blob: ce96c44e74044fc18320ebc047eccec44a45fc05 [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
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
Willy Tarreau86512dd2021-04-09 17:10:39 +02007 2021/04/09
Willy Tarreau6a06a402007-07-15 20:15:28 +02008
9
10This document covers the configuration language as implemented in the version
Davor Ocelice9ed2812017-12-25 17:49:28 +010011specified above. It does not provide any hints, examples, or advice. For such
Willy Tarreau0ba27502007-12-24 16:55:16 +010012documentation, please refer to the Reference Manual or the Architecture Manual.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013The summary below is meant to help you find sections by name and navigate
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014through the document.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016Note to documentation contributors :
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017 This document is formatted with 80 columns per line, with even number of
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018 spaces for indentation and without tabs. Please follow these rules strictly
19 so that it remains easily printable everywhere. If a line needs to be
20 printed verbatim and does not fit, please end each line with a backslash
Willy Tarreau62a36c42010-08-17 15:53:10 +020021 ('\') and continue on next line, indented by two characters. It is also
Davor Ocelice9ed2812017-12-25 17:49:28 +010022 sometimes useful to prefix all output lines (logs, console outputs) with 3
23 closing angle brackets ('>>>') in order to emphasize the difference between
24 inputs and outputs when they may be ambiguous. If you add sections,
Willy Tarreau62a36c42010-08-17 15:53:10 +020025 please update the summary below for easier searching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020026
27
28Summary
29-------
30
311. Quick reminder about HTTP
321.1. The HTTP transaction model
331.2. HTTP request
Davor Ocelice9ed2812017-12-25 17:49:28 +0100341.2.1. The request line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200351.2.2. The request headers
361.3. HTTP response
Davor Ocelice9ed2812017-12-25 17:49:28 +0100371.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200381.3.2. The response headers
39
402. Configuring HAProxy
412.1. Configuration file format
William Lallemandf9873ba2015-05-05 17:37:14 +0200422.2. Quoting and escaping
William Lallemandb2f07452015-05-12 14:27:13 +0200432.3. Environment variables
Willy Tarreau4b103022021-02-12 17:59:10 +0100442.4. Conditional blocks
452.5. Time format
462.6. Examples
Willy Tarreauc57f0e22009-05-10 13:12:33 +020047
483. Global parameters
493.1. Process management and security
503.2. Performance tuning
513.3. Debugging
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +0100523.4. Userlists
Cyril Bontédc4d9032012-04-08 21:57:39 +0200533.5. Peers
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200543.6. Mailers
William Lallemandc9515522019-06-12 16:32:11 +0200553.7. Programs
Christopher Faulet76edc0f2020-01-13 15:52:01 +0100563.8. HTTP-errors
Emeric Brun99c453d2020-05-25 15:01:04 +0200573.9. Rings
William Lallemand0217b7b2020-11-18 10:41:24 +0100583.10. Log forwarding
Willy Tarreauc57f0e22009-05-10 13:12:33 +020059
604. Proxies
614.1. Proxy keywords matrix
624.2. Alphabetically sorted keywords reference
63
Davor Ocelice9ed2812017-12-25 17:49:28 +0100645. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200655.1. Bind options
665.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200675.3. Server DNS resolution
685.3.1. Global overview
695.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020070
Julien Pivotto6ccee412019-11-27 15:49:54 +0100716. Cache
726.1. Limitation
736.2. Setup
746.2.1. Cache section
756.2.2. Proxy section
76
Willy Tarreau74ca5042013-06-11 23:12:07 +0200777. Using ACLs and fetching samples
787.1. ACL basics
797.1.1. Matching booleans
807.1.2. Matching integers
817.1.3. Matching strings
827.1.4. Matching regular expressions (regexes)
837.1.5. Matching arbitrary data blocks
847.1.6. Matching IPv4 and IPv6 addresses
857.2. Using ACLs to form conditions
867.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200877.3.1. Converters
887.3.2. Fetching samples from internal states
897.3.3. Fetching samples at Layer 4
907.3.4. Fetching samples at Layer 5
917.3.5. Fetching samples from buffer contents (Layer 6)
927.3.6. Fetching HTTP samples (Layer 7)
Christopher Faulete596d182020-05-05 17:46:34 +0200937.3.7. Fetching samples for developers
Willy Tarreau74ca5042013-06-11 23:12:07 +0200947.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020095
968. Logging
978.1. Log levels
988.2. Log formats
998.2.1. Default log format
1008.2.2. TCP log format
1018.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +01001028.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +01001038.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001048.3. Advanced logging options
1058.3.1. Disabling logging of external tests
1068.3.2. Logging before waiting for the session to terminate
1078.3.3. Raising log level upon errors
1088.3.4. Disabling logging of successful connections
1098.4. Timing events
1108.5. Session state at disconnection
1118.6. Non-printable characters
1128.7. Capturing HTTP cookies
1138.8. Capturing HTTP headers
1148.9. Examples of logs
115
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001169. Supported filters
1179.1. Trace
1189.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001199.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001209.4. Cache
Christopher Fauletb30b3102019-09-12 23:03:09 +02001219.5. fcgi-app
Miroslav Zagoracdc32cd92020-12-13 18:32:57 +01001229.6. OpenTracing
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200123
Christopher Fauletb30b3102019-09-12 23:03:09 +020012410. FastCGI applications
12510.1. Setup
12610.1.1. Fcgi-app section
12710.1.2. Proxy section
12810.1.3. Example
12910.2. Default parameters
13010.3. Limitations
131
Emeric Brunce325c42021-04-02 17:05:09 +020013211. Address formats
13311.1. Address family prefixes
13411.2. Socket type prefixes
13511.3. Protocol prefixes
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200136
1371. Quick reminder about HTTP
138----------------------------
139
Davor Ocelice9ed2812017-12-25 17:49:28 +0100140When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200141fully analyzed and indexed, thus it becomes possible to build matching criteria
142on almost anything found in the contents.
143
144However, it is important to understand how HTTP requests and responses are
145formed, and how HAProxy decomposes them. It will then become easier to write
146correct rules and to debug existing configurations.
147
148
1491.1. The HTTP transaction model
150-------------------------------
151
152The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100153to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100154from the client to the server, a request is sent by the client through the
155connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200156will involve a new connection :
157
158 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
159
160In this mode, called the "HTTP close" mode, there are as many connection
161establishments as there are HTTP transactions. Since the connection is closed
162by the server after the response, the client does not need to know the content
163length.
164
165Due to the transactional nature of the protocol, it was possible to improve it
166to avoid closing a connection between two subsequent transactions. In this mode
167however, it is mandatory that the server indicates the content length for each
168response so that the client does not wait indefinitely. For this, a special
169header is used: "Content-length". This mode is called the "keep-alive" mode :
170
171 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
172
173Its advantages are a reduced latency between transactions, and less processing
174power required on the server side. It is generally better than the close mode,
175but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200176a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200177
Willy Tarreau95c4e142017-11-26 12:18:55 +0100178Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200179keep-alive, but the client does not wait for the first response to send the
180second request. This is useful for fetching large number of images composing a
181page :
182
183 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
184
185This can obviously have a tremendous benefit on performance because the network
186latency is eliminated between subsequent requests. Many HTTP agents do not
187correctly support pipelining since there is no way to associate a response with
188the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100189server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200190
Willy Tarreau95c4e142017-11-26 12:18:55 +0100191The next improvement is the multiplexed mode, as implemented in HTTP/2. This
192time, each transaction is assigned a single stream identifier, and all streams
193are multiplexed over an existing connection. Many requests can be sent in
194parallel by the client, and responses can arrive in any order since they also
195carry the stream identifier.
196
Willy Tarreau70dffda2014-01-30 03:07:23 +0100197By default HAProxy operates in keep-alive mode with regards to persistent
198connections: for each connection it processes each request and response, and
199leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100200start of a new request. When it receives HTTP/2 connections from a client, it
201processes all the requests in parallel and leaves the connection idling,
202waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200203
Christopher Faulet315b39c2018-09-21 16:26:19 +0200204HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100205 - keep alive : all requests and responses are processed (default)
206 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100207 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100208 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200209 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100210
Willy Tarreau95c4e142017-11-26 12:18:55 +0100211
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200212
2131.2. HTTP request
214-----------------
215
216First, let's consider this HTTP request :
217
218 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100219 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200220 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
221 2 Host: www.mydomain.com
222 3 User-agent: my small browser
223 4 Accept: image/jpeg, image/gif
224 5 Accept: image/png
225
226
2271.2.1. The Request line
228-----------------------
229
230Line 1 is the "request line". It is always composed of 3 fields :
231
232 - a METHOD : GET
233 - a URI : /serv/login.php?lang=en&profile=2
234 - a version tag : HTTP/1.1
235
236All of them are delimited by what the standard calls LWS (linear white spaces),
237which are commonly spaces, but can also be tabs or line feeds/carriage returns
238followed by spaces/tabs. The method itself cannot contain any colon (':') and
239is limited to alphabetic letters. All those various combinations make it
240desirable that HAProxy performs the splitting itself rather than leaving it to
241the user to write a complex or inaccurate regular expression.
242
243The URI itself can have several forms :
244
245 - A "relative URI" :
246
247 /serv/login.php?lang=en&profile=2
248
249 It is a complete URL without the host part. This is generally what is
250 received by servers, reverse proxies and transparent proxies.
251
252 - An "absolute URI", also called a "URL" :
253
254 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
255
256 It is composed of a "scheme" (the protocol name followed by '://'), a host
257 name or address, optionally a colon (':') followed by a port number, then
258 a relative URI beginning at the first slash ('/') after the address part.
259 This is generally what proxies receive, but a server supporting HTTP/1.1
260 must accept this form too.
261
262 - a star ('*') : this form is only accepted in association with the OPTIONS
263 method and is not relayable. It is used to inquiry a next hop's
264 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100265
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200266 - an address:port combination : 192.168.0.12:80
267 This is used with the CONNECT method, which is used to establish TCP
268 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
269 other protocols too.
270
271In a relative URI, two sub-parts are identified. The part before the question
272mark is called the "path". It is typically the relative path to static objects
273on the server. The part after the question mark is called the "query string".
274It is mostly used with GET requests sent to dynamic scripts and is very
275specific to the language, framework or application in use.
276
Willy Tarreau95c4e142017-11-26 12:18:55 +0100277HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100278assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100279
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200280
2811.2.2. The request headers
282--------------------------
283
284The headers start at the second line. They are composed of a name at the
285beginning of the line, immediately followed by a colon (':'). Traditionally,
286an LWS is added after the colon but that's not required. Then come the values.
287Multiple identical headers may be folded into one single line, delimiting the
288values with commas, provided that their order is respected. This is commonly
289encountered in the "Cookie:" field. A header may span over multiple lines if
290the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
291define a total of 3 values for the "Accept:" header.
292
Davor Ocelice9ed2812017-12-25 17:49:28 +0100293Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200294their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100295"Connection:" header). In HTTP/2, header names are always sent in lower case,
Willy Tarreau253c2512020-07-07 15:55:23 +0200296as can be seen when running in debug mode. Internally, all header names are
297normalized to lower case so that HTTP/1.x and HTTP/2 use the exact same
298representation, and they are sent as-is on the other side. This explains why an
299HTTP/1.x request typed with camel case is delivered in lower case.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200300
301The end of the headers is indicated by the first empty line. People often say
302that it's a double line feed, which is not exact, even if a double line feed
303is one valid form of empty line.
304
305Fortunately, HAProxy takes care of all these complex combinations when indexing
306headers, checking values and counting them, so there is no reason to worry
307about the way they could be written, but it is important not to accuse an
308application of being buggy if it does unusual, valid things.
309
310Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000311 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200312 in the middle of headers by LWS in order to join multi-line headers. This
313 is necessary for proper analysis and helps less capable HTTP parsers to work
314 correctly and not to be fooled by such complex constructs.
315
316
3171.3. HTTP response
318------------------
319
320An HTTP response looks very much like an HTTP request. Both are called HTTP
321messages. Let's consider this HTTP response :
322
323 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100324 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200325 1 HTTP/1.1 200 OK
326 2 Content-length: 350
327 3 Content-Type: text/html
328
Willy Tarreau816b9792009-09-15 21:25:21 +0200329As a special case, HTTP supports so called "Informational responses" as status
330codes 1xx. These messages are special in that they don't convey any part of the
331response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100332continue to post its request for instance. In the case of a status 100 response
333the requested information will be carried by the next non-100 response message
334following the informational one. This implies that multiple responses may be
335sent to a single request, and that this only works when keep-alive is enabled
336(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
337correctly forward and skip them, and only process the next non-100 response. As
338such, these messages are neither logged nor transformed, unless explicitly
339state otherwise. Status 101 messages indicate that the protocol is changing
340over the same connection and that haproxy must switch to tunnel mode, just as
341if a CONNECT had occurred. Then the Upgrade header would contain additional
342information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200343
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200344
Davor Ocelice9ed2812017-12-25 17:49:28 +01003451.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200346------------------------
347
348Line 1 is the "response line". It is always composed of 3 fields :
349
350 - a version tag : HTTP/1.1
351 - a status code : 200
352 - a reason : OK
353
354The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100355 - 1xx = informational message to be skipped (e.g. 100, 101)
356 - 2xx = OK, content is following (e.g. 200, 206)
357 - 3xx = OK, no content following (e.g. 302, 304)
358 - 4xx = error caused by the client (e.g. 401, 403, 404)
359 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200360
Lukas Tribus23953682017-04-28 13:24:30 +0000361Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100362"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200363found there, but it's a common practice to respect the well-established
364messages. It can be composed of one or multiple words, such as "OK", "Found",
365or "Authentication Required".
366
Davor Ocelice9ed2812017-12-25 17:49:28 +0100367HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200368
369 Code When / reason
370 200 access to stats page, and when replying to monitoring requests
371 301 when performing a redirection, depending on the configured code
372 302 when performing a redirection, depending on the configured code
373 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100374 307 when performing a redirection, depending on the configured code
375 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200376 400 for an invalid or too large request
377 401 when an authentication is required to perform the action (when
378 accessing the stats page)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200379 403 when a request is forbidden by a "http-request deny" rule
Florian Tham9205fea2020-01-08 13:35:30 +0100380 404 when the requested resource could not be found
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200381 408 when the request timeout strikes before the request is complete
Florian Tham272e29b2020-01-08 10:19:05 +0100382 410 when the requested resource is no longer available and will not
383 be available again
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200384 500 when haproxy encounters an unrecoverable internal error, such as a
385 memory allocation failure, which should never happen
Christopher Faulete095f312020-12-07 11:22:24 +0100386 501 when haproxy is unable to satisfy a client request because of an
387 unsupported feature
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200388 502 when the server returns an empty, invalid or incomplete response, or
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200389 when an "http-response deny" rule blocks the response.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200390 503 when no server was available to handle the request, or in response to
391 monitoring requests which match the "monitor fail" condition
392 504 when the response timeout strikes before the server responds
393
394The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3954.2).
396
397
3981.3.2. The response headers
399---------------------------
400
401Response headers work exactly like request headers, and as such, HAProxy uses
402the same parsing function for both. Please refer to paragraph 1.2.2 for more
403details.
404
405
4062. Configuring HAProxy
407----------------------
408
4092.1. Configuration file format
410------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200411
412HAProxy's configuration process involves 3 major sources of parameters :
413
414 - the arguments from the command-line, which always take precedence
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100415 - the configuration file(s), whose format is described here
Thayne McCombscdbcca92021-01-07 21:24:41 -0700416 - the running process's environment, in case some environment variables are
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100417 explicitly referenced
Willy Tarreau6a06a402007-07-15 20:15:28 +0200418
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100419The configuration file follows a fairly simple hierarchical format which obey
420a few basic rules:
Willy Tarreau0ba27502007-12-24 16:55:16 +0100421
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100422 1. a configuration file is an ordered sequence of statements
423
424 2. a statement is a single non-empty line before any unprotected "#" (hash)
425
426 3. a line is a series of tokens or "words" delimited by unprotected spaces or
427 tab characters
428
429 4. the first word or sequence of words of a line is one of the keywords or
430 keyword sequences listed in this document
431
432 5. all other words are all arguments of the first one, some being well-known
433 keywords listed in this document, others being values, references to other
434 parts of the configuration, or expressions
435
436 6. certain keywords delimit a section inside which only a subset of keywords
437 are supported
438
439 7. a section ends at the end of a file or on a special keyword starting a new
440 section
441
442This is all that is needed to know to write a simple but reliable configuration
443generator, but this is not enough to reliably parse any configuration nor to
444figure how to deal with certain corner cases.
445
446First, there are a few consequences of the rules above. Rule 6 and 7 imply that
447the keywords used to define a new section are valid everywhere and cannot have
448a different meaning in a specific section. These keywords are always a single
449word (as opposed to a sequence of words), and traditionally the section that
450follows them is designated using the same name. For example when speaking about
451the "global section", it designates the section of configuration that follows
452the "global" keyword. This usage is used a lot in error messages to help locate
453the parts that need to be addressed.
454
455A number of sections create an internal object or configuration space, which
456requires to be distinguished from other ones. In this case they will take an
457extra word which will set the name of this particular section. For some of them
458the section name is mandatory. For example "frontend foo" will create a new
459section of type "frontend" named "foo". Usually a name is specific to its
460section and two sections of different types may use the same name, but this is
461not recommended as it tends to complexify configuration management.
462
463A direct consequence of rule 7 is that when multiple files are read at once,
464each of them must start with a new section, and the end of each file will end
465a section. A file cannot contain sub-sections nor end an existing section and
466start a new one.
467
468Rule 1 mentioned that ordering matters. Indeed, some keywords create directives
469that can be repeated multiple times to create ordered sequences of rules to be
470applied in a certain order. For example "tcp-request" can be used to alternate
471"accept" and "reject" rules on varying criteria. As such, a configuration file
472processor must always preserve a section's ordering when editing a file. The
473ordering of sections usually does not matter except for the global section
474which must be placed before other sections, but it may be repeated if needed.
475In addition, some automatic identifiers may automatically be assigned to some
476of the created objects (e.g. proxies), and by reordering sections, their
477identifiers will change. These ones appear in the statistics for example. As
478such, the configuration below will assign "foo" ID number 1 and "bar" ID number
4792, which will be swapped if the two sections are reversed:
480
481 listen foo
482 bind :80
483
484 listen bar
485 bind :81
486
487Another important point is that according to rules 2 and 3 above, empty lines,
488spaces, tabs, and comments following and unprotected "#" character are not part
489of the configuration as they are just used as delimiters. This implies that the
490following configurations are strictly equivalent:
491
492 global#this is the global section
493 daemon#daemonize
494 frontend foo
495 mode http # or tcp
496
497and:
498
499 global
500 daemon
501
502 # this is the public web frontend
503 frontend foo
504 mode http
505
506The common practice is to align to the left only the keyword that initiates a
507new section, and indent (i.e. prepend a tab character or a few spaces) all
508other keywords so that it's instantly visible that they belong to the same
509section (as done in the second example above). Placing comments before a new
510section helps the reader decide if it's the desired one. Leaving a blank line
511at the end of a section also visually helps spotting the end when editing it.
512
513Tabs are very convenient for indent but they do not copy-paste well. If spaces
514are used instead, it is recommended to avoid placing too many (2 to 4) so that
515editing in field doesn't become a burden with limited editors that do not
516support automatic indent.
517
518In the early days it used to be common to see arguments split at fixed tab
519positions because most keywords would not take more than two arguments. With
520modern versions featuring complex expressions this practice does not stand
521anymore, and is not recommended.
522
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200523
William Lallemandf9873ba2015-05-05 17:37:14 +02005242.2. Quoting and escaping
525-------------------------
526
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100527In modern configurations, some arguments require the use of some characters
528that were previously considered as pure delimiters. In order to make this
529possible, HAProxy supports character escaping by prepending a backslash ('\')
530in front of the character to be escaped, weak quoting within double quotes
531('"') and strong quoting within single quotes ("'").
William Lallemandf9873ba2015-05-05 17:37:14 +0200532
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100533This is pretty similar to what is done in a number of programming languages and
534very close to what is commonly encountered in Bourne shell. The principle is
535the following: while the configuration parser cuts the lines into words, it
536also takes care of quotes and backslashes to decide whether a character is a
537delimiter or is the raw representation of this character within the current
538word. The escape character is then removed, the quotes are removed, and the
539remaining word is used as-is as a keyword or argument for example.
William Lallemandf9873ba2015-05-05 17:37:14 +0200540
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100541If a backslash is needed in a word, it must either be escaped using itself
542(i.e. double backslash) or be strongly quoted.
543
544Escaping outside quotes is achieved by preceding a special character by a
545backslash ('\'):
William Lallemandf9873ba2015-05-05 17:37:14 +0200546
547 \ to mark a space and differentiate it from a delimiter
548 \# to mark a hash and differentiate it from a comment
549 \\ to use a backslash
550 \' to use a single quote and differentiate it from strong quoting
551 \" to use a double quote and differentiate it from weak quoting
552
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100553In addition, a few non-printable characters may be emitted using their usual
554C-language representation:
555
556 \n to insert a line feed (LF, character \x0a or ASCII 10 decimal)
557 \r to insert a carriage return (CR, character \x0d or ASCII 13 decimal)
558 \t to insert a tab (character \x09 or ASCII 9 decimal)
559 \xNN to insert character having ASCII code hex NN (e.g \x0a for LF).
560
561Weak quoting is achieved by surrounding double quotes ("") around the character
562or sequence of characters to protect. Weak quoting prevents the interpretation
563of:
William Lallemandf9873ba2015-05-05 17:37:14 +0200564
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100565 space or tab as a word separator
William Lallemandf9873ba2015-05-05 17:37:14 +0200566 ' single quote as a strong quoting delimiter
567 # hash as a comment start
568
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100569Weak quoting permits the interpretation of environment variables (which are not
570evaluated outside of quotes) by preceding them with a dollar sign ('$'). If a
571dollar character is needed inside double quotes, it must be escaped using a
572backslash.
William Lallemandb2f07452015-05-12 14:27:13 +0200573
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100574Strong quoting is achieved by surrounding single quotes ('') around the
575character or sequence of characters to protect. Inside single quotes, nothing
576is interpreted, it's the efficient way to quote regular expressions.
William Lallemandf9873ba2015-05-05 17:37:14 +0200577
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100578As a result, here is the matrix indicating how special characters can be
579entered in different contexts (unprintable characters are replaced with their
580name within angle brackets). Note that some characters that may only be
581represented escaped have no possible representation inside single quotes,
582hence the '-' there:
William Lallemandf9873ba2015-05-05 17:37:14 +0200583
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100584 Character | Unquoted | Weakly quoted | Strongly quoted
585 -----------+---------------+-----------------------------+-----------------
586 <TAB> | \<TAB>, \x09 | "<TAB>", "\<TAB>", "\x09" | '<TAB>'
587 <LF> | \n, \x0a | "\n", "\x0a" | -
588 <CR> | \r, \x0d | "\r", "\x0d" | -
589 <SPC> | \<SPC>, \x20 | "<SPC>", "\<SPC>", "\x20" | '<SPC>'
590 " | \", \x22 | "\"", "\x22" | '"'
591 # | \#, \x23 | "#", "\#", "\x23" | '#'
592 $ | $, \$, \x24 | "\$", "\x24" | '$'
593 ' | \', \x27 | "'", "\'", "\x27" | -
594 \ | \\, \x5c | "\\", "\x5c" | '\'
William Lallemandf9873ba2015-05-05 17:37:14 +0200595
596 Example:
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100597 # those are all strictly equivalent:
William Lallemandf9873ba2015-05-05 17:37:14 +0200598 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 log-format "%{+Q}o %t"' %s'\ %{-Q}r
603
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100604There is one particular case where a second level of quoting or escaping may be
605necessary. Some keywords take arguments within parenthesis, sometimes delimited
606by commas. These arguments are commonly integers or predefined words, but when
607they are arbitrary strings, it may be required to perform a separate level of
608escaping to disambiguate the characters that belong to the argument from the
609characters that are used to delimit the arguments themselves. A pretty common
610case is the "regsub" converter. It takes a regular expression in argument, and
611if a closing parenthesis is needed inside, this one will require to have its
612own quotes.
613
614The keyword argument parser is exactly the same as the top-level one regarding
615quotes, except that is will not make special cases of backslashes. But what is
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500616not always obvious is that the delimiters used inside must first be escaped or
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100617quoted so that they are not resolved at the top level.
618
619Let's take this example making use of the "regsub" converter which takes 3
620arguments, one regular expression, one replacement string and one set of flags:
621
622 # replace all occurrences of "foo" with "blah" in the path:
623 http-request set-path %[path,regsub(foo,blah,g)]
624
625Here no special quoting was necessary. But if now we want to replace either
626"foo" or "bar" with "blah", we'll need the regular expression "(foo|bar)". We
627cannot write:
628
629 http-request set-path %[path,regsub((foo|bar),blah,g)]
630
631because we would like the string to cut like this:
632
633 http-request set-path %[path,regsub((foo|bar),blah,g)]
634 |---------|----|-|
635 arg1 _/ / /
636 arg2 __________/ /
637 arg3 ______________/
638
639but actually what is passed is a string between the opening and closing
640parenthesis then garbage:
641
642 http-request set-path %[path,regsub((foo|bar),blah,g)]
643 |--------|--------|
644 arg1=(foo|bar _/ /
645 trailing garbage _________/
646
647The obvious solution here seems to be that the closing parenthesis needs to be
648quoted, but alone this will not work, because as mentioned above, quotes are
649processed by the top-level parser which will resolve them before processing
650this word:
651
652 http-request set-path %[path,regsub("(foo|bar)",blah,g)]
653 ------------ -------- ----------------------------------
654 word1 word2 word3=%[path,regsub((foo|bar),blah,g)]
655
656So we didn't change anything for the argument parser at the second level which
657still sees a truncated regular expression as the only argument, and garbage at
658the end of the string. By escaping the quotes they will be passed unmodified to
659the second level:
660
661 http-request set-path %[path,regsub(\"(foo|bar)\",blah,g)]
662 ------------ -------- ------------------------------------
663 word1 word2 word3=%[path,regsub("(foo|bar)",blah,g)]
664 |---------||----|-|
665 arg1=(foo|bar) _/ / /
666 arg2=blah ___________/ /
667 arg3=g _______________/
668
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500669Another approach consists in using single quotes outside the whole string and
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100670double quotes inside (so that the double quotes are not stripped again):
671
672 http-request set-path '%[path,regsub("(foo|bar)",blah,g)]'
673 ------------ -------- ----------------------------------
674 word1 word2 word3=%[path,regsub("(foo|bar)",blah,g)]
675 |---------||----|-|
676 arg1=(foo|bar) _/ / /
677 arg2 ___________/ /
678 arg3 _______________/
679
680When using regular expressions, it can happen that the dollar ('$') character
681appears in the expression or that a backslash ('\') is used in the replacement
682string. In this case these ones will also be processed inside the double quotes
683thus single quotes are preferred (or double escaping). Example:
684
685 http-request set-path '%[path,regsub("^/(here)(/|$)","my/\1",g)]'
686 ------------ -------- -----------------------------------------
687 word1 word2 word3=%[path,regsub("^/(here)(/|$)","my/\1",g)]
688 |-------------| |-----||-|
689 arg1=(here)(/|$) _/ / /
690 arg2=my/\1 ________________/ /
691 arg3 ______________________/
692
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500693Remember that backslahes are not escape characters within single quotes and
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100694that the whole word3 above is already protected against them using the single
695quotes. Conversely, if double quotes had been used around the whole expression,
696single the dollar character and the backslashes would have been resolved at top
697level, breaking the argument contents at the second level.
698
699When in doubt, simply do not use quotes anywhere, and start to place single or
700double quotes around arguments that require a comma or a closing parenthesis,
701and think about escaping these quotes using a backslash of the string contains
702a dollar or a backslash. Again, this is pretty similar to what is used under
703a Bourne shell when double-escaping a command passed to "eval". For API writers
704the best is probably to place escaped quotes around each and every argument,
705regardless of their contents. Users will probably find that using single quotes
706around the whole expression and double quotes around each argument provides
707more readable configurations.
William Lallemandf9873ba2015-05-05 17:37:14 +0200708
709
William Lallemandb2f07452015-05-12 14:27:13 +02007102.3. Environment variables
711--------------------------
712
713HAProxy's configuration supports environment variables. Those variables are
714interpreted only within double quotes. Variables are expanded during the
715configuration parsing. Variable names must be preceded by a dollar ("$") and
716optionally enclosed with braces ("{}") similarly to what is done in Bourne
717shell. Variable names can contain alphanumerical characters or the character
Amaury Denoyellefa41cb62020-10-01 14:32:35 +0200718underscore ("_") but should not start with a digit. If the variable contains a
719list of several values separated by spaces, it can be expanded as individual
720arguments by enclosing the variable with braces and appending the suffix '[*]'
721before the closing brace.
William Lallemandb2f07452015-05-12 14:27:13 +0200722
723 Example:
724
725 bind "fd@${FD_APP1}"
726
727 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
728
729 user "$HAPROXY_USER"
730
William Lallemand4d03e432019-06-14 15:35:37 +0200731Some variables are defined by HAProxy, they can be used in the configuration
732file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200733
William Lallemand4d03e432019-06-14 15:35:37 +0200734* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
735 name of the local peer. (See "-L" in the management guide.)
736
737* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
738 separated by semicolons. Can be useful in the case you specified a
739 directory.
740
741* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
742
John Roeslerfb2fce12019-07-10 15:45:51 -0500743* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200744 processes, separated by semicolons.
745
John Roeslerfb2fce12019-07-10 15:45:51 -0500746* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200747 CLI, separated by semicolons.
748
749See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200750
Willy Tarreau4b103022021-02-12 17:59:10 +0100751
7522.4. Conditional blocks
753-----------------------
754
755It may sometimes be convenient to be able to conditionally enable or disable
756some arbitrary parts of the configuration, for example to enable/disable SSL or
757ciphers, enable or disable some pre-production listeners without modifying the
758configuration, or adjust the configuration's syntax to support two distinct
759versions of HAProxy during a migration.. HAProxy brings a set of nestable
760preprocessor-like directives which allow to integrate or ignore some blocks of
761text. These directives must be placed on their own line and they act on the
762lines that follow them. Two of them support an expression, the other ones only
763switch to an alternate block or end a current level. The 4 following directives
764are defined to form conditional blocks:
765
766 - .if <condition>
767 - .elif <condition>
768 - .else
769 - .endif
770
771The ".if" directive nests a new level, ".elif" stays at the same level, ".else"
772as well, and ".endif" closes a level. Each ".if" must be terminated by a
773matching ".endif". The ".elif" may only be placed after ".if" or ".elif", and
774there is no limit to the number of ".elif" that may be chained. There may be
775only one ".else" per ".if" and it must always be after the ".if" or the last
776".elif" of a block.
777
778Comments may be placed on the same line if needed after a '#', they will be
779ignored. The directives are tokenized like other configuration directives, and
780as such it is possible to use environment variables in conditions.
781
782The conditions are currently limited to:
783
784 - an empty string, always returns "false"
785 - the integer zero ('0'), always returns "false"
786 - a non-nul integer (e.g. '1'), always returns "true".
787
788Other patterns are not supported yet but the purpose is to bring a few
789functions to test for certain build options and supported features.
790
791Three other directives are provided to report some status:
792
793 - .notice "message" : emit this message at level NOTICE
794 - .warning "message" : emit this message at level WARNING
795 - .alert "message" : emit this message at level ALERT
796
797Messages emitted at level WARNING may cause the process to fail to start if the
798"strict-mode" is enabled. Messages emitted at level ALERT will always cause a
799fatal error. These can be used to detect some inappropriate conditions and
800provide advice to the user.
801
802Example:
803
804 .if "${A}"
805 .if "${B}"
806 .notice "A=1, B=1"
807 .elif "${C}"
808 .notice "A=1, B=0, C=1"
809 .elif "${D}"
810 .warning "A=1, B=0, C=0, D=1"
811 .else
812 .alert "A=1, B=0, C=0, D=0"
813 .endif
814 .else
815 .notice "A=0"
816 .endif
817
818
8192.5. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200820----------------
821
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100822Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100823values are generally expressed in milliseconds (unless explicitly stated
824otherwise) but may be expressed in any other unit by suffixing the unit to the
825numeric value. It is important to consider this because it will not be repeated
826for every keyword. Supported units are :
827
828 - us : microseconds. 1 microsecond = 1/1000000 second
829 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
830 - s : seconds. 1s = 1000ms
831 - m : minutes. 1m = 60s = 60000ms
832 - h : hours. 1h = 60m = 3600s = 3600000ms
833 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
834
835
Willy Tarreau4b103022021-02-12 17:59:10 +01008362.6. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200837-------------
838
839 # Simple configuration for an HTTP proxy listening on port 80 on all
840 # interfaces and forwarding requests to a single backend "servers" with a
841 # single server "server1" listening on 127.0.0.1:8000
842 global
843 daemon
844 maxconn 256
845
846 defaults
847 mode http
848 timeout connect 5000ms
849 timeout client 50000ms
850 timeout server 50000ms
851
852 frontend http-in
853 bind *:80
854 default_backend servers
855
856 backend servers
857 server server1 127.0.0.1:8000 maxconn 32
858
859
860 # The same configuration defined with a single listen block. Shorter but
861 # less expressive, especially in HTTP mode.
862 global
863 daemon
864 maxconn 256
865
866 defaults
867 mode http
868 timeout connect 5000ms
869 timeout client 50000ms
870 timeout server 50000ms
871
872 listen http-in
873 bind *:80
874 server server1 127.0.0.1:8000 maxconn 32
875
876
877Assuming haproxy is in $PATH, test these configurations in a shell with:
878
Willy Tarreauccb289d2010-12-11 20:19:38 +0100879 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200880
881
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008823. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200883--------------------
884
885Parameters in the "global" section are process-wide and often OS-specific. They
886are generally set once for all and do not need being changed once correct. Some
887of them have command-line equivalents.
888
889The following keywords are supported in the "global" section :
890
891 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200892 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200893 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200894 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200895 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200896 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200897 - description
898 - deviceatlas-json-file
899 - deviceatlas-log-level
900 - deviceatlas-separator
901 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900902 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200903 - gid
904 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100905 - hard-stop-after
Christopher Faulet98fbe952019-07-22 16:18:24 +0200906 - h1-case-adjust
907 - h1-case-adjust-file
Willy Tarreaud96f1122019-12-03 07:07:36 +0100908 - insecure-fork-wanted
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100909 - insecure-setuid-wanted
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100910 - issuers-chain-path
Dragan Dosen13cd54c2020-06-18 18:24:05 +0200911 - localpeer
Willy Tarreau6a06a402007-07-15 20:15:28 +0200912 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200913 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100914 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200915 - lua-load
Thierry Fournier59f11be2020-11-29 00:37:41 +0100916 - lua-load-per-thread
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +0100917 - lua-prepend-path
William Lallemand27edc4b2019-05-07 17:49:33 +0200918 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200919 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200920 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200921 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200922 - pidfile
Willy Tarreau119e50e2020-05-22 13:53:29 +0200923 - pp2-never-send-local
Willy Tarreau1d549722016-02-16 12:41:57 +0100924 - presetenv
925 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200926 - uid
927 - ulimit-n
928 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200929 - set-dumpable
Willy Tarreau13d2ba22021-03-26 11:38:08 +0100930 - set-var
Willy Tarreau1d549722016-02-16 12:41:57 +0100931 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200932 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200933 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200934 - ssl-default-bind-ciphersuites
Jerome Magninb203ff62020-04-03 15:28:22 +0200935 - ssl-default-bind-curves
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200936 - ssl-default-bind-options
937 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200938 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200939 - ssl-default-server-options
940 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100941 - ssl-server-verify
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +0200942 - ssl-skip-self-issued-ca
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100943 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100944 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100945 - 51degrees-data-file
946 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200947 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200948 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200949 - wurfl-data-file
950 - wurfl-information-list
951 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200952 - wurfl-cache-size
William Dauchy0fec3ab2019-10-27 20:08:11 +0100953 - strict-limits
Willy Tarreaud72758d2010-01-12 10:42:19 +0100954
Willy Tarreau6a06a402007-07-15 20:15:28 +0200955 * Performance tuning
William Dauchy0a8824f2019-10-27 20:08:09 +0100956 - busy-polling
Willy Tarreau1746eec2014-04-25 10:46:47 +0200957 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200958 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200959 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100960 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100961 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100962 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200963 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200964 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200965 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200966 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200967 - noepoll
968 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000969 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200970 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100971 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300972 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000973 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100974 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200975 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200976 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200977 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000978 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000979 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200980 - tune.buffers.limit
981 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200982 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200983 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100984 - tune.comp.maxlevel
Willy Tarreaubc52bec2020-06-18 08:58:47 +0200985 - tune.fd.edge-triggered
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200986 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200987 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200988 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100989 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200990 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200991 - tune.http.maxhdr
Willy Tarreau76cc6992020-07-01 18:49:24 +0200992 - tune.idle-pool.shared
Willy Tarreau7e312732014-02-12 16:35:14 +0100993 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100994 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100995 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100996 - tune.lua.session-timeout
997 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200998 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100999 - tune.maxaccept
1000 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +02001001 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +02001002 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001003 - tune.pipesize
Willy Tarreaua8e2d972020-07-01 18:27:16 +02001004 - tune.pool-high-fd-ratio
1005 - tune.pool-low-fd-ratio
Willy Tarreaue803de22010-01-21 17:43:04 +01001006 - tune.rcvbuf.client
1007 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +01001008 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +02001009 - tune.runqueue-depth
Willy Tarreaue7723bd2020-06-24 11:11:02 +02001010 - tune.sched.low-latency
Willy Tarreaue803de22010-01-21 17:43:04 +01001011 - tune.sndbuf.client
1012 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001013 - tune.ssl.cachesize
William Lallemand7d42ef52020-07-06 11:41:30 +02001014 - tune.ssl.keylog
Willy Tarreaubfd59462013-02-21 07:46:09 +01001015 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +02001016 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +01001017 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001018 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +02001019 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001020 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001021 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +01001022 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001023 - tune.vars.reqres-max-size
1024 - tune.vars.sess-max-size
1025 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +01001026 - tune.zlib.memlevel
1027 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +01001028
Willy Tarreau6a06a402007-07-15 20:15:28 +02001029 * Debugging
Willy Tarreau6a06a402007-07-15 20:15:28 +02001030 - quiet
Willy Tarreau3eb10b82020-04-15 16:42:39 +02001031 - zero-warning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001032
1033
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010343.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +02001035------------------------------------
1036
Emeric Brunc8e8d122012-10-02 18:42:10 +02001037ca-base <dir>
1038 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +01001039 relative path is used with "ca-file", "ca-verify-file" or "crl-file"
1040 directives. Absolute locations specified in "ca-file", "ca-verify-file" and
1041 "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +02001042
Willy Tarreau6a06a402007-07-15 20:15:28 +02001043chroot <jail dir>
1044 Changes current directory to <jail dir> and performs a chroot() there before
1045 dropping privileges. This increases the security level in case an unknown
1046 vulnerability would be exploited, since it would make it very hard for the
1047 attacker to exploit the system. This only works when the process is started
1048 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +01001049 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +01001050
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001051cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
1052 On Linux 2.6 and above, it is possible to bind a process or a thread to a
1053 specific CPU set. This means that the process or the thread will never run on
1054 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
1055 sets. The first argument is a process set, eventually followed by a thread
1056 set. These sets have the format
1057
1058 all | odd | even | number[-[number]]
1059
1060 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +01001061 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001062 ignored. It is possible to specify a range with two such number delimited by
1063 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +01001064 "all", only odd numbers using "odd" or even numbers using "even", just like
1065 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +01001066 CPU sets. Each CPU set is either a unique number between 0 and 31 or 63 or a
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +01001067 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001068 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001069 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001070 specified. Each "cpu-map" directive will replace the previous ones when they
1071 overlap. A thread will be bound on the intersection of its mapping and the
1072 one of the process on which it is attached. If the intersection is null, no
1073 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +01001074
Christopher Fauletff4121f2017-11-22 16:38:49 +01001075 Ranges can be partially defined. The higher bound can be omitted. In such
1076 case, it is replaced by the corresponding maximum value, 32 or 64 depending
1077 on the machine's word size.
1078
Christopher Faulet26028f62017-11-22 15:01:51 +01001079 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001080 automatically bind a process or a thread to a CPU by incrementing
1081 process/thread and CPU sets. To be valid, both sets must have the same
1082 size. No matter the declaration order of the CPU sets, it will be bound from
1083 the lowest to the highest bound. Having a process and a thread range with the
1084 "auto:" prefix is not supported. Only one range is supported, the other one
1085 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +01001086
1087 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001088 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
1089
1090 cpu-map 1/all 0-3 # bind all threads of the first process on the
1091 # first 4 CPUs
1092
1093 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
1094 # or "cpu-map 1-32 0-31" depending on the machine's
1095 # word size.
1096
Christopher Faulet26028f62017-11-22 15:01:51 +01001097 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001098 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +01001099 cpu-map auto:1-4 0-3
1100 cpu-map auto:1-4 0-1 2-3
1101 cpu-map auto:1-4 3 2 1 0
1102
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001103 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
1104 # and so on.
1105 cpu-map auto:1/1-4 0-3
1106 cpu-map auto:1/1-4 0-1 2-3
1107 cpu-map auto:1/1-4 3 2 1 0
1108
Davor Ocelice9ed2812017-12-25 17:49:28 +01001109 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +01001110 cpu-map auto:all 0-63
1111 cpu-map auto:even 0-31
1112 cpu-map auto:odd 32-63
1113
1114 # invalid cpu-map because process and CPU sets have different sizes.
1115 cpu-map auto:1-4 0 # invalid
1116 cpu-map auto:1 0-3 # invalid
1117
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001118 # invalid cpu-map because automatic binding is used with a process range
1119 # and a thread range.
1120 cpu-map auto:all/all 0 # invalid
1121 cpu-map auto:all/1-4 0 # invalid
1122 cpu-map auto:1-4/all 0 # invalid
1123
Emeric Brunc8e8d122012-10-02 18:42:10 +02001124crt-base <dir>
1125 Assigns a default directory to fetch SSL certificates from when a relative
William Dauchy238ea3b2020-01-11 13:09:12 +01001126 path is used with "crtfile" or "crt" directives. Absolute locations specified
1127 prevail and ignore "crt-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +02001128
Willy Tarreau6a06a402007-07-15 20:15:28 +02001129daemon
1130 Makes the process fork into background. This is the recommended mode of
1131 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +01001132 disabled by the command line "-db" argument. This option is ignored in
1133 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001134
David Carlier8167f302015-06-01 13:50:06 +02001135deviceatlas-json-file <path>
1136 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001137 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +02001138
1139deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001140 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +02001141 optional and set to 0 by default if not set.
