20291 - "backend_queue" is the total number of requests which were processed before
20292 this one in the backend's global queue. It is zero when the request has not
20293 gone through the global queue. It makes it possible to estimate the average
20294 queue length, which easily translates into a number of missing servers when
20295 divided by a server's "maxconn" parameter. It is worth noting that if a
20296 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010020297 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020298 through both the server queue and the backend queue unless a redispatch
20299 occurs.
20300
20301
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200203028.2.3. HTTP log format
20303----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020304
20305The HTTP format is the most complete and the best suited for HTTP proxies. It
20306is enabled by when "option httplog" is specified in the frontend. It provides
20307the same level of information as the TCP format with additional features which
20308are specific to the HTTP protocol. Just like the TCP format, the log is usually
20309emitted at the end of the session, unless "option logasap" is specified, which
20310generally only makes sense for download sites. A session which matches the
20311"monitor" rules will never logged. It is also possible not to log sessions for
20312which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020020313frontend. Successful connections will not be logged if "option dontlog-normal"
20314is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020315
20316Most fields are shared with the TCP log, some being different. A few fields may
20317slightly vary depending on some configuration options. Those ones are marked
20318with a star ('*') after the field name below.
20319
20320 Example :
20321 frontend http-in
20322 mode http
20323 option httplog
20324 log global
20325 default_backend bck
20326
20327 backend static
20328 server srv1 127.0.0.1:8000
20329
20330 >>> Feb 6 12:14:14 localhost \
20331 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
20332 static/srv1 10/0/30/69/109 200 2750 - - ---- 1/1/1/1/0 0/0 {1wt.eu} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020333 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020334
20335 Field Format Extract from the example above
20336 1 process_name '[' pid ']:' haproxy[14389]:
20337 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020338 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020339 4 frontend_name http-in
20340 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020341 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020342 7 status_code 200
20343 8 bytes_read* 2750
20344 9 captured_request_cookie -
20345 10 captured_response_cookie -
20346 11 termination_state ----
20347 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
20348 13 srv_queue '/' backend_queue 0/0
20349 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
20350 15 '{' captured_response_headers* '}' {}
20351 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010020352
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020353Detailed fields description :
20354 - "client_ip" is the IP address of the client which initiated the TCP
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020355 connection to HAProxy. If the connection was accepted on a UNIX socket
Willy Tarreauceb24bc2010-11-09 12:46:41 +010020356 instead, the IP address would be replaced with the word "unix". Note that
20357 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010020358 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010020359 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010020360 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020361
20362 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010020363 If the connection was accepted on a UNIX socket instead, the port would be
20364 replaced with the ID of the accepting socket, which is also reported in the
20365 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020366
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020367 - "request_date" is the exact date when the first byte of the HTTP request
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020368 was received by HAProxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020369
20370 - "frontend_name" is the name of the frontend (or listener) which received
20371 and processed the connection.
20372
20373 - "backend_name" is the name of the backend (or listener) which was selected
20374 to manage the connection to the server. This will be the same as the
20375 frontend if no switching rule has been applied.
20376
20377 - "server_name" is the name of the last server to which the connection was
20378 sent, which might differ from the first one if there were connection errors
20379 and a redispatch occurred. Note that this server belongs to the backend
20380 which processed the request. If the request was aborted before reaching a
20381 server, "<NOSRV>" is indicated instead of a server name. If the request was
20382 intercepted by the stats subsystem, "<STATS>" is indicated instead.
20383
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020384 - "TR" is the total time in milliseconds spent waiting for a full HTTP
20385 request from the client (not counting body) after the first byte was
20386 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050020387 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020388 always be very small because a request generally fits in one single packet.
20389 Large times here generally indicate network issues between the client and
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020390 HAProxy or requests being typed by hand. See section 8.4 "Timing Events"
Willy Tarreau590a0512018-09-05 11:56:48 +020020391 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020392
20393 - "Tw" is the total time in milliseconds spent waiting in the various queues.
20394 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020020395 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020396
20397 - "Tc" is the total time in milliseconds spent waiting for the connection to
20398 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020020399 request was aborted before a connection could be established. See section
20400 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020401
20402 - "Tr" is the total time in milliseconds spent waiting for the server to send
20403 a full HTTP response, not counting data. It can be "-1" if the request was
20404 aborted before a complete response could be received. It generally matches
20405 the server's processing time for the request, though it may be altered by
20406 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020020407 "GET" requests generally indicate an overloaded server. See section 8.4
20408 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020409
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020410 - "Ta" is the time the request remained active in HAProxy, which is the total
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020411 time in milliseconds elapsed between the first byte of the request was
20412 received and the last byte of response was sent. It covers all possible
20413 processing except the handshake (see Th) and idle time (see Ti). There is
20414 one exception, if "option logasap" was specified, then the time counting
20415 stops at the moment the log is emitted. In this case, a '+' sign is
20416 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020020417 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020418
20419 - "status_code" is the HTTP status code returned to the client. This status
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020420 is generally set by the server, but it might also be set by HAProxy when
20421 the server cannot be reached or when its response is blocked by HAProxy.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020422
20423 - "bytes_read" is the total number of bytes transmitted to the client when
20424 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050020425 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020426 the final one may be larger. Please note that this value is a 64-bit
20427 counter, so log analysis tools must be able to handle it without
20428 overflowing.
20429
20430 - "captured_request_cookie" is an optional "name=value" entry indicating that
20431 the client had this cookie in the request. The cookie name and its maximum
20432 length are defined by the "capture cookie" statement in the frontend
20433 configuration. The field is a single dash ('-') when the option is not
20434 set. Only one cookie may be captured, it is generally used to track session
20435 ID exchanges between a client and a server to detect session crossing
20436 between clients due to application bugs. For more details, please consult
20437 the section "Capturing HTTP headers and cookies" below.
20438
20439 - "captured_response_cookie" is an optional "name=value" entry indicating
20440 that the server has returned a cookie with its response. The cookie name
20441 and its maximum length are defined by the "capture cookie" statement in the
20442 frontend configuration. The field is a single dash ('-') when the option is
20443 not set. Only one cookie may be captured, it is generally used to track
20444 session ID exchanges between a client and a server to detect session
20445 crossing between clients due to application bugs. For more details, please
20446 consult the section "Capturing HTTP headers and cookies" below.
20447
20448 - "termination_state" is the condition the session was in when the session
20449 ended. This indicates the session state, which side caused the end of
20450 session to happen, for what reason (timeout, error, ...), just like in TCP
20451 logs, and information about persistence operations on cookies in the last
20452 two characters. The normal flags should begin with "--", indicating the
20453 session was closed by either end with no data remaining in buffers. See
20454 below "Session state at disconnection" for more details.
20455
20456 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020457 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020458 limits have been reached. For instance, if actconn is close to 512 or 1024
20459 when multiple connection errors occur, chances are high that the system
20460 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020461 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020462 system.
20463
20464 - "feconn" is the total number of concurrent connections on the frontend when
20465 the session was logged. It is useful to estimate the amount of resource
20466 required to sustain high loads, and to detect when the frontend's "maxconn"
20467 has been reached. Most often when this value increases by huge jumps, it is
20468 because there is congestion on the backend servers, but sometimes it can be
20469 caused by a denial of service attack.
20470
20471 - "beconn" is the total number of concurrent connections handled by the
20472 backend when the session was logged. It includes the total number of
20473 concurrent connections active on servers as well as the number of
20474 connections pending in queues. It is useful to estimate the amount of
20475 additional servers needed to support high loads for a given application.
20476 Most often when this value increases by huge jumps, it is because there is
20477 congestion on the backend servers, but sometimes it can be caused by a
20478 denial of service attack.
20479
20480 - "srv_conn" is the total number of concurrent connections still active on
20481 the server when the session was logged. It can never exceed the server's
20482 configured "maxconn" parameter. If this value is very often close or equal
20483 to the server's "maxconn", it means that traffic regulation is involved a
20484 lot, meaning that either the server's maxconn value is too low, or that
20485 there aren't enough servers to process the load with an optimal response
20486 time. When only one of the server's "srv_conn" is high, it usually means
20487 that this server has some trouble causing the requests to take longer to be
20488 processed than on other servers.
20489
20490 - "retries" is the number of connection retries experienced by this session
20491 when trying to connect to the server. It must normally be zero, unless a
20492 server is being stopped at the same moment the connection was attempted.
20493 Frequent retries generally indicate either a network problem between
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020494 HAProxy and the server, or a misconfigured system backlog on the server
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020495 preventing new connections from being queued. This field may optionally be
20496 prefixed with a '+' sign, indicating that the session has experienced a
20497 redispatch after the maximal retry count has been reached on the initial
20498 server. In this case, the server name appearing in the log is the one the
20499 connection was redispatched to, and not the first one, though both may
20500 sometimes be the same in case of hashing for instance. So as a general rule
20501 of thumb, when a '+' is present in front of the retry count, this count
20502 should not be attributed to the logged server.
20503
20504 - "srv_queue" is the total number of requests which were processed before
20505 this one in the server queue. It is zero when the request has not gone
20506 through the server queue. It makes it possible to estimate the approximate
20507 server's response time by dividing the time spent in queue by the number of
20508 requests in the queue. It is worth noting that if a session experiences a
20509 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010020510 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020511 backend queue unless a redispatch occurs.