1142
1143deviceatlas-separator <char>
1144 Sets the character separator for the API properties results. This directive
1145 is optional and set to | by default if not set.
1146
Cyril Bonté0306c4a2015-10-26 22:37:38 +01001147deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001148 Sets the client cookie's name used for the detection if the DeviceAtlas
1149 Client-side component was used during the request. This directive is optional
1150 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +01001151
Simon Horman98637e52014-06-20 12:30:16 +09001152external-check
Willy Tarreaud96f1122019-12-03 07:07:36 +01001153 Allows the use of an external agent to perform health checks. This is
1154 disabled by default as a security precaution, and even when enabled, checks
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001155 may still fail unless "insecure-fork-wanted" is enabled as well. If the
1156 program launched makes use of a setuid executable (it should really not),
1157 you may also need to set "insecure-setuid-wanted" in the global section.
1158 See "option external-check", and "insecure-fork-wanted", and
1159 "insecure-setuid-wanted".
Simon Horman98637e52014-06-20 12:30:16 +09001160
Willy Tarreau6a06a402007-07-15 20:15:28 +02001161gid <number>
Thayne McCombscdbcca92021-01-07 21:24:41 -07001162 Changes the process's group ID to <number>. It is recommended that the group
Willy Tarreau6a06a402007-07-15 20:15:28 +02001163 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1164 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +01001165 Note that if haproxy is started from a user having supplementary groups, it
1166 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001167 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +01001168
Willy Tarreau11770ce2019-12-03 08:29:22 +01001169group <group name>
1170 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
1171 See also "gid" and "user".
1172
Cyril Bonté203ec5a2017-03-23 22:44:13 +01001173hard-stop-after <time>
1174 Defines the maximum time allowed to perform a clean soft-stop.
1175
1176 Arguments :
1177 <time> is the maximum time (by default in milliseconds) for which the
1178 instance will remain alive when a soft-stop is received via the
1179 SIGUSR1 signal.
1180
1181 This may be used to ensure that the instance will quit even if connections
1182 remain opened during a soft-stop (for example with long timeouts for a proxy
1183 in tcp mode). It applies both in TCP and HTTP mode.
1184
1185 Example:
1186 global
1187 hard-stop-after 30s
1188
Christopher Faulet98fbe952019-07-22 16:18:24 +02001189h1-case-adjust <from> <to>
1190 Defines the case adjustment to apply, when enabled, to the header name
1191 <from>, to change it to <to> before sending it to HTTP/1 clients or
1192 servers. <from> must be in lower case, and <from> and <to> must not differ
1193 except for their case. It may be repeated if several header names need to be
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05001194 adjusted. Duplicate entries are not allowed. If a lot of header names have to
Christopher Faulet98fbe952019-07-22 16:18:24 +02001195 be adjusted, it might be more convenient to use "h1-case-adjust-file".
1196 Please note that no transformation will be applied unless "option
1197 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
1198 specified in a proxy.
1199
1200 There is no standard case for header names because, as stated in RFC7230,
1201 they are case-insensitive. So applications must handle them in a case-
1202 insensitive manner. But some bogus applications violate the standards and
1203 erroneously rely on the cases most commonly used by browsers. This problem
1204 becomes critical with HTTP/2 because all header names must be exchanged in
1205 lower case, and HAProxy follows the same convention. All header names are
1206 sent in lower case to clients and servers, regardless of the HTTP version.
1207
1208 Applications which fail to properly process requests or responses may require
1209 to temporarily use such workarounds to adjust header names sent to them for
1210 the time it takes the application to be fixed. Please note that an
1211 application which requires such workarounds might be vulnerable to content
1212 smuggling attacks and must absolutely be fixed.
1213
1214 Example:
1215 global
1216 h1-case-adjust content-length Content-Length
1217
1218 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
1219 "option h1-case-adjust-bogus-server".
1220
1221h1-case-adjust-file <hdrs-file>
1222 Defines a file containing a list of key/value pairs used to adjust the case
1223 of some header names before sending them to HTTP/1 clients or servers. The
1224 file <hdrs-file> must contain 2 header names per line. The first one must be
1225 in lower case and both must not differ except for their case. Lines which
1226 start with '#' are ignored, just like empty lines. Leading and trailing tabs
1227 and spaces are stripped. Duplicate entries are not allowed. Please note that
1228 no transformation will be applied unless "option h1-case-adjust-bogus-client"
1229 or "option h1-case-adjust-bogus-server" is specified in a proxy.
1230
1231 If this directive is repeated, only the last one will be processed. It is an
1232 alternative to the directive "h1-case-adjust" if a lot of header names need
1233 to be adjusted. Please read the risks associated with using this.
1234
1235 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
1236 "option h1-case-adjust-bogus-server".
1237
Willy Tarreaud96f1122019-12-03 07:07:36 +01001238insecure-fork-wanted
1239 By default haproxy tries hard to prevent any thread and process creation
1240 after it starts. Doing so is particularly important when using Lua files of
1241 uncertain origin, and when experimenting with development versions which may
1242 still contain bugs whose exploitability is uncertain. And generally speaking
1243 it's good hygiene to make sure that no unexpected background activity can be
1244 triggered by traffic. But this prevents external checks from working, and may
1245 break some very specific Lua scripts which actively rely on the ability to
1246 fork. This option is there to disable this protection. Note that it is a bad
1247 idea to disable it, as a vulnerability in a library or within haproxy itself
1248 will be easier to exploit once disabled. In addition, forking from Lua or
1249 anywhere else is not reliable as the forked process may randomly embed a lock
1250 set by another thread and never manage to finish an operation. As such it is
1251 highly recommended that this option is never used and that any workload
1252 requiring such a fork be reconsidered and moved to a safer solution (such as
1253 agents instead of external checks). This option supports the "no" prefix to
1254 disable it.
1255
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001256insecure-setuid-wanted
1257 HAProxy doesn't need to call executables at run time (except when using
1258 external checks which are strongly recommended against), and is even expected
1259 to isolate itself into an empty chroot. As such, there basically is no valid
1260 reason to allow a setuid executable to be called without the user being fully
1261 aware of the risks. In a situation where haproxy would need to call external
1262 checks and/or disable chroot, exploiting a vulnerability in a library or in
1263 haproxy itself could lead to the execution of an external program. On Linux
1264 it is possible to lock the process so that any setuid bit present on such an
1265 executable is ignored. This significantly reduces the risk of privilege
1266 escalation in such a situation. This is what haproxy does by default. In case
1267 this causes a problem to an external check (for example one which would need
1268 the "ping" command), then it is possible to disable this protection by
1269 explicitly adding this directive in the global section. If enabled, it is
1270 possible to turn it back off by prefixing it with the "no" keyword.
1271
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +01001272issuers-chain-path <dir>
1273 Assigns a directory to load certificate chain for issuer completion. All
1274 files must be in PEM format. For certificates loaded with "crt" or "crt-list",
1275 if certificate chain is not included in PEM (also commonly known as
1276 intermediate certificate), haproxy will complete chain if the issuer of the
1277 certificate corresponds to the first certificate of the chain loaded with
1278 "issuers-chain-path".
1279 A "crt" file with PrivateKey+Certificate+IntermediateCA2+IntermediateCA1
1280 could be replaced with PrivateKey+Certificate. HAProxy will complete the
1281 chain if a file with IntermediateCA2+IntermediateCA1 is present in
1282 "issuers-chain-path" directory. All other certificates with the same issuer
1283 will share the chain in memory.
1284
Dragan Dosen13cd54c2020-06-18 18:24:05 +02001285localpeer <name>
1286 Sets the local instance's peer name. It will be ignored if the "-L"
1287 command line argument is specified or if used after "peers" section
1288 definitions. In such cases, a warning message will be emitted during
1289 the configuration parsing.
1290
1291 This option will also set the HAPROXY_LOCALPEER environment variable.
1292 See also "-L" in the management guide and "peers" section below.
1293
Jan Wagner3e678602020-12-17 22:22:32 +01001294log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001295 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +01001296 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +01001297 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +01001298 configured with "log global".
1299
1300 <address> can be one of:
1301
Willy Tarreau2769aa02007-12-27 18:26:09 +01001302 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +01001303 no port is specified, 514 is used by default (the standard syslog
1304 port).
1305
David du Colombier24bb5f52011-03-17 10:40:23 +01001306 - An IPv6 address followed by a colon and optionally a UDP port. If
1307 no port is specified, 514 is used by default (the standard syslog
1308 port).
1309
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001310 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +01001311 considerations for chroot (be sure the path is accessible inside
1312 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +01001313 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +01001314
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001315 - A file descriptor number in the form "fd@<number>", which may point
1316 to a pipe, terminal, or socket. In this case unbuffered logs are used
1317 and one writev() call per log is performed. This is a bit expensive
1318 but acceptable for most workloads. Messages sent this way will not be
1319 truncated but may be dropped, in which case the DroppedLogs counter
1320 will be incremented. The writev() call is atomic even on pipes for
1321 messages up to PIPE_BUF size, which POSIX recommends to be at least
1322 512 and which is 4096 bytes on most modern operating systems. Any
1323 larger message may be interleaved with messages from other processes.
1324 Exceptionally for debugging purposes the file descriptor may also be
1325 directed to a file, but doing so will significantly slow haproxy down
1326 as non-blocking calls will be ignored. Also there will be no way to
1327 purge nor rotate this file without restarting the process. Note that
1328 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +01001329 for use with a TCP syslog server. See also the "short" and "raw"
1330 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001331
1332 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
1333 "fd@2", see above.
1334
Willy Tarreauc046d162019-08-30 15:24:59 +02001335 - A ring buffer in the form "ring@<name>", which will correspond to an
1336 in-memory ring buffer accessible over the CLI using the "show events"
1337 command, which will also list existing rings and their sizes. Such
1338 buffers are lost on reload or restart but when used as a complement
1339 this can help troubleshooting by having the logs instantly available.
1340
William Lallemandb2f07452015-05-12 14:27:13 +02001341 You may want to reference some environment variables in the address
1342 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001343
Willy Tarreau18324f52014-06-27 18:10:07 +02001344 <length> is an optional maximum line length. Log lines larger than this value
1345 will be truncated before being sent. The reason is that syslog
1346 servers act differently on log line length. All servers support the
1347 default value of 1024, but some servers simply drop larger lines
1348 while others do log them. If a server supports long lines, it may
1349 make sense to set this value here in order to avoid truncating long
1350 lines. Similarly, if a server drops long lines, it is preferable to
1351 truncate them before sending them. Accepted values are 80 to 65535
1352 inclusive. The default value of 1024 is generally fine for all
1353 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +01001354 JSON-formatted logs may require larger values. You may also need to
1355 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +02001356
Dragan Dosen7ad31542015-09-28 17:16:47 +02001357 <format> is the log format used when generating syslog messages. It may be
1358 one of the following :
1359
Emeric Brun0237c4e2020-11-27 16:24:34 +01001360 local Analog to rfc3164 syslog message format except that hostname
1361 field is stripped. This is the default.
1362 Note: option "log-send-hostname" switches the default to
1363 rfc3164.
1364
1365 rfc3164 The RFC3164 syslog message format.
Dragan Dosen7ad31542015-09-28 17:16:47 +02001366 (https://tools.ietf.org/html/rfc3164)
1367
1368 rfc5424 The RFC5424 syslog message format.
1369 (https://tools.ietf.org/html/rfc5424)
1370
Emeric Brun54648852020-07-06 15:54:06 +02001371 priority A message containing only a level plus syslog facility between
1372 angle brackets such as '<63>', followed by the text. The PID,
1373 date, time, process name and system name are omitted. This is
1374 designed to be used with a local log server.
1375
Willy Tarreaue8746a02018-11-12 08:45:00 +01001376 short A message containing only a level between angle brackets such as
1377 '<3>', followed by the text. The PID, date, time, process name
1378 and system name are omitted. This is designed to be used with a
1379 local log server. This format is compatible with what the systemd
1380 logger consumes.
1381
Emeric Brun54648852020-07-06 15:54:06 +02001382 timed A message containing only a level between angle brackets such as
1383 '<3>', followed by ISO date and by the text. The PID, process
1384 name and system name are omitted. This is designed to be
1385 used with a local log server.
1386
1387 iso A message containing only the ISO date, followed by the text.
1388 The PID, process name and system name are omitted. This is
1389 designed to be used with a local log server.
1390
Willy Tarreauc1b06452018-11-12 11:57:56 +01001391 raw A message containing only the text. The level, PID, date, time,
1392 process name and system name are omitted. This is designed to be
1393 used in containers or during development, where the severity only
1394 depends on the file descriptor used (stdout/stderr).
1395
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001396 <ranges> A list of comma-separated ranges to identify the logs to sample.
1397 This is used to balance the load of the logs to send to the log
1398 server. The limits of the ranges cannot be null. They are numbered
1399 from 1. The size or period (in number of logs) of the sample must be
1400 set with <sample_size> parameter.
1401
1402 <sample_size>
1403 The size of the sample in number of logs to consider when balancing
1404 their logging loads. It is used to balance the load of the logs to
1405 send to the syslog server. This size must be greater or equal to the
1406 maximum of the high limits of the ranges.
1407 (see also <ranges> parameter).
1408
Robert Tsai81ae1952007-12-05 10:47:29 +01001409 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001410
Willy Tarreaue8746a02018-11-12 08:45:00 +01001411 kern user mail daemon auth syslog lpr news
1412 uucp cron auth2 ftp ntp audit alert cron2
1413 local0 local1 local2 local3 local4 local5 local6 local7
1414
Willy Tarreauc1b06452018-11-12 11:57:56 +01001415 Note that the facility is ignored for the "short" and "raw"
1416 formats, but still required as a positional field. It is
1417 recommended to use "daemon" in this case to make it clear that
1418 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001419
1420 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001421 all messages are sent. If a maximum level is specified, only messages with a
1422 severity at least as important as this level will be sent. An optional minimum
1423 level can be specified. If it is set, logs emitted with a more severe level
1424 than this one will be capped to this level. This is used to avoid sending
1425 "emerg" messages on all terminals on some default syslog configurations.
1426 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001427
Cyril Bontédc4d9032012-04-08 21:57:39 +02001428 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001429
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001430log-send-hostname [<string>]
1431 Sets the hostname field in the syslog header. If optional "string" parameter
1432 is set the header is set to the string contents, otherwise uses the hostname
1433 of the system. Generally used if one is not relaying logs through an
1434 intermediate syslog server or for simply customizing the hostname printed in
1435 the logs.
1436
Kevinm48936af2010-12-22 16:08:21 +00001437log-tag <string>
1438 Sets the tag field in the syslog header to this string. It defaults to the
1439 program name as launched from the command line, which usually is "haproxy".
1440 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001441 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001442
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001443lua-load <file>
Thierry Fournier59f11be2020-11-29 00:37:41 +01001444 This global directive loads and executes a Lua file in the shared context
1445 that is visible to all threads. Any variable set in such a context is visible
1446 from any thread. This is the easiest and recommended way to load Lua programs
1447 but it will not scale well if a lot of Lua calls are performed, as only one
1448 thread may be running on the global state at a time. A program loaded this
1449 way will always see 0 in the "core.thread" variable. This directive can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001450 used multiple times.
1451
Thierry Fournier59f11be2020-11-29 00:37:41 +01001452lua-load-per-thread <file>
1453 This global directive loads and executes a Lua file into each started thread.
1454 Any global variable has a thread-local visibility so that each thread could
1455 see a different value. As such it is strongly recommended not to use global
1456 variables in programs loaded this way. An independent copy is loaded and
1457 initialized for each thread, everything is done sequentially and in the
1458 thread's numeric order from 1 to nbthread. If some operations need to be
1459 performed only once, the program should check the "core.thread" variable to
1460 figure what thread is being initialized. Programs loaded this way will run
1461 concurrently on all threads and will be highly scalable. This is the
1462 recommended way to load simple functions that register sample-fetches,
1463 converters, actions or services once it is certain the program doesn't depend
1464 on global variables. For the sake of simplicity, the directive is available
1465 even if only one thread is used and even if threads are disabled (in which
1466 case it will be equivalent to lua-load). This directive can be used multiple
1467 times.
1468
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +01001469lua-prepend-path <string> [<type>]
1470 Prepends the given string followed by a semicolon to Lua's package.<type>
1471 variable.
1472 <type> must either be "path" or "cpath". If <type> is not given it defaults
1473 to "path".
1474
1475 Lua's paths are semicolon delimited lists of patterns that specify how the
1476 `require` function attempts to find the source file of a library. Question
1477 marks (?) within a pattern will be replaced by module name. The path is
1478 evaluated left to right. This implies that paths that are prepended later
1479 will be checked earlier.
1480
1481 As an example by specifying the following path:
1482
1483 lua-prepend-path /usr/share/haproxy-lua/?/init.lua
1484 lua-prepend-path /usr/share/haproxy-lua/?.lua
1485
1486 When `require "example"` is being called Lua will first attempt to load the
1487 /usr/share/haproxy-lua/example.lua script, if that does not exist the
1488 /usr/share/haproxy-lua/example/init.lua will be attempted and the default
1489 paths if that does not exist either.
1490
1491 See https://www.lua.org/pil/8.1.html for the details within the Lua
1492 documentation.
1493
William Lallemand4cfede82017-11-24 22:02:34 +01001494master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001495 Master-worker mode. It is equivalent to the command line "-W" argument.
1496 This mode will launch a "master" which will monitor the "workers". Using
1497 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001498 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001499 or daemon mode. It is recommended to use this mode with multiprocess and
1500 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001501 By default, if a worker exits with a bad return code, in the case of a
1502 segfault for example, all workers will be killed, and the master will leave.
1503 It is convenient to combine this behavior with Restart=on-failure in a
1504 systemd unit file in order to relaunch the whole process. If you don't want
1505 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001506
William Lallemand4cfede82017-11-24 22:02:34 +01001507 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001508
William Lallemand27edc4b2019-05-07 17:49:33 +02001509mworker-max-reloads <number>
1510 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001511 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001512 number of reloads is greater than this number, the worker will receive a
1513 SIGTERM. This option helps to keep under control the number of workers.
1514 See also "show proc" in the Management Guide.
1515
Willy Tarreauf42d7942020-10-20 11:54:49 +02001516nbproc <number> (deprecated)
Willy Tarreau6a06a402007-07-15 20:15:28 +02001517 Creates <number> processes when going daemon. This requires the "daemon"
1518 mode. By default, only one process is created, which is the recommended mode
1519 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001520 process, it may be needed to fork multiple daemons. When set to a value
1521 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreauf42d7942020-10-20 11:54:49 +02001522 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. This directive is deprecated
1523 and scheduled for removal in 2.5. Please use "nbthread" instead. See also
1524 "daemon" and "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001525
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001526nbthread <number>
1527 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001528 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1529 "nbproc" historically used to be the only way to use multiple processors, it
1530 also involved a number of shortcomings related to the lack of synchronization
1531 between processes (health-checks, peers, stick-tables, stats, ...) which do
1532 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001533 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1534 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1535 when nbproc is not used, the default "nbthread" value is automatically set to
1536 the number of CPUs the process is bound to upon startup. This means that the
1537 thread count can easily be adjusted from the calling process using commands
1538 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1539 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001540
Willy Tarreau6a06a402007-07-15 20:15:28 +02001541pidfile <pidfile>
MIZUTA Takeshic32f3942020-08-26 13:46:19 +09001542 Writes PIDs of all daemons into file <pidfile> when daemon mode or writes PID
1543 of master process into file <pidfile> when master-worker mode. This option is
1544 equivalent to the "-p" command line argument. The file must be accessible to
1545 the user starting the process. See also "daemon" and "master-worker".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001546
Willy Tarreau119e50e2020-05-22 13:53:29 +02001547pp2-never-send-local
1548 A bug in the PROXY protocol v2 implementation was present in HAProxy up to
1549 version 2.1, causing it to emit a PROXY command instead of a LOCAL command
1550 for health checks. This is particularly minor but confuses some servers'
1551 logs. Sadly, the bug was discovered very late and revealed that some servers
1552 which possibly only tested their PROXY protocol implementation against
1553 HAProxy fail to properly handle the LOCAL command, and permanently remain in
1554 the "down" state when HAProxy checks them. When this happens, it is possible
1555 to enable this global option to revert to the older (bogus) behavior for the
1556 time it takes to contact the affected components' vendors and get them fixed.
1557 This option is disabled by default and acts on all servers having the
1558 "send-proxy-v2" statement.
1559
Willy Tarreau1d549722016-02-16 12:41:57 +01001560presetenv <name> <value>
1561 Sets environment variable <name> to value <value>. If the variable exists, it
1562 is NOT overwritten. The changes immediately take effect so that the next line
1563 in the configuration file sees the new value. See also "setenv", "resetenv",
1564 and "unsetenv".
1565
1566resetenv [<name> ...]
1567 Removes all environment variables except the ones specified in argument. It
1568 allows to use a clean controlled environment before setting new values with
1569 setenv or unsetenv. Please note that some internal functions may make use of
1570 some environment variables, such as time manipulation functions, but also
1571 OpenSSL or even external checks. This must be used with extreme care and only
1572 after complete validation. The changes immediately take effect so that the
1573 next line in the configuration file sees the new environment. See also
1574 "setenv", "presetenv", and "unsetenv".
1575
Christopher Fauletff4121f2017-11-22 16:38:49 +01001576stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001577 Limits the stats socket to a certain set of processes numbers. By default the
1578 stats socket is bound to all processes, causing a warning to be emitted when
1579 nbproc is greater than 1 because there is no way to select the target process
1580 when connecting. However, by using this setting, it becomes possible to pin
1581 the stats socket to a specific set of processes, typically the first one. The
1582 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001583 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001584 word size (32 or 64). Ranges can be partially defined. The higher bound can
1585 be omitted. In such case, it is replaced by the corresponding maximum
1586 value. A better option consists in using the "process" setting of the "stats
1587 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001588
Baptiste Assmann5626f482015-08-23 10:00:10 +02001589server-state-base <directory>
1590 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001591 file names which do not start with a '/'. See also "server-state-file",
1592 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001593
1594server-state-file <file>
1595 Specifies the path to the file containing state of servers. If the path starts
1596 with a slash ('/'), it is considered absolute, otherwise it is considered
1597 relative to the directory specified using "server-state-base" (if set) or to
1598 the current directory. Before reloading HAProxy, it is possible to save the
1599 servers' current state using the stats command "show servers state". The
1600 output of this command must be written in the file pointed by <file>. When
1601 starting up, before handling traffic, HAProxy will read, load and apply state
1602 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001603 configuration. See also "server-state-base" and "show servers state",
1604 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001605
Willy Tarreau13d2ba22021-03-26 11:38:08 +01001606set-var <var-name> <expr>
1607 Sets the process-wide variable '<var-name>' to the result of the evaluation
1608 of the sample expression <expr>. The variable '<var-name>' may only be a
1609 process-wide variable (using the 'proc.' prefix). It works exactly like the
1610 'set-var' action in TCP or HTTP rules except that the expression is evaluated
1611 at configuration parsing time and that the variable is instantly set. The
1612 sample fetch functions and converters permitted in the expression are only
1613 those using internal data, typically 'int(value)' or 'str(value)'. It's is
1614 possible to reference previously allocated variables as well. These variables
1615 will then be readable (and modifiable) from the regular rule sets.
1616
1617 Example:
1618 global
1619 set-var proc.current_state str(primary)
1620 set-var proc.prio int(100)
1621 set-var proc.threshold int(200),sub(proc.prio)
1622
Willy Tarreau1d549722016-02-16 12:41:57 +01001623setenv <name> <value>
1624 Sets environment variable <name> to value <value>. If the variable exists, it
1625 is overwritten. The changes immediately take effect so that the next line in
1626 the configuration file sees the new value. See also "presetenv", "resetenv",
1627 and "unsetenv".
1628
Willy Tarreau636848a2019-04-15 19:38:50 +02001629set-dumpable
1630 This option is better left disabled by default and enabled only upon a
William Dauchyec730982019-10-27 20:08:10 +01001631 developer's request. If it has been enabled, it may still be forcibly
1632 disabled by prefixing it with the "no" keyword. It has no impact on
1633 performance nor stability but will try hard to re-enable core dumps that were
1634 possibly disabled by file size limitations (ulimit -f), core size limitations
1635 (ulimit -c), or "dumpability" of a process after changing its UID/GID (such
1636 as /proc/sys/fs/suid_dumpable on Linux). Core dumps might still be limited by
1637 the current directory's permissions (check what directory the file is started
1638 from), the chroot directory's permission (it may be needed to temporarily
1639 disable the chroot directive or to move it to a dedicated writable location),
1640 or any other system-specific constraint. For example, some Linux flavours are
1641 notorious for replacing the default core file with a path to an executable
1642 not even installed on the system (check /proc/sys/kernel/core_pattern). Often,
1643 simply writing "core", "core.%p" or "/var/log/core/core.%p" addresses the
1644 issue. When trying to enable this option waiting for a rare issue to
1645 re-appear, it's often a good idea to first try to obtain such a dump by
1646 issuing, for example, "kill -11" to the haproxy process and verify that it
1647 leaves a core where expected when dying.
Willy Tarreau636848a2019-04-15 19:38:50 +02001648
Willy Tarreau610f04b2014-02-13 11:36:41 +01001649ssl-default-bind-ciphers <ciphers>
1650 This setting is only available when support for OpenSSL was built in. It sets
1651 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001652 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001653 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001654 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1655 information and recommendations see e.g.
1656 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1657 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1658 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1659 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001660
1661ssl-default-bind-ciphersuites <ciphersuites>
1662 This setting is only available when support for OpenSSL was built in and
1663 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1664 describing the list of cipher algorithms ("cipher suite") that are negotiated
1665 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1666 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001667 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1668 cipher configuration for TLSv1.2 and earlier, please check the
1669 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001670 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001671
Jerome Magninb203ff62020-04-03 15:28:22 +02001672ssl-default-bind-curves <curves>
1673 This setting is only available when support for OpenSSL was built in. It sets
1674 the default string describing the list of elliptic curves algorithms ("curve
1675 suite") that are negotiated during the SSL/TLS handshake with ECDHE. The format
1676 of the string is a colon-delimited list of curve name.
1677 Please check the "bind" keyword for more information.
1678
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001679ssl-default-bind-options [<option>]...
1680 This setting is only available when support for OpenSSL was built in. It sets
1681 default ssl-options to force on all "bind" lines. Please check the "bind"
1682 keyword to see available options.
1683
1684 Example:
1685 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001686 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001687
Willy Tarreau610f04b2014-02-13 11:36:41 +01001688ssl-default-server-ciphers <ciphers>
1689 This setting is only available when support for OpenSSL was built in. It
1690 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001691 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001692 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001693 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1694 information and recommendations see e.g.
1695 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1696 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1697 For TLSv1.3 cipher configuration, please check the
1698 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1699 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001700
1701ssl-default-server-ciphersuites <ciphersuites>
1702 This setting is only available when support for OpenSSL was built in and
1703 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1704 string describing the list of cipher algorithms that are negotiated during
1705 the TLSv1.3 handshake with the server, for all "server" lines which do not
1706 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001707 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1708 cipher configuration for TLSv1.2 and earlier, please check the
1709 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1710 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001711
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001712ssl-default-server-options [<option>]...
1713 This setting is only available when support for OpenSSL was built in. It sets
1714 default ssl-options to force on all "server" lines. Please check the "server"
1715 keyword to see available options.
1716
Remi Gacogne47783ef2015-05-29 15:53:22 +02001717ssl-dh-param-file <file>
1718 This setting is only available when support for OpenSSL was built in. It sets
1719 the default DH parameters that are used during the SSL/TLS handshake when
1720 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001721 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001722 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001723 are not specified either by using ssl-dh-param-file or by setting them
1724 directly in the certificate file, pre-generated DH parameters of the size
1725 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1726 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001727 Custom DH parameters may be generated by using the OpenSSL command
1728 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1729 parameters should not be considered secure anymore.
1730
William Lallemand8e8581e2020-10-20 17:36:46 +02001731ssl-load-extra-del-ext
1732 This setting allows to configure the way HAProxy does the lookup for the
1733 extra SSL files. By default HAProxy adds a new extension to the filename.
William Lallemand089c1382020-10-23 17:35:12 +02001734 (ex: with "foobar.crt" load "foobar.crt.key"). With this option enabled,
William Lallemand8e8581e2020-10-20 17:36:46 +02001735 HAProxy removes the extension before adding the new one (ex: with
William Lallemand089c1382020-10-23 17:35:12 +02001736 "foobar.crt" load "foobar.key").
1737
1738 Your crt file must have a ".crt" extension for this option to work.
William Lallemand8e8581e2020-10-20 17:36:46 +02001739
1740 This option is not compatible with bundle extensions (.ecdsa, .rsa. .dsa)
1741 and won't try to remove them.
1742
1743 This option is disabled by default. See also "ssl-load-extra-files".
1744
William Lallemand4c5adbf2020-02-24 14:23:22 +01001745ssl-load-extra-files <none|all|bundle|sctl|ocsp|issuer|key>*
William Lallemand3af48e72020-02-03 17:15:52 +01001746 This setting alters the way HAProxy will look for unspecified files during
Jerome Magnin587be9c2020-09-07 11:55:57 +02001747 the loading of the SSL certificates associated to "bind" lines. It does not
1748 apply to certificates used for client authentication on "server" lines.
William Lallemand3af48e72020-02-03 17:15:52 +01001749
1750 By default, HAProxy discovers automatically a lot of files not specified in
1751 the configuration, and you may want to disable this behavior if you want to
1752 optimize the startup time.
1753
1754 "none": Only load the files specified in the configuration. Don't try to load
1755 a certificate bundle if the file does not exist. In the case of a directory,
1756 it won't try to bundle the certificates if they have the same basename.
1757
1758 "all": This is the default behavior, it will try to load everything,
William Lallemand4c5adbf2020-02-24 14:23:22 +01001759 bundles, sctl, ocsp, issuer, key.
William Lallemand3af48e72020-02-03 17:15:52 +01001760
1761 "bundle": When a file specified in the configuration does not exist, HAProxy
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001762 will try to load a "cert bundle".
1763
1764 Starting from HAProxy 2.3, the bundles are not loaded in the same OpenSSL
1765 certificate store, instead it will loads each certificate in a separate
1766 store which is equivalent to declaring multiple "crt". OpenSSL 1.1.1 is
1767 required to achieve this. Which means that bundles are now used only for
1768 backward compatibility and are not mandatory anymore to do an hybrid RSA/ECC
1769 bind configuration..
1770
1771 To associate these PEM files into a "cert bundle" that is recognized by
1772 haproxy, they must be named in the following way: All PEM files that are to
1773 be bundled must have the same base name, with a suffix indicating the key
1774 type. Currently, three suffixes are supported: rsa, dsa and ecdsa. For
1775 example, if www.example.com has two PEM files, an RSA file and an ECDSA
1776 file, they must be named: "example.pem.rsa" and "example.pem.ecdsa". The
1777 first part of the filename is arbitrary; only the suffix matters. To load
1778 this bundle into haproxy, specify the base name only:
1779
1780 Example : bind :8443 ssl crt example.pem
1781
1782 Note that the suffix is not given to haproxy; this tells haproxy to look for
1783 a cert bundle.
1784
1785 HAProxy will load all PEM files in the bundle as if they were configured
1786 separately in several "crt".
1787
1788 The bundle loading does not have an impact anymore on the directory loading
1789 since files are loading separately.
1790
1791 On the CLI, bundles are seen as separate files, and the bundle extension is
1792 required to commit them.
1793
William Dauchy57dd6f12020-10-06 15:22:37 +02001794 OCSP files (.ocsp), issuer files (.issuer), Certificate Transparency (.sctl)
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001795 as well as private keys (.key) are supported with multi-cert bundling.
William Lallemand3af48e72020-02-03 17:15:52 +01001796
1797 "sctl": Try to load "<basename>.sctl" for each crt keyword.
1798
1799 "ocsp": Try to load "<basename>.ocsp" for each crt keyword.
1800
1801 "issuer": Try to load "<basename>.issuer" if the issuer of the OCSP file is
1802 not provided in the PEM file.
1803
William Lallemand4c5adbf2020-02-24 14:23:22 +01001804 "key": If the private key was not provided by the PEM file, try to load a
1805 file "<basename>.key" containing a private key.
1806
William Lallemand3af48e72020-02-03 17:15:52 +01001807 The default behavior is "all".
1808
1809 Example:
1810 ssl-load-extra-files bundle sctl
1811 ssl-load-extra-files sctl ocsp issuer
1812 ssl-load-extra-files none
1813
1814 See also: "crt", section 5.1 about bind options.
1815
Emeric Brun850efd52014-01-29 12:24:34 +01001816ssl-server-verify [none|required]
1817 The default behavior for SSL verify on servers side. If specified to 'none',
1818 servers certificates are not verified. The default is 'required' except if
1819 forced using cmdline option '-dV'.
1820
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001821ssl-skip-self-issued-ca
Daniel Corbett67a82712020-07-06 23:01:19 -04001822 Self issued CA, aka x509 root CA, is the anchor for chain validation: as a
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001823 server is useless to send it, client must have it. Standard configuration
1824 need to not include such CA in PEM file. This option allows you to keep such
1825 CA in PEM file without sending it to the client. Use case is to provide
1826 issuer for ocsp without the need for '.issuer' file and be able to share it
1827 with 'issuers-chain-path'. This concerns all certificates without intermediate
1828 certificates. It's useless for BoringSSL, .issuer is ignored because ocsp
William Lallemand9a1d8392020-08-10 17:28:23 +02001829 bits does not need it. Requires at least OpenSSL 1.0.2.
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001830
Willy Tarreauabb175f2012-09-24 12:43:26 +02001831stats socket [<address:port>|<path>] [param*]
1832 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1833 Connections to this socket will return various statistics outputs and even
1834 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001835 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001836 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001837
Willy Tarreauabb175f2012-09-24 12:43:26 +02001838 All parameters supported by "bind" lines are supported, for instance to
1839 restrict access to some users or their access rights. Please consult
1840 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001841
1842stats timeout <timeout, in milliseconds>
1843 The default timeout on the stats socket is set to 10 seconds. It is possible
1844 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001845 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001846
1847stats maxconn <connections>
1848 By default, the stats socket is limited to 10 concurrent connections. It is
1849 possible to change this value with "stats maxconn".
1850
Willy Tarreau6a06a402007-07-15 20:15:28 +02001851uid <number>
Thayne McCombscdbcca92021-01-07 21:24:41 -07001852 Changes the process's user ID to <number>. It is recommended that the user ID
Willy Tarreau6a06a402007-07-15 20:15:28 +02001853 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1854 be started with superuser privileges in order to be able to switch to another
1855 one. See also "gid" and "user".
1856
1857ulimit-n <number>
1858 Sets the maximum number of per-process file-descriptors to <number>. By
1859 default, it is automatically computed, so it is recommended not to use this
1860 option.
1861
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001862unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1863 [ group <group> ] [ gid <gid> ]
1864
1865 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1866 This is mainly used to simplify declaration of those UNIX sockets and reduce
1867 the risk of errors, since those settings are most commonly required but are
1868 also process-specific. The <prefix> setting can be used to force all socket
1869 path to be relative to that directory. This might be needed to access another
1870 component's chroot. Note that those paths are resolved before haproxy chroots
1871 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1872 all have the same meaning as their homonyms used by the "bind" statement. If
1873 both are specified, the "bind" statement has priority, meaning that the
1874 "unix-bind" settings may be seen as process-wide default settings.
1875
Willy Tarreau1d549722016-02-16 12:41:57 +01001876unsetenv [<name> ...]
1877 Removes environment variables specified in arguments. This can be useful to
1878 hide some sensitive information that are occasionally inherited from the
1879 user's environment during some operations. Variables which did not exist are
1880 silently ignored so that after the operation, it is certain that none of
1881 these variables remain. The changes immediately take effect so that the next
1882 line in the configuration file will not see these variables. See also
1883 "setenv", "presetenv", and "resetenv".
1884
Willy Tarreau6a06a402007-07-15 20:15:28 +02001885user <user name>
1886 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1887 See also "uid" and "group".
1888
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001889node <name>
1890 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1891
1892 This statement is useful in HA configurations where two or more processes or
1893 servers share the same IP address. By setting a different node-name on all
1894 nodes, it becomes easy to immediately spot what server is handling the
1895 traffic.
1896
1897description <text>
1898 Add a text that describes the instance.
1899
1900 Please note that it is required to escape certain characters (# for example)
1901 and this text is inserted into a html page so you should avoid using
1902 "<" and ">" characters.
1903
Thomas Holmesdb04f192015-05-18 13:21:39 +0100190451degrees-data-file <file path>
1905 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001906 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001907
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001908 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001909 compiled with USE_51DEGREES.
1910
Ben Shillitof25e8e52016-12-02 14:25:37 +0000191151degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001912 A list of 51Degrees property names to be load from the dataset. A full list
1913 of names is available on the 51Degrees website:
1914 https://51degrees.com/resources/property-dictionary
1915
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001916 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001917 compiled with USE_51DEGREES.
1918
Dragan Dosen93b38d92015-06-29 16:43:25 +0200191951degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001920 A char that will be appended to every property value in a response header
1921 containing 51Degrees results. If not set that will be set as ','.
1922
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001923 Please note that this option is only available when haproxy has been
1924 compiled with USE_51DEGREES.
1925
192651degrees-cache-size <number>
1927 Sets the size of the 51Degrees converter cache to <number> entries. This
1928 is an LRU cache which reminds previous device detections and their results.
1929 By default, this cache is disabled.
1930
1931 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001932 compiled with USE_51DEGREES.
1933
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001934wurfl-data-file <file path>
1935 The path of the WURFL data file to provide device detection services. The
1936 file should be accessible by HAProxy with relevant permissions.
1937
1938 Please note that this option is only available when haproxy has been compiled
1939 with USE_WURFL=1.
1940
1941wurfl-information-list [<capability>]*
1942 A space-delimited list of WURFL capabilities, virtual capabilities, property
1943 names we plan to use in injected headers. A full list of capability and
1944 virtual capability names is available on the Scientiamobile website :
1945
1946 https://www.scientiamobile.com/wurflCapability
1947
1948 Valid WURFL properties are:
1949 - wurfl_id Contains the device ID of the matched device.
1950
1951 - wurfl_root_id Contains the device root ID of the matched
1952 device.
1953
1954 - wurfl_isdevroot Tells if the matched device is a root device.
1955 Possible values are "TRUE" or "FALSE".
1956
1957 - wurfl_useragent The original useragent coming with this
1958 particular web request.
1959
1960 - wurfl_api_version Contains a string representing the currently
1961 used Libwurfl API version.
1962
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001963 - wurfl_info A string containing information on the parsed
1964 wurfl.xml and its full path.
1965
1966 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1967 WURFL has been loaded successfully.
1968
1969 - wurfl_normalized_useragent The normalized useragent.
1970
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001971 Please note that this option is only available when haproxy has been compiled
1972 with USE_WURFL=1.
1973
1974wurfl-information-list-separator <char>
1975 A char that will be used to separate values in a response header containing
1976 WURFL results. If not set that a comma (',') will be used by default.
1977
1978 Please note that this option is only available when haproxy has been compiled
1979 with USE_WURFL=1.
1980
1981wurfl-patch-file [<file path>]
1982 A list of WURFL patch file paths. Note that patches are loaded during startup
1983 thus before the chroot.
1984
1985 Please note that this option is only available when haproxy has been compiled
1986 with USE_WURFL=1.
1987
paulborilebad132c2019-04-18 11:57:04 +02001988wurfl-cache-size <size>
1989 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1990 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001991 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001992 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001993
1994 Please note that this option is only available when haproxy has been compiled
1995 with USE_WURFL=1.
1996
William Dauchy0fec3ab2019-10-27 20:08:11 +01001997strict-limits
William Dauchya5194602020-03-28 19:29:58 +01001998 Makes process fail at startup when a setrlimit fails. Haproxy tries to set the
1999 best setrlimit according to what has been calculated. If it fails, it will
2000 emit a warning. This option is here to guarantee an explicit failure of
2001 haproxy when those limits fail. It is enabled by default. It may still be
2002 forcibly disabled by prefixing it with the "no" keyword.
William Dauchy0fec3ab2019-10-27 20:08:11 +01002003
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020043.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02002005-----------------------
2006
Willy Tarreaubeb859a2018-11-22 18:07:59 +01002007busy-polling
2008 In some situations, especially when dealing with low latency on processors
2009 supporting a variable frequency or when running inside virtual machines, each
2010 time the process waits for an I/O using the poller, the processor goes back
2011 to sleep or is offered to another VM for a long time, and it causes
2012 excessively high latencies. This option provides a solution preventing the
2013 processor from sleeping by always using a null timeout on the pollers. This
2014 results in a significant latency reduction (30 to 100 microseconds observed)
2015 at the expense of a risk to overheat the processor. It may even be used with
2016 threads, in which case improperly bound threads may heavily conflict,
2017 resulting in a worse performance and high values for the CPU stolen fields
2018 in "show info" output, indicating which threads are misconfigured. It is
2019 important not to let the process run on the same processor as the network
2020 interrupts when this option is used. It is also better to avoid using it on
2021 multiple CPU threads sharing the same core. This option is disabled by
2022 default. If it has been enabled, it may still be forcibly disabled by
2023 prefixing it with the "no" keyword. It is ignored by the "select" and
2024 "poll" pollers.
2025
William Dauchy3894d972019-12-28 15:36:02 +01002026 This option is automatically disabled on old processes in the context of
2027 seamless reload; it avoids too much cpu conflicts when multiple processes
2028 stay around for some time waiting for the end of their current connections.
2029
Willy Tarreau1746eec2014-04-25 10:46:47 +02002030max-spread-checks <delay in milliseconds>
2031 By default, haproxy tries to spread the start of health checks across the
2032 smallest health check interval of all the servers in a farm. The principle is
2033 to avoid hammering services running on the same server. But when using large
2034 check intervals (10 seconds or more), the last servers in the farm take some
2035 time before starting to be tested, which can be a problem. This parameter is
2036 used to enforce an upper bound on delay between the first and the last check,
2037 even if the servers' check intervals are larger. When servers run with
2038 shorter intervals, their intervals will be respected though.
2039
Willy Tarreau6a06a402007-07-15 20:15:28 +02002040maxconn <number>
2041 Sets the maximum per-process number of concurrent connections to <number>. It
2042 is equivalent to the command-line argument "-n". Proxies will stop accepting
2043 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02002044 automatically adjusted according to this value. See also "ulimit-n". Note:
2045 the "select" poller cannot reliably use more than 1024 file descriptors on
2046 some platforms. If your platform only supports select and reports "select
2047 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01002048 below 500 in general). If this value is not set, it will automatically be
2049 calculated based on the current file descriptors limit reported by the
2050 "ulimit -n" command, possibly reduced to a lower value if a memory limit
2051 is enforced, based on the buffer size, memory allocated to compression, SSL
2052 cache size, and use or not of SSL and the associated maxsslconn (which can
2053 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02002054
Willy Tarreau81c25d02011-09-07 15:17:21 +02002055maxconnrate <number>
2056 Sets the maximum per-process number of connections per second to <number>.
2057 Proxies will stop accepting connections when this limit is reached. It can be
2058 used to limit the global capacity regardless of each frontend capacity. It is
2059 important to note that this can only be used as a service protection measure,
2060 as there will not necessarily be a fair share between frontends when the
2061 limit is reached, so it's a good idea to also limit each frontend to some
2062 value close to its expected share. Also, lowering tune.maxaccept can improve
2063 fairness.
2064
William Lallemandd85f9172012-11-09 17:05:39 +01002065maxcomprate <number>
2066 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01002067 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01002068 level will be decreased during the session. If the maximum is reached at the
2069 beginning of a session, the session will not compress at all. If the maximum
2070 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01002071 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01002072 default value.
2073
William Lallemand072a2bf2012-11-20 17:01:01 +01002074maxcompcpuusage <number>
2075 Sets the maximum CPU usage HAProxy can reach before stopping the compression
2076 for new requests or decreasing the compression level of current requests.
2077 It works like 'maxcomprate' but measures CPU usage instead of incoming data
2078 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
2079 case of multiple processes (nbproc > 1), each process manages its individual
2080 usage. A value of 100 disable the limit. The default value is 100. Setting
2081 a lower value will prevent the compression work from slowing the whole
2082 process down and from introducing high latencies.
2083
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002084maxpipes <number>
2085 Sets the maximum per-process number of pipes to <number>. Currently, pipes
2086 are only used by kernel-based tcp splicing. Since a pipe contains two file
2087 descriptors, the "ulimit-n" value will be increased accordingly. The default
2088 value is maxconn/4, which seems to be more than enough for most heavy usages.
2089 The splice code dynamically allocates and releases pipes, and can fall back
2090 to standard copy, so setting this value too low may only impact performance.
2091
Willy Tarreau93e7c002013-10-07 18:51:07 +02002092maxsessrate <number>
2093 Sets the maximum per-process number of sessions per second to <number>.
2094 Proxies will stop accepting connections when this limit is reached. It can be
2095 used to limit the global capacity regardless of each frontend capacity. It is
2096 important to note that this can only be used as a service protection measure,
2097 as there will not necessarily be a fair share between frontends when the
2098 limit is reached, so it's a good idea to also limit each frontend to some
2099 value close to its expected share. Also, lowering tune.maxaccept can improve
2100 fairness.
2101
Willy Tarreau403edff2012-09-06 11:58:37 +02002102maxsslconn <number>
2103 Sets the maximum per-process number of concurrent SSL connections to
2104 <number>. By default there is no SSL-specific limit, which means that the
2105 global maxconn setting will apply to all connections. Setting this limit
2106 avoids having openssl use too much memory and crash when malloc returns NULL
2107 (since it unfortunately does not reliably check for such conditions). Note
2108 that the limit applies both to incoming and outgoing connections, so one
2109 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01002110 If this value is not set, but a memory limit is enforced, this value will be
2111 automatically computed based on the memory limit, maxconn, the buffer size,
2112 memory allocated to compression, SSL cache size, and use of SSL in either
2113 frontends, backends or both. If neither maxconn nor maxsslconn are specified
2114 when there is a memory limit, haproxy will automatically adjust these values
2115 so that 100% of the connections can be made over SSL with no risk, and will
2116 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02002117
Willy Tarreaue43d5322013-10-07 20:01:52 +02002118maxsslrate <number>
2119 Sets the maximum per-process number of SSL sessions per second to <number>.
2120 SSL listeners will stop accepting connections when this limit is reached. It
2121 can be used to limit the global SSL CPU usage regardless of each frontend
2122 capacity. It is important to note that this can only be used as a service
2123 protection measure, as there will not necessarily be a fair share between
2124 frontends when the limit is reached, so it's a good idea to also limit each
2125 frontend to some value close to its expected share. It is also important to
2126 note that the sessions are accounted before they enter the SSL stack and not
2127 after, which also protects the stack against bad handshakes. Also, lowering
2128 tune.maxaccept can improve fairness.
2129
William Lallemand9d5f5482012-11-07 16:12:57 +01002130maxzlibmem <number>
2131 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
2132 When the maximum amount is reached, future sessions will not compress as long
2133 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01002134 The default value is 0. The value is available in bytes on the UNIX socket
2135 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
2136 "ZlibMemUsage" in bytes.
2137
Willy Tarreau6a06a402007-07-15 20:15:28 +02002138noepoll
2139 Disables the use of the "epoll" event polling system on Linux. It is
2140 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01002141 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02002142
2143nokqueue
2144 Disables the use of the "kqueue" event polling system on BSD. It is
2145 equivalent to the command-line argument "-dk". The next polling system
2146 used will generally be "poll". See also "nopoll".
2147
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00002148noevports
2149 Disables the use of the event ports event polling system on SunOS systems
2150 derived from Solaris 10 and later. It is equivalent to the command-line
2151 argument "-dv". The next polling system used will generally be "poll". See
2152 also "nopoll".
2153
Willy Tarreau6a06a402007-07-15 20:15:28 +02002154nopoll
2155 Disables the use of the "poll" event polling system. It is equivalent to the
2156 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002157 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00002158 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
2159 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02002160
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002161nosplice
2162 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002163 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002164 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002165 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002166 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
2167 be used. This option makes it easier to globally disable kernel splicing in
2168 case of doubt. See also "option splice-auto", "option splice-request" and
2169 "option splice-response".
2170
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002171nogetaddrinfo
2172 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
2173 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
2174
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00002175noreuseport
2176 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
2177 command line argument "-dR".
2178
Willy Tarreaud2d33482019-04-25 17:09:07 +02002179profiling.tasks { auto | on | off }
2180 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
2181 the profiling automatically turns on a thread when it starts to suffer from
2182 an average latency of 1000 microseconds or higher as reported in the
2183 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05002184 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02002185 loops so it does not vary quickly and tends to significantly smooth short
2186 spikes). It may also spontaneously trigger from time to time on overloaded
2187 systems, containers, or virtual machines, or when the system swaps (which
2188 must absolutely never happen on a load balancer).
2189
2190 CPU profiling per task can be very convenient to report where the time is
2191 spent and which requests have what effect on which other request. Enabling
2192 it will typically affect the overall's performance by less than 1%, thus it
2193 is recommended to leave it to the default 'auto' value so that it only
2194 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01002195 supporting the clock_gettime(2) syscall with clock identifiers
2196 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
2197 be zero. This option may be changed at run time using "set profiling" on the
2198 CLI.
2199
Willy Tarreaufe255b72007-10-14 23:09:26 +02002200spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09002201 Sometimes it is desirable to avoid sending agent and health checks to
2202 servers at exact intervals, for instance when many logical servers are
2203 located on the same physical server. With the help of this parameter, it
2204 becomes possible to add some randomness in the check interval between 0
2205 and +/- 50%. A value between 2 and 5 seems to show good results. The
2206 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02002207
Davor Ocelice9ed2812017-12-25 17:49:28 +01002208ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00002209 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01002210 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00002211 multiple times, it will simply enable multiple crypto engines. Referencing an
2212 unsupported engine will prevent haproxy from starting. Note that many engines
2213 will lead to lower HTTPS performance than pure software with recent
2214 processors. The optional command "algo" sets the default algorithms an ENGINE
2215 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01002216 of "ALL" uses the engine for all cryptographic operations. If no list of
2217 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00002218 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
2219 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
2220 openssl configuration file uses:
2221 https://www.openssl.org/docs/man1.0.2/apps/config.html
2222
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00002223ssl-mode-async
2224 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02002225 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00002226 implementation supports a maximum of 32 engines. The Openssl ASYNC API
2227 doesn't support moving read/write buffers and is not compliant with
2228 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05002229 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00002230 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00002231
Willy Tarreau33cb0652014-12-23 22:52:37 +01002232tune.buffers.limit <number>
2233 Sets a hard limit on the number of buffers which may be allocated per process.
2234 The default value is zero which means unlimited. The minimum non-zero value
2235 will always be greater than "tune.buffers.reserve" and should ideally always
2236 be about twice as large. Forcing this value can be particularly useful to
2237 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01002238 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01002239 another one to be released by another session. Since buffers are dynamically
2240 allocated and released, the waiting time is very short and not perceptible
2241 provided that limits remain reasonable. In fact sometimes reducing the limit
2242 may even increase performance by increasing the CPU cache's efficiency. Tests
2243 have shown good results on average HTTP traffic with a limit to 1/10 of the
2244 expected global maxconn setting, which also significantly reduces memory
2245 usage. The memory savings come from the fact that a number of connections
2246 will not allocate 2*tune.bufsize. It is best not to touch this value unless
2247 advised to do so by an haproxy core developer.
2248
Willy Tarreau1058ae72014-12-23 22:40:40 +01002249tune.buffers.reserve <number>
2250 Sets the number of buffers which are pre-allocated and reserved for use only
2251 during memory shortage conditions resulting in failed memory allocations. The
2252 minimum value is 2 and is also the default. There is no reason a user would
2253 want to change this value, it's mostly aimed at haproxy core developers.
2254
Willy Tarreau27a674e2009-08-17 07:23:33 +02002255tune.bufsize <number>
2256 Sets the buffer size to this size (in bytes). Lower values allow more
2257 sessions to coexist in the same amount of RAM, and higher values allow some
2258 applications with very large cookies to work. The default value is 16384 and
2259 can be changed at build time. It is strongly recommended not to change this
2260 from the default value, as very low values will break some services such as
2261 statistics, and values larger than default size will increase memory usage,
2262 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01002263 parameter should be decreased by the same factor as this one is increased. In
2264 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
2265 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04002266 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01002267 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
2268 value set using this parameter will automatically be rounded up to the next
2269 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02002270
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +01002271tune.chksize <number> (deprecated)
2272 This option is deprecated and ignored.
Willy Tarreau43961d52010-10-04 20:39:20 +02002273
William Lallemandf3747832012-11-09 12:33:10 +01002274tune.comp.maxlevel <number>
2275 Sets the maximum compression level. The compression level affects CPU
2276 usage during compression. This value affects CPU usage during compression.
2277 Each session using compression initializes the compression algorithm with
2278 this value. The default value is 1.
2279
Willy Tarreauc299e1e2019-02-27 11:35:12 +01002280tune.fail-alloc
2281 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
2282 allocation attempt fails. Must be between 0 (no failure) and 100 (no
2283 success). This is useful to debug and make sure memory failures are handled
2284 gracefully.
2285
Willy Tarreaubc52bec2020-06-18 08:58:47 +02002286tune.fd.edge-triggered { on | off } [ EXPERIMENTAL ]
2287 Enables ('on') or disables ('off') the edge-triggered polling mode for FDs
2288 that support it. This is currently only support with epoll. It may noticeably
2289 reduce the number of epoll_ctl() calls and slightly improve performance in
2290 certain scenarios. This is still experimental, it may result in frozen
2291 connections if bugs are still present, and is disabled by default.
2292
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02002293tune.h2.header-table-size <number>
2294 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
2295 cannot be larger than 65536 bytes. A larger value may help certain clients
2296 send more compact requests, depending on their capabilities. This amount of
2297 memory is consumed for each HTTP/2 connection. It is recommended not to
2298 change it.
2299
Willy Tarreaue6baec02017-07-27 11:45:11 +02002300tune.h2.initial-window-size <number>
2301 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01002302 can upload before waiting for an acknowledgment from haproxy. This setting
2303 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02002304 The default value is 65535, which roughly allows up to 5 Mbps of upload
2305 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
2306 over a 1-ms local network. It can make sense to increase this value to allow
2307 faster uploads, or to reduce it to increase fairness when dealing with many
2308 clients. It doesn't affect resource usage.
2309
Willy Tarreau5242ef82017-07-27 11:47:28 +02002310tune.h2.max-concurrent-streams <number>
2311 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
2312 number of outstanding requests on a single connection). The default value is
2313 100. A larger one may slightly improve page load time for complex sites when
2314 visited over high latency networks, but increases the amount of resources a
2315 single client may allocate. A value of zero disables the limit so a single
2316 client may create as many streams as allocatable by haproxy. It is highly
2317 recommended not to change this value.
2318
Willy Tarreaua24b35c2019-02-21 13:24:36 +01002319tune.h2.max-frame-size <number>
2320 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
2321 receive to its peers. The default value is the largest between 16384 and the
2322 buffer size (tune.bufsize). In any case, haproxy will not announce support
2323 for frame sizes larger than buffers. The main purpose of this setting is to
2324 allow to limit the maximum frame size setting when using large buffers. Too
2325 large frame sizes might have performance impact or cause some peers to
2326 misbehave. It is highly recommended not to change this value.
2327
Willy Tarreau193b8c62012-11-22 00:17:38 +01002328tune.http.cookielen <number>
2329 Sets the maximum length of captured cookies. This is the maximum value that
2330 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
2331 will automatically be truncated to this one. It is important not to set too
2332 high a value because all cookie captures still allocate this size whatever
2333 their configured value (they share a same pool). This value is per request
2334 per response, so the memory allocated is twice this value per connection.
2335 When not specified, the limit is set to 63 characters. It is recommended not
2336 to change this value.
2337
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002338tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002339 Sets the maximum length of request URI in logs. This prevents truncating long
2340 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002341 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002342 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002343 configuration directives too.
2344 The default value is 1024.
2345
Willy Tarreauac1932d2011-10-24 19:14:41 +02002346tune.http.maxhdr <number>
2347 Sets the maximum number of headers in a request. When a request comes with a
2348 number of headers greater than this value (including the first line), it is
2349 rejected with a "400 Bad Request" status code. Similarly, too large responses
2350 are blocked with "502 Bad Gateway". The default value is 101, which is enough
2351 for all usages, considering that the widely deployed Apache server uses the
2352 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02002353 a buggy application to work by the time it gets fixed. The accepted range is
2354 1..32767. Keep in mind that each new header consumes 32bits of memory for
2355 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02002356
Willy Tarreau76cc6992020-07-01 18:49:24 +02002357tune.idle-pool.shared { on | off }
2358 Enables ('on') or disables ('off') sharing of idle connection pools between
2359 threads for a same server. The default is to share them between threads in
2360 order to minimize the number of persistent connections to a server, and to
2361 optimize the connection reuse rate. But to help with debugging or when
2362 suspecting a bug in HAProxy around connection reuse, it can be convenient to
2363 forcefully disable this idle pool sharing between multiple threads, and force
Willy Tarreau0784db82021-02-19 11:45:22 +01002364 this option to "off". The default is on. It is strongly recommended against
2365 disabling this option without setting a conservative value on "pool-low-conn"
2366 for all servers relying on connection reuse to achieve a high performance
2367 level, otherwise connections might be closed very often as the thread count
2368 increases.
Willy Tarreau76cc6992020-07-01 18:49:24 +02002369
Willy Tarreau7e312732014-02-12 16:35:14 +01002370tune.idletimer <timeout>
2371 Sets the duration after which haproxy will consider that an empty buffer is
2372 probably associated with an idle stream. This is used to optimally adjust
2373 some packet sizes while forwarding large and small data alternatively. The
2374 decision to use splice() or to send large buffers in SSL is modulated by this
2375 parameter. The value is in milliseconds between 0 and 65535. A value of zero
2376 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01002377 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05002378 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01002379 tune.ssl.maxrecord below.
2380
Willy Tarreau7ac908b2019-02-27 12:02:18 +01002381tune.listener.multi-queue { on | off }
2382 Enables ('on') or disables ('off') the listener's multi-queue accept which
2383 spreads the incoming traffic to all threads a "bind" line is allowed to run
2384 on instead of taking them for itself. This provides a smoother traffic
2385 distribution and scales much better, especially in environments where threads
2386 may be unevenly loaded due to external activity (network interrupts colliding
2387 with one thread for example). This option is enabled by default, but it may
2388 be forcefully disabled for troubleshooting or for situations where it is
2389 estimated that the operating system already provides a good enough
2390 distribution and connections are extremely short-lived.
2391
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002392tune.lua.forced-yield <number>
2393 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01002394 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002395 HAProxy scheduler to process other tasks like accepting connections or
2396 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01002397 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002398 lowered. If the Lua code is quite long and its result is absolutely required
2399 to process the data, the <number> can be increased.
2400
Willy Tarreau32f61e22015-03-18 17:54:59 +01002401tune.lua.maxmem
2402 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
2403 default it is zero which means unlimited. It is important to set a limit to
2404 ensure that a bug in a script will not result in the system running out of
2405 memory.
2406
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002407tune.lua.session-timeout <timeout>
2408 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002409 preventing infinite loops or spending too much time in Lua. This timeout
2410 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002411 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002412
2413tune.lua.task-timeout <timeout>
2414 Purpose is the same as "tune.lua.session-timeout", but this timeout is
2415 dedicated to the tasks. By default, this timeout isn't set because a task may
2416 remain alive during of the lifetime of HAProxy. For example, a task used to
2417 check servers.
2418
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002419tune.lua.service-timeout <timeout>
2420 This is the execution timeout for the Lua services. This is useful for
2421 preventing infinite loops or spending too much time in Lua. This timeout
2422 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002423 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002424
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002425tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01002426 Sets the maximum number of consecutive connections a process may accept in a
2427 row before switching to other work. In single process mode, higher numbers
Willy Tarreau66161322021-02-19 15:50:27 +01002428 used to give better performance at high connection rates, though this is not
2429 the case anymore with the multi-queue. This value applies individually to
2430 each listener, so that the number of processes a listener is bound to is
2431 taken into account. This value defaults to 4 which showed best results. If a
2432 significantly higher value was inherited from an ancient config, it might be
2433 worth removing it as it will both increase performance and lower response
2434 time. In multi-process mode, it is divided by twice the number of processes
2435 the listener is bound to. Setting this value to -1 completely disables the
2436 limitation. It should normally not be needed to tweak this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002437
2438tune.maxpollevents <number>
2439 Sets the maximum amount of events that can be processed at once in a call to
2440 the polling system. The default value is adapted to the operating system. It
2441 has been noticed that reducing it below 200 tends to slightly decrease
2442 latency at the expense of network bandwidth, and increasing it above 200
2443 tends to trade latency for slightly increased bandwidth.
2444
Willy Tarreau27a674e2009-08-17 07:23:33 +02002445tune.maxrewrite <number>
2446 Sets the reserved buffer space to this size in bytes. The reserved space is
2447 used for header rewriting or appending. The first reads on sockets will never
2448 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
2449 bufsize, though that does not make much sense since there are rarely large
2450 numbers of headers to add. Setting it too high prevents processing of large
2451 requests or responses. Setting it too low prevents addition of new headers
2452 to already large requests or to POST requests. It is generally wise to set it
2453 to about 1024. It is automatically readjusted to half of bufsize if it is
2454 larger than that. This means you don't have to worry about it when changing
2455 bufsize.
2456
Willy Tarreauf3045d22015-04-29 16:24:50 +02002457tune.pattern.cache-size <number>
2458 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
2459 cache which reminds previous lookups and their results. It is used by ACLs
2460 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
2461 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
2462 strings. It applies to pattern expressions which means that it will be able
2463 to memorize the result of a lookup among all the patterns specified on a
2464 configuration line (including all those loaded from files). It automatically
2465 invalidates entries which are updated using HTTP actions or on the CLI. The
2466 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau403bfbb2019-10-23 06:59:31 +02002467 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
2468 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02002469 risk of collision in this cache, which is in the order of the size of the
2470 cache divided by 2^64. Typically, at 10000 requests per second with the
2471 default cache size of 10000 entries, there's 1% chance that a brute force
2472 attack could cause a single collision after 60 years, or 0.1% after 6 years.
2473 This is considered much lower than the risk of a memory corruption caused by
2474 aging components. If this is not acceptable, the cache can be disabled by
2475 setting this parameter to 0.
2476
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02002477tune.pipesize <number>
2478 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
2479 are the default size for the system. But sometimes when using TCP splicing,
2480 it can improve performance to increase pipe sizes, especially if it is
2481 suspected that pipes are not filled and that many calls to splice() are
2482 performed. This has an impact on the kernel's memory footprint, so this must
2483 not be changed if impacts are not understood.
2484
Olivier Houchard88698d92019-04-16 19:07:22 +02002485tune.pool-high-fd-ratio <number>
2486 This setting sets the max number of file descriptors (in percentage) used by
2487 haproxy globally against the maximum number of file descriptors haproxy can
2488 use before we start killing idle connections when we can't reuse a connection
2489 and we have to create a new one. The default is 25 (one quarter of the file
2490 descriptor will mean that roughly half of the maximum front connections can
2491 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05002492 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02002493
Willy Tarreau83ca3052020-07-01 18:30:16 +02002494tune.pool-low-fd-ratio <number>
2495 This setting sets the max number of file descriptors (in percentage) used by
2496 haproxy globally against the maximum number of file descriptors haproxy can
2497 use before we stop putting connection into the idle pool for reuse. The
2498 default is 20.
2499
Willy Tarreaue803de22010-01-21 17:43:04 +01002500tune.rcvbuf.client <number>
2501tune.rcvbuf.server <number>
2502 Forces the kernel socket receive buffer size on the client or the server side
2503 to the specified value in bytes. This value applies to all TCP/HTTP frontends
2504 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002505 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002506 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002507 order to save kernel memory by preventing it from buffering too large amounts
2508 of received data. Lower values will significantly increase CPU usage though.
2509
Willy Tarreaub22fc302015-12-14 12:04:35 +01002510tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002511 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01002512 socket buffers. One of them is that a read returns more than <recv_enough>
2513 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
2514 may be changed by this setting to better deal with workloads involving lots
2515 of short messages such as telnet or SSH sessions.
2516
Olivier Houchard1599b802018-05-24 18:59:04 +02002517tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05002518 Sets the maximum amount of task that can be processed at once when running
Willy Tarreau060a7612021-03-10 11:06:26 +01002519 tasks. The default value depends on the number of threads but sits between 35
2520 and 280, which tend to show the highest request rates and lowest latencies.
2521 Increasing it may incur latency when dealing with I/Os, making it too small
2522 can incur extra overhead. Higher thread counts benefit from lower values.
2523 When experimenting with much larger values, it may be useful to also enable
2524 tune.sched.low-latency and possibly tune.fd.edge-triggered to limit the
2525 maximum latency to the lowest possible.
Willy Tarreaue7723bd2020-06-24 11:11:02 +02002526
2527tune.sched.low-latency { on | off }
2528 Enables ('on') or disables ('off') the low-latency task scheduler. By default
2529 haproxy processes tasks from several classes one class at a time as this is
2530 the most efficient. But when running with large values of tune.runqueue-depth
2531 this can have a measurable effect on request or connection latency. When this
2532 low-latency setting is enabled, tasks of lower priority classes will always
2533 be executed before other ones if they exist. This will permit to lower the
2534 maximum latency experienced by new requests or connections in the middle of
2535 massive traffic, at the expense of a higher impact on this large traffic.
2536 For regular usage it is better to leave this off. The default value is off.
Olivier Houchard1599b802018-05-24 18:59:04 +02002537
Willy Tarreaue803de22010-01-21 17:43:04 +01002538tune.sndbuf.client <number>
2539tune.sndbuf.server <number>
2540 Forces the kernel socket send buffer size on the client or the server side to
2541 the specified value in bytes. This value applies to all TCP/HTTP frontends
2542 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002543 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002544 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002545 order to save kernel memory by preventing it from buffering too large amounts
2546 of received data. Lower values will significantly increase CPU usage though.
2547 Another use case is to prevent write timeouts with extremely slow clients due
2548 to the kernel waiting for a large part of the buffer to be read before
2549 notifying haproxy again.
2550
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002551tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01002552 Sets the size of the global SSL session cache, in a number of blocks. A block
William Dauchy9a4bbfe2021-02-12 15:58:46 +01002553 is large enough to contain an encoded session without peer certificate. An
2554 encoded session with peer certificate is stored in multiple blocks depending
2555 on the size of the peer certificate. A block uses approximately 200 bytes of
2556 memory (based on `sizeof(struct sh_ssl_sess_hdr) + SHSESS_BLOCK_MIN_SIZE`
2557 calculation used for `shctx_init` function). The default value may be forced
2558 at build time, otherwise defaults to 20000. When the cache is full, the most
2559 idle entries are purged and reassigned. Higher values reduce the occurrence
2560 of such a purge, hence the number of CPU-intensive SSL handshakes by ensuring
2561 that all users keep their session as long as possible. All entries are
2562 pre-allocated upon startup and are shared between all processes if "nbproc"
2563 is greater than 1. Setting this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002564
Emeric Brun8dc60392014-05-09 13:52:00 +02002565tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02002566 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02002567 should normally not be used since it will force many renegotiations due to
2568 clients hitting a random process. But it may be required on some operating
2569 systems where none of the SSL cache synchronization method may be used. In
2570 this case, adding a first layer of hash-based load balancing before the SSL
2571 layer might limit the impact of the lack of session sharing.
2572
William Lallemand7d42ef52020-07-06 11:41:30 +02002573tune.ssl.keylog { on | off }
2574 This option activates the logging of the TLS keys. It should be used with
2575 care as it will consume more memory per SSL session and could decrease
2576 performances. This is disabled by default.
2577
2578 These sample fetches should be used to generate the SSLKEYLOGFILE that is
2579 required to decipher traffic with wireshark.
2580
2581 https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/Key_Log_Format
2582
2583 The SSLKEYLOG is a series of lines which are formatted this way:
2584
2585 <Label> <space> <ClientRandom> <space> <Secret>
2586
2587 The ClientRandom is provided by the %[ssl_fc_client_random,hex] sample
2588 fetch, the secret and the Label could be find in the array below. You need
2589 to generate a SSLKEYLOGFILE with all the labels in this array.
2590
2591 The following sample fetches are hexadecimal strings and does not need to be
2592 converted.
2593
2594 SSLKEYLOGFILE Label | Sample fetches for the Secrets
2595 --------------------------------|-----------------------------------------
2596 CLIENT_EARLY_TRAFFIC_SECRET | %[ssl_fc_client_early_traffic_secret]
2597 CLIENT_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_client_handshake_traffic_secret]
2598 SERVER_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_server_handshake_traffic_secret]
2599 CLIENT_TRAFFIC_SECRET_0 | %[ssl_fc_client_traffic_secret_0]
2600 SERVER_TRAFFIC_SECRET_0 | %[ssl_fc_server_traffic_secret_0]
William Lallemandd742b6c2020-07-07 10:14:56 +02002601 EXPORTER_SECRET | %[ssl_fc_exporter_secret]
2602 EARLY_EXPORTER_SECRET | %[ssl_fc_early_exporter_secret]
William Lallemand7d42ef52020-07-06 11:41:30 +02002603
2604 This is only available with OpenSSL 1.1.1, and useful with TLS1.3 session.
2605
2606 If you want to generate the content of a SSLKEYLOGFILE with TLS < 1.3, you
2607 only need this line:
2608
2609 "CLIENT_RANDOM %[ssl_fc_client_random,hex] %[ssl_fc_session_key,hex]"
2610
Emeric Brun4f65bff2012-11-16 15:11:00 +01002611tune.ssl.lifetime <timeout>
2612 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002613 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01002614 does not guarantee that sessions will last that long, because if the cache is
2615 full, the longest idle sessions will be purged despite their configured
2616 lifetime. The real usefulness of this setting is to prevent sessions from
2617 being used for too long.
2618
Willy Tarreaubfd59462013-02-21 07:46:09 +01002619tune.ssl.maxrecord <number>
2620 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
2621 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
2622 data only once it has received a full record. With large records, it means
2623 that clients might have to download up to 16kB of data before starting to
2624 process them. Limiting the value can improve page load times on browsers
2625 located over high latency or low bandwidth networks. It is suggested to find
2626 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
2627 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
2628 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
2629 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002630 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01002631 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01002632
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002633tune.ssl.default-dh-param <number>
2634 Sets the maximum size of the Diffie-Hellman parameters used for generating
2635 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
2636 final size will try to match the size of the server's RSA (or DSA) key (e.g,
2637 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
Willy Tarreau3ba77d22020-05-08 09:31:18 +02002638 this maximum value. Default value if 2048. Only 1024 or higher values are
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002639 allowed. Higher values will increase the CPU load, and values greater than
2640 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02002641 used if static Diffie-Hellman parameters are supplied either directly
2642 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002643
Christopher Faulet31af49d2015-06-09 17:29:50 +02002644tune.ssl.ssl-ctx-cache-size <number>
2645 Sets the size of the cache used to store generated certificates to <number>
2646 entries. This is a LRU cache. Because generating a SSL certificate
2647 dynamically is expensive, they are cached. The default cache size is set to
2648 1000 entries.
2649
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01002650tune.ssl.capture-cipherlist-size <number>
2651 Sets the maximum size of the buffer used for capturing client-hello cipher
2652 list. If the value is 0 (default value) the capture is disabled, otherwise
2653 a buffer is allocated for each SSL/TLS connection.
2654
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002655tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002656tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002657tune.vars.reqres-max-size <size>
2658tune.vars.sess-max-size <size>
2659tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002660 These five tunes help to manage the maximum amount of memory used by the
2661 variables system. "global" limits the overall amount of memory available for
2662 all scopes. "proc" limits the memory for the process scope, "sess" limits the
2663 memory for the session scope, "txn" for the transaction scope, and "reqres"
2664 limits the memory for each request or response processing.
2665 Memory accounting is hierarchical, meaning more coarse grained limits include
2666 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
2667 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002668
Daniel Schneller0b547052016-03-21 20:46:57 +01002669 For example, when "tune.vars.sess-max-size" is limited to 100,
2670 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
2671 100 either. If we create a variable "txn.var" that contains 100 bytes,
2672 all available space is consumed.
2673 Notice that exceeding the limits at runtime will not result in an error
2674 message, but values might be cut off or corrupted. So make sure to accurately
2675 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002676
William Lallemanda509e4c2012-11-07 16:54:34 +01002677tune.zlib.memlevel <number>
2678 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002679 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002680 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002681 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002682 between 1 and 9. The default value is 8.
2683
2684tune.zlib.windowsize <number>
2685 Sets the window size (the size of the history buffer) as a parameter of the
2686 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002687 in better compression at the expense of memory usage. Can be a value between
2688 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002689
Willy Tarreauc57f0e22009-05-10 13:12:33 +020026903.3. Debugging
2691--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002692
Willy Tarreau6a06a402007-07-15 20:15:28 +02002693quiet
2694 Do not display any message during startup. It is equivalent to the command-
2695 line argument "-q".
2696
Willy Tarreau3eb10b82020-04-15 16:42:39 +02002697zero-warning
2698 When this option is set, haproxy will refuse to start if any warning was
2699 emitted while processing the configuration. It is highly recommended to set
2700 this option on configurations that are not changed often, as it helps detect
2701 subtle mistakes and keep the configuration clean and forward-compatible. Note
2702 that "haproxy -c" will also report errors in such a case. This option is
2703 equivalent to command line argument "-dW".
2704
Emeric Brunf099e792010-09-27 12:05:28 +02002705
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010027063.4. Userlists
2707--------------
2708It is possible to control access to frontend/backend/listen sections or to
2709http stats by allowing only authenticated and authorized users. To do this,
2710it is required to create at least one userlist and to define users.
2711
2712userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002713 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002714 used to store authentication & authorization data for independent customers.
2715
2716group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002717 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002718 attach users to this group by using a comma separated list of names
2719 proceeded by "users" keyword.
2720
Cyril Bontéf0c60612010-02-06 14:44:47 +01002721user <username> [password|insecure-password <password>]
2722 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002723 Adds user <username> to the current userlist. Both secure (encrypted) and
2724 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002725 evaluated using the crypt(3) function, so depending on the system's
2726 capabilities, different algorithms are supported. For example, modern Glibc
2727 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2728 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002729
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002730 Attention: Be aware that using encrypted passwords might cause significantly
2731 increased CPU usage, depending on the number of requests, and the algorithm
2732 used. For any of the hashed variants, the password for each request must
2733 be processed through the chosen algorithm, before it can be compared to the
2734 value specified in the config file. Most current algorithms are deliberately
2735 designed to be expensive to compute to achieve resistance against brute
2736 force attacks. They do not simply salt/hash the clear text password once,
2737 but thousands of times. This can quickly become a major factor in haproxy's
2738 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002739
2740 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002741 userlist L1
2742 group G1 users tiger,scott
2743 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002744
Cyril Bontéf0c60612010-02-06 14:44:47 +01002745 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2746 user scott insecure-password elgato
2747 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002748
Cyril Bontéf0c60612010-02-06 14:44:47 +01002749 userlist L2
2750 group G1
2751 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002752
Cyril Bontéf0c60612010-02-06 14:44:47 +01002753 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2754 user scott insecure-password elgato groups G1,G2
2755 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002756
2757 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002758
Emeric Brunf099e792010-09-27 12:05:28 +02002759
27603.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002761----------
Emeric Brun94900952015-06-11 18:25:54 +02002762It is possible to propagate entries of any data-types in stick-tables between
2763several haproxy instances over TCP connections in a multi-master fashion. Each
2764instance pushes its local updates and insertions to remote peers. The pushed
2765values overwrite remote ones without aggregation. Interrupted exchanges are
2766automatically detected and recovered from the last known point.
2767In addition, during a soft restart, the old process connects to the new one
2768using such a TCP connection to push all its entries before the new process
2769tries to connect to other peers. That ensures very fast replication during a
2770reload, it typically takes a fraction of a second even for large tables.
2771Note that Server IDs are used to identify servers remotely, so it is important
2772that configurations look similar or at least that the same IDs are forced on
2773each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002774
2775peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002776 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002777 which is referenced by one or more stick-tables.
2778
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002779bind [<address>]:<port_range> [, ...] [param*]
2780 Defines the binding parameters of the local peer of this "peers" section.
2781 Such lines are not supported with "peer" line in the same "peers" section.
2782
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002783disabled
2784 Disables a peers section. It disables both listening and any synchronization
2785 related to this section. This is provided to disable synchronization of stick
2786 tables without having to comment out all "peers" references.
2787
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002788default-bind [param*]
2789 Defines the binding parameters for the local peer, excepted its address.
2790
2791default-server [param*]
2792 Change default options for a server in a "peers" section.
2793
2794 Arguments:
2795 <param*> is a list of parameters for this server. The "default-server"
2796 keyword accepts an important number of options and has a complete
2797 section dedicated to it. Please refer to section 5 for more
2798 details.
2799
2800
2801 See also: "server" and section 5 about server options
2802
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002803enable
2804 This re-enables a disabled peers section which was previously disabled.
2805
Jan Wagner3e678602020-12-17 22:22:32 +01002806log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailleb6f759b2019-11-05 09:57:45 +01002807 <facility> [<level> [<minlevel>]]
2808 "peers" sections support the same "log" keyword as for the proxies to
2809 log information about the "peers" listener. See "log" option for proxies for
2810 more details.
2811
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002812peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002813 Defines a peer inside a peers section.
2814 If <peername> is set to the local peer name (by default hostname, or forced
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002815 using "-L" command line option or "localpeer" global configuration setting),
2816 haproxy will listen for incoming remote peer connection on <ip>:<port>.
2817 Otherwise, <ip>:<port> defines where to connect to in order to join the
2818 remote peer, and <peername> is used at the protocol level to identify and
2819 validate the remote peer on the server side.
Emeric Brunf099e792010-09-27 12:05:28 +02002820
2821 During a soft restart, local peer <ip>:<port> is used by the old instance to
2822 connect the new one and initiate a complete replication (teaching process).
2823
2824 It is strongly recommended to have the exact same peers declaration on all
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002825 peers and to only rely on the "-L" command line argument or the "localpeer"
2826 global configuration setting to change the local peer name. This makes it
2827 easier to maintain coherent configuration files across all peers.
Emeric Brunf099e792010-09-27 12:05:28 +02002828
William Lallemandb2f07452015-05-12 14:27:13 +02002829 You may want to reference some environment variables in the address
2830 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002831
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002832 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2833 "server" keyword explanation below).
2834
2835server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002836 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002837 with a support for all "server" parameters found in 5.2 paragraph.
2838 If the underlying peer is local, <ip>:<port> parameters must not be present.
2839 These parameters must be provided on a "bind" line (see "bind" keyword
2840 of this "peers" section).
2841 Some of these parameters are irrelevant for "peers" sections.
2842
2843
Cyril Bontédc4d9032012-04-08 21:57:39 +02002844 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002845 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002846 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002847 peer haproxy1 192.168.0.1:1024
2848 peer haproxy2 192.168.0.2:1024
2849 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002850
2851 backend mybackend
2852 mode tcp
2853 balance roundrobin
2854 stick-table type ip size 20k peers mypeers
2855 stick on src
2856
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002857 server srv1 192.168.0.30:80
2858 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002859
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002860 Example:
2861 peers mypeers
2862 bind 127.0.0.11:10001 ssl crt mycerts/pem
2863 default-server ssl verify none
2864 server hostA 127.0.0.10:10000
2865 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002866
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002867
2868table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2869 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2870
2871 Configure a stickiness table for the current section. This line is parsed
2872 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002873 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002874 mandatory first parameter to designate the stick-table. Contrary to others
2875 sections, there may be several "table" lines in "peers" sections (see also
2876 "stick-table" keyword).
2877
2878 Also be aware of the fact that "peers" sections have their own stick-table
2879 namespaces to avoid collisions between stick-table names identical in
2880 different "peers" section. This is internally handled prepending the "peers"
2881 sections names to the name of the stick-tables followed by a '/' character.
2882 If somewhere else in the configuration file you have to refer to such
2883 stick-tables declared in "peers" sections you must use the prefixed version
2884 of the stick-table name as follows:
2885
2886 peers mypeers
2887 peer A ...
2888 peer B ...
2889 table t1 ...
2890
2891 frontend fe1
2892 tcp-request content track-sc0 src table mypeers/t1
2893
2894 This is also this prefixed version of the stick-table names which must be
2895 used to refer to stick-tables through the CLI.
2896
2897 About "peers" protocol, as only "peers" belonging to the same section may
2898 communicate with each others, there is no need to do such a distinction.
2899 Several "peers" sections may declare stick-tables with the same name.
2900 This is shorter version of the stick-table name which is sent over the network.
2901 There is only a '/' character as prefix to avoid stick-table name collisions between
2902 stick-tables declared as backends and stick-table declared in "peers" sections
2903 as follows in this weird but supported configuration:
2904
2905 peers mypeers
2906 peer A ...
2907 peer B ...
2908 table t1 type string size 10m store gpc0
2909
2910 backend t1
2911 stick-table type string size 10m store gpc0 peers mypeers
2912
Daniel Corbett67a82712020-07-06 23:01:19 -04002913 Here "t1" table declared in "mypeers" section has "mypeers/t1" as global name.
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002914 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2915 level the former table is named "/t1", the latter is again named "t1".
2916
Simon Horman51a1cf62015-02-03 13:00:44 +090029173.6. Mailers
2918------------
2919It is possible to send email alerts when the state of servers changes.
2920If configured email alerts are sent to each mailer that is configured
2921in a mailers section. Email is sent to mailers using SMTP.
2922
Pieter Baauw386a1272015-08-16 15:26:24 +02002923mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002924 Creates a new mailer list with the name <mailersect>. It is an
2925 independent section which is referenced by one or more proxies.
2926
2927mailer <mailername> <ip>:<port>
2928 Defines a mailer inside a mailers section.
2929
2930 Example:
2931 mailers mymailers
2932 mailer smtp1 192.168.0.1:587
2933 mailer smtp2 192.168.0.2:587
2934
2935 backend mybackend
2936 mode tcp
2937 balance roundrobin
2938
2939 email-alert mailers mymailers
2940 email-alert from test1@horms.org
2941 email-alert to test2@horms.org
2942
2943 server srv1 192.168.0.30:80
2944 server srv2 192.168.0.31:80
2945
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002946timeout mail <time>
2947 Defines the time available for a mail/connection to be made and send to
2948 the mail-server. If not defined the default value is 10 seconds. To allow
2949 for at least two SYN-ACK packets to be send during initial TCP handshake it
2950 is advised to keep this value above 4 seconds.
2951
2952 Example:
2953 mailers mymailers
2954 timeout mail 20s
2955 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002956
William Lallemandc9515522019-06-12 16:32:11 +020029573.7. Programs
2958-------------
2959In master-worker mode, it is possible to launch external binaries with the
2960master, these processes are called programs. These programs are launched and
2961managed the same way as the workers.
2962
2963During a reload of HAProxy, those processes are dealing with the same
2964sequence as a worker:
2965
2966 - the master is re-executed
2967 - the master sends a SIGUSR1 signal to the program
2968 - if "option start-on-reload" is not disabled, the master launches a new
2969 instance of the program
2970
2971During a stop, or restart, a SIGTERM is sent to the programs.
2972
2973program <name>
2974 This is a new program section, this section will create an instance <name>
2975 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2976 the management guide).
2977
2978command <command> [arguments*]
2979 Define the command to start with optional arguments. The command is looked
2980 up in the current PATH if it does not include an absolute path. This is a
2981 mandatory option of the program section. Arguments containing spaces must
2982 be enclosed in quotes or double quotes or be prefixed by a backslash.
2983
Andrew Heberle97236962019-07-12 11:50:26 +08002984user <user name>
2985 Changes the executed command user ID to the <user name> from /etc/passwd.
2986 See also "group".
2987
2988group <group name>
2989 Changes the executed command group ID to the <group name> from /etc/group.
2990 See also "user".
2991
William Lallemandc9515522019-06-12 16:32:11 +02002992option start-on-reload
2993no option start-on-reload
2994 Start (or not) a new instance of the program upon a reload of the master.
2995 The default is to start a new instance. This option may only be used in a
2996 program section.