20512
20513 - "backend_queue" is the total number of requests which were processed before
20514 this one in the backend's global queue. It is zero when the request has not
20515 gone through the global queue. It makes it possible to estimate the average
20516 queue length, which easily translates into a number of missing servers when
20517 divided by a server's "maxconn" parameter. It is worth noting that if a
20518 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010020519 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020520 through both the server queue and the backend queue unless a redispatch
20521 occurs.
20522
20523 - "captured_request_headers" is a list of headers captured in the request due
20524 to the presence of the "capture request header" statement in the frontend.
20525 Multiple headers can be captured, they will be delimited by a vertical bar
20526 ('|'). When no capture is enabled, the braces do not appear, causing a
20527 shift of remaining fields. It is important to note that this field may
20528 contain spaces, and that using it requires a smarter log parser than when
20529 it's not used. Please consult the section "Capturing HTTP headers and
20530 cookies" below for more details.
20531
20532 - "captured_response_headers" is a list of headers captured in the response
20533 due to the presence of the "capture response header" statement in the
20534 frontend. Multiple headers can be captured, they will be delimited by a
20535 vertical bar ('|'). When no capture is enabled, the braces do not appear,
20536 causing a shift of remaining fields. It is important to note that this
20537 field may contain spaces, and that using it requires a smarter log parser
20538 than when it's not used. Please consult the section "Capturing HTTP headers
20539 and cookies" below for more details.
20540
20541 - "http_request" is the complete HTTP request line, including the method,
20542 request and HTTP version string. Non-printable characters are encoded (see
20543 below the section "Non-printable characters"). This is always the last
20544 field, and it is always delimited by quotes and is the only one which can
20545 contain quotes. If new fields are added to the log format, they will be
20546 added before this field. This field might be truncated if the request is
20547 huge and does not fit in the standard syslog buffer (1024 characters). This
20548 is the reason why this field must always remain the last one.
20549
20550
Cyril Bontédc4d9032012-04-08 21:57:39 +0200205518.2.4. Custom log format
20552------------------------
William Lallemand48940402012-01-30 16:47:22 +010020553
Willy Tarreau2beef582012-12-20 17:22:52 +010020554The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010020555mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010020556
Davor Ocelice9ed2812017-12-25 17:49:28 +010020557HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010020558Variables can take arguments using braces ('{}'), and multiple arguments are
20559separated by commas within the braces. Flags may be added or removed by
20560prefixing them with a '+' or '-' sign.
20561
20562Special variable "%o" may be used to propagate its flags to all other
20563variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010020564("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010020565
Willy Tarreauc8368452012-12-21 00:09:23 +010020566If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020020567as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010020568less common information such as the client's SSL certificate's DN, or to log
20569the key that would be used to store an entry into a stick table.
20570
Dragan Dosen1e3b16f2020-06-23 18:16:44 +020020571Note: spaces must be escaped. In configuration directives "log-format",
20572"log-format-sd" and "unique-id-format", spaces are considered as
20573delimiters and are merged. In order to emit a verbatim '%', it must be
20574preceded by another '%' resulting in '%%'.
William Lallemand48940402012-01-30 16:47:22 +010020575
Dragan Dosen835b9212016-02-12 13:23:03 +010020576Note: when using the RFC5424 syslog message format, the characters '"',
20577'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
20578https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
20579such cases, the use of the flag "E" should be considered.
20580
William Lallemand48940402012-01-30 16:47:22 +010020581Flags are :
20582 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020583 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010020584 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
20585 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010020586
20587 Example:
20588
20589 log-format %T\ %t\ Some\ Text
20590 log-format %{+Q}o\ %t\ %s\ %{-Q}r
20591
Dragan Dosen835b9212016-02-12 13:23:03 +010020592 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
20593
William Lallemand48940402012-01-30 16:47:22 +010020594At the moment, the default HTTP format is defined this way :
20595
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020596 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
20597 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010020598
William Lallemandbddd4fd2012-02-27 11:23:10 +010020599the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010020600
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020601 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
20602 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
20603 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010020604
William Lallemandbddd4fd2012-02-27 11:23:10 +010020605and the default TCP format is defined this way :
20606
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020607 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
20608 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010020609
William Lallemand48940402012-01-30 16:47:22 +010020610Please refer to the table below for currently defined variables :
20611
William Lallemandbddd4fd2012-02-27 11:23:10 +010020612 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020613 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020614 +---+------+-----------------------------------------------+-------------+
20615 | | %o | special variable, apply flags on all next var | |
20616 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010020617 | | %B | bytes_read (from server to client) | numeric |
20618 | H | %CC | captured_request_cookie | string |
20619 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020020620 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000020621 | H | %HM | HTTP method (ex: POST) | string |
Maciej Zdeb21acc332020-11-26 10:45:52 +000020622 | H | %HP | HTTP request URI without query string | string |
Maciej Zdebfcdfd852020-11-30 18:27:47 +000020623 | H | %HPO | HTTP path only (without host nor query string)| string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000020624 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000020625 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
20626 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010020627 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020020628 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020020629 | | %T | gmt_date_time | date |
Christopher Faulet3f177162021-12-03 10:48:36 +010020630 | H | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020631 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020020632 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080020633 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020634 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
20635 | H | %Ti | idle time before the HTTP request | numeric |
20636 | H | %Tq | Th + Ti + TR | numeric |
20637 | H | %TR | time to receive the full request from 1st byte| numeric |
20638 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020020639 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020640 | | %Tt | Tt | numeric |
Damien Claisse57c8eb92020-04-28 12:09:19 +000020641 | | %Tu | Tu | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020642 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010020643 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020644 | | %ac | actconn | numeric |
20645 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020646 | | %bc | beconn (backend concurrent connections) | numeric |
20647 | | %bi | backend_source_ip (connecting address) | IP |
20648 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020649 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010020650 | | %ci | client_ip (accepted address) | IP |
20651 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020652 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020653 | | %fc | feconn (frontend concurrent connections) | numeric |
20654 | | %fi | frontend_ip (accepting address) | IP |
20655 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020020656 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020020657 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020020658 | | %hr | captured_request_headers default style | string |
20659 | | %hrl | captured_request_headers CLF style | string list |
20660 | | %hs | captured_response_headers default style | string |
20661 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020020662 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020020663 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020664 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020665 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010020666 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020667 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020668 | | %sc | srv_conn (server concurrent connections) | numeric |
20669 | | %si | server_IP (target address) | IP |
20670 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020671 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020672 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
20673 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020674 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020675 | H | %tr | date_time of HTTP request | date |
20676 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020020677 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020678 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020679 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020680 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010020681
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020682 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010020683
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010020684
206858.2.5. Error log format
20686-----------------------
20687
20688When an incoming connection fails due to an SSL handshake or an invalid PROXY
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020689protocol header, HAProxy will log the event using a shorter, fixed line format.
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010020690By default, logs are emitted at the LOG_INFO level, unless the option
20691"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010020692will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010020693logged if the "dontlognull" option is set.
20694
20695The format looks like this :
20696
20697 >>> Dec 3 18:27:14 localhost \
20698 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
20699 Connection error during SSL handshake
20700
20701 Field Format Extract from the example above
20702 1 process_name '[' pid ']:' haproxy[6103]:
20703 2 client_ip ':' client_port 127.0.0.1:56059
20704 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
20705 4 frontend_name "/" bind_name ":" frt/f1:
20706 5 message Connection error during SSL handshake
20707
20708These fields just provide minimal information to help debugging connection
20709failures.
20710
20711
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200207128.3. Advanced logging options
20713-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020714
20715Some advanced logging options are often looked for but are not easy to find out
20716just by looking at the various options. Here is an entry point for the few
20717options which can enable better logging. Please refer to the keywords reference
20718for more information about their usage.
20719
20720
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200207218.3.1. Disabling logging of external tests
20722------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020723
20724It is quite common to have some monitoring tools perform health checks on
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020725HAProxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020726commercial load-balancer, and sometimes it will simply be a more complete
20727monitoring system such as Nagios. When the tests are very frequent, users often
20728ask how to disable logging for those checks. There are three possibilities :
20729
20730 - if connections come from everywhere and are just TCP probes, it is often
20731 desired to simply disable logging of connections without data exchange, by
20732 setting "option dontlognull" in the frontend. It also disables logging of
20733 port scans, which may or may not be desired.
20734
Willy Tarreau9e9919d2020-10-14 15:55:23 +020020735 - it is possible to use the "http-request set-log-level silent" action using
20736 a variety of conditions (source networks, paths, user-agents, etc).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020737
20738 - if the tests are performed on a known URI, use "monitor-uri" to declare
20739 this URI as dedicated to monitoring. Any host sending this request will
20740 only get the result of a health-check, and the request will not be logged.
20741
20742
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200207438.3.2. Logging before waiting for the session to terminate
20744----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020745
20746The problem with logging at end of connection is that you have no clue about
20747what is happening during very long sessions, such as remote terminal sessions
20748or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010020749"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020750just before data transfer begins. This means that in case of TCP, it will still
20751log the connection status to the server, and in case of HTTP, it will log just
20752after processing the server headers. In this case, the number of bytes reported
20753is the number of header bytes sent to the client. In order to avoid confusion
20754with normal logs, the total time field and the number of bytes are prefixed
20755with a '+' sign which means that real numbers are certainly larger.
20756
20757
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200207588.3.3. Raising log level upon errors
20759------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020020760
20761Sometimes it is more convenient to separate normal traffic from errors logs,
20762for instance in order to ease error monitoring from log files. When the option
20763"log-separate-errors" is used, connections which experience errors, timeouts,
20764retries, redispatches or HTTP status codes 5xx will see their syslog level
20765raised from "info" to "err". This will help a syslog daemon store the log in
20766a separate file. It is very important to keep the errors in the normal traffic
20767file too, so that log ordering is not altered. You should also be careful if
20768you already have configured your syslog daemon to store all logs higher than
20769"notice" in an "admin" file, because the "err" level is higher than "notice".
20770
20771
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200207728.3.4. Disabling logging of successful connections
20773--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020020774
20775Although this may sound strange at first, some large sites have to deal with
20776multiple thousands of logs per second and are experiencing difficulties keeping
20777them intact for a long time or detecting errors within them. If the option
20778"dontlog-normal" is set on the frontend, all normal connections will not be
20779logged. In this regard, a normal connection is defined as one without any
20780error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
20781and a response with a status 5xx is not considered normal and will be logged
20782too. Of course, doing is is really discouraged as it will remove most of the
20783useful information from the logs. Do this only if you have no other
20784alternative.
20785
20786
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200207878.4. Timing events
20788------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020789
20790Timers provide a great help in troubleshooting network problems. All values are
20791reported in milliseconds (ms). These timers should be used in conjunction with
20792the session termination flags. In TCP mode with "option tcplog" set on the
20793frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020794mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
20795addition, three other measures are provided, "Th", "Ti", and "Tq".
20796
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010020797Timings events in HTTP mode:
20798
20799 first request 2nd request
20800 |<-------------------------------->|<-------------- ...
20801 t tr t tr ...
20802 ---|----|----|----|----|----|----|----|----|--
20803 : Th Ti TR Tw Tc Tr Td : Ti ...
20804 :<---- Tq ---->: :
20805 :<-------------- Tt -------------->:
Damien Claisse57c8eb92020-04-28 12:09:19 +000020806 :<-- -----Tu--------------->:
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010020807 :<--------- Ta --------->:
20808
20809Timings events in TCP mode:
20810
20811 TCP session
20812 |<----------------->|
20813 t t
20814 ---|----|----|----|----|---
20815 | Th Tw Tc Td |
20816 |<------ Tt ------->|
20817
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020818 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010020819 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020820 only happen once during the whole connection's lifetime. A large time here
20821 may indicate that the client only pre-established the connection without
20822 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010020823 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020020824 SSL handshake was very expensive to compute. Please note that this time is
20825 reported only before the first request, so it is safe to average it over
20826 all request to calculate the amortized value. The second and subsequent
20827 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020828
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020829 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
20830 counts between the end of the handshakes and the first byte of the HTTP
20831 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020020832 to count after the end of the transmission the previous response. When a
20833 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
20834 after the previous request. Some browsers pre-establish connections to a
20835 server in order to reduce the latency of a future request, and keep them
20836 pending until they need it. This delay will be reported as the idle time. A
20837 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020838
20839 - TR: total time to get the client request (HTTP mode only). It's the time
20840 elapsed between the first bytes received and the moment the proxy received
20841 the empty line marking the end of the HTTP headers. The value "-1"
20842 indicates that the end of headers has never been seen. This happens when
20843 the client closes prematurely or times out. This time is usually very short
20844 since most requests fit in a single packet. A large time may indicate a
20845 request typed by hand during a test.
20846
20847 - Tq: total time to get the client request from the accept date or since the
20848 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010020849 exactly equal to Th + Ti + TR unless any of them is -1, in which case it
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020850 returns -1 as well. This timer used to be very useful before the arrival of
20851 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
20852 it in favor of TR nowadays, as the idle time adds a lot of noise to the
20853 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020854
20855 - Tw: total time spent in the queues waiting for a connection slot. It
20856 accounts for backend queue as well as the server queues, and depends on the
20857 queue size, and the time needed for the server to complete previous
20858 requests. The value "-1" means that the request was killed before reaching
20859 the queue, which is generally what happens with invalid or denied requests.
20860
20861 - Tc: total time to establish the TCP connection to the server. It's the time
20862 elapsed between the moment the proxy sent the connection request, and the
20863 moment it was acknowledged by the server, or between the TCP SYN packet and
20864 the matching SYN/ACK packet in return. The value "-1" means that the
20865 connection never established.
20866
20867 - Tr: server response time (HTTP mode only). It's the time elapsed between
20868 the moment the TCP connection was established to the server and the moment
20869 the server sent its complete response headers. It purely shows its request
20870 processing time, without the network overhead due to the data transmission.
20871 It is worth noting that when the client has data to send to the server, for
20872 instance during a POST request, the time already runs, and this can distort
20873 apparent response time. For this reason, it's generally wise not to trust
20874 too much this field for POST requests initiated from clients behind an
20875 untrusted network. A value of "-1" here means that the last the response
20876 header (empty line) was never seen, most likely because the server timeout
20877 stroke before the server managed to process the request.
20878
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020879 - Ta: total active time for the HTTP request, between the moment the proxy
20880 received the first byte of the request header and the emission of the last
20881 byte of the response body. The exception is when the "logasap" option is
20882 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
20883 a '+' sign. From this field, we can deduce "Td", the data transmission time,
20884 by subtracting other timers when valid :
20885
20886 Td = Ta - (TR + Tw + Tc + Tr)
20887
20888 Timers with "-1" values have to be excluded from this equation. Note that
20889 "Ta" can never be negative.
20890
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020891 - Tt: total session duration time, between the moment the proxy accepted it
20892 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020893 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
20894 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020895 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020896
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020897 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020898
20899 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020900 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
20901 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020902
Damien Claisse57c8eb92020-04-28 12:09:19 +000020903 - Tu: total estimated time as seen from client, between the moment the proxy
20904 accepted it and the moment both ends were closed, without idle time.
20905 This is useful to roughly measure end-to-end time as a user would see it,
20906 without idle time pollution from keep-alive time between requests. This
20907 timer in only an estimation of time seen by user as it assumes network
20908 latency is the same in both directions. The exception is when the "logasap"
20909 option is specified. In this case, it only equals (Th+TR+Tw+Tc+Tr), and is
20910 prefixed with a '+' sign.
20911
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020912These timers provide precious indications on trouble causes. Since the TCP
20913protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
20914that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020915due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
20916"Tt" is close to a timeout value specified in the configuration, it often means
20917that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020918
20919Most common cases :
20920
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020921 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
20922 the client and the proxy. This is very rare on local networks but might
20923 happen when clients are on far remote networks and send large requests. It
20924 may happen that values larger than usual appear here without any network
20925 cause. Sometimes, during an attack or just after a resource starvation has
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020926 ended, HAProxy may accept thousands of connections in a few milliseconds.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020927 The time spent accepting these connections will inevitably slightly delay
20928 processing of other connections, and it can happen that request times in the
20929 order of a few tens of milliseconds are measured after a few thousands of
20930 new connections have been accepted at once. Using one of the keep-alive
20931 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020020932 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020933
20934 - If "Tc" is close to 3000, a packet has probably been lost between the
20935 server and the proxy during the server connection phase. This value should
20936 always be very low, such as 1 ms on local networks and less than a few tens
20937 of ms on remote networks.
20938
Willy Tarreau55165fe2009-05-10 12:02:55 +020020939 - If "Tr" is nearly always lower than 3000 except some rare values which seem
20940 to be the average majored by 3000, there are probably some packets lost
20941 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020942
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020943 - If "Ta" is large even for small byte counts, it generally is because
20944 neither the client nor the server decides to close the connection while
Daniel Corbett9f0843f2021-05-08 10:50:37 -040020945 HAProxy is running in tunnel mode and both have agreed on a keep-alive
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020946 connection mode. In order to solve this issue, it will be needed to specify
20947 one of the HTTP options to manipulate keep-alive or close options on either
20948 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
20949 important when connection regulation is used with the "maxconn" option on
20950 the servers, since no new connection will be sent to the server until
20951 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020952
20953Other noticeable HTTP log cases ('xx' means any value to be ignored) :
20954
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020955 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020956 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020957 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020958
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020959 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020960 or it aborted too early. Check the session termination flags
20961 then "timeout http-request" and "timeout client" settings.
20962
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020963 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020964 servers were out of order, because the request was invalid
20965 or forbidden by ACL rules. Check the session termination
20966 flags.
20967
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020968 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
20969 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020970 Check the session termination flags, then check the
20971 "timeout connect" setting. Note that the tarpit action might
20972 return similar-looking patterns, with "Tw" equal to the time
20973 the client connection was maintained open.
20974
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020975 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020976 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020977 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020978 termination flags, then check the "timeout server" setting.
20979
20980
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200209818.5. Session state at disconnection
20982-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020983
20984TCP and HTTP logs provide a session termination indicator in the
20985"termination_state" field, just before the number of active connections. It is
209862-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
20987each of which has a special meaning :
20988
20989 - On the first character, a code reporting the first event which caused the
20990 session to terminate :
20991
20992 C : the TCP session was unexpectedly aborted by the client.
20993
20994 S : the TCP session was unexpectedly aborted by the server, or the
20995 server explicitly refused it.
20996
20997 P : the session was prematurely aborted by the proxy, because of a
20998 connection limit enforcement, because a DENY filter was matched,
20999 because of a security check which detected and blocked a dangerous
21000 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010021001 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020021002
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021003 L : the session was locally processed by HAProxy and was not passed to
Willy Tarreau570f2212013-06-10 16:42:09 +020021004 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021005
21006 R : a resource on the proxy has been exhausted (memory, sockets, source
21007 ports, ...). Usually, this appears during the connection phase, and
21008 system logs should contain a copy of the precise error. If this
21009 happens, it must be considered as a very serious anomaly which
21010 should be fixed as soon as possible by any means.