2997
2998
Christopher Faulet76edc0f2020-01-13 15:52:01 +010029993.8. HTTP-errors
3000----------------
3001
3002It is possible to globally declare several groups of HTTP errors, to be
3003imported afterwards in any proxy section. Same group may be referenced at
3004several places and can be fully or partially imported.
3005
3006http-errors <name>
3007 Create a new http-errors group with the name <name>. It is an independent
3008 section that may be referenced by one or more proxies using its name.
3009
3010errorfile <code> <file>
3011 Associate a file contents to an HTTP error code
3012
3013 Arguments :
3014 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02003015 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01003016 425, 429, 500, 501, 502, 503, and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003017
3018 <file> designates a file containing the full HTTP response. It is
3019 recommended to follow the common practice of appending ".http" to
3020 the filename so that people do not confuse the response with HTML
3021 error pages, and to use absolute paths, since files are read
3022 before any chroot is performed.
3023
3024 Please referrers to "errorfile" keyword in section 4 for details.
3025
3026 Example:
3027 http-errors website-1
3028 errorfile 400 /etc/haproxy/errorfiles/site1/400.http
3029 errorfile 404 /etc/haproxy/errorfiles/site1/404.http
3030 errorfile 408 /dev/null # work around Chrome pre-connect bug
3031
3032 http-errors website-2
3033 errorfile 400 /etc/haproxy/errorfiles/site2/400.http
3034 errorfile 404 /etc/haproxy/errorfiles/site2/404.http
3035 errorfile 408 /dev/null # work around Chrome pre-connect bug
3036
Emeric Brun99c453d2020-05-25 15:01:04 +020030373.9. Rings
3038----------
3039
3040It is possible to globally declare ring-buffers, to be used as target for log
3041servers or traces.
3042
3043ring <ringname>
3044 Creates a new ring-buffer with name <ringname>.
3045
3046description <text>
Daniel Corbett67a82712020-07-06 23:01:19 -04003047 The description is an optional description string of the ring. It will
Emeric Brun99c453d2020-05-25 15:01:04 +02003048 appear on CLI. By default, <name> is reused to fill this field.
3049
3050format <format>
3051 Format used to store events into the ring buffer.
3052
3053 Arguments:
3054 <format> is the log format used when generating syslog messages. It may be
3055 one of the following :
3056
3057 iso A message containing only the ISO date, followed by the text.
3058 The PID, process name and system name are omitted. This is
3059 designed to be used with a local log server.
3060
Emeric Brun0237c4e2020-11-27 16:24:34 +01003061 local Analog to rfc3164 syslog message format except that hostname
3062 field is stripped. This is the default.
3063 Note: option "log-send-hostname" switches the default to
3064 rfc3164.
3065
Emeric Brun99c453d2020-05-25 15:01:04 +02003066 raw A message containing only the text. The level, PID, date, time,
3067 process name and system name are omitted. This is designed to be
3068 used in containers or during development, where the severity
3069 only depends on the file descriptor used (stdout/stderr). This
3070 is the default.
3071
Emeric Brun0237c4e2020-11-27 16:24:34 +01003072 rfc3164 The RFC3164 syslog message format.
Emeric Brun99c453d2020-05-25 15:01:04 +02003073 (https://tools.ietf.org/html/rfc3164)
3074
3075 rfc5424 The RFC5424 syslog message format.
3076 (https://tools.ietf.org/html/rfc5424)
3077
3078 short A message containing only a level between angle brackets such as
3079 '<3>', followed by the text. The PID, date, time, process name
3080 and system name are omitted. This is designed to be used with a
3081 local log server. This format is compatible with what the systemd
3082 logger consumes.
3083
Emeric Brun54648852020-07-06 15:54:06 +02003084 priority A message containing only a level plus syslog facility between angle
3085 brackets such as '<63>', followed by the text. The PID, date, time,
3086 process name and system name are omitted. This is designed to be used
3087 with a local log server.
3088
Emeric Brun99c453d2020-05-25 15:01:04 +02003089 timed A message containing only a level between angle brackets such as
3090 '<3>', followed by ISO date and by the text. The PID, process
3091 name and system name are omitted. This is designed to be
3092 used with a local log server.
3093
3094maxlen <length>
3095 The maximum length of an event message stored into the ring,
3096 including formatted header. If an event message is longer than
3097 <length>, it will be truncated to this length.
3098
Emeric Brun494c5052020-05-28 11:13:15 +02003099server <name> <address> [param*]
3100 Used to configure a syslog tcp server to forward messages from ring buffer.
3101 This supports for all "server" parameters found in 5.2 paragraph. Some of
3102 these parameters are irrelevant for "ring" sections. Important point: there
3103 is little reason to add more than one server to a ring, because all servers
3104 will receive the exact same copy of the ring contents, and as such the ring
3105 will progress at the speed of the slowest server. If one server does not
3106 respond, it will prevent old messages from being purged and may block new
3107 messages from being inserted into the ring. The proper way to send messages
3108 to multiple servers is to use one distinct ring per log server, not to
Emeric Brun97556472020-05-30 01:42:45 +02003109 attach multiple servers to the same ring. Note that specific server directive
3110 "log-proto" is used to set the protocol used to send messages.
Emeric Brun494c5052020-05-28 11:13:15 +02003111
Emeric Brun99c453d2020-05-25 15:01:04 +02003112size <size>
3113 This is the optional size in bytes for the ring-buffer. Default value is
3114 set to BUFSIZE.
3115
Emeric Brun494c5052020-05-28 11:13:15 +02003116timeout connect <timeout>
3117 Set the maximum time to wait for a connection attempt to a server to succeed.
3118
3119 Arguments :
3120 <timeout> is the timeout value specified in milliseconds by default, but
3121 can be in any other unit if the number is suffixed by the unit,
3122 as explained at the top of this document.
3123
3124timeout server <timeout>
3125 Set the maximum time for pending data staying into output buffer.
3126
3127 Arguments :
3128 <timeout> is the timeout value specified in milliseconds by default, but
3129 can be in any other unit if the number is suffixed by the unit,
3130 as explained at the top of this document.
3131
Emeric Brun99c453d2020-05-25 15:01:04 +02003132 Example:
3133 global
3134 log ring@myring local7
3135
3136 ring myring
3137 description "My local buffer"
3138 format rfc3164
3139 maxlen 1200
3140 size 32764
Emeric Brun494c5052020-05-28 11:13:15 +02003141 timeout connect 5s
3142 timeout server 10s
Emeric Brun97556472020-05-30 01:42:45 +02003143 server mysyslogsrv 127.0.0.1:6514 log-proto octet-count
Emeric Brun99c453d2020-05-25 15:01:04 +02003144
Emeric Brun12941c82020-07-07 14:19:42 +020031453.10. Log forwarding
3146-------------------
3147
3148It is possible to declare one or multiple log forwarding section,
3149haproxy will forward all received log messages to a log servers list.
3150
3151log-forward <name>
3152 Creates a new log forwarder proxy identified as <name>.
3153
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003154backlog <conns>
3155 Give hints to the system about the approximate listen backlog desired size
3156 on connections accept.
3157
3158bind <addr> [param*]
3159 Used to configure a stream log listener to receive messages to forward.
Emeric Brunda46c1c2020-10-08 08:39:02 +02003160 This supports the "bind" parameters found in 5.1 paragraph including
3161 those about ssl but some statements such as "alpn" may be irrelevant for
3162 syslog protocol over TCP.
3163 Those listeners support both "Octet Counting" and "Non-Transparent-Framing"
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003164 modes as defined in rfc-6587.
3165
Willy Tarreau76aaa7f2020-09-16 15:07:22 +02003166dgram-bind <addr> [param*]
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003167 Used to configure a datagram log listener to receive messages to forward.
3168 Addresses must be in IPv4 or IPv6 form,followed by a port. This supports
3169 for some of the "bind" parameters found in 5.1 paragraph among which
3170 "interface", "namespace" or "transparent", the other ones being
Willy Tarreau26ff5da2020-09-16 15:22:19 +02003171 silently ignored as irrelevant for UDP/syslog case.
Emeric Brun12941c82020-07-07 14:19:42 +02003172
3173log global
Jan Wagner3e678602020-12-17 22:22:32 +01003174log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Emeric Brun12941c82020-07-07 14:19:42 +02003175 <facility> [<level> [<minlevel>]]
3176 Used to configure target log servers. See more details on proxies
3177 documentation.
3178 If no format specified, haproxy tries to keep the incoming log format.
3179 Configured facility is ignored, except if incoming message does not
3180 present a facility but one is mandatory on the outgoing format.
3181 If there is no timestamp available in the input format, but the field
3182 exists in output format, haproxy will use the local date.
3183
3184 Example:
3185 global
3186 log stderr format iso local7
3187
3188 ring myring
3189 description "My local buffer"
3190 format rfc5424
3191 maxlen 1200
3192 size 32764
3193 timeout connect 5s
3194 timeout server 10s
3195 # syslog tcp server
3196 server mysyslogsrv 127.0.0.1:514 log-proto octet-count
3197
3198 log-forward sylog-loadb
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003199 dgram-bind 127.0.0.1:1514
3200 bind 127.0.0.1:1514
Emeric Brun12941c82020-07-07 14:19:42 +02003201 # all messages on stderr
3202 log global
3203 # all messages on local tcp syslog server
3204 log ring@myring local0
3205 # load balance messages on 4 udp syslog servers
3206 log 127.0.0.1:10001 sample 1:4 local0
3207 log 127.0.0.1:10002 sample 2:4 local0
3208 log 127.0.0.1:10003 sample 3:4 local0
3209 log 127.0.0.1:10004 sample 4:4 local0
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003210
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003211maxconn <conns>
3212 Fix the maximum number of concurrent connections on a log forwarder.
3213 10 is the default.
3214
3215timeout client <timeout>
3216 Set the maximum inactivity time on the client side.
3217
Willy Tarreauc57f0e22009-05-10 13:12:33 +020032184. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02003219----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003220
Willy Tarreau6a06a402007-07-15 20:15:28 +02003221Proxy configuration can be located in a set of sections :
Willy Tarreau7c0b4d82021-02-12 14:58:08 +01003222 - defaults [<name>] [ from <defaults_name> ]
3223 - frontend <name> [ from <defaults_name> ]
3224 - backend <name> [ from <defaults_name> ]
3225 - listen <name> [ from <defaults_name> ]
Willy Tarreau6a06a402007-07-15 20:15:28 +02003226
3227A "frontend" section describes a set of listening sockets accepting client
3228connections.
3229
3230A "backend" section describes a set of servers to which the proxy will connect
3231to forward incoming connections.
3232
3233A "listen" section defines a complete proxy with its frontend and backend
3234parts combined in one section. It is generally useful for TCP-only traffic.
3235
Willy Tarreau7c0b4d82021-02-12 14:58:08 +01003236A "defaults" section resets all settings to the documented ones and presets new
3237ones for use by subsequent sections. All of "frontend", "backend" and "listen"
3238sections always take their initial settings from a defaults section, by default
3239the latest one that appears before the newly created section. It is possible to
3240explicitly designate a specific "defaults" section to load the initial settings
3241from by indicating its name on the section line after the optional keyword
3242"from". While "defaults" section do not impose a name, this use is encouraged
3243for better readability. It is also the only way to designate a specific section
3244to use instead of the default previous one. Since "defaults" section names are
3245optional, by default a very permissive check is applied on their name and these
3246are even permitted to overlap. However if a "defaults" section is referenced by
3247any other section, its name must comply with the syntax imposed on all proxy
3248names, and this name must be unique among the defaults sections. Please note
3249that regardless of what is currently permitted, it is recommended to avoid
3250duplicate section names in general and to respect the same syntax as for proxy
3251names. This rule might be enforced in a future version.
3252
3253Note that it is even possible for a defaults section to take its initial
3254settings from another one, and as such, inherit settings across multiple levels
3255of defaults sections. This can be convenient to establish certain configuration
3256profiles to carry groups of default settings (e.g. TCP vs HTTP or short vs long
3257timeouts) but can quickly become confusing to follow.
3258
Willy Tarreau0ba27502007-12-24 16:55:16 +01003259All proxy names must be formed from upper and lower case letters, digits,
3260'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
3261case-sensitive, which means that "www" and "WWW" are two different proxies.
3262
3263Historically, all proxy names could overlap, it just caused troubles in the
3264logs. Since the introduction of content switching, it is mandatory that two
3265proxies with overlapping capabilities (frontend/backend) have different names.
3266However, it is still permitted that a frontend and a backend share the same
3267name, as this configuration seems to be commonly encountered.
3268
3269Right now, two major proxy modes are supported : "tcp", also known as layer 4,
3270and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003271bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003272protocol, and can interact with it by allowing, blocking, switching, adding,
3273modifying, or removing arbitrary contents in requests or responses, based on
3274arbitrary criteria.
3275
Willy Tarreau70dffda2014-01-30 03:07:23 +01003276In HTTP mode, the processing applied to requests and responses flowing over
3277a connection depends in the combination of the frontend's HTTP options and
Julien Pivotto21ad3152019-12-10 13:11:17 +01003278the backend's. HAProxy supports 3 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01003279
3280 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
3281 requests and responses are processed, and connections remain open but idle
3282 between responses and new requests.
3283
Willy Tarreau70dffda2014-01-30 03:07:23 +01003284 - SCL: server close ("option http-server-close") : the server-facing
3285 connection is closed after the end of the response is received, but the
3286 client-facing connection remains open.
3287
Christopher Faulet315b39c2018-09-21 16:26:19 +02003288 - CLO: close ("option httpclose"): the connection is closed after the end of
3289 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003290
3291The effective mode that will be applied to a connection passing through a
3292frontend and a backend can be determined by both proxy modes according to the
3293following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02003294weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003295
Christopher Faulet315b39c2018-09-21 16:26:19 +02003296 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01003297
Christopher Faulet315b39c2018-09-21 16:26:19 +02003298 | KAL | SCL | CLO
3299 ----+-----+-----+----
3300 KAL | KAL | SCL | CLO
3301 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02003302 mode SCL | SCL | SCL | CLO
3303 ----+-----+-----+----
3304 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01003305
Christopher Faulet4d37e532021-03-26 14:44:00 +01003306It is possible to chain a TCP frontend to an HTTP backend. It is pointless if
3307only HTTP traffic is handled. But It may be used to handle several protocols
3308into the same frontend. It this case, the client's connection is first handled
3309as a raw tcp connection before being upgraded to HTTP. Before the upgrade, the
3310content processings are performend on raw data. Once upgraded, data are parsed
3311and stored using an internal representation called HTX and it is no longer
3312possible to rely on raw representation. There is no way to go back.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003313
Christopher Faulet4d37e532021-03-26 14:44:00 +01003314There are two kind of upgrades, in-place upgrades and destructive upgrades. The
3315first ones concern the TCP to HTTP/1 upgrades. In HTTP/1, the request
3316processings are serialized, thus the applicative stream can be preserved. The
3317second ones concern the TCP to HTTP/2 upgrades. Because it is a multiplexed
3318protocol, the applicative stream cannot be associated to any HTTP/2 stream and
3319is destroyed. New applicative streams are then created when HAProxy receives
3320new HTTP/2 streams at the lower level, in the H2 multiplexer. It is important
3321to understand this difference because that drastically change the way to
3322process data. When an HTTP/1 upgrade is performed, the content processings
3323already performed on raw data are neither lost nor reexecuted while for an
3324HTTP/2 upgrade, applicative streams are distinct and all frontend rules are
3325evaluated systematically on each one. And as said, the first stream, the TCP
3326one, is destroyed, but only after the frontend rules were evaluated.
3327
3328There is another importnat point to understand when HTTP processings are
3329performed from a TCP proxy. While HAProxy is able to parse HTTP/1 in-fly from
3330tcp-request content rules, it is not possible for HTTP/2. Only the HTTP/2
3331preface can be parsed. This is a huge limitation regarding the HTTP content
3332analysis in TCP. Concretely it is only possible to know if received data are
3333HTTP. For instance, it is not possible to choose a backend based on the Host
3334header value while it is trivial in HTTP/1. Hopefully, there is a solution to
3335mitigate this drawback.
3336
3337It exists two way to perform HTTP upgrades. The first one, the historical
3338method, is to select an HTTP backend. The upgrade happens when the backend is
3339set. Thus, for in-place upgrades, only the backend configuration is considered
3340in the HTTP data processing. For destructive upgrades, the applicative stream
3341is destroyed, thus its processing is stopped. With this method, possibilities
3342to choose a backend with an HTTP/2 connection are really limited, as mentioned
3343above, and a bit useless because the stream is destroyed. The second method is
3344to upgrade during the tcp-request content rules evaluation, thanks to the
3345"switch-mode http" action. In this case, the upgrade is performed in the
3346frontend context and it is possible to define HTTP directives in this
3347frontend. For in-place upgrades, it offers all the power of the HTTP analysis
3348as soon as possible. It is not that far from an HTTP frontend. For destructive
3349upgrades, it does not change anything except it is useless to choose a backend
3350on limited information. It is of course the recommended method. Thus, testing
3351the request protocol from the tcp-request content rules to perform an HTTP
3352upgrade is enough. All the remaining HTTP manipulation may be moved to the
3353frontend http-request ruleset. But keep in mind that tcp-request content rules
3354remains evaluated on each streams, that can't be changed.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003355
Willy Tarreauc57f0e22009-05-10 13:12:33 +020033564.1. Proxy keywords matrix
3357--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003358
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003359The following list of keywords is supported. Most of them may only be used in a
3360limited set of section types. Some of them are marked as "deprecated" because
3361they are inherited from an old syntax which may be confusing or functionally
3362limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01003363marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003364option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02003365and must be disabled for a specific instance. Such options may also be prefixed
3366with "default" in order to restore default settings regardless of what has been
3367specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003368
Willy Tarreau6a06a402007-07-15 20:15:28 +02003369
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003370 keyword defaults frontend listen backend
3371------------------------------------+----------+----------+---------+---------
3372acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003373backlog X X X -
3374balance X - X X
3375bind - X X -
3376bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003377capture cookie - X X -
3378capture request header - X X -
3379capture response header - X X -
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003380clitcpka-cnt X X X -
3381clitcpka-idle X X X -
3382clitcpka-intvl X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02003383compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003384cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003385declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003386default-server X - X X
3387default_backend X X X -
3388description - X X X
3389disabled X X X X
3390dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09003391email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09003392email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09003393email-alert mailers X X X X
3394email-alert myhostname X X X X
3395email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003396enabled X X X X
3397errorfile X X X X
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003398errorfiles X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003399errorloc X X X X
3400errorloc302 X X X X
3401-- keyword -------------------------- defaults - frontend - listen -- backend -
3402errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003403force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003404filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003405fullconn X - X X
3406grace X X X X
3407hash-type X - X X
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01003408http-after-response - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003409http-check comment X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003410http-check connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003411http-check disable-on-404 X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003412http-check expect X - X X
Peter Gervai8912ae62020-06-11 18:26:36 +02003413http-check send X - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02003414http-check send-state X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003415http-check set-var X - X X
3416http-check unset-var X - X X
Christopher Faulet3b967c12020-05-15 15:47:44 +02003417http-error X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003418http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003419http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02003420http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02003421http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003422id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003423ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003424load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02003425log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01003426log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02003427log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01003428log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02003429max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003430maxconn X X X -
3431mode X X X X
3432monitor fail - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003433monitor-uri X X X -
3434option abortonclose (*) X - X X
3435option accept-invalid-http-request (*) X X X -
3436option accept-invalid-http-response (*) X - X X
3437option allbackups (*) X - X X
3438option checkcache (*) X - X X
3439option clitcpka (*) X X X -
3440option contstats (*) X X X -
Christopher Faulet89aed322020-06-02 17:33:56 +02003441option disable-h2-upgrade (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003442option dontlog-normal (*) X X X -
3443option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003444-- keyword -------------------------- defaults - frontend - listen -- backend -
3445option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02003446option h1-case-adjust-bogus-client (*) X X X -
3447option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02003448option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02003449option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01003450option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02003451option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02003452option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003453option http-server-close (*) X X X X
3454option http-use-proxy-header (*) X X X -
3455option httpchk X - X X
3456option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01003457option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003458option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003459option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02003460option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003461option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003462option log-health-checks (*) X - X X
3463option log-separate-errors (*) X X X -
3464option logasap (*) X X X -
3465option mysql-check X - X X
3466option nolinger (*) X X X X
3467option originalto X X X X
3468option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02003469option pgsql-check X - X X
3470option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003471option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02003472option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003473option smtpchk X - X X
3474option socket-stats (*) X X X -
3475option splice-auto (*) X X X X
3476option splice-request (*) X X X X
3477option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01003478option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003479option srvtcpka (*) X - X X
3480option ssl-hello-chk X - X X
3481-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01003482option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003483option tcp-smart-accept (*) X X X -
3484option tcp-smart-connect (*) X - X X
3485option tcpka X X X X
3486option tcplog X X X X
3487option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003488external-check command X - X X
3489external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003490persist rdp-cookie X - X X
3491rate-limit sessions X X X -
3492redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003493-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003494retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02003495retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003496server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003497server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02003498server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003499source X - X X
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003500srvtcpka-cnt X - X X
3501srvtcpka-idle X - X X
3502srvtcpka-intvl X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02003503stats admin - X X X
3504stats auth X X X X
3505stats enable X X X X
3506stats hide-version X X X X
3507stats http-request - X X X
3508stats realm X X X X
3509stats refresh X X X X
3510stats scope X X X X
3511stats show-desc X X X X
3512stats show-legends X X X X
3513stats show-node X X X X
3514stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003515-- keyword -------------------------- defaults - frontend - listen -- backend -
3516stick match - - X X
3517stick on - - X X
3518stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02003519stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01003520stick-table - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003521tcp-check comment X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003522tcp-check connect X - X X
3523tcp-check expect X - X X
3524tcp-check send X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003525tcp-check send-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003526tcp-check send-binary X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003527tcp-check send-binary-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003528tcp-check set-var X - X X
3529tcp-check unset-var X - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02003530tcp-request connection - X X -
3531tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02003532tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02003533tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02003534tcp-response content - - X X
3535tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003536timeout check X - X X
3537timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02003538timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003539timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003540timeout http-keep-alive X X X X
3541timeout http-request X X X X
3542timeout queue X - X X
3543timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02003544timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003545timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02003546timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003547transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01003548unique-id-format X X X -
3549unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003550use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02003551use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02003552use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003553------------------------------------+----------+----------+---------+---------
3554 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02003555
Willy Tarreau0ba27502007-12-24 16:55:16 +01003556
Willy Tarreauc57f0e22009-05-10 13:12:33 +020035574.2. Alphabetically sorted keywords reference
3558---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003559
3560This section provides a description of each keyword and its usage.
3561
3562
3563acl <aclname> <criterion> [flags] [operator] <value> ...
3564 Declare or complete an access list.
3565 May be used in sections : defaults | frontend | listen | backend
3566 no | yes | yes | yes
3567 Example:
3568 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
3569 acl invalid_src src_port 0:1023
3570 acl local_dst hdr(host) -i localhost
3571
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003572 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003573
3574
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003575backlog <conns>
3576 Give hints to the system about the approximate listen backlog desired size
3577 May be used in sections : defaults | frontend | listen | backend
3578 yes | yes | yes | no
3579 Arguments :
3580 <conns> is the number of pending connections. Depending on the operating
3581 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02003582 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003583
3584 In order to protect against SYN flood attacks, one solution is to increase
3585 the system's SYN backlog size. Depending on the system, sometimes it is just
3586 tunable via a system parameter, sometimes it is not adjustable at all, and
3587 sometimes the system relies on hints given by the application at the time of
3588 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
3589 to the listen() syscall. On systems which can make use of this value, it can
3590 sometimes be useful to be able to specify a different value, hence this
3591 backlog parameter.
3592
3593 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
3594 used as a hint and the system accepts up to the smallest greater power of
3595 two, and never more than some limits (usually 32768).
3596
3597 See also : "maxconn" and the target operating system's tuning guide.
3598
3599
Willy Tarreau0ba27502007-12-24 16:55:16 +01003600balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02003601balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003602 Define the load balancing algorithm to be used in a backend.
3603 May be used in sections : defaults | frontend | listen | backend
3604 yes | no | yes | yes
3605 Arguments :
3606 <algorithm> is the algorithm used to select a server when doing load
3607 balancing. This only applies when no persistence information
3608 is available, or when a connection is redispatched to another
3609 server. <algorithm> may be one of the following :
3610
3611 roundrobin Each server is used in turns, according to their weights.
3612 This is the smoothest and fairest algorithm when the server's
3613 processing time remains equally distributed. This algorithm
3614 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02003615 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08003616 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02003617 large farms, when a server becomes up after having been down
3618 for a very short time, it may sometimes take a few hundreds
3619 requests for it to be re-integrated into the farm and start
3620 receiving traffic. This is normal, though very rare. It is
3621 indicated here in case you would have the chance to observe
3622 it, so that you don't worry.
3623
3624 static-rr Each server is used in turns, according to their weights.
3625 This algorithm is as similar to roundrobin except that it is
3626 static, which means that changing a server's weight on the
3627 fly will have no effect. On the other hand, it has no design
3628 limitation on the number of servers, and when a server goes
3629 up, it is always immediately reintroduced into the farm, once
3630 the full map is recomputed. It also uses slightly less CPU to
3631 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003632
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003633 leastconn The server with the lowest number of connections receives the
3634 connection. Round-robin is performed within groups of servers
3635 of the same load to ensure that all servers will be used. Use
3636 of this algorithm is recommended where very long sessions are
3637 expected, such as LDAP, SQL, TSE, etc... but is not very well
3638 suited for protocols using short sessions such as HTTP. This
3639 algorithm is dynamic, which means that server weights may be
Willy Tarreau8c855f62020-10-22 17:41:45 +02003640 adjusted on the fly for slow starts for instance. It will
3641 also consider the number of queued connections in addition to
3642 the established ones in order to minimize queuing.
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003643
Willy Tarreauf09c6602012-02-13 17:12:08 +01003644 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003645 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01003646 identifier to the highest (see server parameter "id"), which
3647 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02003648 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01003649 not make sense to use this algorithm without setting maxconn.
3650 The purpose of this algorithm is to always use the smallest
3651 number of servers so that extra servers can be powered off
3652 during non-intensive hours. This algorithm ignores the server
3653 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02003654 or IMAP than HTTP, though it can be useful there too. In
3655 order to use this algorithm efficiently, it is recommended
3656 that a cloud controller regularly checks server usage to turn
3657 them off when unused, and regularly checks backend queue to
3658 turn new servers on when the queue inflates. Alternatively,
3659 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01003660
Willy Tarreau0ba27502007-12-24 16:55:16 +01003661 source The source IP address is hashed and divided by the total
3662 weight of the running servers to designate which server will
3663 receive the request. This ensures that the same client IP
3664 address will always reach the same server as long as no
3665 server goes down or up. If the hash result changes due to the
3666 number of running servers changing, many clients will be
3667 directed to a different server. This algorithm is generally
3668 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003669 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01003670 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003671 static by default, which means that changing a server's
3672 weight on the fly will have no effect, but this can be
3673 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003674
Oskar Stolc8dc41842012-05-19 10:19:54 +01003675 uri This algorithm hashes either the left part of the URI (before
3676 the question mark) or the whole URI (if the "whole" parameter
3677 is present) and divides the hash value by the total weight of
3678 the running servers. The result designates which server will
3679 receive the request. This ensures that the same URI will
3680 always be directed to the same server as long as no server
3681 goes up or down. This is used with proxy caches and
3682 anti-virus proxies in order to maximize the cache hit rate.
3683 Note that this algorithm may only be used in an HTTP backend.
3684 This algorithm is static by default, which means that
3685 changing a server's weight on the fly will have no effect,
3686 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003687
Oskar Stolc8dc41842012-05-19 10:19:54 +01003688 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003689 "depth", both followed by a positive integer number. These
3690 options may be helpful when it is needed to balance servers
3691 based on the beginning of the URI only. The "len" parameter
3692 indicates that the algorithm should only consider that many
3693 characters at the beginning of the URI to compute the hash.
3694 Note that having "len" set to 1 rarely makes sense since most
3695 URIs start with a leading "/".
3696
3697 The "depth" parameter indicates the maximum directory depth
3698 to be used to compute the hash. One level is counted for each
3699 slash in the request. If both parameters are specified, the
3700 evaluation stops when either is reached.
3701
Willy Tarreau57a37412020-09-23 08:56:29 +02003702 A "path-only" parameter indicates that the hashing key starts
3703 at the first '/' of the path. This can be used to ignore the
3704 authority part of absolute URIs, and to make sure that HTTP/1
3705 and HTTP/2 URIs will provide the same hash.
3706
Willy Tarreau0ba27502007-12-24 16:55:16 +01003707 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003708 the query string of each HTTP GET request.
3709
3710 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02003711 request entity will be searched for the parameter argument,
3712 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02003713 ('?') in the URL. The message body will only start to be
3714 analyzed once either the advertised amount of data has been
3715 received or the request buffer is full. In the unlikely event
3716 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02003717 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02003718 be randomly balanced if at all. This keyword used to support
3719 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003720
3721 If the parameter is found followed by an equal sign ('=') and
3722 a value, then the value is hashed and divided by the total
3723 weight of the running servers. The result designates which
3724 server will receive the request.
3725
3726 This is used to track user identifiers in requests and ensure
3727 that a same user ID will always be sent to the same server as
3728 long as no server goes up or down. If no value is found or if
3729 the parameter is not found, then a round robin algorithm is
3730 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003731 backend. This algorithm is static by default, which means
3732 that changing a server's weight on the fly will have no
3733 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003734
Cyril Bontédc4d9032012-04-08 21:57:39 +02003735 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
3736 request. Just as with the equivalent ACL 'hdr()' function,
3737 the header name in parenthesis is not case sensitive. If the
3738 header is absent or if it does not contain any value, the
3739 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01003740
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003741 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01003742 reducing the hash algorithm to the main domain part with some
3743 specific headers such as 'Host'. For instance, in the Host
3744 value "haproxy.1wt.eu", only "1wt" will be considered.
3745
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003746 This algorithm is static by default, which means that
3747 changing a server's weight on the fly will have no effect,
3748 but this can be changed using "hash-type".
3749
Willy Tarreau21c741a2019-01-14 18:14:27 +01003750 random
3751 random(<draws>)
3752 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02003753 hashing function. This means that the servers' weights are
3754 respected, dynamic weight changes immediately take effect, as
3755 well as new server additions. Random load balancing can be
3756 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01003757 or removed as it may avoid the hammering effect that could
3758 result from roundrobin or leastconn in this situation. The
3759 hash-balance-factor directive can be used to further improve
3760 fairness of the load balancing, especially in situations
3761 where servers show highly variable response times. When an
3762 argument <draws> is present, it must be an integer value one
3763 or greater, indicating the number of draws before selecting
3764 the least loaded of these servers. It was indeed demonstrated
3765 that picking the least loaded of two servers is enough to
3766 significantly improve the fairness of the algorithm, by
3767 always avoiding to pick the most loaded server within a farm
3768 and getting rid of any bias that could be induced by the
3769 unfair distribution of the consistent list. Higher values N
3770 will take away N-1 of the highest loaded servers at the
3771 expense of performance. With very high values, the algorithm
3772 will converge towards the leastconn's result but much slower.
3773 The default value is 2, which generally shows very good
3774 distribution and performance. This algorithm is also known as
3775 the Power of Two Random Choices and is described here :
3776 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02003777
Emeric Brun736aa232009-06-30 17:56:00 +02003778 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02003779 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02003780 The RDP cookie <name> (or "mstshash" if omitted) will be
3781 looked up and hashed for each incoming TCP request. Just as
3782 with the equivalent ACL 'req_rdp_cookie()' function, the name
3783 is not case-sensitive. This mechanism is useful as a degraded
3784 persistence mode, as it makes it possible to always send the
3785 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003786 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02003787 used instead.
3788
3789 Note that for this to work, the frontend must ensure that an
3790 RDP cookie is already present in the request buffer. For this
3791 you must use 'tcp-request content accept' rule combined with
3792 a 'req_rdp_cookie_cnt' ACL.
3793
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003794 This algorithm is static by default, which means that
3795 changing a server's weight on the fly will have no effect,
3796 but this can be changed using "hash-type".
3797
Cyril Bontédc4d9032012-04-08 21:57:39 +02003798 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09003799
Willy Tarreau0ba27502007-12-24 16:55:16 +01003800 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003801 algorithms. Right now, only "url_param" and "uri" support an
3802 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003803
Willy Tarreau3cd9af22009-03-15 14:06:41 +01003804 The load balancing algorithm of a backend is set to roundrobin when no other
3805 algorithm, mode nor option have been set. The algorithm may only be set once
3806 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003807
Lukas Tribus80512b12018-10-27 20:07:40 +02003808 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05003809 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02003810 to be routed to different backend servers, breaking the invalid assumptions
3811 NTLM relies on.
3812
Willy Tarreau0ba27502007-12-24 16:55:16 +01003813 Examples :
3814 balance roundrobin
3815 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003816 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01003817 balance hdr(User-Agent)
3818 balance hdr(host)
3819 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003820
3821 Note: the following caveats and limitations on using the "check_post"
3822 extension with "url_param" must be considered :
3823
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003824 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003825 to determine if the parameters will be found in the body or entity which
3826 may contain binary data. Therefore another method may be required to
3827 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02003828 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003829
3830 - using a <max_wait> value larger than the request buffer size does not
3831 make sense and is useless. The buffer size is set at build time, and
3832 defaults to 16 kB.
3833
3834 - Content-Encoding is not supported, the parameter search will probably
3835 fail; and load balancing will fall back to Round Robin.
3836
3837 - Expect: 100-continue is not supported, load balancing will fall back to
3838 Round Robin.
3839
Lukas Tribus23953682017-04-28 13:24:30 +00003840 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003841 If the entire parameter value is not present in the first chunk, the
3842 selection of server is undefined (actually, defined by how little
3843 actually appeared in the first chunk).
3844
3845 - This feature does not support generation of a 100, 411 or 501 response.
3846
3847 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003848 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003849 white space or control characters are found, indicating the end of what
3850 might be a URL parameter list. This is probably not a concern with SGML
3851 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003852
Willy Tarreau294d0f02015-08-10 19:40:12 +02003853 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003854
3855
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003856bind [<address>]:<port_range> [, ...] [param*]
3857bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003858 Define one or several listening addresses and/or ports in a frontend.
3859 May be used in sections : defaults | frontend | listen | backend
3860 no | yes | yes | no
3861 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003862 <address> is optional and can be a host name, an IPv4 address, an IPv6
3863 address, or '*'. It designates the address the frontend will
3864 listen on. If unset, all IPv4 addresses of the system will be
3865 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01003866 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01003867 Optionally, an address family prefix may be used before the
3868 address to force the family regardless of the address format,
3869 which can be useful to specify a path to a unix socket with
3870 no slash ('/'). Currently supported prefixes are :
3871 - 'ipv4@' -> address is always IPv4
3872 - 'ipv6@' -> address is always IPv6
Emeric Brun3835c0d2020-07-07 09:46:09 +02003873 - 'udp@' -> address is resolved as IPv4 or IPv6 and
Emeric Brun12941c82020-07-07 14:19:42 +02003874 protocol UDP is used. Currently those listeners are
3875 supported only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003876 - 'udp4@' -> address is always IPv4 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003877 is used. Currently those listeners are supported
3878 only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003879 - 'udp6@' -> address is always IPv6 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003880 is used. Currently those listeners are supported
3881 only in log-forward sections.
Willy Tarreau24709282013-03-10 21:32:12 +01003882 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02003883 - 'abns@' -> address is in abstract namespace (Linux only).
3884 Note: since abstract sockets are not "rebindable", they
3885 do not cope well with multi-process mode during
3886 soft-restart, so it is better to avoid them if
3887 nbproc is greater than 1. The effect is that if the
3888 new process fails to start, only one of the old ones
3889 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01003890 - 'fd@<n>' -> use file descriptor <n> inherited from the
3891 parent. The fd must be bound and may or may not already
3892 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02003893 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
3894 connected unix socket or of a socketpair. The bind waits
3895 to receive a FD over the unix socket and uses it as if it
3896 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02003897 You may want to reference some environment variables in the
3898 address parameter, see section 2.3 about environment
3899 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003900
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003901 <port_range> is either a unique TCP port, or a port range for which the
3902 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003903 above. The port is mandatory for TCP listeners. Note that in
3904 the case of an IPv6 address, the port is always the number
3905 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003906 - a numerical port (ex: '80')
3907 - a dash-delimited ports range explicitly stating the lower
3908 and upper bounds (ex: '2000-2100') which are included in
3909 the range.
3910
3911 Particular care must be taken against port ranges, because
3912 every <address:port> couple consumes one socket (= a file
3913 descriptor), so it's easy to consume lots of descriptors
3914 with a simple range, and to run out of sockets. Also, each
3915 <address:port> couple must be used only once among all
3916 instances running on a same system. Please note that binding
3917 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003918 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003919 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003920
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003921 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01003922 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003923 receive UNIX connections on the socket located at this place.
3924 The path must begin with a slash and by default is absolute.
3925 It can be relative to the prefix defined by "unix-bind" in
3926 the global section. Note that the total length of the prefix
3927 followed by the socket path cannot exceed some system limits
3928 for UNIX sockets, which commonly are set to 107 characters.
3929
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003930 <param*> is a list of parameters common to all sockets declared on the
3931 same line. These numerous parameters depend on OS and build
3932 options and have a complete section dedicated to them. Please
3933 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003934
Willy Tarreau0ba27502007-12-24 16:55:16 +01003935 It is possible to specify a list of address:port combinations delimited by
3936 commas. The frontend will then listen on all of these addresses. There is no
3937 fixed limit to the number of addresses and ports which can be listened on in
3938 a frontend, as well as there is no limit to the number of "bind" statements
3939 in a frontend.
3940
3941 Example :
3942 listen http_proxy
3943 bind :80,:443
3944 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003945 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01003946
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003947 listen http_https_proxy
3948 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02003949 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003950
Willy Tarreau24709282013-03-10 21:32:12 +01003951 listen http_https_proxy_explicit
3952 bind ipv6@:80
3953 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
3954 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
3955
Willy Tarreaudad36a32013-03-11 01:20:04 +01003956 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02003957 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01003958
Willy Tarreau55dcaf62015-09-27 15:03:15 +02003959 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
3960 sun_path length is used for the address length. Some other programs
3961 such as socat use the string length only by default. Pass the option
3962 ",unix-tightsocklen=0" to any abstract socket definition in socat to
3963 make it compatible with HAProxy's.
3964
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003965 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003966 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003967
3968
Christopher Fauletff4121f2017-11-22 16:38:49 +01003969bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003970 Limit visibility of an instance to a certain set of processes numbers.
3971 May be used in sections : defaults | frontend | listen | backend
3972 yes | yes | yes | yes
3973 Arguments :
3974 all All process will see this instance. This is the default. It
3975 may be used to override a default value.
3976
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003977 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003978 option may be combined with other numbers.
3979
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003980 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003981 option may be combined with other numbers. Do not use it
3982 with less than 2 processes otherwise some instances might be
3983 missing from all processes.
3984
Christopher Fauletff4121f2017-11-22 16:38:49 +01003985 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003986 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01003987 the machine's word size. Ranges can be partially defined. The
3988 higher bound can be omitted. In such case, it is replaced by
3989 the corresponding maximum value. If a proxy is bound to
3990 process numbers greater than the configured global.nbproc, it
3991 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003992 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003993
3994 This keyword limits binding of certain instances to certain processes. This
3995 is useful in order not to have too many processes listening to the same
3996 ports. For instance, on a dual-core machine, it might make sense to set
3997 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3998 and 'even' instances.
3999
Willy Tarreaua9db57e2013-01-18 11:29:29 +01004000 At the moment, it is not possible to reference more than 32 or 64 processes
4001 using this keyword, but this should be more than enough for most setups.
4002 Please note that 'all' really means all processes regardless of the machine's
4003 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004004
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02004005 Each "bind" line may further be limited to a subset of the proxy's processes,
4006 please consult the "process" bind keyword in section 5.1.
4007
Willy Tarreaub369a042014-09-16 13:21:03 +02004008 When a frontend has no explicit "bind-process" line, it tries to bind to all
4009 the processes referenced by its "bind" lines. That means that frontends can
4010 easily adapt to their listeners' processes.
4011
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004012 If some backends are referenced by frontends bound to other processes, the
4013 backend automatically inherits the frontend's processes.
4014
4015 Example :
4016 listen app_ip1
4017 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02004018 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004019
4020 listen app_ip2
4021 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02004022 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004023
4024 listen management
4025 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02004026 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004027
Willy Tarreau110ecc12012-11-15 17:50:01 +01004028 listen management
4029 bind 10.0.0.4:80
4030 bind-process 1-4
4031
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02004032 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004033
4034
Willy Tarreau0ba27502007-12-24 16:55:16 +01004035capture cookie <name> len <length>
4036 Capture and log a cookie in the request and in the response.
4037 May be used in sections : defaults | frontend | listen | backend
4038 no | yes | yes | no
4039 Arguments :
4040 <name> is the beginning of the name of the cookie to capture. In order
4041 to match the exact name, simply suffix the name with an equal
4042 sign ('='). The full name will appear in the logs, which is
4043 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01004044 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01004045
4046 <length> is the maximum number of characters to report in the logs, which
4047 include the cookie name, the equal sign and the value, all in the
4048 standard "name=value" form. The string will be truncated on the
4049 right if it exceeds <length>.
4050
4051 Only the first cookie is captured. Both the "cookie" request headers and the
4052 "set-cookie" response headers are monitored. This is particularly useful to
4053 check for application bugs causing session crossing or stealing between
4054 users, because generally the user's cookies can only change on a login page.
4055
4056 When the cookie was not presented by the client, the associated log column
4057 will report "-". When a request does not cause a cookie to be assigned by the
4058 server, a "-" is reported in the response column.