21011
21012 I : an internal error was identified by the proxy during a self-check.
21013 This should NEVER happen, and you are encouraged to report any log
21014 containing this, because this would almost certainly be a bug. It
21015 would be wise to preventively restart the process after such an
21016 event too, in case it would be caused by memory corruption.
21017
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021018 D : the session was killed by HAProxy because the server was detected
Simon Horman752dc4a2011-06-21 14:34:59 +090021019 as down and was configured to kill all connections when going down.
21020
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021021 U : the session was killed by HAProxy on this backup server because an
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070021022 active server was detected as up and was configured to kill all
21023 backup connections when going up.
21024
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021025 K : the session was actively killed by an admin operating on HAProxy.
Willy Tarreaua2a64e92011-09-07 23:01:56 +020021026
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021027 c : the client-side timeout expired while waiting for the client to
21028 send or receive data.
21029
21030 s : the server-side timeout expired while waiting for the server to
21031 send or receive data.
21032
21033 - : normal session completion, both the client and the server closed
21034 with nothing left in the buffers.
21035
21036 - on the second character, the TCP or HTTP session state when it was closed :
21037
Willy Tarreauf7b30a92010-12-06 22:59:17 +010021038 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021039 (HTTP mode only). Nothing was sent to any server.
21040
21041 Q : the proxy was waiting in the QUEUE for a connection slot. This can
21042 only happen when servers have a 'maxconn' parameter set. It can
21043 also happen in the global queue after a redispatch consecutive to
21044 a failed attempt to connect to a dying server. If no redispatch is
21045 reported, then no connection attempt was made to any server.
21046
21047 C : the proxy was waiting for the CONNECTION to establish on the
21048 server. The server might at most have noticed a connection attempt.
21049
21050 H : the proxy was waiting for complete, valid response HEADERS from the
21051 server (HTTP only).
21052
21053 D : the session was in the DATA phase.
21054
21055 L : the proxy was still transmitting LAST data to the client while the
21056 server had already finished. This one is very rare as it can only
21057 happen when the client dies while receiving the last packets.
21058
21059 T : the request was tarpitted. It has been held open with the client
21060 during the whole "timeout tarpit" duration or until the client
21061 closed, both of which will be reported in the "Tw" timer.
21062
21063 - : normal session completion after end of data transfer.
21064
21065 - the third character tells whether the persistence cookie was provided by
21066 the client (only in HTTP mode) :
21067
21068 N : the client provided NO cookie. This is usually the case for new
21069 visitors, so counting the number of occurrences of this flag in the
21070 logs generally indicate a valid trend for the site frequentation.
21071
21072 I : the client provided an INVALID cookie matching no known server.
21073 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020021074 cookies between HTTP/HTTPS sites, persistence conditionally
21075 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021076
21077 D : the client provided a cookie designating a server which was DOWN,
21078 so either "option persist" was used and the client was sent to
21079 this server, or it was not set and the client was redispatched to
21080 another server.
21081
Willy Tarreau996a92c2010-10-13 19:30:47 +020021082 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021083 server.
21084
Willy Tarreau996a92c2010-10-13 19:30:47 +020021085 E : the client provided a valid cookie, but with a last date which was
21086 older than what is allowed by the "maxidle" cookie parameter, so
21087 the cookie is consider EXPIRED and is ignored. The request will be
21088 redispatched just as if there was no cookie.
21089
21090 O : the client provided a valid cookie, but with a first date which was
21091 older than what is allowed by the "maxlife" cookie parameter, so
21092 the cookie is consider too OLD and is ignored. The request will be
21093 redispatched just as if there was no cookie.
21094
Willy Tarreauc89ccb62012-04-05 21:18:22 +020021095 U : a cookie was present but was not used to select the server because
21096 some other server selection mechanism was used instead (typically a
21097 "use-server" rule).
21098
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021099 - : does not apply (no cookie set in configuration).
21100
21101 - the last character reports what operations were performed on the persistence
21102 cookie returned by the server (only in HTTP mode) :
21103
21104 N : NO cookie was provided by the server, and none was inserted either.
21105
21106 I : no cookie was provided by the server, and the proxy INSERTED one.
21107 Note that in "cookie insert" mode, if the server provides a cookie,
21108 it will still be overwritten and reported as "I" here.
21109
Willy Tarreau996a92c2010-10-13 19:30:47 +020021110 U : the proxy UPDATED the last date in the cookie that was presented by
21111 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030021112 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020021113 date indicated in the cookie. If any other change happens, such as
21114 a redispatch, then the cookie will be marked as inserted instead.
21115
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021116 P : a cookie was PROVIDED by the server and transmitted as-is.
21117
21118 R : the cookie provided by the server was REWRITTEN by the proxy, which
21119 happens in "cookie rewrite" or "cookie prefix" modes.
21120
21121 D : the cookie provided by the server was DELETED by the proxy.
21122
21123 - : does not apply (no cookie set in configuration).
21124
Willy Tarreau996a92c2010-10-13 19:30:47 +020021125The combination of the two first flags gives a lot of information about what
21126was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021127helpful to detect server saturation, network troubles, local system resource
21128starvation, attacks, etc...
21129
21130The most common termination flags combinations are indicated below. They are
21131alphabetically sorted, with the lowercase set just after the upper case for
21132easier finding and understanding.
21133
21134 Flags Reason
21135
21136 -- Normal termination.
21137
21138 CC The client aborted before the connection could be established to the
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021139 server. This can happen when HAProxy tries to connect to a recently
21140 dead (or unchecked) server, and the client aborts while HAProxy is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021141 waiting for the server to respond or for "timeout connect" to expire.
21142
21143 CD The client unexpectedly aborted during data transfer. This can be
21144 caused by a browser crash, by an intermediate equipment between the
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021145 client and HAProxy which decided to actively break the connection,
21146 by network routing issues between the client and HAProxy, or by a
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021147 keep-alive session between the server and the client terminated first
21148 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010021149
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021150 cD The client did not send nor acknowledge any data for as long as the
21151 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020021152 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021153
21154 CH The client aborted while waiting for the server to start responding.
21155 It might be the server taking too long to respond or the client
21156 clicking the 'Stop' button too fast.
21157
21158 cH The "timeout client" stroke while waiting for client data during a
21159 POST request. This is sometimes caused by too large TCP MSS values
21160 for PPPoE networks which cannot transport full-sized packets. It can
21161 also happen when client timeout is smaller than server timeout and
21162 the server takes too long to respond.
21163
21164 CQ The client aborted while its session was queued, waiting for a server
21165 with enough empty slots to accept it. It might be that either all the
21166 servers were saturated or that the assigned server was taking too
21167 long a time to respond.
21168
21169 CR The client aborted before sending a full HTTP request. Most likely
21170 the request was typed by hand using a telnet client, and aborted
21171 too early. The HTTP status code is likely a 400 here. Sometimes this
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021172 might also be caused by an IDS killing the connection between HAProxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020021173 and the client. "option http-ignore-probes" can be used to ignore
21174 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021175
21176 cR The "timeout http-request" stroke before the client sent a full HTTP
21177 request. This is sometimes caused by too large TCP MSS values on the
21178 client side for PPPoE networks which cannot transport full-sized
21179 packets, or by clients sending requests by hand and not typing fast
21180 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020021181 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020021182 some browsers started to implement a "pre-connect" feature consisting
21183 in speculatively connecting to some recently visited web sites just
21184 in case the user would like to visit them. This results in many
21185 connections being established to web sites, which end up in 408
21186 Request Timeout if the timeout strikes first, or 400 Bad Request when
21187 the browser decides to close them first. These ones pollute the log
21188 and feed the error counters. Some versions of some browsers have even
21189 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010021190 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020021191 http-ignore-probes" in the frontend, resulting in connections with
21192 zero data transfer to be totally ignored. This will definitely hide
21193 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021194
21195 CT The client aborted while its session was tarpitted. It is important to
21196 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020021197 wrong tarpit rules have been written. If a lot of them happen, it
21198 might make sense to lower the "timeout tarpit" value to something
21199 closer to the average reported "Tw" timer, in order not to consume
21200 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021201
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021202 LR The request was intercepted and locally handled by HAProxy. Generally
Willy Tarreau570f2212013-06-10 16:42:09 +020021203 it means that this was a redirect or a stats request.
21204
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021205 SC The server or an equipment between it and HAProxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021206 the TCP connection (the proxy received a TCP RST or an ICMP message
21207 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010021208 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021209 or no ARP response on local network). When this happens in HTTP mode,
21210 the status code is likely a 502 or 503 here.
21211
21212 sC The "timeout connect" stroke before a connection to the server could
21213 complete. When this happens in HTTP mode, the status code is likely a
21214 503 or 504 here.
21215
21216 SD The connection to the server died with an error during the data
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021217 transfer. This usually means that HAProxy has received an RST from
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021218 the server or an ICMP message from an intermediate equipment while
21219 exchanging data with the server. This can be caused by a server crash
21220 or by a network issue on an intermediate equipment.
21221
21222 sD The server did not send nor acknowledge any data for as long as the
21223 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010021224 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021225 load-balancers, ...), as well as keep-alive sessions maintained
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021226 between the client and the server expiring first on HAProxy.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021227
21228 SH The server aborted before sending its full HTTP response headers, or
21229 it crashed while processing the request. Since a server aborting at
21230 this moment is very rare, it would be wise to inspect its logs to
21231 control whether it crashed and why. The logged request may indicate a
21232 small set of faulty requests, demonstrating bugs in the application.
21233 Sometimes this might also be caused by an IDS killing the connection
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021234 between HAProxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021235
21236 sH The "timeout server" stroke before the server could return its
21237 response headers. This is the most common anomaly, indicating too
21238 long transactions, probably caused by server or database saturation.
21239 The immediate workaround consists in increasing the "timeout server"
21240 setting, but it is important to keep in mind that the user experience
21241 will suffer from these long response times. The only long term
21242 solution is to fix the application.
21243
21244 sQ The session spent too much time in queue and has been expired. See
21245 the "timeout queue" and "timeout connect" settings to find out how to
21246 fix this if it happens too often. If it often happens massively in
21247 short periods, it may indicate general problems on the affected
21248 servers due to I/O or database congestion, or saturation caused by
21249 external attacks.
21250
21251 PC The proxy refused to establish a connection to the server because the
Thayne McCombscdbcca92021-01-07 21:24:41 -070021252 process's socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020021253 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021254 so that it does not happen anymore. This status is very rare and
21255 might happen when the global "ulimit-n" parameter is forced by hand.
21256
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010021257 PD The proxy blocked an incorrectly formatted chunked encoded message in
21258 a request or a response, after the server has emitted its headers. In
21259 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010021260 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020021261 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010021262
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021263 PH The proxy blocked the server's response, because it was invalid,
21264 incomplete, dangerous (cache control), or matched a security filter.
21265 In any case, an HTTP 502 error is sent to the client. One possible
21266 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010021267 containing unauthorized characters. It is also possible but quite
21268 rare, that the proxy blocked a chunked-encoding request from the
21269 client due to an invalid syntax, before the server responded. In this
21270 case, an HTTP 400 error is sent to the client and reported in the
Christopher Fauletae72ae42022-05-05 12:27:07 +020021271 logs. Finally, it may be due to an HTTP header rewrite failure on the
21272 response. In this case, an HTTP 500 error is sent (see
21273 "tune.maxrewrite" and "http-response strict-mode" for more
21274 inforomation).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021275
21276 PR The proxy blocked the client's HTTP request, either because of an
21277 invalid HTTP syntax, in which case it returned an HTTP 400 error to
21278 the client, or because a deny filter matched, in which case it
Christopher Fauletae72ae42022-05-05 12:27:07 +020021279 returned an HTTP 403 error. It may also be due to an HTTP header
21280 rewrite failure on the request. In this case, an HTTP 500 error is
21281 sent (see "tune.maxrewrite" and "http-request strict-mode" for more
21282 inforomation).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021283
21284 PT The proxy blocked the client's request and has tarpitted its
21285 connection before returning it a 500 server error. Nothing was sent
21286 to the server. The connection was maintained open for as long as
21287 reported by the "Tw" timer field.
21288
21289 RC A local resource has been exhausted (memory, sockets, source ports)
21290 preventing the connection to the server from establishing. The error
21291 logs will tell precisely what was missing. This is very rare and can
21292 only be solved by proper system tuning.
21293
Willy Tarreau996a92c2010-10-13 19:30:47 +020021294The combination of the two last flags gives a lot of information about how
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021295persistence was handled by the client, the server and by HAProxy. This is very
Willy Tarreau996a92c2010-10-13 19:30:47 +020021296important to troubleshoot disconnections, when users complain they have to
21297re-authenticate. The commonly encountered flags are :
21298
21299 -- Persistence cookie is not enabled.
21300
21301 NN No cookie was provided by the client, none was inserted in the
21302 response. For instance, this can be in insert mode with "postonly"
21303 set on a GET request.
21304
21305 II A cookie designating an invalid server was provided by the client,
21306 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040021307 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020021308 value can be presented by a client when no other server knows it.
21309
21310 NI No cookie was provided by the client, one was inserted in the
21311 response. This typically happens for first requests from every user
21312 in "insert" mode, which makes it an easy way to count real users.
21313
21314 VN A cookie was provided by the client, none was inserted in the
21315 response. This happens for most responses for which the client has
21316 already got a cookie.
21317
21318 VU A cookie was provided by the client, with a last visit date which is
21319 not completely up-to-date, so an updated cookie was provided in
21320 response. This can also happen if there was no date at all, or if
21321 there was a date but the "maxidle" parameter was not set, so that the
21322 cookie can be switched to unlimited time.
21323
21324 EI A cookie was provided by the client, with a last visit date which is
21325 too old for the "maxidle" parameter, so the cookie was ignored and a
21326 new cookie was inserted in the response.
21327
21328 OI A cookie was provided by the client, with a first visit date which is
21329 too old for the "maxlife" parameter, so the cookie was ignored and a
21330 new cookie was inserted in the response.
21331
21332 DI The server designated by the cookie was down, a new server was
21333 selected and a new cookie was emitted in the response.
21334
21335 VI The server designated by the cookie was not marked dead but could not
21336 be reached. A redispatch happened and selected another one, which was
21337 then advertised in the response.
21338
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021339
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200213408.6. Non-printable characters
21341-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021342
21343In order not to cause trouble to log analysis tools or terminals during log
21344consulting, non-printable characters are not sent as-is into log files, but are
21345converted to the two-digits hexadecimal representation of their ASCII code,
21346prefixed by the character '#'. The only characters that can be logged without
21347being escaped are comprised between 32 and 126 (inclusive). Obviously, the
21348escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
21349is the same for the character '"' which becomes "#22", as well as '{', '|' and
21350'}' when logging headers.
21351
21352Note that the space character (' ') is not encoded in headers, which can cause
21353issues for tools relying on space count to locate fields. A typical header
21354containing spaces is "User-Agent".
21355
21356Last, it has been observed that some syslog daemons such as syslog-ng escape
21357the quote ('"') with a backslash ('\'). The reverse operation can safely be
21358performed since no quote may appear anywhere else in the logs.
21359
21360
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200213618.7. Capturing HTTP cookies
21362---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021363
21364Cookie capture simplifies the tracking a complete user session. This can be
21365achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021366section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021367cookie will simultaneously be checked in the request ("Cookie:" header) and in
21368the response ("Set-Cookie:" header). The respective values will be reported in
21369the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021370locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021371not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
21372user switches to a new session for example, because the server will reassign it
21373a new cookie. It is also possible to detect if a server unexpectedly sets a
21374wrong cookie to a client, leading to session crossing.
21375
21376 Examples :
21377 # capture the first cookie whose name starts with "ASPSESSION"
21378 capture cookie ASPSESSION len 32
21379
21380 # capture the first cookie whose name is exactly "vgnvisitor"
21381 capture cookie vgnvisitor= len 32
21382
21383
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200213848.8. Capturing HTTP headers
21385---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021386
21387Header captures are useful to track unique request identifiers set by an upper
21388proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
21389the response, one can search for information about the response length, how the
21390server asked the cache to behave, or an object location during a redirection.
21391
21392Header captures are performed using the "capture request header" and "capture
21393response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021394section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021395
21396It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010021397time. Non-existent headers are logged as empty strings, and if one header
21398appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021399are grouped within braces '{' and '}' in the same order as they were declared,
21400and delimited with a vertical bar '|' without any space. Response headers
21401follow the same representation, but are displayed after a space following the
21402request headers block. These blocks are displayed just before the HTTP request
21403in the logs.
21404
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020021405As a special case, it is possible to specify an HTTP header capture in a TCP
21406frontend. The purpose is to enable logging of headers which will be parsed in
21407an HTTP backend if the request is then switched to this HTTP backend.
21408
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021409 Example :
21410 # This instance chains to the outgoing proxy
21411 listen proxy-out
21412 mode http
21413 option httplog
21414 option logasap
21415 log global
21416 server cache1 192.168.1.1:3128
21417
21418 # log the name of the virtual server
21419 capture request header Host len 20
21420
21421 # log the amount of data uploaded during a POST
21422 capture request header Content-Length len 10
21423
21424 # log the beginning of the referrer
21425 capture request header Referer len 20
21426
21427 # server name (useful for outgoing proxies only)
21428 capture response header Server len 20
21429
21430 # logging the content-length is useful with "option logasap"
21431 capture response header Content-Length len 10
21432
Davor Ocelice9ed2812017-12-25 17:49:28 +010021433 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021434 capture response header Cache-Control len 8
21435
21436 # the Via header will report the next proxy's name
21437 capture response header Via len 20
21438
21439 # log the URL location during a redirection
21440 capture response header Location len 20
21441
21442 >>> Aug 9 20:26:09 localhost \
21443 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
21444 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
21445 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
21446 "GET http://fr.adserver.yahoo.com/"
21447
21448 >>> Aug 9 20:30:46 localhost \
21449 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
21450 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
21451 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010021452 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021453
21454 >>> Aug 9 20:30:46 localhost \
21455 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
21456 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
21457 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
21458 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010021459 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021460
21461
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200214628.9. Examples of logs
21463---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021464
21465These are real-world examples of logs accompanied with an explanation. Some of
21466them have been made up by hand. The syslog part has been removed for better
21467reading. Their sole purpose is to explain how to decipher them.