4059
4060 The capture is performed in the frontend only because it is necessary that
4061 the log format does not change for a given frontend depending on the
4062 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01004063 "capture cookie" statement in a frontend. The maximum capture length is set
4064 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
4065 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004066
4067 Example:
4068 capture cookie ASPSESSION len 32
4069
4070 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004071 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004072
4073
4074capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01004075 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004076 May be used in sections : defaults | frontend | listen | backend
4077 no | yes | yes | no
4078 Arguments :
4079 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004080 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01004081 appear in the requests, with the first letter of each word in
4082 upper case. The header name will not appear in the logs, only the
4083 value is reported, but the position in the logs is respected.
4084
4085 <length> is the maximum number of characters to extract from the value and
4086 report in the logs. The string will be truncated on the right if
4087 it exceeds <length>.
4088
Willy Tarreau4460d032012-11-21 23:37:37 +01004089 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01004090 value will be added to the logs between braces ('{}'). If multiple headers
4091 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004092 in the same order they were declared in the configuration. Non-existent
4093 headers will be logged just as an empty string. Common uses for request
4094 header captures include the "Host" field in virtual hosting environments, the
4095 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004096 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004097 environments to find where the request came from.
4098
4099 Note that when capturing headers such as "User-agent", some spaces may be
4100 logged, making the log analysis more difficult. Thus be careful about what
4101 you log if you know your log parser is not smart enough to rely on the
4102 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004103
Willy Tarreau0900abb2012-11-22 00:21:46 +01004104 There is no limit to the number of captured request headers nor to their
4105 length, though it is wise to keep them low to limit memory usage per session.
4106 In order to keep log format consistent for a same frontend, header captures
4107 can only be declared in a frontend. It is not possible to specify a capture
4108 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004109
4110 Example:
4111 capture request header Host len 15
4112 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01004113 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01004114
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004115 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01004116 about logging.
4117
4118
4119capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01004120 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004121 May be used in sections : defaults | frontend | listen | backend
4122 no | yes | yes | no
4123 Arguments :
4124 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004125 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01004126 appear in the response, with the first letter of each word in
4127 upper case. The header name will not appear in the logs, only the
4128 value is reported, but the position in the logs is respected.
4129
4130 <length> is the maximum number of characters to extract from the value and
4131 report in the logs. The string will be truncated on the right if
4132 it exceeds <length>.
4133
Willy Tarreau4460d032012-11-21 23:37:37 +01004134 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01004135 result will be added to the logs between braces ('{}') after the captured
4136 request headers. If multiple headers are captured, they will be delimited by
4137 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004138 the configuration. Non-existent headers will be logged just as an empty
4139 string. Common uses for response header captures include the "Content-length"
4140 header which indicates how many bytes are expected to be returned, the
4141 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004142
Willy Tarreau0900abb2012-11-22 00:21:46 +01004143 There is no limit to the number of captured response headers nor to their
4144 length, though it is wise to keep them low to limit memory usage per session.
4145 In order to keep log format consistent for a same frontend, header captures
4146 can only be declared in a frontend. It is not possible to specify a capture
4147 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004148
4149 Example:
4150 capture response header Content-length len 9
4151 capture response header Location len 15
4152
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004153 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01004154 about logging.
4155
4156
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004157clitcpka-cnt <count>
4158 Sets the maximum number of keepalive probes TCP should send before dropping
4159 the connection on the client side.
4160 May be used in sections : defaults | frontend | listen | backend
4161 yes | yes | yes | no
4162 Arguments :
4163 <count> is the maximum number of keepalive probes.
4164
4165 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
4166 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004167 The availability of this setting depends on the operating system. It is
4168 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004169
4170 See also : "option clitcpka", "clitcpka-idle", "clitcpka-intvl".
4171
4172
4173clitcpka-idle <timeout>
4174 Sets the time the connection needs to remain idle before TCP starts sending
4175 keepalive probes, if enabled the sending of TCP keepalive packets on the
4176 client side.
4177 May be used in sections : defaults | frontend | listen | backend
4178 yes | yes | yes | no
4179 Arguments :
4180 <timeout> is the time the connection needs to remain idle before TCP starts
4181 sending keepalive probes. It is specified in seconds by default,
4182 but can be in any other unit if the number is suffixed by the
4183 unit, as explained at the top of this document.
4184
4185 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
4186 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004187 The availability of this setting depends on the operating system. It is
4188 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004189
4190 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-intvl".
4191
4192
4193clitcpka-intvl <timeout>
4194 Sets the time between individual keepalive probes on the client side.
4195 May be used in sections : defaults | frontend | listen | backend
4196 yes | yes | yes | no
4197 Arguments :
4198 <timeout> is the time between individual keepalive probes. It is specified
4199 in seconds by default, but can be in any other unit if the number
4200 is suffixed by the unit, as explained at the top of this
4201 document.
4202
4203 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
4204 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004205 The availability of this setting depends on the operating system. It is
4206 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004207
4208 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-idle".
4209
4210
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004211compression algo <algorithm> ...
4212compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02004213compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02004214 Enable HTTP compression.
4215 May be used in sections : defaults | frontend | listen | backend
4216 yes | yes | yes | yes
4217 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004218 algo is followed by the list of supported compression algorithms.
4219 type is followed by the list of MIME types that will be compressed.
4220 offload makes haproxy work as a compression offloader only (see notes).
4221
4222 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01004223 identity this is mostly for debugging, and it was useful for developing
4224 the compression feature. Identity does not apply any change on
4225 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004226
Willy Tarreauc91840a2015-03-28 17:00:39 +01004227 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01004228 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01004229
4230 deflate same as "gzip", but with deflate algorithm and zlib format.
4231 Note that this algorithm has ambiguous support on many
4232 browsers and no support at all from recent ones. It is
4233 strongly recommended not to use it for anything else than
4234 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01004235 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004236
Willy Tarreauc91840a2015-03-28 17:00:39 +01004237 raw-deflate same as "deflate" without the zlib wrapper, and used as an
4238 alternative when the browser wants "deflate". All major
4239 browsers understand it and despite violating the standards,
4240 it is known to work better than "deflate", at least on MSIE
4241 and some versions of Safari. Do not use it in conjunction
4242 with "deflate", use either one or the other since both react
4243 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01004244 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004245
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04004246 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004247 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004248 If backend servers support HTTP compression, these directives
4249 will be no-op: haproxy will see the compressed response and will not
4250 compress again. If backend servers do not support HTTP compression and
4251 there is Accept-Encoding header in request, haproxy will compress the
4252 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02004253
4254 The "offload" setting makes haproxy remove the Accept-Encoding header to
4255 prevent backend servers from compressing responses. It is strongly
4256 recommended not to do this because this means that all the compression work
4257 will be done on the single point where haproxy is located. However in some
4258 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004259 with broken HTTP compression implementation which can't be turned off.
4260 In that case haproxy can be used to prevent that gateway from emitting
4261 invalid payloads. In this case, simply removing the header in the
4262 configuration does not work because it applies before the header is parsed,
4263 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02004264 then be used for such scenarios. Note: for now, the "offload" setting is
4265 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02004266
William Lallemand05097442012-11-20 12:14:28 +01004267 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01004268 * the request does not advertise a supported compression algorithm in the
4269 "Accept-Encoding" header
Julien Pivottoff80c822021-03-29 12:41:40 +02004270 * the response message is not HTTP/1.1 or above
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01004271 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01004272 * response contain neither a "Content-Length" header nor a
4273 "Transfer-Encoding" whose last value is "chunked"
4274 * response contains a "Content-Type" header whose first value starts with
4275 "multipart"
4276 * the response contains the "no-transform" value in the "Cache-control"
4277 header
4278 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
4279 and later
4280 * The response contains a "Content-Encoding" header, indicating that the
4281 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01004282 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01004283
Tim Duesterhusb229f012019-01-29 16:38:56 +01004284 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01004285
William Lallemand82fe75c2012-10-23 10:25:10 +02004286 Examples :
4287 compression algo gzip
4288 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01004289
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004290
Willy Tarreau55165fe2009-05-10 12:02:55 +02004291cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02004292 [ postonly ] [ preserve ] [ httponly ] [ secure ]
4293 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Christopher Faulet2f533902020-01-21 11:06:48 +01004294 [ dynamic ] [ attr <value> ]*
Willy Tarreau0ba27502007-12-24 16:55:16 +01004295 Enable cookie-based persistence in a backend.
4296 May be used in sections : defaults | frontend | listen | backend
4297 yes | no | yes | yes
4298 Arguments :
4299 <name> is the name of the cookie which will be monitored, modified or
4300 inserted in order to bring persistence. This cookie is sent to
4301 the client via a "Set-Cookie" header in the response, and is
4302 brought back by the client in a "Cookie" header in all requests.
4303 Special care should be taken to choose a name which does not
4304 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01004305 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004306 HTTP/HTTPS), care should be taken to use different cookie names
4307 between all backends if persistence between them is not desired.
4308
4309 rewrite This keyword indicates that the cookie will be provided by the
4310 server and that haproxy will have to modify its value to set the
4311 server's identifier in it. This mode is handy when the management
4312 of complex combinations of "Set-cookie" and "Cache-control"
4313 headers is left to the application. The application can then
4314 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004315 cookie. Since all responses should be monitored, this mode
4316 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01004317 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004318 start with this mode for new deployments. This keyword is
4319 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004320
4321 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02004322 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004323
Willy Tarreaua79094d2010-08-31 22:54:15 +02004324 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004325 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02004326 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01004327 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004328 existing configurations running in the "rewrite" mode. The cookie
4329 will only be a session cookie and will not be stored on the
4330 client's disk. By default, unless the "indirect" option is added,
4331 the server will see the cookies emitted by the client. Due to
4332 caching effects, it is generally wise to add the "nocache" or
4333 "postonly" keywords (see below). The "insert" keyword is not
4334 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004335
4336 prefix This keyword indicates that instead of relying on a dedicated
4337 cookie for the persistence, an existing one will be completed.
4338 This may be needed in some specific environments where the client
4339 does not support more than one single cookie and the application
4340 already needs it. In this case, whenever the server sets a cookie
4341 named <name>, it will be prefixed with the server's identifier
4342 and a delimiter. The prefix will be removed from all client
4343 requests so that the server still finds the cookie it emitted.
4344 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004345 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02004346 not compatible with "rewrite" and "insert". Note: it is highly
4347 recommended not to use "indirect" with "prefix", otherwise server
4348 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004349
Willy Tarreaua79094d2010-08-31 22:54:15 +02004350 indirect When this option is specified, no cookie will be emitted to a
4351 client which already has a valid one for the server which has
4352 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004353 it will be removed, unless the "preserve" option is also set. In
4354 "insert" mode, this will additionally remove cookies from the
4355 requests transmitted to the server, making the persistence
4356 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02004357 Note: it is highly recommended not to use "indirect" with
4358 "prefix", otherwise server cookie updates would not be sent to
4359 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004360
4361 nocache This option is recommended in conjunction with the insert mode
4362 when there is a cache between the client and HAProxy, as it
4363 ensures that a cacheable response will be tagged non-cacheable if
4364 a cookie needs to be inserted. This is important because if all
4365 persistence cookies are added on a cacheable home page for
4366 instance, then all customers will then fetch the page from an
4367 outer cache and will all share the same persistence cookie,
4368 leading to one server receiving much more traffic than others.
4369 See also the "insert" and "postonly" options.
4370
4371 postonly This option ensures that cookie insertion will only be performed
4372 on responses to POST requests. It is an alternative to the
4373 "nocache" option, because POST responses are not cacheable, so
4374 this ensures that the persistence cookie will never get cached.
4375 Since most sites do not need any sort of persistence before the
4376 first POST which generally is a login request, this is a very
4377 efficient method to optimize caching without risking to find a
4378 persistence cookie in the cache.
4379 See also the "insert" and "nocache" options.
4380
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004381 preserve This option may only be used with "insert" and/or "indirect". It
4382 allows the server to emit the persistence cookie itself. In this
4383 case, if a cookie is found in the response, haproxy will leave it
4384 untouched. This is useful in order to end persistence after a
4385 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01004386 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004387 the past. By combining this mechanism with the "disable-on-404"
4388 check option, it is possible to perform a completely graceful
4389 shutdown because users will definitely leave the server after
4390 they logout.
4391
Willy Tarreau4992dd22012-05-31 21:02:17 +02004392 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
4393 when a cookie is inserted. This attribute is used so that a
4394 user agent doesn't share the cookie with non-HTTP components.
4395 Please check RFC6265 for more information on this attribute.
4396
4397 secure This option tells haproxy to add a "Secure" cookie attribute when
4398 a cookie is inserted. This attribute is used so that a user agent
4399 never emits this cookie over non-secure channels, which means
4400 that a cookie learned with this flag will be presented only over
4401 SSL/TLS connections. Please check RFC6265 for more information on
4402 this attribute.
4403
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02004404 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004405 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01004406 name. If the domain begins with a dot, the browser is allowed to
4407 use it for any host ending with that name. It is also possible to
4408 specify several domain names by invoking this option multiple
4409 times. Some browsers might have small limits on the number of
4410 domains, so be careful when doing that. For the record, sending
4411 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02004412
Willy Tarreau996a92c2010-10-13 19:30:47 +02004413 maxidle This option allows inserted cookies to be ignored after some idle
4414 time. It only works with insert-mode cookies. When a cookie is
4415 sent to the client, the date this cookie was emitted is sent too.
4416 Upon further presentations of this cookie, if the date is older
4417 than the delay indicated by the parameter (in seconds), it will
4418 be ignored. Otherwise, it will be refreshed if needed when the
4419 response is sent to the client. This is particularly useful to
4420 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004421 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02004422 this option is set and a cookie has no date, it is always
4423 accepted, but gets refreshed in the response. This maintains the
4424 ability for admins to access their sites. Cookies that have a
4425 date in the future further than 24 hours are ignored. Doing so
4426 lets admins fix timezone issues without risking kicking users off
4427 the site.
4428
4429 maxlife This option allows inserted cookies to be ignored after some life
4430 time, whether they're in use or not. It only works with insert
4431 mode cookies. When a cookie is first sent to the client, the date
4432 this cookie was emitted is sent too. Upon further presentations
4433 of this cookie, if the date is older than the delay indicated by
4434 the parameter (in seconds), it will be ignored. If the cookie in
4435 the request has no date, it is accepted and a date will be set.
4436 Cookies that have a date in the future further than 24 hours are
4437 ignored. Doing so lets admins fix timezone issues without risking
4438 kicking users off the site. Contrary to maxidle, this value is
4439 not refreshed, only the first visit date counts. Both maxidle and
4440 maxlife may be used at the time. This is particularly useful to
4441 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004442 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02004443 is stronger than the maxidle method in that it forces a
4444 redispatch after some absolute delay.
4445
Olivier Houchard4e694042017-03-14 20:01:29 +01004446 dynamic Activate dynamic cookies. When used, a session cookie is
4447 dynamically created for each server, based on the IP and port
4448 of the server, and a secret key, specified in the
4449 "dynamic-cookie-key" backend directive.
4450 The cookie will be regenerated each time the IP address change,
4451 and is only generated for IPv4/IPv6.
4452
Christopher Faulet2f533902020-01-21 11:06:48 +01004453 attr This option tells haproxy to add an extra attribute when a
4454 cookie is inserted. The attribute value can contain any
4455 characters except control ones or ";". This option may be
4456 repeated.
4457
Willy Tarreau0ba27502007-12-24 16:55:16 +01004458 There can be only one persistence cookie per HTTP backend, and it can be
4459 declared in a defaults section. The value of the cookie will be the value
4460 indicated after the "cookie" keyword in a "server" statement. If no cookie
4461 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02004462
Willy Tarreau0ba27502007-12-24 16:55:16 +01004463 Examples :
4464 cookie JSESSIONID prefix
4465 cookie SRV insert indirect nocache
4466 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02004467 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01004468
Willy Tarreau294d0f02015-08-10 19:40:12 +02004469 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004470
Willy Tarreau983e01e2010-01-11 18:42:06 +01004471
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004472declare capture [ request | response ] len <length>
4473 Declares a capture slot.
4474 May be used in sections : defaults | frontend | listen | backend
4475 no | yes | yes | no
4476 Arguments:
4477 <length> is the length allowed for the capture.
4478
4479 This declaration is only available in the frontend or listen section, but the
4480 reserved slot can be used in the backends. The "request" keyword allocates a
4481 capture slot for use in the request, and "response" allocates a capture slot
4482 for use in the response.
4483
4484 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02004485 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004486 "http-request capture" and "http-response capture".
4487
4488
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004489default-server [param*]
4490 Change default options for a server in a backend
4491 May be used in sections : defaults | frontend | listen | backend
4492 yes | no | yes | yes
4493 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01004494 <param*> is a list of parameters for this server. The "default-server"
4495 keyword accepts an important number of options and has a complete
4496 section dedicated to it. Please refer to section 5 for more
4497 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004498
Willy Tarreau983e01e2010-01-11 18:42:06 +01004499 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004500 default-server inter 1000 weight 13
4501
4502 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01004503
Willy Tarreau983e01e2010-01-11 18:42:06 +01004504
Willy Tarreau0ba27502007-12-24 16:55:16 +01004505default_backend <backend>
4506 Specify the backend to use when no "use_backend" rule has been matched.
4507 May be used in sections : defaults | frontend | listen | backend
4508 yes | yes | yes | no
4509 Arguments :
4510 <backend> is the name of the backend to use.
4511
4512 When doing content-switching between frontend and backends using the
4513 "use_backend" keyword, it is often useful to indicate which backend will be
4514 used when no rule has matched. It generally is the dynamic backend which
4515 will catch all undetermined requests.
4516
Willy Tarreau0ba27502007-12-24 16:55:16 +01004517 Example :
4518
4519 use_backend dynamic if url_dyn
4520 use_backend static if url_css url_img extension_img
4521 default_backend dynamic
4522
Willy Tarreau98d04852015-05-26 12:18:29 +02004523 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004524
Willy Tarreau0ba27502007-12-24 16:55:16 +01004525
Baptiste Assmann27f51342013-10-09 06:51:49 +02004526description <string>
4527 Describe a listen, frontend or backend.
4528 May be used in sections : defaults | frontend | listen | backend
4529 no | yes | yes | yes
4530 Arguments : string
4531
4532 Allows to add a sentence to describe the related object in the HAProxy HTML
4533 stats page. The description will be printed on the right of the object name
4534 it describes.
4535 No need to backslash spaces in the <string> arguments.
4536
4537
Willy Tarreau0ba27502007-12-24 16:55:16 +01004538disabled
4539 Disable a proxy, frontend or backend.
4540 May be used in sections : defaults | frontend | listen | backend
4541 yes | yes | yes | yes
4542 Arguments : none
4543
4544 The "disabled" keyword is used to disable an instance, mainly in order to
4545 liberate a listening port or to temporarily disable a service. The instance
4546 will still be created and its configuration will be checked, but it will be
4547 created in the "stopped" state and will appear as such in the statistics. It
4548 will not receive any traffic nor will it send any health-checks or logs. It
4549 is possible to disable many instances at once by adding the "disabled"
4550 keyword in a "defaults" section.
4551
4552 See also : "enabled"
4553
4554
Willy Tarreau5ce94572010-06-07 14:35:41 +02004555dispatch <address>:<port>
4556 Set a default server address
4557 May be used in sections : defaults | frontend | listen | backend
4558 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004559 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02004560
4561 <address> is the IPv4 address of the default server. Alternatively, a
4562 resolvable hostname is supported, but this name will be resolved
4563 during start-up.
4564
4565 <ports> is a mandatory port specification. All connections will be sent
4566 to this port, and it is not permitted to use port offsets as is
4567 possible with normal servers.
4568
Willy Tarreau787aed52011-04-15 06:45:37 +02004569 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02004570 server can take the connection. In the past it was used to forward non
4571 persistent connections to an auxiliary load balancer. Due to its simple
4572 syntax, it has also been used for simple TCP relays. It is recommended not to
4573 use it for more clarity, and to use the "server" directive instead.
4574
4575 See also : "server"
4576
Olivier Houchard4e694042017-03-14 20:01:29 +01004577
4578dynamic-cookie-key <string>
4579 Set the dynamic cookie secret key for a backend.
4580 May be used in sections : defaults | frontend | listen | backend
4581 yes | no | yes | yes
4582 Arguments : The secret key to be used.
4583
4584 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01004585 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01004586 specified on the "server" line), using a hash of the IP address of the
4587 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004588 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01004589 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02004590
Willy Tarreau0ba27502007-12-24 16:55:16 +01004591enabled
4592 Enable a proxy, frontend or backend.
4593 May be used in sections : defaults | frontend | listen | backend
4594 yes | yes | yes | yes
4595 Arguments : none
4596
4597 The "enabled" keyword is used to explicitly enable an instance, when the
4598 defaults has been set to "disabled". This is very rarely used.
4599
4600 See also : "disabled"
4601
4602
4603errorfile <code> <file>
4604 Return a file contents instead of errors generated by HAProxy
4605 May be used in sections : defaults | frontend | listen | backend
4606 yes | yes | yes | yes
4607 Arguments :
4608 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004609 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004610 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004611
4612 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004613 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01004614 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01004615 error pages, and to use absolute paths, since files are read
4616 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004617
4618 It is important to understand that this keyword is not meant to rewrite
4619 errors returned by the server, but errors detected and returned by HAProxy.
4620 This is why the list of supported errors is limited to a small set.
4621
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004622 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4623
Christopher Faulet70170672020-05-18 17:42:48 +02004624 The files are parsed when HAProxy starts and must be valid according to the
4625 HTTP specification. They should not exceed the configured buffer size
4626 (BUFSIZE), which generally is 16 kB, otherwise an internal error will be
4627 returned. It is also wise not to put any reference to local contents
4628 (e.g. images) in order to avoid loops between the client and HAProxy when all
4629 servers are down, causing an error to be returned instead of an
4630 image. Finally, The response cannot exceed (tune.bufsize - tune.maxrewrite)
4631 so that "http-after-response" rules still have room to operate (see
4632 "tune.maxrewrite").
Willy Tarreau59140a22009-02-22 12:02:30 +01004633
Willy Tarreau0ba27502007-12-24 16:55:16 +01004634 The files are read at the same time as the configuration and kept in memory.
4635 For this reason, the errors continue to be returned even when the process is
4636 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01004637 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01004638 403 status code and interrogating a blocked URL.
4639
Christopher Faulet3b967c12020-05-15 15:47:44 +02004640 See also : "http-error", "errorloc", "errorloc302", "errorloc303"
Willy Tarreau0ba27502007-12-24 16:55:16 +01004641
Willy Tarreau59140a22009-02-22 12:02:30 +01004642 Example :
4643 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01004644 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01004645 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
4646 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
4647
Willy Tarreau2769aa02007-12-27 18:26:09 +01004648
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004649errorfiles <name> [<code> ...]
4650 Import, fully or partially, the error files defined in the <name> http-errors
4651 section.
4652 May be used in sections : defaults | frontend | listen | backend
4653 yes | yes | yes | yes
4654 Arguments :
4655 <name> is the name of an existing http-errors section.
4656
4657 <code> is a HTTP status code. Several status code may be listed.
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004658 Currently, HAProxy is capable of generating codes 200, 400, 401,
Christopher Faulete095f312020-12-07 11:22:24 +01004659 403, 404, 405, 407, 408, 410, 413, 425, 429, 500, 501, 502, 503,
4660 and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004661
4662 Errors defined in the http-errors section with the name <name> are imported
4663 in the current proxy. If no status code is specified, all error files of the
4664 http-errors section are imported. Otherwise, only error files associated to
4665 the listed status code are imported. Those error files override the already
4666 defined custom errors for the proxy. And they may be overridden by following
Daniel Corbett67a82712020-07-06 23:01:19 -04004667 ones. Functionally, it is exactly the same as declaring all error files by
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004668 hand using "errorfile" directives.
4669
Christopher Faulet3b967c12020-05-15 15:47:44 +02004670 See also : "http-error", "errorfile", "errorloc", "errorloc302" ,
4671 "errorloc303" and section 3.8 about http-errors.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004672
4673 Example :
4674 errorfiles generic
4675 errorfiles site-1 403 404
4676
4677
Willy Tarreau2769aa02007-12-27 18:26:09 +01004678errorloc <code> <url>
4679errorloc302 <code> <url>
4680 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4681 May be used in sections : defaults | frontend | listen | backend
4682 yes | yes | yes | yes
4683 Arguments :
4684 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004685 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004686 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004687
4688 <url> it is the exact contents of the "Location" header. It may contain
4689 either a relative URI to an error page hosted on the same site,
4690 or an absolute URI designating an error page on another site.
4691 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004692 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004693
4694 It is important to understand that this keyword is not meant to rewrite
4695 errors returned by the server, but errors detected and returned by HAProxy.
4696 This is why the list of supported errors is limited to a small set.
4697
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004698 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4699
Willy Tarreau2769aa02007-12-27 18:26:09 +01004700 Note that both keyword return the HTTP 302 status code, which tells the
4701 client to fetch the designated URL using the same HTTP method. This can be
4702 quite problematic in case of non-GET methods such as POST, because the URL
4703 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01004704 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01004705 status code, indicating to the client that the URL must be fetched with a GET
4706 request.
4707
Christopher Faulet3b967c12020-05-15 15:47:44 +02004708 See also : "http-error", "errorfile", "errorloc303"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004709
4710
4711errorloc303 <code> <url>
4712 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4713 May be used in sections : defaults | frontend | listen | backend
4714 yes | yes | yes | yes
4715 Arguments :
4716 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004717 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004718 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004719
4720 <url> it is the exact contents of the "Location" header. It may contain
4721 either a relative URI to an error page hosted on the same site,
4722 or an absolute URI designating an error page on another site.
4723 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004724 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004725
4726 It is important to understand that this keyword is not meant to rewrite
4727 errors returned by the server, but errors detected and returned by HAProxy.
4728 This is why the list of supported errors is limited to a small set.
4729
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004730 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4731
Willy Tarreau2769aa02007-12-27 18:26:09 +01004732 Note that both keyword return the HTTP 303 status code, which tells the
4733 client to fetch the designated URL using the same HTTP GET method. This
4734 solves the usual problems associated with "errorloc" and the 302 code. It is
4735 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004736 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004737
Christopher Faulet3b967c12020-05-15 15:47:44 +02004738 See also : "http-error", "errorfile", "errorloc", "errorloc302"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004739
4740
Simon Horman51a1cf62015-02-03 13:00:44 +09004741email-alert from <emailaddr>
4742 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01004743 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09004744 May be used in sections: defaults | frontend | listen | backend
4745 yes | yes | yes | yes
4746
4747 Arguments :
4748
4749 <emailaddr> is the from email address to use when sending email alerts
4750
4751 Also requires "email-alert mailers" and "email-alert to" to be set
4752 and if so sending email alerts is enabled for the proxy.
4753
Simon Horman64e34162015-02-06 11:11:57 +09004754 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02004755 "email-alert myhostname", "email-alert to", section 3.6 about
4756 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09004757
4758
4759email-alert level <level>
4760 Declare the maximum log level of messages for which email alerts will be
4761 sent. This acts as a filter on the sending of email alerts.
4762 May be used in sections: defaults | frontend | listen | backend
4763 yes | yes | yes | yes
4764
4765 Arguments :
4766
4767 <level> One of the 8 syslog levels:
4768 emerg alert crit err warning notice info debug
4769 The above syslog levels are ordered from lowest to highest.
4770
4771 By default level is alert
4772
4773 Also requires "email-alert from", "email-alert mailers" and
4774 "email-alert to" to be set and if so sending email alerts is enabled
4775 for the proxy.
4776
Simon Horman1421e212015-04-30 13:10:35 +09004777 Alerts are sent when :
4778
4779 * An un-paused server is marked as down and <level> is alert or lower
4780 * A paused server is marked as down and <level> is notice or lower
4781 * A server is marked as up or enters the drain state and <level>
4782 is notice or lower
4783 * "option log-health-checks" is enabled, <level> is info or lower,
4784 and a health check status update occurs
4785
Simon Horman64e34162015-02-06 11:11:57 +09004786 See also : "email-alert from", "email-alert mailers",
4787 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09004788 section 3.6 about mailers.
4789
4790
4791email-alert mailers <mailersect>
4792 Declare the mailers to be used when sending email alerts
4793 May be used in sections: defaults | frontend | listen | backend
4794 yes | yes | yes | yes
4795
4796 Arguments :
4797
4798 <mailersect> is the name of the mailers section to send email alerts.
4799
4800 Also requires "email-alert from" and "email-alert to" to be set
4801 and if so sending email alerts is enabled for the proxy.
4802
Simon Horman64e34162015-02-06 11:11:57 +09004803 See also : "email-alert from", "email-alert level", "email-alert myhostname",
4804 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004805
4806
4807email-alert myhostname <hostname>
4808 Declare the to hostname address to be used when communicating with
4809 mailers.
4810 May be used in sections: defaults | frontend | listen | backend
4811 yes | yes | yes | yes
4812
4813 Arguments :
4814
Baptiste Assmann738bad92015-12-21 15:27:53 +01004815 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09004816
4817 By default the systems hostname is used.
4818
4819 Also requires "email-alert from", "email-alert mailers" and
4820 "email-alert to" to be set and if so sending email alerts is enabled
4821 for the proxy.
4822
Simon Horman64e34162015-02-06 11:11:57 +09004823 See also : "email-alert from", "email-alert level", "email-alert mailers",
4824 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004825
4826
4827email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01004828 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09004829 header of email alerts. This is the address that email alerts are sent to.
4830 May be used in sections: defaults | frontend | listen | backend
4831 yes | yes | yes | yes
4832
4833 Arguments :
4834
4835 <emailaddr> is the to email address to use when sending email alerts
4836
4837 Also requires "email-alert mailers" and "email-alert to" to be set
4838 and if so sending email alerts is enabled for the proxy.
4839
Simon Horman64e34162015-02-06 11:11:57 +09004840 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09004841 "email-alert myhostname", section 3.6 about mailers.
4842
4843
Willy Tarreau4de91492010-01-22 19:10:05 +01004844force-persist { if | unless } <condition>
4845 Declare a condition to force persistence on down servers
4846 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01004847 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01004848
4849 By default, requests are not dispatched to down servers. It is possible to
4850 force this using "option persist", but it is unconditional and redispatches
4851 to a valid server if "option redispatch" is set. That leaves with very little
4852 possibilities to force some requests to reach a server which is artificially
4853 marked down for maintenance operations.
4854
4855 The "force-persist" statement allows one to declare various ACL-based
4856 conditions which, when met, will cause a request to ignore the down status of
4857 a server and still try to connect to it. That makes it possible to start a
4858 server, still replying an error to the health checks, and run a specially
4859 configured browser to test the service. Among the handy methods, one could
4860 use a specific source IP address, or a specific cookie. The cookie also has
4861 the advantage that it can easily be added/removed on the browser from a test
4862 page. Once the service is validated, it is then possible to open the service
4863 to the world by returning a valid response to health checks.
4864
4865 The forced persistence is enabled when an "if" condition is met, or unless an
4866 "unless" condition is met. The final redispatch is always disabled when this
4867 is used.
4868
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004869 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02004870 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01004871
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004872
4873filter <name> [param*]
4874 Add the filter <name> in the filter list attached to the proxy.
4875 May be used in sections : defaults | frontend | listen | backend
4876 no | yes | yes | yes
4877 Arguments :
4878 <name> is the name of the filter. Officially supported filters are
4879 referenced in section 9.
4880
Tim Düsterhus4896c442016-11-29 02:15:19 +01004881 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004882 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01004883 filter. Please refer to the documentation of the corresponding
4884 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004885
4886 Multiple occurrences of the filter line can be used for the same proxy. The
4887 same filter can be referenced many times if needed.
4888
4889 Example:
4890 listen
4891 bind *:80
4892
4893 filter trace name BEFORE-HTTP-COMP
4894 filter compression
4895 filter trace name AFTER-HTTP-COMP
4896
4897 compression algo gzip
4898 compression offload
4899
4900 server srv1 192.168.0.1:80
4901
4902 See also : section 9.
4903
Willy Tarreau4de91492010-01-22 19:10:05 +01004904
Willy Tarreau2769aa02007-12-27 18:26:09 +01004905fullconn <conns>
4906 Specify at what backend load the servers will reach their maxconn
4907 May be used in sections : defaults | frontend | listen | backend
4908 yes | no | yes | yes
4909 Arguments :
4910 <conns> is the number of connections on the backend which will make the
4911 servers use the maximal number of connections.
4912
Willy Tarreau198a7442008-01-17 12:05:32 +01004913 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01004914 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01004915 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01004916 load. The server will then always accept at least <minconn> connections,
4917 never more than <maxconn>, and the limit will be on the ramp between both
4918 values when the backend has less than <conns> concurrent connections. This
4919 makes it possible to limit the load on the servers during normal loads, but
4920 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004921 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004922
Willy Tarreaufbb78422011-06-05 15:38:35 +02004923 Since it's hard to get this value right, haproxy automatically sets it to
4924 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01004925 backend (based on "use_backend" and "default_backend" rules). That way it's
4926 safe to leave it unset. However, "use_backend" involving dynamic names are
4927 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02004928
Willy Tarreau2769aa02007-12-27 18:26:09 +01004929 Example :
4930 # The servers will accept between 100 and 1000 concurrent connections each
4931 # and the maximum of 1000 will be reached when the backend reaches 10000
4932 # connections.
4933 backend dynamic
4934 fullconn 10000
4935 server srv1 dyn1:80 minconn 100 maxconn 1000
4936 server srv2 dyn2:80 minconn 100 maxconn 1000
4937
4938 See also : "maxconn", "server"
4939
4940
Willy Tarreauab0a5192020-10-09 19:07:01 +02004941grace <time> (deprecated)
Willy Tarreau2769aa02007-12-27 18:26:09 +01004942 Maintain a proxy operational for some time after a soft stop
4943 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01004944 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01004945 Arguments :
4946 <time> is the time (by default in milliseconds) for which the instance
4947 will remain operational with the frontend sockets still listening
4948 when a soft-stop is received via the SIGUSR1 signal.
4949
4950 This may be used to ensure that the services disappear in a certain order.
4951 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004952 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01004953 needed by the equipment to detect the failure.
4954
4955 Note that currently, there is very little benefit in using this parameter,
4956 and it may in fact complicate the soft-reconfiguration process more than
4957 simplify it.
4958
Willy Tarreau0ba27502007-12-24 16:55:16 +01004959
Andrew Rodland17be45e2016-10-25 17:04:12 -04004960hash-balance-factor <factor>
4961 Specify the balancing factor for bounded-load consistent hashing
4962 May be used in sections : defaults | frontend | listen | backend
4963 yes | no | no | yes
4964 Arguments :
4965 <factor> is the control for the maximum number of concurrent requests to
4966 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01004967 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04004968
4969 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
4970 enables an algorithm that prevents any one server from getting too many
4971 requests at once, even if some hash buckets receive many more requests than
4972 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
4973 <factor> is a percentage greater than 100. For example, if <factor> is 150,
4974 then no server will be allowed to have a load more than 1.5 times the average.
4975 If server weights are used, they will be respected.
4976
4977 If the first-choice server is disqualified, the algorithm will choose another
4978 server based on the request hash, until a server with additional capacity is
4979 found. A higher <factor> allows more imbalance between the servers, while a
4980 lower <factor> means that more servers will be checked on average, affecting
4981 performance. Reasonable values are from 125 to 200.
4982
Willy Tarreau760e81d2018-05-03 07:20:40 +02004983 This setting is also used by "balance random" which internally relies on the
4984 consistent hashing mechanism.
4985
Andrew Rodland17be45e2016-10-25 17:04:12 -04004986 See also : "balance" and "hash-type".
4987
4988
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004989hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004990 Specify a method to use for mapping hashes to servers
4991 May be used in sections : defaults | frontend | listen | backend
4992 yes | no | yes | yes
4993 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04004994 <method> is the method used to select a server from the hash computed by
4995 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004996
Bhaskar98634f02013-10-29 23:30:51 -04004997 map-based the hash table is a static array containing all alive servers.
4998 The hashes will be very smooth, will consider weights, but
4999 will be static in that weight changes while a server is up
5000 will be ignored. This means that there will be no slow start.
5001 Also, since a server is selected by its position in the array,
5002 most mappings are changed when the server count changes. This
5003 means that when a server goes up or down, or when a server is
5004 added to a farm, most connections will be redistributed to
5005 different servers. This can be inconvenient with caches for
5006 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01005007
Bhaskar98634f02013-10-29 23:30:51 -04005008 consistent the hash table is a tree filled with many occurrences of each
5009 server. The hash key is looked up in the tree and the closest
5010 server is chosen. This hash is dynamic, it supports changing
5011 weights while the servers are up, so it is compatible with the
5012 slow start feature. It has the advantage that when a server
5013 goes up or down, only its associations are moved. When a
5014 server is added to the farm, only a few part of the mappings
5015 are redistributed, making it an ideal method for caches.
5016 However, due to its principle, the distribution will never be
5017 very smooth and it may sometimes be necessary to adjust a
5018 server's weight or its ID to get a more balanced distribution.
5019 In order to get the same distribution on multiple load
5020 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005021 same IDs. Note: consistent hash uses sdbm and avalanche if no
5022 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04005023
5024 <function> is the hash function to be used :
5025
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005026 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04005027 reimplementation of ndbm) database library. It was found to do
5028 well in scrambling bits, causing better distribution of the keys
5029 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005030 function with good distribution, unless the total server weight
5031 is a multiple of 64, in which case applying the avalanche
5032 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04005033
5034 djb2 this function was first proposed by Dan Bernstein many years ago
5035 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005036 function provides a better distribution than sdbm. It generally
5037 works well with text-based inputs though it can perform extremely
5038 poorly with numeric-only input or when the total server weight is
5039 a multiple of 33, unless the avalanche modifier is also used.
5040
Willy Tarreaua0f42712013-11-14 14:30:35 +01005041 wt6 this function was designed for haproxy while testing other
5042 functions in the past. It is not as smooth as the other ones, but
5043 is much less sensible to the input data set or to the number of
5044 servers. It can make sense as an alternative to sdbm+avalanche or
5045 djb2+avalanche for consistent hashing or when hashing on numeric
5046 data such as a source IP address or a visitor identifier in a URL
5047 parameter.
5048
Willy Tarreau324f07f2015-01-20 19:44:50 +01005049 crc32 this is the most common CRC32 implementation as used in Ethernet,
5050 gzip, PNG, etc. It is slower than the other ones but may provide
5051 a better distribution or less predictable results especially when
5052 used on strings.
5053
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005054 <modifier> indicates an optional method applied after hashing the key :
5055
5056 avalanche This directive indicates that the result from the hash
5057 function above should not be used in its raw form but that
5058 a 4-byte full avalanche hash must be applied first. The
5059 purpose of this step is to mix the resulting bits from the
5060 previous hash in order to avoid any undesired effect when
5061 the input contains some limited values or when the number of
5062 servers is a multiple of one of the hash's components (64
5063 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
5064 result less predictable, but it's also not as smooth as when
5065 using the original function. Some testing might be needed
5066 with some workloads. This hash is one of the many proposed
5067 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005068
Bhaskar98634f02013-10-29 23:30:51 -04005069 The default hash type is "map-based" and is recommended for most usages. The
5070 default function is "sdbm", the selection of a function should be based on
5071 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005072
Andrew Rodland17be45e2016-10-25 17:04:12 -04005073 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005074
5075
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005076http-after-response <action> <options...> [ { if | unless } <condition> ]
5077 Access control for all Layer 7 responses (server, applet/service and internal
5078 ones).
5079
5080 May be used in sections: defaults | frontend | listen | backend
5081 no | yes | yes | yes
5082
5083 The http-after-response statement defines a set of rules which apply to layer
5084 7 processing. The rules are evaluated in their declaration order when they
5085 are met in a frontend, listen or backend section. Any rule may optionally be
5086 followed by an ACL-based condition, in which case it will only be evaluated
5087 if the condition is true. Since these rules apply on responses, the backend
5088 rules are applied first, followed by the frontend's rules.
5089
5090 Unlike http-response rules, these ones are applied on all responses, the
5091 server ones but also to all responses generated by HAProxy. These rules are
5092 evaluated at the end of the responses analysis, before the data forwarding.
5093
5094 The first keyword is the rule's action. The supported actions are described
5095 below.
5096
5097 There is no limit to the number of http-after-response statements per
5098 instance.
5099
Christopher Fauletd5ac6de2020-12-02 08:40:14 +01005100 Note: Errors emitted in early stage of the request parsing are handled by the
5101 multiplexer at a lower level, before any http analysis. Thus no
5102 http-after-response ruleset is evaluated on these errors.
5103
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005104 Example:
5105 http-after-response set-header Strict-Transport-Security "max-age=31536000"
5106 http-after-response set-header Cache-Control "no-store,no-cache,private"
5107 http-after-response set-header Pragma "no-cache"
5108
5109http-after-response add-header <name> <fmt> [ { if | unless } <condition> ]
5110
5111 This appends an HTTP header field whose name is specified in <name> and whose
5112 value is defined by <fmt> which follows the log-format rules (see Custom Log
5113 Format in section 8.2.4). This may be used to send a cookie to a client for
5114 example, or to pass some internal information.
5115 This rule is not final, so it is possible to add other similar rules.
5116 Note that header addition is performed immediately, so one rule might reuse
5117 the resulting header from a previous rule.
5118
5119http-after-response allow [ { if | unless } <condition> ]
5120
5121 This stops the evaluation of the rules and lets the response pass the check.
5122 No further "http-after-response" rules are evaluated.
5123
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005124http-after-response del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005125
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005126 This removes all HTTP header fields whose name is specified in <name>. <meth>
5127 is the matching method, applied on the header name. Supported matching methods
5128 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
5129 (substring match) and "reg" (regex match). If not specified, exact matching
5130 method is used.
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005131
5132http-after-response replace-header <name> <regex-match> <replace-fmt>
5133 [ { if | unless } <condition> ]
5134
5135 This works like "http-response replace-header".
5136
5137 Example:
5138 http-after-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
5139
5140 # applied to:
5141 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
5142
5143 # outputs:
5144 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
5145
5146 # assuming the backend IP is 192.168.1.20.
5147
5148http-after-response replace-value <name> <regex-match> <replace-fmt>
5149 [ { if | unless } <condition> ]
5150
5151 This works like "http-response replace-value".
5152
5153 Example:
5154 http-after-response replace-value Cache-control ^public$ private
5155
5156 # applied to:
5157 Cache-Control: max-age=3600, public
5158
5159 # outputs:
5160 Cache-Control: max-age=3600, private
5161
5162http-after-response set-header <name> <fmt> [ { if | unless } <condition> ]
5163
5164 This does the same as "add-header" except that the header name is first
5165 removed if it existed. This is useful when passing security information to
5166 the server, where the header must not be manipulated by external users.