21468
21469 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
21470 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
21471 "HEAD / HTTP/1.0"
21472
21473 => long request (6.5s) entered by hand through 'telnet'. The server replied
21474 in 147 ms, and the session ended normally ('----')
21475
21476 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
21477 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
21478 0/9 "HEAD / HTTP/1.0"
21479
21480 => Idem, but the request was queued in the global queue behind 9 other
21481 requests, and waited there for 1230 ms.
21482
21483 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
21484 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
21485 "GET /image.iso HTTP/1.0"
21486
21487 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010021488 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021489 14 ms, 243 bytes of headers were sent to the client, and total time from
21490 accept to first data byte is 30 ms.
21491
21492 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
21493 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
21494 "GET /cgi-bin/bug.cgi? HTTP/1.0"
21495
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020021496 => the proxy blocked a server response either because of an "http-response
21497 deny" rule, or because the response was improperly formatted and not
21498 HTTP-compliant, or because it blocked sensitive information which risked
21499 being cached. In this case, the response is replaced with a "502 bad
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021500 gateway". The flags ("PH--") tell us that it was HAProxy who decided to
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020021501 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021502
21503 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010021504 px-http/<NOSRV> -1/-1/-1/-1/8490 -1 0 - - CR-- 2/2/2/0/0 0/0 ""
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021505
21506 => the client never completed its request and aborted itself ("C---") after
21507 8.5s, while the proxy was waiting for the request headers ("-R--").
21508 Nothing was sent to any server.
21509
21510 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
21511 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
21512
21513 => The client never completed its request, which was aborted by the
21514 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010021515 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021516 send a 408 return code to the client.
21517
21518 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
21519 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
21520
21521 => This log was produced with "option tcplog". The client timed out after
21522 5 seconds ("c----").
21523
21524 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
21525 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010021526 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021527
21528 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021529 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021530 (config says 'retries 3'), and no redispatch (otherwise we would have
21531 seen "/+3"). Status code 503 was returned to the client. There were 115
21532 connections on this server, 202 connections on this proxy, and 205 on
21533 the global process. It is possible that the server refused the
21534 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010021535
Willy Tarreau52b2d222011-09-07 23:48:48 +020021536
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200215379. Supported filters
21538--------------------
21539
21540Here are listed officially supported filters with the list of parameters they
21541accept. Depending on compile options, some of these filters might be
21542unavailable. The list of available filters is reported in haproxy -vv.
21543
21544See also : "filter"
21545
215469.1. Trace
21547----------
21548
Christopher Fauletc41d8bd2020-11-17 10:43:26 +010021549filter trace [name <name>] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021550
21551 Arguments:
21552 <name> is an arbitrary name that will be reported in
21553 messages. If no name is provided, "TRACE" is used.
21554
Christopher Faulet96a577a2020-11-17 10:45:05 +010021555 <quiet> inhibits trace messages.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021556
Davor Ocelice9ed2812017-12-25 17:49:28 +010021557 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021558 default, this filter forwards all previously parsed
21559 data. With this parameter, it only forwards a random
21560 amount of the parsed data.
21561
Davor Ocelice9ed2812017-12-25 17:49:28 +010021562 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010021563
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021564This filter can be used as a base to develop new filters. It defines all
21565callbacks and print a message on the standard error stream (stderr) with useful
21566information for all of them. It may be useful to debug the activity of other
21567filters or, quite simply, HAProxy's activity.
21568
21569Using <random-parsing> and/or <random-forwarding> parameters is a good way to
21570tests the behavior of a filter that parses data exchanged between a client and
21571a server by adding some latencies in the processing.
21572
21573
215749.2. HTTP compression
21575---------------------
21576
21577filter compression
21578
21579The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
21580keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020021581when no other filter is used, it is enough. When used with the cache or the
21582fcgi-app enabled, it is also enough. In this case, the compression is always
21583done after the response is stored in the cache. But it is mandatory to
21584explicitly use a filter line to enable the HTTP compression when at least one
21585filter other than the cache or the fcgi-app is used for the same
21586listener/frontend/backend. This is important to know the filters evaluation
21587order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021588
Christopher Fauletb30b3102019-09-12 23:03:09 +020021589See also : "compression", section 9.4 about the cache filter and section 9.5
21590 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021591
21592
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200215939.3. Stream Processing Offload Engine (SPOE)
21594--------------------------------------------
21595
21596filter spoe [engine <name>] config <file>
21597
21598 Arguments :
21599
21600 <name> is the engine name that will be used to find the right scope in
21601 the configuration file. If not provided, all the file will be
21602 parsed.
21603
21604 <file> is the path of the engine configuration file. This file can
21605 contain configuration of several engines. In this case, each
21606 part must be placed in its own scope.
21607
21608The Stream Processing Offload Engine (SPOE) is a filter communicating with
21609external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010021610streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020021611exchanged with them are configured in dedicated files, for the main part. It
21612also requires dedicated backends, defined in HAProxy configuration.
21613
21614SPOE communicates with external components using an in-house binary protocol,
21615the Stream Processing Offload Protocol (SPOP).
21616
Tim Düsterhus4896c442016-11-29 02:15:19 +010021617For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020021618"doc/SPOE.txt".
21619
Christopher Faulet99a17a22018-12-11 09:18:27 +0100216209.4. Cache
21621----------
21622
21623filter cache <name>
21624
21625 Arguments :
21626
21627 <name> is name of the cache section this filter will use.
21628
21629The cache uses a filter to store cacheable responses. The HTTP rules
21630"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050021631cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020021632other filters than fcgi-app or compression are used, it is enough. In such
21633case, the compression filter is always evaluated after the cache filter. But it
21634is mandatory to explicitly use a filter line to use a cache when at least one
21635filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010021636listener/frontend/backend. This is important to know the filters evaluation
21637order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010021638
Christopher Fauletb30b3102019-09-12 23:03:09 +020021639See also : section 9.2 about the compression filter, section 9.5 about the
21640 fcgi-app filter and section 6 about cache.
21641
21642
216439.5. Fcgi-app
21644-------------
21645
Daniel Corbett67a82712020-07-06 23:01:19 -040021646filter fcgi-app <name>
Christopher Fauletb30b3102019-09-12 23:03:09 +020021647
21648 Arguments :
21649
21650 <name> is name of the fcgi-app section this filter will use.
21651
21652The FastCGI application uses a filter to evaluate all custom parameters on the
21653request path, and to process the headers on the response path. the <name> must
21654reference an existing fcgi-app section. The directive "use-fcgi-app" should be
21655used to define the application to use. By default the corresponding filter is
21656implicitly defined. And when no other filters than cache or compression are
21657used, it is enough. But it is mandatory to explicitly use a filter line to a
21658fcgi-app when at least one filter other than the compression or the cache is
21659used for the same backend. This is important to know the filters evaluation
21660order.
21661
21662See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
21663 about the cache filter and section 10 about FastCGI application.
21664
21665
Miroslav Zagoracdc32cd92020-12-13 18:32:57 +0100216669.6. OpenTracing
21667----------------
21668
21669The OpenTracing filter adds native support for using distributed tracing in
21670HAProxy. This is enabled by sending an OpenTracing compliant request to one
21671of the supported tracers such as Datadog, Jaeger, Lightstep and Zipkin tracers.
21672Please note: tracers are not listed by any preference, but alphabetically.
21673
Daniel Corbett9f0843f2021-05-08 10:50:37 -040021674This feature is only enabled when HAProxy was built with USE_OT=1.
Miroslav Zagoracdc32cd92020-12-13 18:32:57 +010021675
21676The OpenTracing filter activation is done explicitly by specifying it in the
21677HAProxy configuration. If this is not done, the OpenTracing filter in no way
21678participates in the work of HAProxy.
21679
21680filter opentracing [id <id>] config <file>
21681
21682 Arguments :
21683
21684 <id> is the OpenTracing filter id that will be used to find the
21685 right scope in the configuration file. If no filter id is
21686 specified, 'ot-filter' is used as default. If scope is not
21687 specified in the configuration file, it applies to all defined
21688 OpenTracing filters.
21689
21690 <file> is the path of the OpenTracing configuration file. The same
21691 file can contain configurations for multiple OpenTracing
21692 filters simultaneously. In that case we do not need to define
21693 scope so the same configuration applies to all filters or each
21694 filter must have its own scope defined.
21695
21696More detailed documentation related to the operation, configuration and use
Willy Tarreaua63d1a02021-04-02 17:16:46 +020021697of the filter can be found in the addons/ot directory.
Miroslav Zagoracdc32cd92020-12-13 18:32:57 +010021698
21699
Christopher Fauletb30b3102019-09-12 23:03:09 +02002170010. FastCGI applications
21701-------------------------
21702
21703HAProxy is able to send HTTP requests to Responder FastCGI applications. This
21704feature was added in HAProxy 2.1. To do so, servers must be configured to use
21705the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
21706FastCGI application must be configured and used by the backend managing these
21707servers (using the keyword "use-fcgi-app" into the proxy section). Several
21708FastCGI applications may be defined, but only one can be used at a time by a
21709backend.
21710
21711HAProxy implements all features of the FastCGI specification for Responder
21712application. Especially it is able to multiplex several requests on a simple
21713connection.
21714
2171510.1. Setup
21716-----------
21717
2171810.1.1. Fcgi-app section
21719--------------------------
21720
21721fcgi-app <name>
21722 Declare a FastCGI application named <name>. To be valid, at least the
21723 document root must be defined.
21724
21725acl <aclname> <criterion> [flags] [operator] <value> ...
21726 Declare or complete an access list.
21727
21728 See "acl" keyword in section 4.2 and section 7 about ACL usage for
21729 details. ACLs defined for a FastCGI application are private. They cannot be
21730 used by any other application or by any proxy. In the same way, ACLs defined
21731 in any other section are not usable by a FastCGI application. However,
21732 Pre-defined ACLs are available.
21733
21734docroot <path>
21735 Define the document root on the remote host. <path> will be used to build
21736 the default value of FastCGI parameters SCRIPT_FILENAME and
21737 PATH_TRANSLATED. It is a mandatory setting.
21738
21739index <script-name>
21740 Define the script name that will be appended after an URI that ends with a
21741 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
21742 is an optional setting.
21743
21744 Example :
21745 index index.php
21746
21747log-stderr global
21748log-stderr <address> [len <length>] [format <format>]
Jan Wagner3e678602020-12-17 22:22:32 +010021749 [sample <ranges>:<sample_size>] <facility> [<level> [<minlevel>]]
Christopher Fauletb30b3102019-09-12 23:03:09 +020021750 Enable logging of STDERR messages reported by the FastCGI application.
21751
21752 See "log" keyword in section 4.2 for details. It is an optional setting. By
21753 default STDERR messages are ignored.
21754
21755pass-header <name> [ { if | unless } <condition> ]
21756 Specify the name of a request header which will be passed to the FastCGI
21757 application. It may optionally be followed by an ACL-based condition, in
21758 which case it will only be evaluated if the condition is true.
21759
21760 Most request headers are already available to the FastCGI application,
21761 prefixed with "HTTP_". Thus, this directive is only required to pass headers
21762 that are purposefully omitted. Currently, the headers "Authorization",
21763 "Proxy-Authorization" and hop-by-hop headers are omitted.
21764
21765 Note that the headers "Content-type" and "Content-length" are never passed to
21766 the FastCGI application because they are already converted into parameters.
21767
21768path-info <regex>
Christopher Faulet28cb3662020-02-14 14:47:37 +010021769 Define a regular expression to extract the script-name and the path-info from
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010021770 the URL-decoded path. Thus, <regex> may have two captures: the first one to
21771 capture the script name and the second one to capture the path-info. The
21772 first one is mandatory, the second one is optional. This way, it is possible
21773 to extract the script-name from the path ignoring the path-info. It is an
21774 optional setting. If it is not defined, no matching is performed on the
21775 path. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not
21776 filled.
Christopher Faulet28cb3662020-02-14 14:47:37 +010021777
21778 For security reason, when this regular expression is defined, the newline and
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050021779 the null characters are forbidden from the path, once URL-decoded. The reason
Christopher Faulet28cb3662020-02-14 14:47:37 +010021780 to such limitation is because otherwise the matching always fails (due to a
21781 limitation one the way regular expression are executed in HAProxy). So if one
21782 of these two characters is found in the URL-decoded path, an error is
21783 returned to the client. The principle of least astonishment is applied here.
Christopher Fauletb30b3102019-09-12 23:03:09 +020021784
21785 Example :
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010021786 path-info ^(/.+\.php)(/.*)?$ # both script-name and path-info may be set
21787 path-info ^(/.+\.php) # the path-info is ignored
Christopher Fauletb30b3102019-09-12 23:03:09 +020021788
21789option get-values
21790no option get-values
21791 Enable or disable the retrieve of variables about connection management.
21792
Daniel Corbett67a82712020-07-06 23:01:19 -040021793 HAProxy is able to send the record FCGI_GET_VALUES on connection
Christopher Fauletb30b3102019-09-12 23:03:09 +020021794 establishment to retrieve the value for following variables:
21795
21796 * FCGI_MAX_REQS The maximum number of concurrent requests this
21797 application will accept.
21798
William Lallemand93e548e2019-09-30 13:54:02 +020021799 * FCGI_MPXS_CONNS "0" if this application does not multiplex connections,
21800 "1" otherwise.
Christopher Fauletb30b3102019-09-12 23:03:09 +020021801
21802 Some FastCGI applications does not support this feature. Some others close
Ilya Shipitsin11057a32020-06-21 21:18:27 +050021803 the connection immediately after sending their response. So, by default, this
Christopher Fauletb30b3102019-09-12 23:03:09 +020021804 option is disabled.
21805
21806 Note that the maximum number of concurrent requests accepted by a FastCGI
21807 application is a connection variable. It only limits the number of streams
21808 per connection. If the global load must be limited on the application, the
21809 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
21810 an application does not support connection multiplexing, the maximum number
21811 of concurrent requests is automatically set to 1.
21812
21813option keep-conn
21814no option keep-conn
21815 Instruct the FastCGI application to keep the connection open or not after
21816 sending a response.
21817
21818 If disabled, the FastCGI application closes the connection after responding
21819 to this request. By default, this option is enabled.
21820
21821option max-reqs <reqs>
21822 Define the maximum number of concurrent requests this application will
21823 accept.
21824
21825 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
21826 during connection establishment. Furthermore, if the application does not
21827 support connection multiplexing, this option will be ignored. By default set
21828 to 1.
21829
21830option mpxs-conns
21831no option mpxs-conns
21832 Enable or disable the support of connection multiplexing.
21833
21834 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
21835 during connection establishment. It is disabled by default.
21836
21837set-param <name> <fmt> [ { if | unless } <condition> ]
21838 Set a FastCGI parameter that should be passed to this application. Its
21839 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
21840 "Custom Log format"). It may optionally be followed by an ACL-based
21841 condition, in which case it will only be evaluated if the condition is true.
21842
21843 With this directive, it is possible to overwrite the value of default FastCGI
21844 parameters. If the value is evaluated to an empty string, the rule is
21845 ignored. These directives are evaluated in their declaration order.
21846
21847 Example :
21848 # PHP only, required if PHP was built with --enable-force-cgi-redirect
21849 set-param REDIRECT_STATUS 200
21850
21851 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
21852
21853
2185410.1.2. Proxy section
21855---------------------
21856
21857use-fcgi-app <name>
21858 Define the FastCGI application to use for the backend.
21859
21860 Arguments :
21861 <name> is the name of the FastCGI application to use.
21862
21863 This keyword is only available for HTTP proxies with the backend capability
21864 and with at least one FastCGI server. However, FastCGI servers can be mixed
21865 with HTTP servers. But except there is a good reason to do so, it is not
21866 recommended (see section 10.3 about the limitations for details). Only one
21867 application may be defined at a time per backend.
21868
21869 Note that, once a FastCGI application is referenced for a backend, depending
21870 on the configuration some processing may be done even if the request is not
21871 sent to a FastCGI server. Rules to set parameters or pass headers to an
21872 application are evaluated.
21873
21874
2187510.1.3. Example
21876---------------
21877
21878 frontend front-http
21879 mode http
21880 bind *:80
21881 bind *:
21882
21883 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
21884 default_backend back-static
21885
21886 backend back-static
21887 mode http
21888 server www A.B.C.D:80
21889
21890 backend back-dynamic
21891 mode http
21892 use-fcgi-app php-fpm
21893 server php-fpm A.B.C.D:9000 proto fcgi
21894
21895 fcgi-app php-fpm
21896 log-stderr global
21897 option keep-conn
21898
21899 docroot /var/www/my-app
21900 index index.php
21901 path-info ^(/.+\.php)(/.*)?$
21902
21903
2190410.2. Default parameters
21905------------------------
21906
21907A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
21908the CGI/1.1 specification (RFC3875), several variables must be passed to the
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050021909script. So HAProxy set them and some others commonly used by FastCGI
Christopher Fauletb30b3102019-09-12 23:03:09 +020021910applications. All these variables may be overwritten, with caution though.