5167
5168http-after-response set-status <status> [reason <str>]
5169 [ { if | unless } <condition> ]
5170
5171 This replaces the response status code with <status> which must be an integer
5172 between 100 and 999. Optionally, a custom reason text can be provided defined
5173 by <str>, or the default reason for the specified code will be used as a
5174 fallback.
5175
5176 Example:
5177 # return "431 Request Header Fields Too Large"
5178 http-response set-status 431
5179 # return "503 Slow Down", custom reason
5180 http-response set-status 503 reason "Slow Down"
5181
5182http-after-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5183
5184 This is used to set the contents of a variable. The variable is declared
5185 inline.
5186
5187 Arguments:
5188 <var-name> The name of the variable starts with an indication about its
5189 scope. The scopes allowed are:
5190 "proc" : the variable is shared with the whole process
5191 "sess" : the variable is shared with the whole session
5192 "txn" : the variable is shared with the transaction
5193 (request and response)
5194 "req" : the variable is shared only during request
5195 processing
5196 "res" : the variable is shared only during response
5197 processing
5198 This prefix is followed by a name. The separator is a '.'.
5199 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5200 and '_'.
5201
5202 <expr> Is a standard HAProxy expression formed by a sample-fetch
5203 followed by some converters.
5204
5205 Example:
5206 http-after-response set-var(sess.last_redir) res.hdr(location)
5207
5208http-after-response strict-mode { on | off }
5209
5210 This enables or disables the strict rewriting mode for following rules. It
5211 does not affect rules declared before it and it is only applicable on rules
5212 performing a rewrite on the responses. When the strict mode is enabled, any
5213 rewrite failure triggers an internal error. Otherwise, such errors are
5214 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005215 rewrites optional while others must be performed to continue the response
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005216 processing.
5217
5218 By default, the strict rewriting mode is enabled. Its value is also reset
5219 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04005220 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005221 rules evaluation.
5222
5223http-after-response unset-var(<var-name>) [ { if | unless } <condition> ]
5224
5225 This is used to unset a variable. See "http-after-response set-var" for
5226 details about <var-name>.
5227
5228 Example:
5229 http-after-response unset-var(sess.last_redir)
5230
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005231
5232http-check comment <string>
5233 Defines a comment for the following the http-check rule, reported in logs if
5234 it fails.
5235 May be used in sections : defaults | frontend | listen | backend
5236 yes | no | yes | yes
5237
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005238 Arguments :
5239 <string> is the comment message to add in logs if the following http-check
5240 rule fails.
5241
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005242 It only works for connect, send and expect rules. It is useful to make
5243 user-friendly error reporting.
5244
Daniel Corbett67a82712020-07-06 23:01:19 -04005245 See also : "option httpchk", "http-check connect", "http-check send" and
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005246 "http-check expect".
5247
5248
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005249http-check connect [default] [port <expr>] [addr <ip>] [send-proxy]
5250 [via-socks4] [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +02005251 [proto <name>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02005252 Opens a new connection to perform an HTTP health check
5253 May be used in sections : defaults | frontend | listen | backend
5254 yes | no | yes | yes
5255
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005256 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005257 comment <msg> defines a message to report if the rule evaluation fails.
5258
Christopher Faulete5870d82020-04-15 11:32:03 +02005259 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -04005260 checks. The server options are used only if not redefined.
Christopher Faulete5870d82020-04-15 11:32:03 +02005261
5262 port <expr> if not set, check port or server port is used.
5263 It tells HAProxy where to open the connection to.
5264 <port> must be a valid TCP port source integer, from 1 to
5265 65535 or an sample-fetch expression.
5266
5267 addr <ip> defines the IP address to do the health check.
5268
5269 send-proxy send a PROXY protocol string
5270
5271 via-socks4 enables outgoing health checks using upstream socks4 proxy.
5272
5273 ssl opens a ciphered connection
5274
5275 sni <sni> specifies the SNI to use to do health checks over SSL.
5276
5277 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
5278 list consists in a comma-delimited list of protocol names,
5279 for instance: "h2,http/1.1". If it is not set, the server ALPN
5280 is used.
5281
Christopher Fauletedc6ed92020-04-23 16:27:59 +02005282 proto <name> forces the multiplexer's protocol to use for this connection.
5283 It must be an HTTP mux protocol and it must be usable on the
5284 backend side. The list of available protocols is reported in
5285 haproxy -vv.
5286
Christopher Faulete5870d82020-04-15 11:32:03 +02005287 linger cleanly close the connection instead of using a single RST.
5288
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005289 Just like tcp-check health checks, it is possible to configure the connection
5290 to use to perform HTTP health check. This directive should also be used to
5291 describe a scenario involving several request/response exchanges, possibly on
5292 different ports or with different servers.
5293
5294 When there are no TCP port configured on the server line neither server port
5295 directive, then the first step of the http-check sequence must be to specify
5296 the port with a "http-check connect".
5297
5298 In an http-check ruleset a 'connect' is required, it is also mandatory to start
5299 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
5300 do.
5301
5302 When a connect must start the ruleset, if may still be preceded by set-var,
5303 unset-var or comment rules.
5304
5305 Examples :
Christopher Faulete5870d82020-04-15 11:32:03 +02005306 # check HTTP and HTTPs services on a server.
5307 # first open port 80 thanks to server line port directive, then
5308 # tcp-check opens port 443, ciphered and run a request on it:
5309 option httpchk
5310
5311 http-check connect
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02005312 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005313 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02005314 http-check connect port 443 ssl sni haproxy.1wt.eu
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02005315 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005316 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02005317
5318 server www 10.0.0.1 check port 80
5319
5320 See also : "option httpchk", "http-check send", "http-check expect"
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005321
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005322
Willy Tarreau0ba27502007-12-24 16:55:16 +01005323http-check disable-on-404
5324 Enable a maintenance mode upon HTTP/404 response to health-checks
5325 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01005326 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01005327 Arguments : none
5328
5329 When this option is set, a server which returns an HTTP code 404 will be
5330 excluded from further load-balancing, but will still receive persistent
5331 connections. This provides a very convenient method for Web administrators
5332 to perform a graceful shutdown of their servers. It is also important to note
5333 that a server which is detected as failed while it was in this mode will not
5334 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
5335 will immediately be reinserted into the farm. The status on the stats page
5336 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01005337 option only works in conjunction with the "httpchk" option. If this option
5338 is used with "http-check expect", then it has precedence over it so that 404
Christopher Fauletfa8b89a2020-11-20 18:54:13 +01005339 responses will still be considered as soft-stop. Note also that a stopped
5340 server will stay stopped even if it replies 404s. This option is only
5341 evaluated for running servers.
Willy Tarreaubd741542010-03-16 18:46:54 +01005342
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005343 See also : "option httpchk" and "http-check expect".
Willy Tarreaubd741542010-03-16 18:46:54 +01005344
5345
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005346http-check expect [min-recv <int>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02005347 [ok-status <st>] [error-status <st>] [tout-status <st>]
5348 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
5349 [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005350 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01005351 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02005352 yes | no | yes | yes
Christopher Faulete5870d82020-04-15 11:32:03 +02005353
Willy Tarreaubd741542010-03-16 18:46:54 +01005354 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005355 comment <msg> defines a message to report if the rule evaluation fails.
5356
Christopher Faulete5870d82020-04-15 11:32:03 +02005357 min-recv is optional and can define the minimum amount of data required to
5358 evaluate the current expect rule. If the number of received bytes
5359 is under this limit, the check will wait for more data. This
5360 option can be used to resolve some ambiguous matching rules or to
5361 avoid executing costly regex matches on content known to be still
5362 incomplete. If an exact string is used, the minimum between the
5363 string length and this parameter is used. This parameter is
5364 ignored if it is set to -1. If the expect rule does not match,
5365 the check will wait for more data. If set to 0, the evaluation
5366 result is always conclusive.
5367
5368 ok-status <st> is optional and can be used to set the check status if
5369 the expect rule is successfully evaluated and if it is
5370 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005371 "L6OK" and "L4OK" are supported :
5372 - L7OK : check passed on layer 7
Christopher Faulet83662b52020-11-20 17:47:47 +01005373 - L7OKC : check conditionally passed on layer 7, set
5374 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005375 - L6OK : check passed on layer 6
5376 - L4OK : check passed on layer 4
5377 By default "L7OK" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02005378
5379 error-status <st> is optional and can be used to set the check status if
5380 an error occurred during the expect rule evaluation.
Christopher Faulet83662b52020-11-20 17:47:47 +01005381 "L7OKC", "L7RSP", "L7STS", "L6RSP" and "L4CON" are
5382 supported :
5383 - L7OKC : check conditionally passed on layer 7, set
5384 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005385 - L7RSP : layer 7 invalid response - protocol error
5386 - L7STS : layer 7 response error, for example HTTP 5xx
5387 - L6RSP : layer 6 invalid response - protocol error
5388 - L4CON : layer 1-4 connection problem
5389 By default "L7RSP" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02005390
5391 tout-status <st> is optional and can be used to set the check status if
5392 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005393 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
5394 - L7TOUT : layer 7 (HTTP/SMTP) timeout
5395 - L6TOUT : layer 6 (SSL) timeout
5396 - L4TOUT : layer 1-4 timeout
Christopher Faulete5870d82020-04-15 11:32:03 +02005397 By default "L7TOUT" is used.
5398
5399 on-success <fmt> is optional and can be used to customize the
5400 informational message reported in logs if the expect
5401 rule is successfully evaluated and if it is the last rule
5402 in the tcp-check ruleset. <fmt> is a log-format string.
5403
5404 on-error <fmt> is optional and can be used to customize the
5405 informational message reported in logs if an error
5406 occurred during the expect rule evaluation. <fmt> is a
5407 log-format string.
5408
Willy Tarreaubd741542010-03-16 18:46:54 +01005409 <match> is a keyword indicating how to look for a specific pattern in the
Christopher Fauletb5594262020-05-05 20:23:13 +02005410 response. The keyword may be one of "status", "rstatus", "hdr",
5411 "fhdr", "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01005412 exclamation mark ("!") to negate the match. Spaces are allowed
5413 between the exclamation mark and the keyword. See below for more
5414 details on the supported keywords.
5415
Christopher Faulet39708192020-05-05 10:47:36 +02005416 <pattern> is the pattern to look for. It may be a string, a regular
5417 expression or a more complex pattern with several arguments. If
5418 the string pattern contains spaces, they must be escaped with the
5419 usual backslash ('\').
Willy Tarreaubd741542010-03-16 18:46:54 +01005420
5421 By default, "option httpchk" considers that response statuses 2xx and 3xx
5422 are valid, and that others are invalid. When "http-check expect" is used,
5423 it defines what is considered valid or invalid. Only one "http-check"
5424 statement is supported in a backend. If a server fails to respond or times
5425 out, the check obviously fails. The available matches are :
5426
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005427 status <codes> : test the status codes found parsing <codes> string. it
5428 must be a comma-separated list of status codes or range
5429 codes. A health check response will be considered as
5430 valid if the response's status code matches any status
5431 code or is inside any range of the list. If the "status"
5432 keyword is prefixed with "!", then the response will be
5433 considered invalid if the status code matches.
Willy Tarreaubd741542010-03-16 18:46:54 +01005434
5435 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005436 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005437 response's status code matches the expression. If the
5438 "rstatus" keyword is prefixed with "!", then the response
5439 will be considered invalid if the status code matches.
5440 This is mostly used to check for multiple codes.
5441
Christopher Fauletb5594262020-05-05 20:23:13 +02005442 hdr { name | name-lf } [ -m <meth> ] <name>
5443 [ { value | value-lf } [ -m <meth> ] <value> :
Christopher Faulet39708192020-05-05 10:47:36 +02005444 test the specified header pattern on the HTTP response
5445 headers. The name pattern is mandatory but the value
5446 pattern is optional. If not specified, only the header
5447 presence is verified. <meth> is the matching method,
5448 applied on the header name or the header value. Supported
5449 matching methods are "str" (exact match), "beg" (prefix
5450 match), "end" (suffix match), "sub" (substring match) or
5451 "reg" (regex match). If not specified, exact matching
Christopher Fauletb5594262020-05-05 20:23:13 +02005452 method is used. If the "name-lf" parameter is used,
5453 <name> is evaluated as a log-format string. If "value-lf"
5454 parameter is used, <value> is evaluated as a log-format
5455 string. These parameters cannot be used with the regex
5456 matching method. Finally, the header value is considered
5457 as comma-separated list. Note that matchings are case
5458 insensitive on the header names.
5459
5460 fhdr { name | name-lf } [ -m <meth> ] <name>
5461 [ { value | value-lf } [ -m <meth> ] <value> :
5462 test the specified full header pattern on the HTTP
5463 response headers. It does exactly the same than "hdr"
5464 keyword, except the full header value is tested, commas
5465 are not considered as delimiters.
Christopher Faulet39708192020-05-05 10:47:36 +02005466
Willy Tarreaubd741542010-03-16 18:46:54 +01005467 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005468 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005469 response's body contains this exact string. If the
5470 "string" keyword is prefixed with "!", then the response
5471 will be considered invalid if the body contains this
5472 string. This can be used to look for a mandatory word at
5473 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005474 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01005475 trace).
5476
5477 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005478 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005479 response's body matches this expression. If the "rstring"
5480 keyword is prefixed with "!", then the response will be
5481 considered invalid if the body matches the expression.
5482 This can be used to look for a mandatory word at the end
5483 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01005484 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01005485
Christopher Fauletaaab0832020-05-05 15:54:22 +02005486 string-lf <fmt> : test a log-format string match in the HTTP response body.
5487 A health check response will be considered valid if the
5488 response's body contains the string resulting of the
5489 evaluation of <fmt>, which follows the log-format rules.
5490 If prefixed with "!", then the response will be
5491 considered invalid if the body contains the string.
5492
Willy Tarreaubd741542010-03-16 18:46:54 +01005493 It is important to note that the responses will be limited to a certain size
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +01005494 defined by the global "tune.bufsize" option, which defaults to 16384 bytes.
Willy Tarreaubd741542010-03-16 18:46:54 +01005495 Thus, too large responses may not contain the mandatory pattern when using
5496 "string" or "rstring". If a large response is absolutely required, it is
5497 possible to change the default max size by setting the global variable.
5498 However, it is worth keeping in mind that parsing very large responses can
5499 waste some CPU cycles, especially when regular expressions are used, and that
5500 it is always better to focus the checks on smaller resources.
5501
Christopher Faulete5870d82020-04-15 11:32:03 +02005502 In an http-check ruleset, the last expect rule may be implicit. If no expect
5503 rule is specified after the last "http-check send", an implicit expect rule
5504 is defined to match on 2xx or 3xx status codes. It means this rule is also
5505 defined if there is no "http-check" rule at all, when only "option httpchk"
5506 is set.
Cyril Bonté32602d22015-01-30 00:07:07 +01005507
Willy Tarreaubd741542010-03-16 18:46:54 +01005508 Last, if "http-check expect" is combined with "http-check disable-on-404",
5509 then this last one has precedence when the server responds with 404.
5510
5511 Examples :
5512 # only accept status 200 as valid
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005513 http-check expect status 200,201,300-310
Willy Tarreaubd741542010-03-16 18:46:54 +01005514
Christopher Faulet39708192020-05-05 10:47:36 +02005515 # be sure a sessid coookie is set
5516 http-check expect header name "set-cookie" value -m beg "sessid="
5517
Willy Tarreaubd741542010-03-16 18:46:54 +01005518 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005519 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01005520
5521 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005522 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01005523
5524 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005525 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01005526
Christopher Faulete5870d82020-04-15 11:32:03 +02005527 See also : "option httpchk", "http-check connect", "http-check disable-on-404"
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005528 and "http-check send".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005529
5530
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005531http-check send [meth <method>] [{ uri <uri> | uri-lf <fmt> }>] [ver <version>]
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005532 [hdr <name> <fmt>]* [{ body <string> | body-lf <fmt> }]
5533 [comment <msg>]
Christopher Faulet8acb1282020-04-09 08:44:06 +02005534 Add a possible list of headers and/or a body to the request sent during HTTP
5535 health checks.
5536 May be used in sections : defaults | frontend | listen | backend
5537 yes | no | yes | yes
5538 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005539 comment <msg> defines a message to report if the rule evaluation fails.
5540
Christopher Faulete5870d82020-04-15 11:32:03 +02005541 meth <method> is the optional HTTP method used with the requests. When not
5542 set, the "OPTIONS" method is used, as it generally requires
5543 low server processing and is easy to filter out from the
5544 logs. Any method may be used, though it is not recommended
5545 to invent non-standard ones.
5546
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005547 uri <uri> is optional and set the URI referenced in the HTTP requests
5548 to the string <uri>. It defaults to "/" which is accessible
5549 by default on almost any server, but may be changed to any
5550 other URI. Query strings are permitted.
5551
5552 uri-lf <fmt> is optional and set the URI referenced in the HTTP requests
5553 using the log-format string <fmt>. It defaults to "/" which
5554 is accessible by default on almost any server, but may be
5555 changed to any other URI. Query strings are permitted.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005556
Christopher Faulet907701b2020-04-28 09:37:00 +02005557 ver <version> is the optional HTTP version string. It defaults to
Christopher Faulete5870d82020-04-15 11:32:03 +02005558 "HTTP/1.0" but some servers might behave incorrectly in HTTP
Daniel Corbett67a82712020-07-06 23:01:19 -04005559 1.0, so turning it to HTTP/1.1 may sometimes help. Note that
Christopher Faulete5870d82020-04-15 11:32:03 +02005560 the Host field is mandatory in HTTP/1.1, use "hdr" argument
5561 to add it.
5562
5563 hdr <name> <fmt> adds the HTTP header field whose name is specified in
5564 <name> and whose value is defined by <fmt>, which follows
5565 to the log-format rules.
5566
5567 body <string> add the body defined by <string> to the request sent during
5568 HTTP health checks. If defined, the "Content-Length" header
5569 is thus automatically added to the request.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005570
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005571 body-lf <fmt> add the body defined by the log-format string <fmt> to the
5572 request sent during HTTP health checks. If defined, the
5573 "Content-Length" header is thus automatically added to the
5574 request.
5575
Christopher Faulet8acb1282020-04-09 08:44:06 +02005576 In addition to the request line defined by the "option httpchk" directive,
5577 this one is the valid way to add some headers and optionally a body to the
5578 request sent during HTTP health checks. If a body is defined, the associate
Christopher Faulet9df910c2020-04-29 14:20:47 +02005579 "Content-Length" header is automatically added. Thus, this header or
5580 "Transfer-encoding" header should not be present in the request provided by
5581 "http-check send". If so, it will be ignored. The old trick consisting to add
5582 headers after the version string on the "option httpchk" line is now
Amaury Denoyelle6d975f02020-12-22 14:08:52 +01005583 deprecated.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005584
Christopher Faulete5870d82020-04-15 11:32:03 +02005585 Also "http-check send" doesn't support HTTP keep-alive. Keep in mind that it
Amaury Denoyelle6d975f02020-12-22 14:08:52 +01005586 will automatically append a "Connection: close" header, unless a Connection
5587 header has already already been configured via a hdr entry.
Christopher Faulet9df910c2020-04-29 14:20:47 +02005588
5589 Note that the Host header and the request authority, when both defined, are
5590 automatically synchronized. It means when the HTTP request is sent, when a
5591 Host is inserted in the request, the request authority is accordingly
5592 updated. Thus, don't be surprised if the Host header value overwrites the
5593 configured request authority.
5594
5595 Note also for now, no Host header is automatically added in HTTP/1.1 or above
5596 requests. You should add it explicitly.
Christopher Faulete5870d82020-04-15 11:32:03 +02005597
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005598 See also : "option httpchk", "http-check send-state" and "http-check expect".
Christopher Faulet8acb1282020-04-09 08:44:06 +02005599
5600
Willy Tarreauef781042010-01-27 11:53:01 +01005601http-check send-state
5602 Enable emission of a state header with HTTP health checks
5603 May be used in sections : defaults | frontend | listen | backend
5604 yes | no | yes | yes
5605 Arguments : none
5606
5607 When this option is set, haproxy will systematically send a special header
5608 "X-Haproxy-Server-State" with a list of parameters indicating to each server
5609 how they are seen by haproxy. This can be used for instance when a server is
5610 manipulated without access to haproxy and the operator needs to know whether
5611 haproxy still sees it up or not, or if the server is the last one in a farm.
5612
5613 The header is composed of fields delimited by semi-colons, the first of which
5614 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
5615 checks on the total number before transition, just as appears in the stats
5616 interface. Next headers are in the form "<variable>=<value>", indicating in
5617 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08005618 - a variable "address", containing the address of the backend server.
5619 This corresponds to the <address> field in the server declaration. For
5620 unix domain sockets, it will read "unix".
5621
5622 - a variable "port", containing the port of the backend server. This
5623 corresponds to the <port> field in the server declaration. For unix
5624 domain sockets, it will read "unix".
5625
Willy Tarreauef781042010-01-27 11:53:01 +01005626 - a variable "name", containing the name of the backend followed by a slash
5627 ("/") then the name of the server. This can be used when a server is
5628 checked in multiple backends.
5629
5630 - a variable "node" containing the name of the haproxy node, as set in the
5631 global "node" variable, otherwise the system's hostname if unspecified.
5632
5633 - a variable "weight" indicating the weight of the server, a slash ("/")
5634 and the total weight of the farm (just counting usable servers). This
5635 helps to know if other servers are available to handle the load when this
5636 one fails.
5637
5638 - a variable "scur" indicating the current number of concurrent connections
5639 on the server, followed by a slash ("/") then the total number of
5640 connections on all servers of the same backend.
5641
5642 - a variable "qcur" indicating the current number of requests in the
5643 server's queue.
5644
5645 Example of a header received by the application server :
5646 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
5647 scur=13/22; qcur=0
5648
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005649 See also : "option httpchk", "http-check disable-on-404" and
5650 "http-check send".
Willy Tarreauef781042010-01-27 11:53:01 +01005651
Christopher Faulete5870d82020-04-15 11:32:03 +02005652
5653http-check set-var(<var-name>) <expr>
Christopher Faulete5870d82020-04-15 11:32:03 +02005654 This operation sets the content of a variable. The variable is declared inline.
Christopher Faulete5870d82020-04-15 11:32:03 +02005655 May be used in sections: defaults | frontend | listen | backend
5656 yes | no | yes | yes
5657
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005658 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005659 <var-name> The name of the variable starts with an indication about its
5660 scope. The scopes allowed for http-check are:
5661 "proc" : the variable is shared with the whole process.
5662 "sess" : the variable is shared with the tcp-check session.
5663 "check": the variable is declared for the lifetime of the tcp-check.
5664 This prefix is followed by a name. The separator is a '.'.
5665 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5666 and '-'.
5667
5668 <expr> Is a sample-fetch expression potentially followed by converters.
5669
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005670 Examples :
5671 http-check set-var(check.port) int(1234)
Christopher Faulete5870d82020-04-15 11:32:03 +02005672
5673
5674http-check unset-var(<var-name>)
Christopher Faulete5870d82020-04-15 11:32:03 +02005675 Free a reference to a variable within its scope.
Christopher Faulete5870d82020-04-15 11:32:03 +02005676 May be used in sections: defaults | frontend | listen | backend
5677 yes | no | yes | yes
5678
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005679 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005680 <var-name> The name of the variable starts with an indication about its
5681 scope. The scopes allowed for http-check are:
5682 "proc" : the variable is shared with the whole process.
5683 "sess" : the variable is shared with the tcp-check session.
5684 "check": the variable is declared for the lifetime of the tcp-check.
5685 This prefix is followed by a name. The separator is a '.'.
5686 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5687 and '-'.
5688
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005689 Examples :
5690 http-check unset-var(check.port)
Christopher Faulete5870d82020-04-15 11:32:03 +02005691
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005692
Christopher Faulet3b967c12020-05-15 15:47:44 +02005693http-error status <code> [content-type <type>]
5694 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5695 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5696 [ hdr <name> <fmt> ]*
5697 Defines a custom error message to use instead of errors generated by HAProxy.
5698 May be used in sections : defaults | frontend | listen | backend
5699 yes | yes | yes | yes
5700 Arguments :
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005701 status <code> is the HTTP status code. It must be specified.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005702 Currently, HAProxy is capable of generating codes
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005703 200, 400, 401, 403, 404, 405, 407, 408, 410, 413, 425,
Christopher Faulete095f312020-12-07 11:22:24 +01005704 429, 500, 501, 502, 503, and 504.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005705
5706 content-type <type> is the response content type, for instance
5707 "text/plain". This parameter is ignored and should be
5708 omitted when an errorfile is configured or when the
5709 payload is empty. Otherwise, it must be defined.
5710
5711 default-errorfiles Reset the previously defined error message for current
5712 proxy for the status <code>. If used on a backend, the
5713 frontend error message is used, if defined. If used on
5714 a frontend, the default error message is used.
5715
5716 errorfile <file> designates a file containing the full HTTP response.
5717 It is recommended to follow the common practice of
5718 appending ".http" to the filename so that people do
5719 not confuse the response with HTML error pages, and to
5720 use absolute paths, since files are read before any
5721 chroot is performed.
5722
5723 errorfiles <name> designates the http-errors section to use to import
5724 the error message with the status code <code>. If no
5725 such message is found, the proxy's error messages are
5726 considered.
5727
5728 file <file> specifies the file to use as response payload. If the
5729 file is not empty, its content-type must be set as
5730 argument to "content-type", otherwise, any
5731 "content-type" argument is ignored. <file> is
5732 considered as a raw string.
5733
5734 string <str> specifies the raw string to use as response payload.
5735 The content-type must always be set as argument to
5736 "content-type".
5737
5738 lf-file <file> specifies the file to use as response payload. If the
5739 file is not empty, its content-type must be set as
5740 argument to "content-type", otherwise, any
5741 "content-type" argument is ignored. <file> is
5742 evaluated as a log-format string.
5743
5744 lf-string <str> specifies the log-format string to use as response
5745 payload. The content-type must always be set as
5746 argument to "content-type".
5747
5748 hdr <name> <fmt> adds to the response the HTTP header field whose name
5749 is specified in <name> and whose value is defined by
5750 <fmt>, which follows to the log-format rules.
5751 This parameter is ignored if an errorfile is used.
5752
5753 This directive may be used instead of "errorfile", to define a custom error
5754 message. As "errorfile" directive, it is used for errors detected and
5755 returned by HAProxy. If an errorfile is defined, it is parsed when HAProxy
5756 starts and must be valid according to the HTTP standards. The generated
5757 response must not exceed the configured buffer size (BUFFSIZE), otherwise an
5758 internal error will be returned. Finally, if you consider to use some
5759 http-after-response rules to rewrite these errors, the reserved buffer space
5760 should be available (see "tune.maxrewrite").
5761
5762 The files are read at the same time as the configuration and kept in memory.
5763 For this reason, the errors continue to be returned even when the process is
5764 chrooted, and no file change is considered while the process is running.
5765
Christopher Fauletd5ac6de2020-12-02 08:40:14 +01005766 Note: 400/408/500 errors emitted in early stage of the request parsing are
5767 handled by the multiplexer at a lower level. No custom formatting is
5768 supported at this level. Thus only static error messages, defined with
5769 "errorfile" directive, are supported. However, this limitation only
5770 exists during the request headers parsing or between two transactions.
5771
Christopher Faulet3b967c12020-05-15 15:47:44 +02005772 See also : "errorfile", "errorfiles", "errorloc", "errorloc302",
5773 "errorloc303" and section 3.8 about http-errors.
5774
5775
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005776http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005777 Access control for Layer 7 requests
5778
5779 May be used in sections: defaults | frontend | listen | backend
5780 no | yes | yes | yes
5781
Willy Tarreau20b0de52012-12-24 15:45:22 +01005782 The http-request statement defines a set of rules which apply to layer 7
5783 processing. The rules are evaluated in their declaration order when they are
5784 met in a frontend, listen or backend section. Any rule may optionally be
5785 followed by an ACL-based condition, in which case it will only be evaluated
5786 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005787
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005788 The first keyword is the rule's action. The supported actions are described
5789 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005790
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005791 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005792
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005793 Example:
5794 acl nagios src 192.168.129.3
5795 acl local_net src 192.168.0.0/16
5796 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01005797
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005798 http-request allow if nagios
5799 http-request allow if local_net auth_ok
5800 http-request auth realm Gimme if local_net auth_ok
5801 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01005802
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005803 Example:
5804 acl key req.hdr(X-Add-Acl-Key) -m found
5805 acl add path /addacl
5806 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01005807
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005808 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01005809
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005810 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
5811 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02005812
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005813 Example:
5814 acl value req.hdr(X-Value) -m found
5815 acl setmap path /setmap
5816 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005817
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005818 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06005819
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005820 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
5821 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005822
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005823 See also : "stats http-request", section 3.4 about userlists and section 7
5824 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06005825
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005826http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005827
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005828 This is used to add a new entry into an ACL. The ACL must be loaded from a
5829 file (even a dummy empty file). The file name of the ACL to be updated is
5830 passed between parentheses. It takes one argument: <key fmt>, which follows
5831 log-format rules, to collect content of the new entry. It performs a lookup
5832 in the ACL before insertion, to avoid duplicated (or more) values. This
5833 lookup is done by a linear search and can be expensive with large lists!
5834 It is the equivalent of the "add acl" command from the stats socket, but can
5835 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06005836
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005837http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005838
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005839 This appends an HTTP header field whose name is specified in <name> and
5840 whose value is defined by <fmt> which follows the log-format rules (see
5841 Custom Log Format in section 8.2.4). This is particularly useful to pass
5842 connection-specific information to the server (e.g. the client's SSL
5843 certificate), or to combine several headers into one. This rule is not
5844 final, so it is possible to add other similar rules. Note that header
5845 addition is performed immediately, so one rule might reuse the resulting
5846 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06005847
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005848http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005849
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005850 This stops the evaluation of the rules and lets the request pass the check.
5851 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06005852
Sasha Pachev218f0642014-06-16 12:05:59 -06005853
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005854http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005855
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005856 This stops the evaluation of the rules and immediately responds with an
5857 HTTP 401 or 407 error code to invite the user to present a valid user name
5858 and password. No further "http-request" rules are evaluated. An optional
5859 "realm" parameter is supported, it sets the authentication realm that is
5860 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06005861
Christopher Faulet612f2ea2020-05-27 09:57:28 +02005862 The corresponding proxy's error message is used. It may be customized using
5863 an "errorfile" or an "http-error" directive. For 401 responses, all
5864 occurrences of the WWW-Authenticate header are removed and replaced by a new
5865 one with a basic authentication challenge for realm "<realm>". For 407
5866 responses, the same is done on the Proxy-Authenticate header. If the error
5867 message must not be altered, consider to use "http-request return" rule
5868 instead.
5869
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005870 Example:
5871 acl auth_ok http_auth_group(L1) G1
5872 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06005873
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02005874http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005875
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005876 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005877
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005878http-request capture <sample> [ len <length> | id <id> ]
5879 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005880
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005881 This captures sample expression <sample> from the request buffer, and
5882 converts it to a string of at most <len> characters. The resulting string is
5883 stored into the next request "capture" slot, so it will possibly appear next
5884 to some captured HTTP headers. It will then automatically appear in the logs,
5885 and it will be possible to extract it using sample fetch rules to feed it
5886 into headers or anything. The length should be limited given that this size
5887 will be allocated for each capture during the whole session life.
5888 Please check section 7.3 (Fetching samples) and "capture request header" for
5889 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005890
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005891 If the keyword "id" is used instead of "len", the action tries to store the
5892 captured string in a previously declared capture slot. This is useful to run
5893 captures in backends. The slot id can be declared by a previous directive
Baptiste Assmann19a69b32020-01-16 14:34:22 +01005894 "http-request capture" or with the "declare capture" keyword.
5895
5896 When using this action in a backend, double check that the relevant
5897 frontend(s) have the required capture slots otherwise, this rule will be
5898 ignored at run time. This can't be detected at configuration parsing time
5899 due to HAProxy's ability to dynamically resolve backend name at runtime.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005900
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005901http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005902
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005903 This is used to delete an entry from an ACL. The ACL must be loaded from a
5904 file (even a dummy empty file). The file name of the ACL to be updated is
5905 passed between parentheses. It takes one argument: <key fmt>, which follows
5906 log-format rules, to collect content of the entry to delete.
5907 It is the equivalent of the "del acl" command from the stats socket, but can
5908 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005909
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005910http-request del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02005911
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005912 This removes all HTTP header fields whose name is specified in <name>. <meth>
5913 is the matching method, applied on the header name. Supported matching methods
5914 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
5915 (substring match) and "reg" (regex match). If not specified, exact matching
5916 method is used.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02005917
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005918http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02005919
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005920 This is used to delete an entry from a MAP. The MAP must be loaded from a
5921 file (even a dummy empty file). The file name of the MAP to be updated is
5922 passed between parentheses. It takes one argument: <key fmt>, which follows
5923 log-format rules, to collect content of the entry to delete.
5924 It takes one argument: "file name" It is the equivalent of the "del map"
5925 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02005926
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005927http-request deny [deny_status <status>] [ { if | unless } <condition> ]
5928http-request deny [ { status | deny_status } <code>] [content-type <type>]
5929 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5930 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5931 [ hdr <name> <fmt> ]*
5932 [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04005933
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005934 This stops the evaluation of the rules and immediately rejects the request.
5935 By default an HTTP 403 error is returned. But the response may be customized
5936 using same syntax than "http-request return" rules. Thus, see "http-request
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005937 return" for details. For compatibility purpose, when no argument is defined,
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005938 or only "deny_status", the argument "default-errorfiles" is implied. It means
5939 "http-request deny [deny_status <status>]" is an alias of
5940 "http-request deny [status <status>] default-errorfiles".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005941 No further "http-request" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005942 See also "http-request return".
Patrick Hemmer268a7072018-05-11 12:52:31 -04005943
Olivier Houchard602bf7d2019-05-10 13:59:15 +02005944http-request disable-l7-retry [ { if | unless } <condition> ]
5945 This disables any attempt to retry the request if it fails for any other
5946 reason than a connection failure. This can be useful for example to make
5947 sure POST requests aren't retried on failure.
5948
Baptiste Assmann333939c2019-01-21 08:34:50 +01005949http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
5950
5951 This action performs a DNS resolution of the output of <expr> and stores
5952 the result in the variable <var>. It uses the DNS resolvers section
5953 pointed by <resolvers>.
5954 It is possible to choose a resolution preference using the optional
5955 arguments 'ipv4' or 'ipv6'.
5956 When performing the DNS resolution, the client side connection is on
5957 pause waiting till the end of the resolution.
5958 If an IP address can be found, it is stored into <var>. If any kind of
5959 error occurs, then <var> is not set.
5960 One can use this action to discover a server IP address at run time and
5961 based on information found in the request (IE a Host header).
5962 If this action is used to find the server's IP address (using the
5963 "set-dst" action), then the server IP address in the backend must be set
5964 to 0.0.0.0.
5965
5966 Example:
5967 resolvers mydns
5968 nameserver local 127.0.0.53:53
5969 nameserver google 8.8.8.8:53
5970 timeout retry 1s
5971 hold valid 10s
5972 hold nx 3s
5973 hold other 3s
5974 hold obsolete 0s
5975 accepted_payload_size 8192
5976
5977 frontend fe
5978 bind 10.42.0.1:80
5979 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
5980 http-request capture var(txn.myip) len 40
5981
5982 # return 503 when the variable is not set,
5983 # which mean DNS resolution error
5984 use_backend b_503 unless { var(txn.myip) -m found }
5985
5986 default_backend be
5987
5988 backend b_503
5989 # dummy backend used to return 503.
5990 # one can use the errorfile directive to send a nice
5991 # 503 error page to end users
5992
5993 backend be
5994 # rule to prevent HAProxy from reconnecting to services
5995 # on the local network (forged DNS name used to scan the network)
5996 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
5997 http-request set-dst var(txn.myip)
5998 server clear 0.0.0.0:0
5999
6000 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
6001 be used to scan the network or worst won't loop over itself...
6002
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01006003http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
6004
6005 This is used to build an HTTP 103 Early Hints response prior to any other one.
6006 This appends an HTTP header field to this response whose name is specified in
6007 <name> and whose value is defined by <fmt> which follows the log-format rules
6008 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01006009 to the client some Link headers to preload resources required to render the
6010 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01006011
6012 See RFC 8297 for more information.
6013
Tim Duesterhusd371e992021-04-15 21:45:58 +02006014http-request normalize-uri <normalizer> [ { if | unless } <condition> ]
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006015http-request normalize-uri path-merge-slashes [ { if | unless } <condition> ]
6016http-request normalize-uri path-strip-dotdot [ full ] [ { if | unless } <condition> ]
6017http-request normalize-uri percent-to-uppercase [ strict ] [ { if | unless } <condition> ]
6018http-request normalize-uri query-sort-by-name [ { if | unless } <condition> ]
Tim Duesterhusd371e992021-04-15 21:45:58 +02006019
Tim Duesterhusb918a4a2021-04-16 23:52:29 +02006020 Performs normalization of the request's URI.
6021
Tim Duesterhus2963fd32021-04-17 00:24:56 +02006022 URI normalization in HAProxy 2.4 is currently available as an experimental
6023 technical preview. You should be prepared that the behavior of normalizers
6024 might change to fix possible issues, possibly breaking proper request
6025 processing in your infrastructure.
6026
Tim Duesterhusb918a4a2021-04-16 23:52:29 +02006027 Each normalizer handles a single type of normalization to allow for a
6028 fine-grained selection of the level of normalization that is appropriate for
6029 the supported backend.
6030
6031 As an example the "path-strip-dotdot" normalizer might be useful for a static
6032 fileserver that directly maps the requested URI to the path within the local
6033 filesystem. However it might break routing of an API that expects a specific
6034 number of segments in the path.
6035
6036 The following normalizers are available:
Tim Duesterhusd371e992021-04-15 21:45:58 +02006037
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006038 - path-strip-dotdot: Normalizes "/../" segments within the "path" component.
6039 This merges segments that attempt to access the parent directory with
6040 their preceding segment. Empty segments do not receive special treatment.
6041 Use the "path-merge-slashes" normalizer first if this is undesired.
Tim Duesterhus9982fc22021-04-15 21:45:59 +02006042
6043 Example:
6044 - /foo/../ -> /
6045 - /foo/../bar/ -> /bar/
6046 - /foo/bar/../ -> /foo/
6047 - /../bar/ -> /../bar/
Tim Duesterhus560e1a62021-04-15 21:46:00 +02006048 - /bar/../../ -> /../
Tim Duesterhus9982fc22021-04-15 21:45:59 +02006049 - /foo//../ -> /foo/
6050
Tim Duesterhus560e1a62021-04-15 21:46:00 +02006051 If the "full" option is specified then "../" at the beginning will be
6052 removed as well:
6053
6054 Example:
6055 - /../bar/ -> /bar/
6056 - /bar/../../ -> /
6057
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006058 - path-merge-slashes: Merges adjacent slashes within the "path" component
6059 into a single slash.
Tim Duesterhusd371e992021-04-15 21:45:58 +02006060
6061 Example:
6062 - // -> /
6063 - /foo//bar -> /foo/bar
6064
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006065 - percent-to-uppercase: Uppercases letters within percent-encoded sequences
Tim Duesterhusa4071932021-04-15 21:46:02 +02006066 (RFC 3986#6.2.21).
6067
6068 Example:
6069 - /%6f -> /%6F
6070 - /%zz -> /%zz
6071
6072 If the "strict" option is specified then invalid sequences will result
6073 in a HTTP 400 Bad Request being returned.
6074
6075 Example:
6076 - /%zz -> HTTP 400
6077
Tim Duesterhus5be6ab22021-04-17 11:21:10 +02006078 - query-sort-by-name: Sorts the query string parameters by parameter name.
Tim Duesterhusd7b89be2021-04-15 21:46:01 +02006079 Parameters are assumed to be delimited by '&'. Shorter names sort before
6080 longer names and identical parameter names maintain their relative order.
6081
6082 Example:
6083 - /?c=3&a=1&b=2 -> /?a=1&b=2&c=3
6084 - /?aaa=3&a=1&aa=2 -> /?a=1&aa=2&aaa=3
6085 - /?a=3&b=4&a=1&b=5&a=2 -> /?a=3&a=1&a=2&b=4&b=5
6086
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006087http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006088
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006089 This performs an HTTP redirection based on a redirect rule. This is exactly
6090 the same as the "redirect" statement except that it inserts a redirect rule
6091 which can be processed in the middle of other "http-request" rules and that
6092 these rules use the "log-format" strings. See the "redirect" keyword for the
6093 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006094
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006095http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006096
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006097 This stops the evaluation of the rules and immediately closes the connection
6098 without sending any response. It acts similarly to the
6099 "tcp-request content reject" rules. It can be useful to force an immediate
6100 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006101
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006102http-request replace-header <name> <match-regex> <replace-fmt>
6103 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02006104
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006105 This matches the value of all occurrences of header field <name> against
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006106 <match-regex>. Matching is performed case-sensitively. Matching values are
6107 completely replaced by <replace-fmt>. Format characters are allowed in
6108 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
6109 Standard back-references using the backslash ('\') followed by a number are
6110 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02006111
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006112 This action acts on whole header lines, regardless of the number of values
6113 they may contain. Thus it is well-suited to process headers naturally
6114 containing commas in their value, such as If-Modified-Since. Headers that
6115 contain a comma-separated list of values, such as Accept, should be processed
6116 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01006117
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006118 Example:
6119 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
6120
6121 # applied to:
6122 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
6123
6124 # outputs:
6125 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
6126
6127 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02006128
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006129 http-request replace-header User-Agent curl foo
6130
6131 # applied to:
6132 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02006133
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006134 # outputs:
6135 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02006136
Willy Tarreau262c3f12019-12-17 06:52:51 +01006137http-request replace-path <match-regex> <replace-fmt>
6138 [ { if | unless } <condition> ]
6139
6140 This works like "replace-header" except that it works on the request's path
6141 component instead of a header. The path component starts at the first '/'
Christopher Faulet82c83322020-09-02 14:16:59 +02006142 after an optional scheme+authority and ends before the question mark. Thus,
6143 the replacement does not modify the scheme, the authority and the
6144 query-string.