21911
21912 +-------------------+-----------------------------------------------------+
21913 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
21914 | | to authenticate the user. Concretely, only the |
21915 | | BASIC authentication mechanism is supported. |
21916 | | |
21917 +-------------------+-----------------------------------------------------+
21918 | CONTENT_LENGTH | Contains the size of the message-body attached to |
21919 | | the request. It means only requests with a known |
21920 | | size are considered as valid and sent to the |
21921 | | application. |
21922 | | |
21923 +-------------------+-----------------------------------------------------+
21924 | CONTENT_TYPE | Contains the type of the message-body attached to |
21925 | | the request. It may not be set. |
21926 | | |
21927 +-------------------+-----------------------------------------------------+
21928 | DOCUMENT_ROOT | Contains the document root on the remote host under |
21929 | | which the script should be executed, as defined in |
21930 | | the application's configuration. |
21931 | | |
21932 +-------------------+-----------------------------------------------------+
21933 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
21934 | | to communicate with the FastCGI application. |
21935 | | Concretely, it is set to "CGI/1.1". |
21936 | | |
21937 +-------------------+-----------------------------------------------------+
21938 | PATH_INFO | Contains the portion of the URI path hierarchy |
21939 | | following the part that identifies the script |
21940 | | itself. To be set, the directive "path-info" must |
21941 | | be defined. |
21942 | | |
21943 +-------------------+-----------------------------------------------------+
21944 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
21945 | | It is the concatenation of DOCUMENT_ROOT and |
21946 | | PATH_INFO. If PATH_INFO is not set, this parameters |
21947 | | is not set too. |
21948 | | |
21949 +-------------------+-----------------------------------------------------+
21950 | QUERY_STRING | Contains the request's query string. It may not be |
21951 | | set. |
21952 | | |
21953 +-------------------+-----------------------------------------------------+
21954 | REMOTE_ADDR | Contains the network address of the client sending |
21955 | | the request. |
21956 | | |
21957 +-------------------+-----------------------------------------------------+
21958 | REMOTE_USER | Contains the user identification string supplied by |
21959 | | client as part of user authentication. |
21960 | | |
21961 +-------------------+-----------------------------------------------------+
21962 | REQUEST_METHOD | Contains the method which should be used by the |
21963 | | script to process the request. |
21964 | | |
21965 +-------------------+-----------------------------------------------------+
21966 | REQUEST_URI | Contains the request's URI. |
21967 | | |
21968 +-------------------+-----------------------------------------------------+
21969 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
21970 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
21971 | | |
21972 +-------------------+-----------------------------------------------------+
21973 | SCRIPT_NAME | Contains the name of the script. If the directive |
21974 | | "path-info" is defined, it is the first part of the |
21975 | | URI path hierarchy, ending with the script name. |
21976 | | Otherwise, it is the entire URI path. |
21977 | | |
21978 +-------------------+-----------------------------------------------------+
21979 | SERVER_NAME | Contains the name of the server host to which the |
21980 | | client request is directed. It is the value of the |
21981 | | header "Host", if defined. Otherwise, the |
21982 | | destination address of the connection on the client |
21983 | | side. |
21984 | | |
21985 +-------------------+-----------------------------------------------------+
21986 | SERVER_PORT | Contains the destination TCP port of the connection |
21987 | | on the client side, which is the port the client |
21988 | | connected to. |
21989 | | |
21990 +-------------------+-----------------------------------------------------+
21991 | SERVER_PROTOCOL | Contains the request's protocol. |
21992 | | |
21993 +-------------------+-----------------------------------------------------+
Christopher Fauletb2a50292021-06-11 13:34:42 +020021994 | SERVER_SOFTWARE | Contains the string "HAProxy" followed by the |
21995 | | current HAProxy version. |
21996 | | |
21997 +-------------------+-----------------------------------------------------+
Christopher Fauletb30b3102019-09-12 23:03:09 +020021998 | HTTPS | Set to a non-empty value ("on") if the script was |
21999 | | queried through the HTTPS protocol. |
22000 | | |
22001 +-------------------+-----------------------------------------------------+
22002
22003
2200410.3. Limitations
22005------------------
22006
22007The current implementation have some limitations. The first one is about the
22008way some request headers are hidden to the FastCGI applications. This happens
22009during the headers analysis, on the backend side, before the connection
22010establishment. At this stage, HAProxy know the backend is using a FastCGI
22011application but it don't know if the request will be routed to a FastCGI server
22012or not. But to hide request headers, it simply removes them from the HTX
22013message. So, if the request is finally routed to an HTTP server, it never see
22014these headers. For this reason, it is not recommended to mix FastCGI servers
22015and HTTP servers under the same backend.
22016
22017Similarly, the rules "set-param" and "pass-header" are evaluated during the
22018request headers analysis. So the evaluation is always performed, even if the
22019requests is finally forwarded to an HTTP server.
22020
22021About the rules "set-param", when a rule is applied, a pseudo header is added
22022into the HTX message. So, the same way than for HTTP header rewrites, it may
22023fail if the buffer is full. The rules "set-param" will compete with
22024"http-request" ones.
22025
22026Finally, all FastCGI params and HTTP headers are sent into a unique record
22027FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
22028processing error is returned. It means the record FCGI_PARAM, once encoded,
22029must not exceeds the size of a buffer. However, there is no reserve to respect
22030here.
William Lallemand86d0df02017-11-24 21:36:45 +010022031
Emeric Brunce325c42021-04-02 17:05:09 +020022032
2203311. Address formats
22034-------------------
22035
22036Several statements as "bind, "server", "nameserver" and "log" requires an
22037address.
22038
22039This address can be a host name, an IPv4 address, an IPv6 address, or '*'.
22040The '*' is equal to the special address "0.0.0.0" and can be used, in the case
22041of "bind" or "dgram-bind" to listen on all IPv4 of the system.The IPv6
22042equivalent is '::'.
22043
22044Depending of the statement, a port or port range follows the IP address. This
22045is mandatory on 'bind' statement, optional on 'server'.
22046
22047This address can also begin with a slash '/'. It is considered as the "unix"
22048family, and '/' and following characters must be present the path.
22049
22050Default socket type or transport method "datagram" or "stream" depends on the
22051configuration statement showing the address. Indeed, 'bind' and 'server' will
22052use a "stream" socket type by default whereas 'log', 'nameserver' or
22053'dgram-bind' will use a "datagram".
22054
22055Optionally, a prefix could be used to force the address family and/or the
22056socket type and the transport method.
22057
22058
2205911.1 Address family prefixes
22060----------------------------
22061
22062'abns@<name>' following <name> is an abstract namespace (Linux only).
22063
22064'fd@<n>' following address is a file descriptor <n> inherited from the
22065 parent. The fd must be bound and may or may not already be
22066 listening.
22067
22068'ip@<address>[:port1[-port2]]' following <address> is considered as an IPv4 or
22069 IPv6 address depending on the syntax. Depending
22070 on the statement using this address, a port or
22071 a port range may or must be specified.
22072
22073'ipv4@<address>[:port1[-port2]]' following <address> is always considered as
22074 an IPv4 address. Depending on the statement
22075 using this address, a port or a port range
22076 may or must be specified.
22077
22078'ipv6@<address>[:port1[-port2]]' following <address> is always considered as
22079 an IPv6 address. Depending on the statement
22080 using this address, a port or a port range
22081 may or must be specified.
22082
22083'sockpair@<n>' following address is the file descriptor of a connected unix
22084 socket or of a socketpair. During a connection, the initiator
22085 creates a pair of connected sockets, and passes one of them
22086 over the FD to the other end. The listener waits to receive
22087 the FD from the unix socket and uses it as if it were the FD
22088 of an accept(). Should be used carefully.
22089
22090'unix@<path>' following string is considered as a UNIX socket <path>. this
22091 prefix is useful to declare an UNIX socket path which don't
22092 start by slash '/'.
22093
22094
2209511.2 Socket type prefixes
22096-------------------------
22097
22098Previous "Address family prefixes" can also be prefixed to force the socket
22099type and the transport method. The default depends of the statement using
22100this address but in some cases the user may force it to a different one.
22101This is the case for "log" statement where the default is syslog over UDP
22102but we could force to use syslog over TCP.
22103
22104Those prefixes were designed for internal purpose and users should
22105instead use aliases of the next section "11.5.3 Protocol prefixes".
22106
22107If users need one those prefixes to perform what they expect because
22108they can not configure the same using the protocol prefixes, they should
22109report this to the maintainers.
22110
22111'stream+<family>@<address>' forces socket type and transport method
22112 to "stream"
22113
22114'dgram+<family>@<address>' forces socket type and transport method
22115 to "datagram".
22116
22117
2211811.3 Protocol prefixes
22119----------------------
22120
22121'tcp@<address>[:port1[-port2]]' following <address> is considered as an IPv4
22122 or IPv6 address depending of the syntax but
22123 socket type and transport method is forced to
22124 "stream". Depending on the statement using
22125 this address, a port or a port range can or
22126 must be specified. It is considered as an alias
22127 of 'stream+ip@'.
22128
22129'tcp4@<address>[:port1[-port2]]' following <address> is always considered as
22130 an IPv4 address but socket type and transport
22131 method is forced to "stream". Depending on the
22132 statement using this address, a port or port
22133 range can or must be specified.
22134 It is considered as an alias of 'stream+ipv4@'.
22135
22136'tcp6@<address>[:port1[-port2]]' following <address> is always considered as
22137 an IPv6 address but socket type and transport
22138 method is forced to "stream". Depending on the
22139 statement using this address, a port or port
22140 range can or must be specified.
22141 It is considered as an alias of 'stream+ipv4@'.
22142
22143'udp@<address>[:port1[-port2]]' following <address> is considered as an IPv4
22144 or IPv6 address depending of the syntax but
22145 socket type and transport method is forced to
22146 "datagram". Depending on the statement using
22147 this address, a port or a port range can or
22148 must be specified. It is considered as an alias
22149 of 'dgram+ip@'.
22150
22151'udp4@<address>[:port1[-port2]]' following <address> is always considered as
22152 an IPv4 address but socket type and transport
22153 method is forced to "datagram". Depending on
22154 the statement using this address, a port or
22155 port range can or must be specified.
22156 It is considered as an alias of 'stream+ipv4@'.
22157
22158'udp6@<address>[:port1[-port2]]' following <address> is always considered as
22159 an IPv6 address but socket type and transport
22160 method is forced to "datagram". Depending on
22161 the statement using this address, a port or
22162 port range can or must be specified.
22163 It is considered as an alias of 'stream+ipv4@'.
22164
22165'uxdg@<path>' following string is considered as a unix socket <path> but
22166 transport method is forced to "datagram". It is considered as
22167 an alias of 'dgram+unix@'.
22168
22169'uxst@<path>' following string is considered as a unix socket <path> but
22170 transport method is forced to "stream". It is considered as
22171 an alias of 'stream+unix@'.
22172
22173In future versions, other prefixes could be used to specify protocols like
22174QUIC which proposes stream transport based on socket of type "datagram".
22175
Willy Tarreau0ba27502007-12-24 16:55:16 +010022176/*
22177 * Local variables:
22178 * fill-column: 79
22179 * End:
22180 */