Willy Tarreau262c3f12019-12-17 06:52:51 +01006145
6146 It is worth noting that regular expressions may be more expensive to evaluate
6147 than certain ACLs, so rare replacements may benefit from a condition to avoid
6148 performing the evaluation at all if it does not match.
6149
6150 Example:
6151 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
6152 http-request replace-path (.*) /foo\1
6153
Willy Tarreau262c3f12019-12-17 06:52:51 +01006154 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
6155 http-request replace-path /foo/(.*) /\1
6156 # or more efficient if only some requests match :
6157 http-request replace-path /foo/(.*) /\1 if { url_beg /foo/ }
6158
Christopher Faulet312294f2020-09-02 17:17:44 +02006159http-request replace-pathq <match-regex> <replace-fmt>
6160 [ { if | unless } <condition> ]
6161
6162 This does the same as "http-request replace-path" except that the path
6163 contains the query-string if any is present. Thus, the path and the
6164 query-string are replaced.
6165
6166 Example:
6167 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
6168 http-request replace-pathq ([^?]*)(\?(.*))? \1/foo\2
6169
Willy Tarreau33810222019-06-12 17:44:02 +02006170http-request replace-uri <match-regex> <replace-fmt>
6171 [ { if | unless } <condition> ]
6172
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006173 This works like "replace-header" except that it works on the request's URI part
6174 instead of a header. The URI part may contain an optional scheme, authority or
6175 query string. These are considered to be part of the value that is matched
6176 against.
6177
6178 It is worth noting that regular expressions may be more expensive to evaluate
6179 than certain ACLs, so rare replacements may benefit from a condition to avoid
6180 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02006181
Willy Tarreau62b59132019-12-17 06:51:20 +01006182 IMPORTANT NOTE: historically in HTTP/1.x, the vast majority of requests sent
6183 by browsers use the "origin form", which differs from the "absolute form" in
6184 that they do not contain a scheme nor authority in the URI portion. Mostly
6185 only requests sent to proxies, those forged by hand and some emitted by
6186 certain applications use the absolute form. As such, "replace-uri" usually
6187 works fine most of the time in HTTP/1.x with rules starting with a "/". But
6188 with HTTP/2, clients are encouraged to send absolute URIs only, which look
6189 like the ones HTTP/1 clients use to talk to proxies. Such partial replace-uri
6190 rules may then fail in HTTP/2 when they work in HTTP/1. Either the rules need
Willy Tarreau262c3f12019-12-17 06:52:51 +01006191 to be adapted to optionally match a scheme and authority, or replace-path
6192 should be used.
Willy Tarreau33810222019-06-12 17:44:02 +02006193
Willy Tarreau62b59132019-12-17 06:51:20 +01006194 Example:
6195 # rewrite all "http" absolute requests to "https":
6196 http-request replace-uri ^http://(.*) https://\1
Willy Tarreau33810222019-06-12 17:44:02 +02006197
Willy Tarreau62b59132019-12-17 06:51:20 +01006198 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
6199 http-request replace-uri ([^/:]*://[^/]*)?(.*) \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02006200
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006201http-request replace-value <name> <match-regex> <replace-fmt>
6202 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02006203
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006204 This works like "replace-header" except that it matches the regex against
6205 every comma-delimited value of the header field <name> instead of the
6206 entire header. This is suited for all headers which are allowed to carry
6207 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02006208
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006209 Example:
6210 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02006211
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006212 # applied to:
6213 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02006214
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006215 # outputs:
6216 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01006217
Christopher Faulet24231ab2020-01-24 17:44:23 +01006218http-request return [status <code>] [content-type <type>]
6219 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6220 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006221 [ hdr <name> <fmt> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01006222 [ { if | unless } <condition> ]
6223
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006224 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01006225 default status code used for the response is 200. It can be optionally
6226 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04006227 specified as an argument to "content-type". Finally the response itself may
Sébastien Grossab877122020-10-08 10:06:03 +02006228 be defined. It can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006229 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01006230 are followed to create the response :
6231
6232 * If neither the errorfile nor the payload to use is defined, a dummy
6233 response is returned. Only the "status" argument is considered. It can be
6234 any code in the range [200, 599]. The "content-type" argument, if any, is
6235 ignored.
6236
6237 * If "default-errorfiles" argument is set, the proxy's errorfiles are
6238 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04006239 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Christopher Faulete095f312020-12-07 11:22:24 +01006240 425, 429, 500, 501, 502, 503, and 504). The "content-type" argument, if
6241 any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006242
6243 * If a specific errorfile is defined, with an "errorfile" argument, the
6244 corresponding file, containing a full HTTP response, is returned. Only the
6245 "status" argument is considered. It must be one of the status code handled
Christopher Faulete095f312020-12-07 11:22:24 +01006246 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 501,
6247 502, 503, and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006248
6249 * If an http-errors section is defined, with an "errorfiles" argument, the
6250 corresponding file in the specified http-errors section, containing a full
6251 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04006252 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Christopher Faulete095f312020-12-07 11:22:24 +01006253 408, 410, 413, 425, 429, 500, 501, 502, 503, and 504). The "content-type"
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006254 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006255
6256 * If a "file" or a "lf-file" argument is specified, the file's content is
6257 used as the response payload. If the file is not empty, its content-type
6258 must be set as argument to "content-type". Otherwise, any "content-type"
6259 argument is ignored. With a "lf-file" argument, the file's content is
6260 evaluated as a log-format string. With a "file" argument, it is considered
6261 as a raw content.
6262
6263 * If a "string" or "lf-string" argument is specified, the defined string is
6264 used as the response payload. The content-type must always be set as
6265 argument to "content-type". With a "lf-string" argument, the string is
6266 evaluated as a log-format string. With a "string" argument, it is
6267 considered as a raw string.
6268
Sébastien Grossab877122020-10-08 10:06:03 +02006269 When the response is not based on an errorfile, it is possible to append HTTP
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006270 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
6271 arguments are ignored. For each one, the header name is specified in <name>
6272 and its value is defined by <fmt> which follows the log-format rules.
6273
Christopher Faulet24231ab2020-01-24 17:44:23 +01006274 Note that the generated response must be smaller than a buffer. And to avoid
6275 any warning, when an errorfile or a raw file is loaded, the buffer space
Sébastien Grossab877122020-10-08 10:06:03 +02006276 reserved for the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006277
6278 No further "http-request" rules are evaluated.
6279
6280 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006281 http-request return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006282 if { path /ping }
6283
6284 http-request return content-type image/x-icon file /var/www/favicon.ico \
6285 if { path /favicon.ico }
6286
6287 http-request return status 403 content-type text/plain \
6288 lf-string "Access denied. IP %[src] is blacklisted." \
6289 if { src -f /etc/haproxy/blacklist.lst }
6290
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006291http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6292http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006293
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006294 This actions increments the GPC0 or GPC1 counter according with the sticky
6295 counter designated by <sc-id>. If an error occurs, this action silently fails
6296 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006297
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006298http-request sc-set-gpt0(<sc-id>) { <int> | <expr> }
6299 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006300
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006301 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6302 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
6303 boolean. If an error occurs, this action silently fails and the actions
6304 evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006305
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006306http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006307
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006308 This is used to set the destination IP address to the value of specified
6309 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
6310 but provides the correct IP in a HTTP header; or you want to mask the IP for
6311 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
6312 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01006313
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006314 Arguments:
6315 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6316 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01006317
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006318 Example:
6319 http-request set-dst hdr(x-dst)
6320 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01006321
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006322 When possible, set-dst preserves the original destination port as long as the
6323 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006324
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006325http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006326
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006327 This is used to set the destination port address to the value of specified
6328 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
6329 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006330
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006331 Arguments:
6332 <expr> Is a standard HAProxy expression formed by a sample-fetch
6333 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006334
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006335 Example:
6336 http-request set-dst-port hdr(x-port)
6337 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006338
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006339 When possible, set-dst-port preserves the original destination address as
6340 long as the address family supports a port, otherwise it forces the
6341 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02006342
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006343http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02006344
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006345 This does the same as "http-request add-header" except that the header name
6346 is first removed if it existed. This is useful when passing security
6347 information to the server, where the header must not be manipulated by
6348 external users. Note that the new value is computed before the removal so it
6349 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02006350
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006351 Example:
6352 http-request set-header X-Haproxy-Current-Date %T
6353 http-request set-header X-SSL %[ssl_fc]
6354 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
6355 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
6356 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
6357 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
6358 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
6359 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
6360 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02006361
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006362http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02006363
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006364 This is used to change the log level of the current request when a certain
6365 condition is met. Valid levels are the 8 syslog levels (see the "log"
6366 keyword) plus the special level "silent" which disables logging for this
6367 request. This rule is not final so the last matching rule wins. This rule
6368 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006369
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006370http-request set-map(<file-name>) <key fmt> <value fmt>
6371 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006372
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006373 This is used to add a new entry into a MAP. The MAP must be loaded from a
6374 file (even a dummy empty file). The file name of the MAP to be updated is
6375 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
6376 log-format rules, used to collect MAP key, and <value fmt>, which follows
6377 log-format rules, used to collect content for the new entry.
6378 It performs a lookup in the MAP before insertion, to avoid duplicated (or
6379 more) values. This lookup is done by a linear search and can be expensive
6380 with large lists! It is the equivalent of the "set map" command from the
6381 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006382
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006383http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006384
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006385 This is used to set the Netfilter MARK on all packets sent to the client to
6386 the value passed in <mark> on platforms which support it. This value is an
6387 unsigned 32 bit value which can be matched by netfilter and by the routing
6388 table. It can be expressed both in decimal or hexadecimal format (prefixed by
6389 "0x"). This can be useful to force certain packets to take a different route
6390 (for example a cheaper network path for bulk downloads). This works on Linux
6391 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02006392
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006393http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006394
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006395 This rewrites the request method with the result of the evaluation of format
6396 string <fmt>. There should be very few valid reasons for having to do so as
6397 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006398
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006399http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006400
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006401 This sets the "nice" factor of the current request being processed. It only
6402 has effect against the other requests being processed at the same time.
6403 The default value is 0, unless altered by the "nice" setting on the "bind"
6404 line. The accepted range is -1024..1024. The higher the value, the nicest
6405 the request will be. Lower values will make the request more important than
6406 other ones. This can be useful to improve the speed of some requests, or
6407 lower the priority of non-important requests. Using this setting without
6408 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006409
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006410http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02006411
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006412 This rewrites the request path with the result of the evaluation of format
6413 string <fmt>. The query string, if any, is left intact. If a scheme and
6414 authority is found before the path, they are left intact as well. If the
6415 request doesn't have a path ("*"), this one is replaced with the format.
6416 This can be used to prepend a directory component in front of a path for
6417 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02006418
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006419 Example :
6420 # prepend the host name before the path
6421 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006422
Christopher Faulet312294f2020-09-02 17:17:44 +02006423http-request set-pathq <fmt> [ { if | unless } <condition> ]
6424
6425 This does the same as "http-request set-path" except that the query-string is
6426 also rewritten. It may be used to remove the query-string, including the
6427 question mark (it is not possible using "http-request set-query").
6428
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006429http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02006430
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006431 This is used to set the queue priority class of the current request.
6432 The value must be a sample expression which converts to an integer in the
6433 range -2047..2047. Results outside this range will be truncated.
6434 The priority class determines the order in which queued requests are
6435 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006436
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006437http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006438
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006439 This is used to set the queue priority timestamp offset of the current
6440 request. The value must be a sample expression which converts to an integer
6441 in the range -524287..524287. Results outside this range will be truncated.
6442 When a request is queued, it is ordered first by the priority class, then by
6443 the current timestamp adjusted by the given offset in milliseconds. Lower
6444 values have higher priority.
6445 Note that the resulting timestamp is is only tracked with enough precision
6446 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
6447 adjusted timestamp exceeds this value, it will be misidentified as highest
6448 priority. Thus it is important to set "timeout queue" to a value, where when
6449 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006450
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006451http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01006452
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006453 This rewrites the request's query string which appears after the first
6454 question mark ("?") with the result of the evaluation of format string <fmt>.
6455 The part prior to the question mark is left intact. If the request doesn't
6456 contain a question mark and the new value is not empty, then one is added at
6457 the end of the URI, followed by the new value. If a question mark was
6458 present, it will never be removed even if the value is empty. This can be
6459 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08006460
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006461 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006462
6463 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006464 # replace "%3D" with "=" in the query string
6465 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006466
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006467http-request set-src <expr> [ { if | unless } <condition> ]
6468 This is used to set the source IP address to the value of specified
6469 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
6470 provides the correct IP in a HTTP header; or you want to mask source IP for
Olivier Doucet56e31202020-04-21 09:32:56 +02006471 privacy. All subsequent calls to "src" fetch will return this value
6472 (see example).
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006473
6474 Arguments :
6475 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6476 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006477
Olivier Doucet56e31202020-04-21 09:32:56 +02006478 See also "option forwardfor".
6479
Cyril Bonté78caf842010-03-10 22:41:43 +01006480 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006481 http-request set-src hdr(x-forwarded-for)
6482 http-request set-src src,ipmask(24)
6483
Olivier Doucet56e31202020-04-21 09:32:56 +02006484 # After the masking this will track connections
6485 # based on the IP address with the last byte zeroed out.
6486 http-request track-sc0 src
6487
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006488 When possible, set-src preserves the original source port as long as the
6489 address family allows it, otherwise the source port is set to 0.
6490
6491http-request set-src-port <expr> [ { if | unless } <condition> ]
6492
6493 This is used to set the source port address to the value of specified
6494 expression.
6495
6496 Arguments:
6497 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6498 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006499
Willy Tarreau20b0de52012-12-24 15:45:22 +01006500 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006501 http-request set-src-port hdr(x-port)
6502 http-request set-src-port int(4000)
6503
6504 When possible, set-src-port preserves the original source address as long as
6505 the address family supports a port, otherwise it forces the source address to
6506 IPv4 "0.0.0.0" before rewriting the port.
6507
Amaury Denoyelle8d228232020-12-10 13:43:54 +01006508http-request set-timeout server|tunnel { <timeout> | <expr> }
6509 [ { if | unless } <condition> ]
6510
6511 This action overrides the specified "server" or "tunnel" timeout for the
6512 current stream only. The timeout can be specified in millisecond or with any
6513 other unit if the number is suffixed by the unit as explained at the top of
6514 this document. It is also possible to write an expression which must returns
6515 a number interpreted as a timeout in millisecond.
6516
6517 Note that the server/tunnel timeouts are only relevant on the backend side
6518 and thus this rule is only available for the proxies with backend
6519 capabilities. Also the timeout value must be non-null to obtain the expected
6520 results.
6521
6522 Example:
6523 http-request set-timeout server 5s
6524 http-request set-timeout hdr(host),map_int(host.lst)
6525
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006526http-request set-tos <tos> [ { if | unless } <condition> ]
6527
6528 This is used to set the TOS or DSCP field value of packets sent to the client
6529 to the value passed in <tos> on platforms which support this. This value
6530 represents the whole 8 bits of the IP TOS field, and can be expressed both in
6531 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
6532 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
6533 be used to adjust some routing behavior on border routers based on some
6534 information from the request.
6535
6536 See RFC 2474, 2597, 3260 and 4594 for more information.
6537
6538http-request set-uri <fmt> [ { if | unless } <condition> ]
6539
6540 This rewrites the request URI with the result of the evaluation of format
6541 string <fmt>. The scheme, authority, path and query string are all replaced
6542 at once. This can be used to rewrite hosts in front of proxies, or to
6543 perform complex modifications to the URI such as moving parts between the
6544 path and the query string.
6545 See also "http-request set-path" and "http-request set-query".
6546
6547http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
6548
6549 This is used to set the contents of a variable. The variable is declared
6550 inline.
6551
6552 Arguments:
6553 <var-name> The name of the variable starts with an indication about its
6554 scope. The scopes allowed are:
6555 "proc" : the variable is shared with the whole process
6556 "sess" : the variable is shared with the whole session
6557 "txn" : the variable is shared with the transaction
6558 (request and response)
6559 "req" : the variable is shared only during request
6560 processing
6561 "res" : the variable is shared only during response
6562 processing
6563 This prefix is followed by a name. The separator is a '.'.
6564 The name may only contain characters 'a-z', 'A-Z', '0-9'
6565 and '_'.
6566
6567 <expr> Is a standard HAProxy expression formed by a sample-fetch
6568 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01006569
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006570 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006571 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006572
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006573http-request send-spoe-group <engine-name> <group-name>
6574 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006575
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006576 This action is used to trigger sending of a group of SPOE messages. To do so,
6577 the SPOE engine used to send messages must be defined, as well as the SPOE
6578 group to send. Of course, the SPOE engine must refer to an existing SPOE
6579 filter. If not engine name is provided on the SPOE filter line, the SPOE
6580 agent name must be used.
6581
6582 Arguments:
6583 <engine-name> The SPOE engine name.
6584
6585 <group-name> The SPOE group name as specified in the engine
6586 configuration.
6587
6588http-request silent-drop [ { if | unless } <condition> ]
6589
6590 This stops the evaluation of the rules and makes the client-facing connection
6591 suddenly disappear using a system-dependent way that tries to prevent the
6592 client from being notified. The effect it then that the client still sees an
6593 established connection while there's none on HAProxy. The purpose is to
6594 achieve a comparable effect to "tarpit" except that it doesn't use any local
6595 resource at all on the machine running HAProxy. It can resist much higher
6596 loads than "tarpit", and slow down stronger attackers. It is important to
6597 understand the impact of using this mechanism. All stateful equipment placed
6598 between the client and HAProxy (firewalls, proxies, load balancers) will also
6599 keep the established connection for a long time and may suffer from this
6600 action.
6601 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6602 option is used to block the emission of a TCP reset. On other systems, the
6603 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6604 router, though it's still delivered to local networks. Do not use it unless
6605 you fully understand how it works.
6606
Christopher Faulet46f95542019-12-20 10:07:22 +01006607http-request strict-mode { on | off }
6608
6609 This enables or disables the strict rewriting mode for following rules. It
6610 does not affect rules declared before it and it is only applicable on rules
6611 performing a rewrite on the requests. When the strict mode is enabled, any
6612 rewrite failure triggers an internal error. Otherwise, such errors are
6613 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006614 rewrites optional while others must be performed to continue the request
Christopher Faulet46f95542019-12-20 10:07:22 +01006615 processing.
6616
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006617 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006618 when a ruleset evaluation ends. So, for instance, if you change the mode on
6619 the frontend, the default mode is restored when HAProxy starts the backend
6620 rules evaluation.
6621
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006622http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
6623http-request tarpit [ { status | deny_status } <code>] [content-type <type>]
6624 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6625 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6626 [ hdr <name> <fmt> ]*
6627 [ { if | unless } <condition> ]
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006628
6629 This stops the evaluation of the rules and immediately blocks the request
6630 without responding for a delay specified by "timeout tarpit" or
6631 "timeout connect" if the former is not set. After that delay, if the client
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006632 is still connected, a response is returned so that the client does not
6633 suspect it has been tarpitted. Logs will report the flags "PT". The goal of
6634 the tarpit rule is to slow down robots during an attack when they're limited
6635 on the number of concurrent requests. It can be very efficient against very
6636 dumb robots, and will significantly reduce the load on firewalls compared to
6637 a "deny" rule. But when facing "correctly" developed robots, it can make
6638 things worse by forcing haproxy and the front firewall to support insane
6639 number of concurrent connections. By default an HTTP error 500 is returned.
6640 But the response may be customized using same syntax than
6641 "http-request return" rules. Thus, see "http-request return" for details.
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006642 For compatibility purpose, when no argument is defined, or only "deny_status",
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006643 the argument "default-errorfiles" is implied. It means
6644 "http-request tarpit [deny_status <status>]" is an alias of
6645 "http-request tarpit [status <status>] default-errorfiles".
6646 No further "http-request" rules are evaluated.
6647 See also "http-request return" and "http-request silent-drop".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006648
6649http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6650http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6651http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
6652
6653 This enables tracking of sticky counters from current request. These rules do
6654 not stop evaluation and do not change default action. The number of counters
6655 that may be simultaneously tracked by the same connection is set in
6656 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
Matteo Contrini1857b8c2020-10-16 17:35:54 +02006657 so the track-sc number is between 0 and (MAX_SESS_STKCTR-1). The first
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006658 "track-sc0" rule executed enables tracking of the counters of the specified
6659 table as the first set. The first "track-sc1" rule executed enables tracking
6660 of the counters of the specified table as the second set. The first
6661 "track-sc2" rule executed enables tracking of the counters of the specified
6662 table as the third set. It is a recommended practice to use the first set of
6663 counters for the per-frontend counters and the second set for the per-backend
6664 ones. But this is just a guideline, all may be used everywhere.
6665
6666 Arguments :
6667 <key> is mandatory, and is a sample expression rule as described in
6668 section 7.3. It describes what elements of the incoming request or
6669 connection will be analyzed, extracted, combined, and used to
6670 select which table entry to update the counters.
6671
6672 <table> is an optional table to be used instead of the default one, which
6673 is the stick-table declared in the current proxy. All the counters
6674 for the matches and updates for the key will then be performed in
6675 that table until the session ends.
6676
6677 Once a "track-sc*" rule is executed, the key is looked up in the table and if
6678 it is not found, an entry is allocated for it. Then a pointer to that entry
6679 is kept during all the session's life, and this entry's counters are updated
6680 as often as possible, every time the session's counters are updated, and also
6681 systematically when the session ends. Counters are only updated for events
6682 that happen after the tracking has been started. As an exception, connection
6683 counters and request counters are systematically updated so that they reflect
6684 useful information.
6685
6686 If the entry tracks concurrent connection counters, one connection is counted
6687 for as long as the entry is tracked, and the entry will not expire during
6688 that time. Tracking counters also provides a performance advantage over just
6689 checking the keys, because only one table lookup is performed for all ACL
6690 checks that make use of it.
6691
6692http-request unset-var(<var-name>) [ { if | unless } <condition> ]
6693
6694 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006695
6696 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006697 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006698
Christopher Faulet579d83b2019-11-22 15:34:17 +01006699http-request use-service <service-name> [ { if | unless } <condition> ]
6700
6701 This directive executes the configured HTTP service to reply to the request
6702 and stops the evaluation of the rules. An HTTP service may choose to reply by
6703 sending any valid HTTP response or it may immediately close the connection
6704 without sending any response. Outside natives services, for instance the
6705 Prometheus exporter, it is possible to write your own services in Lua. No
6706 further "http-request" rules are evaluated.
6707
6708 Arguments :
6709 <service-name> is mandatory. It is the service to call
6710
6711 Example:
6712 http-request use-service prometheus-exporter if { path /metrics }
6713
Christopher Faulet021a8e42021-03-29 10:46:38 +02006714http-request wait-for-body time <time> [ at-least <bytes> ]
6715 [ { if | unless } <condition> ]
6716
6717 This will delay the processing of the request waiting for the payload for at
6718 most <time> milliseconds. if "at-least" argument is specified, HAProxy stops
6719 to wait the payload when the first <bytes> bytes are received. 0 means no
6720 limit, it is the default value. Regardless the "at-least" argument value,
6721 HAProxy stops to wait if the whole payload is received or if the request
6722 buffer is full. This action may be used as a replacement to "option
6723 http-buffer-request".
6724
6725 Arguments :
6726
6727 <time> is mandatory. It is the maximum time to wait for the body. It
6728 follows the HAProxy time format and is expressed in milliseconds.
6729
6730 <bytes> is optional. It is the minimum payload size to receive to stop to
6731 wait. It fallows the HAProxy size format and is expressed in
6732 bytes.
6733
6734 Example:
6735 http-request wait-for-body time 1s at-least 1k if METH_POST
6736
6737 See also : "option http-buffer-request"
6738
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006739http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006740
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006741 This will delay the processing of the request until the SSL handshake
6742 happened. This is mostly useful to delay processing early data until we're
6743 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006744
Willy Tarreauef781042010-01-27 11:53:01 +01006745
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006746http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006747 Access control for Layer 7 responses
6748
6749 May be used in sections: defaults | frontend | listen | backend
6750 no | yes | yes | yes
6751
6752 The http-response statement defines a set of rules which apply to layer 7
6753 processing. The rules are evaluated in their declaration order when they are
6754 met in a frontend, listen or backend section. Any rule may optionally be
6755 followed by an ACL-based condition, in which case it will only be evaluated
6756 if the condition is true. Since these rules apply on responses, the backend
6757 rules are applied first, followed by the frontend's rules.
6758
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006759 The first keyword is the rule's action. The supported actions are described
6760 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006761
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006762 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006763
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006764 Example:
6765 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02006766
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006767 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06006768
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006769 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
6770 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06006771
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006772 Example:
6773 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06006774
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006775 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06006776
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006777 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
6778 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06006779
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006780 See also : "http-request", section 3.4 about userlists and section 7 about
6781 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06006782
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006783http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006784
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006785 This is used to add a new entry into an ACL. The ACL must be loaded from a
6786 file (even a dummy empty file). The file name of the ACL to be updated is
6787 passed between parentheses. It takes one argument: <key fmt>, which follows
6788 log-format rules, to collect content of the new entry. It performs a lookup
6789 in the ACL before insertion, to avoid duplicated (or more) values.
6790 This lookup is done by a linear search and can be expensive with large lists!
6791 It is the equivalent of the "add acl" command from the stats socket, but can
6792 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06006793
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006794http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006795
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006796 This appends an HTTP header field whose name is specified in <name> and whose
6797 value is defined by <fmt> which follows the log-format rules (see Custom Log
6798 Format in section 8.2.4). This may be used to send a cookie to a client for
6799 example, or to pass some internal information.
6800 This rule is not final, so it is possible to add other similar rules.
6801 Note that header addition is performed immediately, so one rule might reuse
6802 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06006803
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006804http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006805
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006806 This stops the evaluation of the rules and lets the response pass the check.
6807 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06006808
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02006809http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006810
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006811 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06006812
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006813http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006814
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006815 This captures sample expression <sample> from the response buffer, and
6816 converts it to a string. The resulting string is stored into the next request
6817 "capture" slot, so it will possibly appear next to some captured HTTP
6818 headers. It will then automatically appear in the logs, and it will be
6819 possible to extract it using sample fetch rules to feed it into headers or
6820 anything. Please check section 7.3 (Fetching samples) and
6821 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006822
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006823 The keyword "id" is the id of the capture slot which is used for storing the
6824 string. The capture slot must be defined in an associated frontend.
6825 This is useful to run captures in backends. The slot id can be declared by a
6826 previous directive "http-response capture" or with the "declare capture"
6827 keyword.
Baptiste Assmann19a69b32020-01-16 14:34:22 +01006828
6829 When using this action in a backend, double check that the relevant
6830 frontend(s) have the required capture slots otherwise, this rule will be
6831 ignored at run time. This can't be detected at configuration parsing time
6832 due to HAProxy's ability to dynamically resolve backend name at runtime.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006833
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006834http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006835
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006836 This is used to delete an entry from an ACL. The ACL must be loaded from a
6837 file (even a dummy empty file). The file name of the ACL to be updated is
6838 passed between parentheses. It takes one argument: <key fmt>, which follows
6839 log-format rules, to collect content of the entry to delete.
6840 It is the equivalent of the "del acl" command from the stats socket, but can
6841 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02006842
Maciej Zdebebdd4c52020-11-20 13:58:48 +00006843http-response del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02006844
Maciej Zdebebdd4c52020-11-20 13:58:48 +00006845 This removes all HTTP header fields whose name is specified in <name>. <meth>
6846 is the matching method, applied on the header name. Supported matching methods
6847 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
6848 (substring match) and "reg" (regex match). If not specified, exact matching
6849 method is used.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02006850
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006851http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02006852
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006853 This is used to delete an entry from a MAP. The MAP must be loaded from a
6854 file (even a dummy empty file). The file name of the MAP to be updated is
6855 passed between parentheses. It takes one argument: <key fmt>, which follows
6856 log-format rules, to collect content of the entry to delete.
6857 It takes one argument: "file name" It is the equivalent of the "del map"
6858 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006859
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006860http-response deny [deny_status <status>] [ { if | unless } <condition> ]
6861http-response deny [ { status | deny_status } <code>] [content-type <type>]
6862 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6863 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6864 [ hdr <name> <fmt> ]*
6865 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006866
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006867 This stops the evaluation of the rules and immediately rejects the response.
6868 By default an HTTP 502 error is returned. But the response may be customized
6869 using same syntax than "http-response return" rules. Thus, see
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006870 "http-response return" for details. For compatibility purpose, when no
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006871 argument is defined, or only "deny_status", the argument "default-errorfiles"
6872 is implied. It means "http-response deny [deny_status <status>]" is an alias
6873 of "http-response deny [status <status>] default-errorfiles".
Christopher Faulet040c8cd2020-01-13 16:43:45 +01006874 No further "http-response" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006875 See also "http-response return".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006876
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006877http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006878
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006879 This performs an HTTP redirection based on a redirect rule.
6880 This supports a format string similarly to "http-request redirect" rules,
6881 with the exception that only the "location" type of redirect is possible on
6882 the response. See the "redirect" keyword for the rule's syntax. When a
6883 redirect rule is applied during a response, connections to the server are
6884 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006885
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006886http-response replace-header <name> <regex-match> <replace-fmt>
6887 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006888
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006889 This works like "http-request replace-header" except that it works on the
6890 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01006891
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006892 Example:
6893 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02006894
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006895 # applied to:
6896 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006897
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006898 # outputs:
6899 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 +02006900
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006901 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006902
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006903http-response replace-value <name> <regex-match> <replace-fmt>
6904 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006905
Tim Duesterhus6bd909b2020-01-17 15:53:18 +01006906 This works like "http-request replace-value" except that it works on the
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006907 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006908
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006909 Example:
6910 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006911
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006912 # applied to:
6913 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01006914
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006915 # outputs:
6916 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006917
Christopher Faulet24231ab2020-01-24 17:44:23 +01006918http-response return [status <code>] [content-type <type>]
6919 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6920 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006921 [ hdr <name> <value> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01006922 [ { if | unless } <condition> ]
6923
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006924 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01006925 default status code used for the response is 200. It can be optionally
6926 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04006927 specified as an argument to "content-type". Finally the response itself may
Christopher Faulet24231ab2020-01-24 17:44:23 +01006928 be defined. If can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006929 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01006930 are followed to create the response :
6931
6932 * If neither the errorfile nor the payload to use is defined, a dummy
6933 response is returned. Only the "status" argument is considered. It can be
6934 any code in the range [200, 599]. The "content-type" argument, if any, is
6935 ignored.
6936
6937 * If "default-errorfiles" argument is set, the proxy's errorfiles are
6938 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04006939 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Christopher Faulete095f312020-12-07 11:22:24 +01006940 425, 429, 500, 501, 502, 503, and 504). The "content-type" argument, if
6941 any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006942
6943 * If a specific errorfile is defined, with an "errorfile" argument, the
6944 corresponding file, containing a full HTTP response, is returned. Only the
6945 "status" argument is considered. It must be one of the status code handled
Christopher Faulete095f312020-12-07 11:22:24 +01006946 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 501,
6947 502, 503, and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006948
6949 * If an http-errors section is defined, with an "errorfiles" argument, the
6950 corresponding file in the specified http-errors section, containing a full
6951 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04006952 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Christopher Faulete095f312020-12-07 11:22:24 +01006953 408, 410, 413, 425, 429, 500, 501, 502, 503, and 504). The "content-type"
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006954 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006955
6956 * If a "file" or a "lf-file" argument is specified, the file's content is
6957 used as the response payload. If the file is not empty, its content-type
6958 must be set as argument to "content-type". Otherwise, any "content-type"
6959 argument is ignored. With a "lf-file" argument, the file's content is
6960 evaluated as a log-format string. With a "file" argument, it is considered
6961 as a raw content.
6962
6963 * If a "string" or "lf-string" argument is specified, the defined string is
6964 used as the response payload. The content-type must always be set as
6965 argument to "content-type". With a "lf-string" argument, the string is
6966 evaluated as a log-format string. With a "string" argument, it is
6967 considered as a raw string.
6968
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006969 When the response is not based an errorfile, it is possible to appends HTTP
6970 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
6971 arguments are ignored. For each one, the header name is specified in <name>
6972 and its value is defined by <fmt> which follows the log-format rules.
6973
Christopher Faulet24231ab2020-01-24 17:44:23 +01006974 Note that the generated response must be smaller than a buffer. And to avoid
6975 any warning, when an errorfile or a raw file is loaded, the buffer space
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006976 reserved to the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006977
6978 No further "http-response" rules are evaluated.
6979
6980 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006981 http-response return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006982 if { status eq 404 }
6983
6984 http-response return content-type text/plain \
6985 string "This is the end !" \
6986 if { status eq 500 }
6987
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006988http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6989http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08006990
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006991 This action increments the GPC0 or GPC1 counter according with the sticky
6992 counter designated by <sc-id>. If an error occurs, this action silently fails
6993 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02006994
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006995http-response sc-set-gpt0(<sc-id>) { <int> | <expr> }
6996 [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02006997
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006998 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6999 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
7000 boolean. If an error occurs, this action silently fails and the actions
7001 evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01007002
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007003http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02007004
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007005 This action is used to trigger sending of a group of SPOE messages. To do so,
7006 the SPOE engine used to send messages must be defined, as well as the SPOE
7007 group to send. Of course, the SPOE engine must refer to an existing SPOE
7008 filter. If not engine name is provided on the SPOE filter line, the SPOE
7009 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02007010
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007011 Arguments:
7012 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02007013
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007014 <group-name> The SPOE group name as specified in the engine
7015 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02007016
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007017http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02007018
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007019 This does the same as "add-header" except that the header name is first
7020 removed if it existed. This is useful when passing security information to
7021 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02007022
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007023http-response set-log-level <level> [ { if | unless } <condition> ]
7024
7025 This is used to change the log level of the current request when a certain
7026 condition is met. Valid levels are the 8 syslog levels (see the "log"
7027 keyword) plus the special level "silent" which disables logging for this
7028 request. This rule is not final so the last matching rule wins. This rule can
7029 be useful to disable health checks coming from another equipment.
7030
7031http-response set-map(<file-name>) <key fmt> <value fmt>
7032
7033 This is used to add a new entry into a MAP. The MAP must be loaded from a
7034 file (even a dummy empty file). The file name of the MAP to be updated is
7035 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
7036 log-format rules, used to collect MAP key, and <value fmt>, which follows
7037 log-format rules, used to collect content for the new entry. It performs a
7038 lookup in the MAP before insertion, to avoid duplicated (or more) values.
7039 This lookup is done by a linear search and can be expensive with large lists!
7040 It is the equivalent of the "set map" command from the stats socket, but can
7041 be triggered by an HTTP response.
7042
7043http-response set-mark <mark> [ { if | unless } <condition> ]
7044
7045 This is used to set the Netfilter MARK on all packets sent to the client to
7046 the value passed in <mark> on platforms which support it. This value is an
7047 unsigned 32 bit value which can be matched by netfilter and by the routing
7048 table. It can be expressed both in decimal or hexadecimal format (prefixed
7049 by "0x"). This can be useful to force certain packets to take a different
7050 route (for example a cheaper network path for bulk downloads). This works on
7051 Linux kernels 2.6.32 and above and requires admin privileges.
7052
7053http-response set-nice <nice> [ { if | unless } <condition> ]
7054
7055 This sets the "nice" factor of the current request being processed.
7056 It only has effect against the other requests being processed at the same
7057 time. The default value is 0, unless altered by the "nice" setting on the
7058 "bind" line. The accepted range is -1024..1024. The higher the value, the
7059 nicest the request will be. Lower values will make the request more important
7060 than other ones. This can be useful to improve the speed of some requests, or
7061 lower the priority of non-important requests. Using this setting without
7062 prior experimentation can cause some major slowdown.
7063
7064http-response set-status <status> [reason <str>]
7065 [ { if | unless } <condition> ]
7066
7067 This replaces the response status code with <status> which must be an integer
7068 between 100 and 999. Optionally, a custom reason text can be provided defined
7069 by <str>, or the default reason for the specified code will be used as a
7070 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007071
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007072 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007073 # return "431 Request Header Fields Too Large"
7074 http-response set-status 431
7075 # return "503 Slow Down", custom reason
7076 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007077
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007078http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007079
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007080 This is used to set the TOS or DSCP field value of packets sent to the client
7081 to the value passed in <tos> on platforms which support this.
7082 This value represents the whole 8 bits of the IP TOS field, and can be
7083 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
7084 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
7085 always 0. This can be used to adjust some routing behavior on border routers
7086 based on some information from the request.
7087
7088 See RFC 2474, 2597, 3260 and 4594 for more information.
7089
7090http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
7091
7092 This is used to set the contents of a variable. The variable is declared
7093 inline.
7094
7095 Arguments:
7096 <var-name> The name of the variable starts with an indication about its
7097 scope. The scopes allowed are:
7098 "proc" : the variable is shared with the whole process
7099 "sess" : the variable is shared with the whole session
7100 "txn" : the variable is shared with the transaction
7101 (request and response)
7102 "req" : the variable is shared only during request
7103 processing
7104 "res" : the variable is shared only during response
7105 processing
7106 This prefix is followed by a name. The separator is a '.'.
7107 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
7108 and '_'.
7109
7110 <expr> Is a standard HAProxy expression formed by a sample-fetch
7111 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007112
7113 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007114 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007115
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007116http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007117
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007118 This stops the evaluation of the rules and makes the client-facing connection
7119 suddenly disappear using a system-dependent way that tries to prevent the
7120 client from being notified. The effect it then that the client still sees an
7121 established connection while there's none on HAProxy. The purpose is to
7122 achieve a comparable effect to "tarpit" except that it doesn't use any local
7123 resource at all on the machine running HAProxy. It can resist much higher
7124 loads than "tarpit", and slow down stronger attackers. It is important to
7125 understand the impact of using this mechanism. All stateful equipment placed
7126 between the client and HAProxy (firewalls, proxies, load balancers) will also
7127 keep the established connection for a long time and may suffer from this
7128 action.
7129 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
7130 option is used to block the emission of a TCP reset. On other systems, the
7131 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
7132 router, though it's still delivered to local networks. Do not use it unless
7133 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007134
Christopher Faulet46f95542019-12-20 10:07:22 +01007135http-response strict-mode { on | off }
7136
7137 This enables or disables the strict rewriting mode for following rules. It
7138 does not affect rules declared before it and it is only applicable on rules
7139 performing a rewrite on the responses. When the strict mode is enabled, any
7140 rewrite failure triggers an internal error. Otherwise, such errors are
7141 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05007142 rewrites optional while others must be performed to continue the response
Christopher Faulet46f95542019-12-20 10:07:22 +01007143 processing.
7144
Christopher Faulet1aea50e2020-01-17 16:03:53 +01007145 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01007146 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04007147 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet46f95542019-12-20 10:07:22 +01007148 rules evaluation.
7149
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007150http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
7151http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
7152http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02007153
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007154 This enables tracking of sticky counters from current response. Please refer
7155 to "http-request track-sc" for a complete description. The only difference
7156 from "http-request track-sc" is the <key> sample expression can only make use
7157 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
7158 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
7159 supported, haproxy will fail and warn while parsing the config.
7160
7161http-response unset-var(<var-name>) [ { if | unless } <condition> ]
7162
7163 This is used to unset a variable. See "http-response set-var" for details
7164 about <var-name>.
7165
7166 Example:
7167 http-response unset-var(sess.last_redir)
7168
Christopher Faulet021a8e42021-03-29 10:46:38 +02007169http-response wait-for-body time <time> [ at-least <bytes> ]
7170 [ { if | unless } <condition> ]
7171
7172 This will delay the processing of the response waiting for the payload for at
7173 most <time> milliseconds. if "at-least" argument is specified, HAProxy stops
7174 to wait the payload when the first <bytes> bytes are received. 0 means no
7175 limit, it is the default value. Regardless the "at-least" argument value,
7176 HAProxy stops to wait if the whole payload is received or if the response
7177 buffer is full.
7178
7179 Arguments :
7180
7181 <time> is mandatory. It is the maximum time to wait for the body. It
7182 follows the HAProxy time format and is expressed in milliseconds.
7183
7184 <bytes> is optional. It is the minimum payload size to receive to stop to
7185 wait. It fallows the HAProxy size format and is expressed in
7186 bytes.
7187
7188 Example:
7189 http-response wait-for-body time 1s at-least 10k
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02007190
Willy Tarreau30631952015-08-06 15:05:24 +02007191http-reuse { never | safe | aggressive | always }
7192 Declare how idle HTTP connections may be shared between requests
7193
7194 May be used in sections: defaults | frontend | listen | backend
7195 yes | no | yes | yes
7196
7197 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01007198 which is considered safe for reuse is moved back to the server's idle
7199 connections pool so that any other request can make use of it. This is the
7200 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02007201
7202 The argument indicates the desired connection reuse strategy :
7203
Olivier Houchard86006a52018-12-14 19:37:49 +01007204 - "never" : idle connections are never shared between sessions. This mode
7205 may be enforced to cancel a different strategy inherited from
7206 a defaults section or for troubleshooting. For example, if an
7207 old bogus application considers that multiple requests over
7208 the same connection come from the same client and it is not
7209 possible to fix the application, it may be desirable to
7210 disable connection sharing in a single backend. An example of
7211 such an application could be an old haproxy using cookie
7212 insertion in tunnel mode and not checking any request past the
7213 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02007214
Olivier Houchard86006a52018-12-14 19:37:49 +01007215 - "safe" : this is the default and the recommended strategy. The first
7216 request of a session is always sent over its own connection,
7217 and only subsequent requests may be dispatched over other
7218 existing connections. This ensures that in case the server
7219 closes the connection when the request is being sent, the
7220 browser can decide to silently retry it. Since it is exactly
7221 equivalent to regular keep-alive, there should be no side
Amaury Denoyelle27179652020-10-14 18:17:12 +02007222 effects. There is also a special handling for the connections
7223 using protocols subject to Head-of-line blocking (backend with
7224 h2 or fcgi). In this case, when at least one stream is
7225 processed, the used connection is reserved to handle streams
7226 of the same session. When no more streams are processed, the
7227 connection is released and can be reused.
Willy Tarreau30631952015-08-06 15:05:24 +02007228
7229 - "aggressive" : this mode may be useful in webservices environments where
7230 all servers are not necessarily known and where it would be
7231 appreciable to deliver most first requests over existing
7232 connections. In this case, first requests are only delivered
7233 over existing connections that have been reused at least once,
7234 proving that the server correctly supports connection reuse.
7235 It should only be used when it's sure that the client can
7236 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02007237 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02007238 downsides of rare connection failures.
7239
7240 - "always" : this mode is only recommended when the path to the server is
7241 known for never breaking existing connections quickly after
7242 releasing them. It allows the first request of a session to be
7243 sent to an existing connection. This can provide a significant
7244 performance increase over the "safe" strategy when the backend
7245 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01007246 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02007247 sharing. It is recommended that the "http-keep-alive" timeout
7248 remains low in this mode so that no dead connections remain
7249 usable. In most cases, this will lead to the same performance
7250 gains as "aggressive" but with more risks. It should only be
7251 used when it improves the situation over "aggressive".
7252
7253 When http connection sharing is enabled, a great care is taken to respect the
Amaury Denoyelled773a4e2021-01-29 15:18:49 +01007254 connection properties and compatibility. Indeed, some properties are specific
7255 and it is not possibly to reuse it blindly. Those are the SSL SNI, source
7256 and destination address and proxy protocol block. A connection is reused only
7257 if it shares the same set of properties with the request.
Willy Tarreau30631952015-08-06 15:05:24 +02007258
Amaury Denoyelled773a4e2021-01-29 15:18:49 +01007259 Also note that connections with certain bogus authentication schemes (relying
7260 on the connection) like NTLM are marked private and never shared.
Willy Tarreau30631952015-08-06 15:05:24 +02007261
Lukas Tribuse8adfeb2019-11-06 11:50:25 +01007262 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02007263
7264 Note: connection reuse improves the accuracy of the "server maxconn" setting,
7265 because almost no new connection will be established while idle connections
7266 remain available. This is particularly true with the "always" strategy.
7267
7268 See also : "option http-keep-alive", "server maxconn"
7269
7270
Mark Lamourinec2247f02012-01-04 13:02:01 -05007271http-send-name-header [<header>]
7272 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05007273 May be used in sections: defaults | frontend | listen | backend
7274 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05007275 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05007276 <header> The header string to use to send the server name
7277
Willy Tarreau81bef7e2019-10-07 14:58:02 +02007278 The "http-send-name-header" statement causes the header field named <header>
7279 to be set to the name of the target server at the moment the request is about
7280 to be sent on the wire. Any existing occurrences of this header are removed.
7281 Upon retries and redispatches, the header field is updated to always reflect
7282 the server being attempted to connect to. Given that this header is modified
7283 very late in the connection setup, it may have unexpected effects on already
7284 modified headers. For example using it with transport-level header such as
7285 connection, content-length, transfer-encoding and so on will likely result in
7286 invalid requests being sent to the server. Additionally it has been reported
7287 that this directive is currently being used as a way to overwrite the Host
7288 header field in outgoing requests; while this trick has been known to work
7289 as a side effect of the feature for some time, it is not officially supported
7290 and might possibly not work anymore in a future version depending on the
7291 technical difficulties this feature induces. A long-term solution instead
7292 consists in fixing the application which required this trick so that it binds
7293 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05007294
7295 See also : "server"
7296
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01007297id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02007298 Set a persistent ID to a proxy.
7299 May be used in sections : defaults | frontend | listen | backend
7300 no | yes | yes | yes
7301 Arguments : none
7302
7303 Set a persistent ID for the proxy. This ID must be unique and positive.
7304 An unused ID will automatically be assigned if unset. The first assigned
7305 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01007306
7307
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007308ignore-persist { if | unless } <condition>
7309 Declare a condition to ignore persistence
7310 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01007311 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007312
7313 By default, when cookie persistence is enabled, every requests containing
7314 the cookie are unconditionally persistent (assuming the target server is up
7315 and running).
7316
7317 The "ignore-persist" statement allows one to declare various ACL-based
7318 conditions which, when met, will cause a request to ignore persistence.
7319 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007320 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007321 persistence for a specific User-Agent (for example, some web crawler bots).
7322
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007323 The persistence is ignored when an "if" condition is met, or unless an
7324 "unless" condition is met.
7325
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03007326 Example:
7327 acl url_static path_beg /static /images /img /css
7328 acl url_static path_end .gif .png .jpg .css .js
7329 ignore-persist if url_static
7330
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007331 See also : "force-persist", "cookie", and section 7 about ACL usage.
7332
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007333load-server-state-from-file { global | local | none }
7334 Allow seamless reload of HAProxy
7335 May be used in sections: defaults | frontend | listen | backend
7336 yes | no | yes | yes
7337
7338 This directive points HAProxy to a file where server state from previous
7339 running process has been saved. That way, when starting up, before handling
7340 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01007341 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007342 to tell haproxy which file to use. For now, only 2 arguments to either prevent
7343 loading state or load states from a file containing all backends and servers.
7344 The state file can be generated by running the command "show servers state"
7345 over the stats socket and redirect output.
7346
Davor Ocelice9ed2812017-12-25 17:49:28 +01007347 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007348 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02007349 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007350
7351 Arguments:
7352 global load the content of the file pointed by the global directive
7353 named "server-state-file".
7354
7355 local load the content of the file pointed by the directive
7356 "server-state-file-name" if set. If not set, then the backend
7357 name is used as a file name.
7358
7359 none don't load any stat for this backend
7360
7361 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01007362 - server's IP address is preserved across reloads by default, but the
7363 order can be changed thanks to the server's "init-addr" setting. This
7364 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007365 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01007366 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007367
7368 - server's weight is applied from previous running process unless it has
7369 has changed between previous and new configuration files.
7370
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007371 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007372
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007373 global
7374 stats socket /tmp/socket
7375 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007376
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007377 defaults
7378 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007379
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007380 backend bk
7381 server s1 127.0.0.1:22 check weight 11
7382 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007383
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007384
7385 Then one can run :
7386
7387 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
7388
7389 Content of the file /tmp/server_state would be like this:
7390
7391 1
7392 # <field names skipped for the doc example>
7393 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
7394 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
7395
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007396 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007397
7398 global
7399 stats socket /tmp/socket
7400 server-state-base /etc/haproxy/states
7401
7402 defaults
7403 load-server-state-from-file local
7404
7405 backend bk
7406 server s1 127.0.0.1:22 check weight 11
7407 server s2 127.0.0.1:22 check weight 12
7408
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007409
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007410 Then one can run :
7411
7412 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
7413
7414 Content of the file /etc/haproxy/states/bk would be like this:
7415
7416 1
7417 # <field names skipped for the doc example>
7418 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
7419 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
7420
7421 See also: "server-state-file", "server-state-file-name", and
7422 "show servers state"
7423
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007424
Willy Tarreau2769aa02007-12-27 18:26:09 +01007425log global
Jan Wagner3e678602020-12-17 22:22:32 +01007426log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02007427 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02007428no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01007429 Enable per-instance logging of events and traffic.
7430 May be used in sections : defaults | frontend | listen | backend
7431 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02007432
7433 Prefix :
7434 no should be used when the logger list must be flushed. For example,
7435 if you don't want to inherit from the default logger list. This
7436 prefix does not allow arguments.
7437
Willy Tarreau2769aa02007-12-27 18:26:09 +01007438 Arguments :
7439 global should be used when the instance's logging parameters are the
7440 same as the global ones. This is the most common usage. "global"
7441 replaces <address>, <facility> and <level> with those of the log
7442 entries found in the "global" section. Only one "log global"
7443 statement may be used per instance, and this form takes no other
7444 parameter.
7445
7446 <address> indicates where to send the logs. It takes the same format as
7447 for the "global" section's logs, and can be one of :
7448
7449 - An IPv4 address optionally followed by a colon (':') and a UDP
7450 port. If no port is specified, 514 is used by default (the
7451 standard syslog port).
7452
David du Colombier24bb5f52011-03-17 10:40:23 +01007453 - An IPv6 address followed by a colon (':') and optionally a UDP
7454 port. If no port is specified, 514 is used by default (the
7455 standard syslog port).
7456
Willy Tarreau2769aa02007-12-27 18:26:09 +01007457 - A filesystem path to a UNIX domain socket, keeping in mind
7458 considerations for chroot (be sure the path is accessible
7459 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007460 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007461
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007462 - A file descriptor number in the form "fd@<number>", which may
7463 point to a pipe, terminal, or socket. In this case unbuffered
7464 logs are used and one writev() call per log is performed. This
7465 is a bit expensive but acceptable for most workloads. Messages
7466 sent this way will not be truncated but may be dropped, in
7467 which case the DroppedLogs counter will be incremented. The
7468 writev() call is atomic even on pipes for messages up to
7469 PIPE_BUF size, which POSIX recommends to be at least 512 and
7470 which is 4096 bytes on most modern operating systems. Any
7471 larger message may be interleaved with messages from other
7472 processes. Exceptionally for debugging purposes the file
7473 descriptor may also be directed to a file, but doing so will
7474 significantly slow haproxy down as non-blocking calls will be
7475 ignored. Also there will be no way to purge nor rotate this
7476 file without restarting the process. Note that the configured
7477 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01007478 with a TCP syslog server. See also the "short" and "raw"
7479 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007480
7481 - "stdout" / "stderr", which are respectively aliases for "fd@1"
7482 and "fd@2", see above.
7483
Willy Tarreauc046d162019-08-30 15:24:59 +02007484 - A ring buffer in the form "ring@<name>", which will correspond
7485 to an in-memory ring buffer accessible over the CLI using the
7486 "show events" command, which will also list existing rings and
7487 their sizes. Such buffers are lost on reload or restart but
7488 when used as a complement this can help troubleshooting by
7489 having the logs instantly available.
7490
Emeric Brun94aab062021-04-02 10:41:36 +02007491 - An explicit stream address prefix such as "tcp@","tcp6@",
7492 "tcp4@" or "uxst@" will allocate an implicit ring buffer with
7493 a stream forward server targeting the given address.
7494
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007495 You may want to reference some environment variables in the
7496 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01007497
Willy Tarreau18324f52014-06-27 18:10:07 +02007498 <length> is an optional maximum line length. Log lines larger than this
7499 value will be truncated before being sent. The reason is that
7500 syslog servers act differently on log line length. All servers
7501 support the default value of 1024, but some servers simply drop
7502 larger lines while others do log them. If a server supports long
7503 lines, it may make sense to set this value here in order to avoid
7504 truncating long lines. Similarly, if a server drops long lines,
7505 it is preferable to truncate them before sending them. Accepted
7506 values are 80 to 65535 inclusive. The default value of 1024 is
7507 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01007508 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02007509
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02007510 <ranges> A list of comma-separated ranges to identify the logs to sample.
7511 This is used to balance the load of the logs to send to the log
7512 server. The limits of the ranges cannot be null. They are numbered
7513 from 1. The size or period (in number of logs) of the sample must
7514 be set with <sample_size> parameter.
7515
7516 <sample_size>
7517 The size of the sample in number of logs to consider when balancing
7518 their logging loads. It is used to balance the load of the logs to
7519 send to the syslog server. This size must be greater or equal to the
7520 maximum of the high limits of the ranges.
7521 (see also <ranges> parameter).
7522
Willy Tarreauadb345d2018-11-12 07:56:13 +01007523 <format> is the log format used when generating syslog messages. It may be
7524 one of the following :
7525
Emeric Brun0237c4e2020-11-27 16:24:34 +01007526 local Analog to rfc3164 syslog message format except that hostname
7527 field is stripped. This is the default.
7528 Note: option "log-send-hostname" switches the default to
7529 rfc3164.
7530
7531 rfc3164 The RFC3164 syslog message format.
Willy Tarreauadb345d2018-11-12 07:56:13 +01007532 (https://tools.ietf.org/html/rfc3164)
7533
7534 rfc5424 The RFC5424 syslog message format.
7535 (https://tools.ietf.org/html/rfc5424)
7536
Emeric Brun54648852020-07-06 15:54:06 +02007537 priority A message containing only a level plus syslog facility between
7538 angle brackets such as '<63>', followed by the text. The PID,
7539 date, time, process name and system name are omitted. This is
7540 designed to be used with a local log server.
7541
Willy Tarreaue8746a02018-11-12 08:45:00 +01007542 short A message containing only a level between angle brackets such as
7543 '<3>', followed by the text. The PID, date, time, process name
7544 and system name are omitted. This is designed to be used with a
7545 local log server. This format is compatible with what the
7546 systemd logger consumes.
7547
Emeric Brun54648852020-07-06 15:54:06 +02007548 timed A message containing only a level between angle brackets such as
7549 '<3>', followed by ISO date and by the text. The PID, process
7550 name and system name are omitted. This is designed to be
7551 used with a local log server.
7552
7553 iso A message containing only the ISO date, followed by the text.
7554 The PID, process name and system name are omitted. This is
7555 designed to be used with a local log server.
7556
Willy Tarreauc1b06452018-11-12 11:57:56 +01007557 raw A message containing only the text. The level, PID, date, time,
7558 process name and system name are omitted. This is designed to
7559 be used in containers or during development, where the severity
7560 only depends on the file descriptor used (stdout/stderr).
7561
Willy Tarreau2769aa02007-12-27 18:26:09 +01007562 <facility> must be one of the 24 standard syslog facilities :
7563
Willy Tarreaue8746a02018-11-12 08:45:00 +01007564 kern user mail daemon auth syslog lpr news
7565 uucp cron auth2 ftp ntp audit alert cron2
7566 local0 local1 local2 local3 local4 local5 local6 local7
7567
Willy Tarreauc1b06452018-11-12 11:57:56 +01007568 Note that the facility is ignored for the "short" and "raw"
7569 formats, but still required as a positional field. It is
7570 recommended to use "daemon" in this case to make it clear that
7571 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007572
7573 <level> is optional and can be specified to filter outgoing messages. By
7574 default, all messages are sent. If a level is specified, only
7575 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007576 will be sent. An optional minimum level can be specified. If it
7577 is set, logs emitted with a more severe level than this one will
7578 be capped to this level. This is used to avoid sending "emerg"
7579 messages on all terminals on some default syslog configurations.
7580 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007581
7582 emerg alert crit err warning notice info debug
7583
William Lallemand0f99e342011-10-12 17:50:54 +02007584 It is important to keep in mind that it is the frontend which decides what to
7585 log from a connection, and that in case of content switching, the log entries
7586 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01007587
7588 However, backend log declaration define how and where servers status changes
7589 will be logged. Level "notice" will be used to indicate a server going up,
7590 "warning" will be used for termination signals and definitive service
7591 termination, and "alert" will be used for when a server goes down.
7592
7593 Note : According to RFC3164, messages are truncated to 1024 bytes before
7594 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007595
7596 Example :
7597 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01007598 log stdout format short daemon # send log to systemd
7599 log stdout format raw daemon # send everything to stdout
7600 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007601 log 127.0.0.1:514 local0 notice # only send important events
Emeric Brun94aab062021-04-02 10:41:36 +02007602 log tcp@127.0.0.1:514 local0 notice notice # same but limit output
7603 # level and send in tcp
William Lallemandb2f07452015-05-12 14:27:13 +02007604 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01007605
Willy Tarreau2769aa02007-12-27 18:26:09 +01007606
William Lallemand48940402012-01-30 16:47:22 +01007607log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007608 Specifies the log format string to use for traffic logs
7609 May be used in sections: defaults | frontend | listen | backend
7610 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01007611
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007612 This directive specifies the log format string that will be used for all logs
7613 resulting from traffic passing through the frontend using this line. If the
7614 directive is used in a defaults section, all subsequent frontends will use
7615 the same log format. Please see section 8.2.4 which covers the log format
7616 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01007617
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007618 "log-format" directive overrides previous "option tcplog", "log-format" and
7619 "option httplog" directives.
7620
Dragan Dosen7ad31542015-09-28 17:16:47 +02007621log-format-sd <string>
7622 Specifies the RFC5424 structured-data log format string
7623 May be used in sections: defaults | frontend | listen | backend
7624 yes | yes | yes | no
7625
7626 This directive specifies the RFC5424 structured-data log format string that
7627 will be used for all logs resulting from traffic passing through the frontend
7628 using this line. If the directive is used in a defaults section, all
7629 subsequent frontends will use the same log format. Please see section 8.2.4
7630 which covers the log format string in depth.
7631
7632 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
7633 about the RFC5424 structured-data part.
7634
7635 Note : This log format string will be used only for loggers that have set
7636 log format to "rfc5424".
7637
7638 Example :
7639 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
7640
7641
Willy Tarreau094af4e2015-01-07 15:03:42 +01007642log-tag <string>
7643 Specifies the log tag to use for all outgoing logs
7644 May be used in sections: defaults | frontend | listen | backend
7645 yes | yes | yes | yes
7646
7647 Sets the tag field in the syslog header to this string. It defaults to the
7648 log-tag set in the global section, otherwise the program name as launched
7649 from the command line, which usually is "haproxy". Sometimes it can be useful
7650 to differentiate between multiple processes running on the same host, or to
7651 differentiate customer instances running in the same process. In the backend,
7652 logs about servers up/down will use this tag. As a hint, it can be convenient
7653 to set a log-tag related to a hosted customer in a defaults section then put
7654 all the frontends and backends for that customer, then start another customer
7655 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007656
Willy Tarreauc35362a2014-04-25 13:58:37 +02007657max-keep-alive-queue <value>
7658 Set the maximum server queue size for maintaining keep-alive connections
7659 May be used in sections: defaults | frontend | listen | backend
7660 yes | no | yes | yes
7661
7662 HTTP keep-alive tries to reuse the same server connection whenever possible,
7663 but sometimes it can be counter-productive, for example if a server has a lot
7664 of connections while other ones are idle. This is especially true for static
7665 servers.
7666
7667 The purpose of this setting is to set a threshold on the number of queued
7668 connections at which haproxy stops trying to reuse the same server and prefers
7669 to find another one. The default value, -1, means there is no limit. A value
7670 of zero means that keep-alive requests will never be queued. For very close
7671 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007672 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02007673 use a value of 10 or less). For remote servers suffering from a high latency,
7674 higher values might be needed to cover for the latency and/or the cost of
7675 picking a different server.
7676
7677 Note that this has no impact on responses which are maintained to the same
7678 server consecutively to a 401 response. They will still go to the same server
7679 even if they have to be queued.
7680
7681 See also : "option http-server-close", "option prefer-last-server", server
7682 "maxconn" and cookie persistence.
7683
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01007684max-session-srv-conns <nb>
7685 Set the maximum number of outgoing connections we can keep idling for a given
7686 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
7687 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02007688
Willy Tarreau2769aa02007-12-27 18:26:09 +01007689maxconn <conns>
7690 Fix the maximum number of concurrent connections on a frontend
7691 May be used in sections : defaults | frontend | listen | backend
7692 yes | yes | yes | no
7693 Arguments :
7694 <conns> is the maximum number of concurrent connections the frontend will
7695 accept to serve. Excess connections will be queued by the system
7696 in the socket's listen queue and will be served once a connection
7697 closes.
7698
7699 If the system supports it, it can be useful on big sites to raise this limit
7700 very high so that haproxy manages connection queues, instead of leaving the
7701 clients with unanswered connection attempts. This value should not exceed the
7702 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01007703 of tune.bufsize (16kB by default) each, as well as some other data resulting
7704 in about 33 kB of RAM being consumed per established connection. That means
7705 that a medium system equipped with 1GB of RAM can withstand around
7706 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007707
7708 Also, when <conns> is set to large values, it is possible that the servers
7709 are not sized to accept such loads, and for this reason it is generally wise
7710 to assign them some reasonable connection limits.
7711
Willy Tarreauc8d5b952019-02-27 17:25:52 +01007712 When this value is set to zero, which is the default, the global "maxconn"
7713 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02007714
Willy Tarreau2769aa02007-12-27 18:26:09 +01007715 See also : "server", global section's "maxconn", "fullconn"
7716
7717
Willy Tarreau77e0dae2020-10-14 15:44:27 +02007718mode { tcp|http }
Willy Tarreau2769aa02007-12-27 18:26:09 +01007719 Set the running mode or protocol of the instance
7720 May be used in sections : defaults | frontend | listen | backend
7721 yes | yes | yes | yes
7722 Arguments :
7723 tcp The instance will work in pure TCP mode. A full-duplex connection
7724 will be established between clients and servers, and no layer 7
7725 examination will be performed. This is the default mode. It
7726 should be used for SSL, SSH, SMTP, ...
7727
7728 http The instance will work in HTTP mode. The client request will be
7729 analyzed in depth before connecting to any server. Any request
7730 which is not RFC-compliant will be rejected. Layer 7 filtering,
7731 processing and switching will be possible. This is the mode which
7732 brings HAProxy most of its value.
7733
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007734 When doing content switching, it is mandatory that the frontend and the
7735 backend are in the same mode (generally HTTP), otherwise the configuration
7736 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007737
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007738 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007739 defaults http_instances
7740 mode http
7741
Willy Tarreau0ba27502007-12-24 16:55:16 +01007742
Cyril Bontéf0c60612010-02-06 14:44:47 +01007743monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01007744 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007745 May be used in sections : defaults | frontend | listen | backend
7746 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01007747 Arguments :
7748 if <cond> the monitor request will fail if the condition is satisfied,
7749 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007750 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01007751 are met, for instance a low number of servers both in a
7752 backend and its backup.
7753
7754 unless <cond> the monitor request will succeed only if the condition is
7755 satisfied, and will fail otherwise. Such a condition may be
7756 based on a test on the presence of a minimum number of active
7757 servers in a list of backends.
7758
7759 This statement adds a condition which can force the response to a monitor
7760 request to report a failure. By default, when an external component queries
7761 the URI dedicated to monitoring, a 200 response is returned. When one of the
7762 conditions above is met, haproxy will return 503 instead of 200. This is
7763 very useful to report a site failure to an external component which may base
7764 routing advertisements between multiple sites on the availability reported by
7765 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02007766 criterion. Note that "monitor fail" only works in HTTP mode. Both status
7767 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007768
7769 Example:
7770 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01007771 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01007772 acl site_dead nbsrv(dynamic) lt 2
7773 acl site_dead nbsrv(static) lt 2
7774 monitor-uri /site_alive
7775 monitor fail if site_dead
7776
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007777 See also : "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007778
7779
Willy Tarreau2769aa02007-12-27 18:26:09 +01007780monitor-uri <uri>
7781 Intercept a URI used by external components' monitor requests
7782 May be used in sections : defaults | frontend | listen | backend
7783 yes | yes | yes | no
7784 Arguments :
7785 <uri> is the exact URI which we want to intercept to return HAProxy's
7786 health status instead of forwarding the request.
7787
7788 When an HTTP request referencing <uri> will be received on a frontend,
7789 HAProxy will not forward it nor log it, but instead will return either
7790 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
7791 conditions defined with "monitor fail". This is normally enough for any
7792 front-end HTTP probe to detect that the service is UP and running without
7793 forwarding the request to a backend server. Note that the HTTP method, the
7794 version and all headers are ignored, but the request must at least be valid
7795 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
7796
Willy Tarreau721d8e02017-12-01 18:25:08 +01007797 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007798 and even before any "http-request". The only rulesets applied before are the
7799 tcp-request ones. They cannot be logged either, and it is the intended
7800 purpose. They are only used to report HAProxy's health to an upper component,
7801 nothing more. However, it is possible to add any number of conditions using
7802 "monitor fail" and ACLs so that the result can be adjusted to whatever check
7803 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007804
Christopher Faulet6072beb2020-02-18 15:34:58 +01007805 Note: if <uri> starts by a slash ('/'), the matching is performed against the
7806 request's path instead of the request's uri. It is a workaround to let
7807 the HTTP/2 requests match the monitor-uri. Indeed, in HTTP/2, clients
7808 are encouraged to send absolute URIs only.
7809
Willy Tarreau2769aa02007-12-27 18:26:09 +01007810 Example :
7811 # Use /haproxy_test to report haproxy's status
7812 frontend www
7813 mode http
7814 monitor-uri /haproxy_test
7815
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007816 See also : "monitor fail"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007817
Willy Tarreau0ba27502007-12-24 16:55:16 +01007818
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007819option abortonclose
7820no option abortonclose
7821 Enable or disable early dropping of aborted requests pending in queues.
7822 May be used in sections : defaults | frontend | listen | backend
7823 yes | no | yes | yes
7824 Arguments : none
7825
7826 In presence of very high loads, the servers will take some time to respond.
7827 The per-instance connection queue will inflate, and the response time will
7828 increase respective to the size of the queue times the average per-session
7829 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01007830 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007831 the queue, and slowing down other users, and the servers as well, because the
7832 request will eventually be served, then aborted at the first error
7833 encountered while delivering the response.
7834
7835 As there is no way to distinguish between a full STOP and a simple output
7836 close on the client side, HTTP agents should be conservative and consider
7837 that the client might only have closed its output channel while waiting for
7838 the response. However, this introduces risks of congestion when lots of users
7839 do the same, and is completely useless nowadays because probably no client at
7840 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01007841 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007842 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01007843 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007844 of being the single component to break rare but valid traffic is extremely
7845 low, which adds to the temptation to be able to abort a session early while
7846 still not served and not pollute the servers.
7847
Davor Ocelice9ed2812017-12-25 17:49:28 +01007848 In HAProxy, the user can choose the desired behavior using the option
7849 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007850 compliant and aborted requests will be served. But when the option is
7851 specified, a session with an incoming channel closed will be aborted while
7852 it is still possible, either pending in the queue for a connection slot, or
7853 during the connection establishment if the server has not yet acknowledged
7854 the connection request. This considerably reduces the queue size and the load
7855 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01007856 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007857
7858 If this option has been enabled in a "defaults" section, it can be disabled
7859 in a specific instance by prepending the "no" keyword before it.
7860
7861 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
7862
7863
Willy Tarreau4076a152009-04-02 15:18:36 +02007864option accept-invalid-http-request
7865no option accept-invalid-http-request
7866 Enable or disable relaxing of HTTP request parsing
7867 May be used in sections : defaults | frontend | listen | backend
7868 yes | yes | yes | no
7869 Arguments : none
7870
Willy Tarreau91852eb2015-05-01 13:26:00 +02007871 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007872 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007873 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007874 forbidden characters are essentially used to build attacks exploiting server
7875 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7876 server will emit invalid header names for whatever reason (configuration,
7877 implementation) and the issue will not be immediately fixed. In such a case,
7878 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01007879 even if that does not make sense, by specifying this option. Similarly, the
7880 list of characters allowed to appear in a URI is well defined by RFC3986, and
7881 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
7882 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01007883 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02007884 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02007885 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
7886 to pass through (no version specified) and multiple digits for both the major
7887 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007888
7889 This option should never be enabled by default as it hides application bugs
7890 and open security breaches. It should only be deployed after a problem has
7891 been confirmed.
7892
7893 When this option is enabled, erroneous header names will still be accepted in
7894 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01007895 analysis using the "show errors" request on the UNIX stats socket. Similarly,
7896 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02007897 also helps confirming that the issue has been solved.
7898
7899 If this option has been enabled in a "defaults" section, it can be disabled
7900 in a specific instance by prepending the "no" keyword before it.
7901
7902 See also : "option accept-invalid-http-response" and "show errors" on the
7903 stats socket.
7904
7905
7906option accept-invalid-http-response
7907no option accept-invalid-http-response
7908 Enable or disable relaxing of HTTP response parsing
7909 May be used in sections : defaults | frontend | listen | backend
7910 yes | no | yes | yes
7911 Arguments : none
7912
Willy Tarreau91852eb2015-05-01 13:26:00 +02007913 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007914 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007915 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007916 forbidden characters are essentially used to build attacks exploiting server
7917 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7918 server will emit invalid header names for whatever reason (configuration,
7919 implementation) and the issue will not be immediately fixed. In such a case,
7920 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02007921 even if that does not make sense, by specifying this option. This option also
7922 relaxes the test on the HTTP version format, it allows multiple digits for
7923 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007924
7925 This option should never be enabled by default as it hides application bugs
7926 and open security breaches. It should only be deployed after a problem has
7927 been confirmed.
7928
7929 When this option is enabled, erroneous header names will still be accepted in
7930 responses, but the complete response will be captured in order to permit
7931 later analysis using the "show errors" request on the UNIX stats socket.
7932 Doing this also helps confirming that the issue has been solved.
7933
7934 If this option has been enabled in a "defaults" section, it can be disabled
7935 in a specific instance by prepending the "no" keyword before it.
7936
7937 See also : "option accept-invalid-http-request" and "show errors" on the
7938 stats socket.
7939
7940
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007941option allbackups
7942no option allbackups
7943 Use either all backup servers at a time or only the first one
7944 May be used in sections : defaults | frontend | listen | backend
7945 yes | no | yes | yes
7946 Arguments : none
7947
7948 By default, the first operational backup server gets all traffic when normal
7949 servers are all down. Sometimes, it may be preferred to use multiple backups
7950 at once, because one will not be enough. When "option allbackups" is enabled,
7951 the load balancing will be performed among all backup servers when all normal
7952 ones are unavailable. The same load balancing algorithm will be used and the
7953 servers' weights will be respected. Thus, there will not be any priority
7954 order between the backup servers anymore.
7955
7956 This option is mostly used with static server farms dedicated to return a
7957 "sorry" page when an application is completely offline.
7958
7959 If this option has been enabled in a "defaults" section, it can be disabled
7960 in a specific instance by prepending the "no" keyword before it.
7961
7962
7963option checkcache
7964no option checkcache
Godbach7056a352013-12-11 20:01:07 +08007965 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007966 May be used in sections : defaults | frontend | listen | backend
7967 yes | no | yes | yes
7968 Arguments : none
7969
7970 Some high-level frameworks set application cookies everywhere and do not
7971 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007972 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007973 high risk of session crossing or stealing between users traversing the same
7974 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007975 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007976
7977 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007978 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01007979 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007980 response to check if there's a risk of caching a cookie on a client-side
7981 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01007982 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007983 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01007984 - all those with a return code other than 200, 203, 204, 206, 300, 301,
7985 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01007986 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01007987 - all those that result from a request using a method other than GET, HEAD,
7988 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01007989 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007990 - those with a 'Pragma: no-cache' header
7991 - those with a 'Cache-control: private' header
7992 - those with a 'Cache-control: no-store' header
7993 - those with a 'Cache-control: max-age=0' header
7994 - those with a 'Cache-control: s-maxage=0' header
7995 - those with a 'Cache-control: no-cache' header
7996 - those with a 'Cache-control: no-cache="set-cookie"' header
7997 - those with a 'Cache-control: no-cache="set-cookie,' header
7998 (allowing other fields after set-cookie)
7999
8000 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02008001 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
8002 gateway". The session state shows "PH--" meaning that the proxy blocked the
8003 response during headers processing. Additionally, an alert will be sent in
8004 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008005
8006 Due to the high impact on the application, the application should be tested
8007 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008008 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01008009 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008010
8011 If this option has been enabled in a "defaults" section, it can be disabled
8012 in a specific instance by prepending the "no" keyword before it.
8013
8014
8015option clitcpka
8016no option clitcpka
8017 Enable or disable the sending of TCP keepalive packets on the client side
8018 May be used in sections : defaults | frontend | listen | backend
8019 yes | yes | yes | no
8020 Arguments : none
8021
8022 When there is a firewall or any session-aware component between a client and
8023 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01008024 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008025 components decides to expire a session which has remained idle for too long.
8026
8027 Enabling socket-level TCP keep-alives makes the system regularly send packets
8028 to the other end of the connection, leaving it active. The delay between
8029 keep-alive probes is controlled by the system only and depends both on the
8030 operating system and its tuning parameters.
8031
8032 It is important to understand that keep-alive packets are neither emitted nor
8033 received at the application level. It is only the network stacks which sees
8034 them. For this reason, even if one side of the proxy already uses keep-alives
8035 to maintain its connection alive, those keep-alive packets will not be
8036 forwarded to the other side of the proxy.
8037
8038 Please note that this has nothing to do with HTTP keep-alive.
8039
8040 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
8041 client side of a connection, which should help when session expirations are
8042 noticed between HAProxy and a client.
8043
8044 If this option has been enabled in a "defaults" section, it can be disabled
8045 in a specific instance by prepending the "no" keyword before it.
8046
8047 See also : "option srvtcpka", "option tcpka"
8048
8049
Willy Tarreau0ba27502007-12-24 16:55:16 +01008050option contstats
8051 Enable continuous traffic statistics updates
8052 May be used in sections : defaults | frontend | listen | backend
8053 yes | yes | yes | no
8054 Arguments : none
8055
8056 By default, counters used for statistics calculation are incremented
8057 only when a session finishes. It works quite well when serving small
8058 objects, but with big ones (for example large images or archives) or
8059 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01008060 a hedgehog. With this option enabled counters get incremented frequently
8061 along the session, typically every 5 seconds, which is often enough to
8062 produce clean graphs. Recounting touches a hotpath directly so it is not
8063 not enabled by default, as it can cause a lot of wakeups for very large
8064 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008065
Christopher Faulet89aed322020-06-02 17:33:56 +02008066option disable-h2-upgrade
8067no option disable-h2-upgrade
8068 Enable or disable the implicit HTTP/2 upgrade from an HTTP/1.x client
8069 connection.
8070 May be used in sections : defaults | frontend | listen | backend
8071 yes | yes | yes | no
8072 Arguments : none
8073
8074 By default, HAProxy is able to implicitly upgrade an HTTP/1.x client
8075 connection to an HTTP/2 connection if the first request it receives from a
8076 given HTTP connection matches the HTTP/2 connection preface (i.e. the string
8077 "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 +01008078 HTTP/1.x and HTTP/2 clients on a non-SSL connections. This option must be
8079 used to disable the implicit upgrade. Note this implicit upgrade is only
8080 supported for HTTP proxies, thus this option too. Note also it is possible to
8081 force the HTTP/2 on clear connections by specifying "proto h2" on the bind
8082 line. Finally, this option is applied on all bind lines. To disable implicit
8083 HTTP/2 upgrades for a specific bind line, it is possible to use "proto h1".
Christopher Faulet89aed322020-06-02 17:33:56 +02008084
8085 If this option has been enabled in a "defaults" section, it can be disabled
8086 in a specific instance by prepending the "no" keyword before it.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008087
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008088option dontlog-normal
8089no option dontlog-normal
8090 Enable or disable logging of normal, successful connections
8091 May be used in sections : defaults | frontend | listen | backend
8092 yes | yes | yes | no
8093 Arguments : none
8094
8095 There are large sites dealing with several thousand connections per second
8096 and for which logging is a major pain. Some of them are even forced to turn
8097 logs off and cannot debug production issues. Setting this option ensures that
8098 normal connections, those which experience no error, no timeout, no retry nor
8099 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
8100 mode, the response status code is checked and return codes 5xx will still be
8101 logged.
8102
8103 It is strongly discouraged to use this option as most of the time, the key to
8104 complex issues is in the normal logs which will not be logged here. If you
8105 need to separate logs, see the "log-separate-errors" option instead.
8106
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008107 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008108 logging.
8109
8110
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008111option dontlognull
8112no option dontlognull
8113 Enable or disable logging of null connections
8114 May be used in sections : defaults | frontend | listen | backend
8115 yes | yes | yes | no
8116 Arguments : none
8117
8118 In certain environments, there are components which will regularly connect to
8119 various systems to ensure that they are still alive. It can be the case from
8120 another load balancer as well as from monitoring systems. By default, even a
8121 simple port probe or scan will produce a log. If those connections pollute
8122 the logs too much, it is possible to enable option "dontlognull" to indicate
8123 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02008124 which typically corresponds to those probes. Note that errors will still be
8125 returned to the client and accounted for in the stats. If this is not what is
8126 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008127
8128 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01008129 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008130 would not be logged.
8131
8132 If this option has been enabled in a "defaults" section, it can be disabled
8133 in a specific instance by prepending the "no" keyword before it.
8134
Willy Tarreau9e9919d2020-10-14 15:55:23 +02008135 See also : "log", "http-ignore-probes", "monitor-uri", and
Willy Tarreau0f228a02015-05-01 15:37:53 +02008136 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008137
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008138
Willy Tarreau87cf5142011-08-19 22:57:24 +02008139option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01008140 Enable insertion of the X-Forwarded-For header to requests sent to servers
8141 May be used in sections : defaults | frontend | listen | backend
8142 yes | yes | yes | yes
8143 Arguments :
8144 <network> is an optional argument used to disable this option for sources
8145 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02008146 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01008147 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008148
8149 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
8150 their client address. This is sometimes annoying when the client's IP address
8151 is expected in server logs. To solve this problem, the well-known HTTP header
8152 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
8153 This header contains a value representing the client's IP address. Since this
8154 header is always appended at the end of the existing header list, the server
8155 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02008156 the server's manual to find how to enable use of this standard header. Note
8157 that only the last occurrence of the header must be used, since it is really
8158 possible that the client has already brought one.
8159
Willy Tarreaud72758d2010-01-12 10:42:19 +01008160 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02008161 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01008162 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01008163 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008164 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02008165 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01008166
8167 Sometimes, a same HAProxy instance may be shared between a direct client
8168 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
8169 used to decrypt HTTPS traffic). It is possible to disable the addition of the
8170 header for a known source address or network by adding the "except" keyword
8171 followed by the network address. In this case, any source IP matching the
8172 network will not cause an addition of this header. Most common uses are with
Christopher Faulet5d1def62021-02-26 09:19:15 +01008173 private networks or 127.0.0.1. IPv4 and IPv6 are both supported.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008174
Willy Tarreau87cf5142011-08-19 22:57:24 +02008175 Alternatively, the keyword "if-none" states that the header will only be
8176 added if it is not present. This should only be used in perfectly trusted
8177 environment, as this might cause a security issue if headers reaching haproxy
8178 are under the control of the end-user.
8179
Willy Tarreauc27debf2008-01-06 08:57:02 +01008180 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02008181 least one of them uses it, the header will be added. Note that the backend's
8182 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02008183 both are defined. In the case of the "if-none" argument, if at least one of
8184 the frontend or the backend does not specify it, it wants the addition to be
8185 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008186
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02008187 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01008188 # Public HTTP address also used by stunnel on the same machine
8189 frontend www
8190 mode http
8191 option forwardfor except 127.0.0.1 # stunnel already adds the header
8192
Ross Westaf72a1d2008-08-03 10:51:45 +02008193 # Those servers want the IP Address in X-Client
8194 backend www
8195 mode http
8196 option forwardfor header X-Client
8197
Willy Tarreau87cf5142011-08-19 22:57:24 +02008198 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02008199 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01008200
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008201
Christopher Faulet98fbe952019-07-22 16:18:24 +02008202option h1-case-adjust-bogus-client
8203no option h1-case-adjust-bogus-client
8204 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
8205 May be used in sections : defaults | frontend | listen | backend
8206 yes | yes | yes | no
8207 Arguments : none
8208
8209 There is no standard case for header names because, as stated in RFC7230,
8210 they are case-insensitive. So applications must handle them in a case-
8211 insensitive manner. But some bogus applications violate the standards and
8212 erroneously rely on the cases most commonly used by browsers. This problem
8213 becomes critical with HTTP/2 because all header names must be exchanged in
8214 lower case, and HAProxy follows the same convention. All header names are
8215 sent in lower case to clients and servers, regardless of the HTTP version.
8216
8217 When HAProxy receives an HTTP/1 response, its header names are converted to
8218 lower case and manipulated and sent this way to the clients. If a client is
8219 known to violate the HTTP standards and to fail to process a response coming
8220 from HAProxy, it is possible to transform the lower case header names to a
8221 different format when the response is formatted and sent to the client, by
8222 enabling this option and specifying the list of headers to be reformatted
8223 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
8224 must only be a temporary workaround for the time it takes the client to be
8225 fixed, because clients which require such workarounds might be vulnerable to
8226 content smuggling attacks and must absolutely be fixed.
8227
8228 Please note that this option will not affect standards-compliant clients.
8229
8230 If this option has been enabled in a "defaults" section, it can be disabled
8231 in a specific instance by prepending the "no" keyword before it.
8232
8233 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
8234 "h1-case-adjust-file".
8235
8236
8237option h1-case-adjust-bogus-server
8238no option h1-case-adjust-bogus-server
8239 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
8240 May be used in sections : defaults | frontend | listen | backend
8241 yes | no | yes | yes
8242 Arguments : none
8243
8244 There is no standard case for header names because, as stated in RFC7230,
8245 they are case-insensitive. So applications must handle them in a case-
8246 insensitive manner. But some bogus applications violate the standards and
8247 erroneously rely on the cases most commonly used by browsers. This problem
8248 becomes critical with HTTP/2 because all header names must be exchanged in
8249 lower case, and HAProxy follows the same convention. All header names are
8250 sent in lower case to clients and servers, regardless of the HTTP version.
8251
8252 When HAProxy receives an HTTP/1 request, its header names are converted to
8253 lower case and manipulated and sent this way to the servers. If a server is
8254 known to violate the HTTP standards and to fail to process a request coming
8255 from HAProxy, it is possible to transform the lower case header names to a
8256 different format when the request is formatted and sent to the server, by
8257 enabling this option and specifying the list of headers to be reformatted
8258 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
8259 must only be a temporary workaround for the time it takes the server to be
8260 fixed, because servers which require such workarounds might be vulnerable to
8261 content smuggling attacks and must absolutely be fixed.
8262
8263 Please note that this option will not affect standards-compliant servers.
8264
8265 If this option has been enabled in a "defaults" section, it can be disabled
8266 in a specific instance by prepending the "no" keyword before it.
8267
8268 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
8269 "h1-case-adjust-file".
8270
8271
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008272option http-buffer-request
8273no option http-buffer-request
8274 Enable or disable waiting for whole HTTP request body before proceeding
8275 May be used in sections : defaults | frontend | listen | backend
8276 yes | yes | yes | yes
8277 Arguments : none
8278
8279 It is sometimes desirable to wait for the body of an HTTP request before
8280 taking a decision. This is what is being done by "balance url_param" for
8281 example. The first use case is to buffer requests from slow clients before
8282 connecting to the server. Another use case consists in taking the routing
8283 decision based on the request body's contents. This option placed in a
8284 frontend or backend forces the HTTP processing to wait until either the whole
Christopher Faulet6db8a2e2019-11-19 16:27:25 +01008285 body is received or the request buffer is full. It can have undesired side
8286 effects with some applications abusing HTTP by expecting unbuffered
8287 transmissions between the frontend and the backend, so this should definitely
8288 not be used by default.
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008289
Christopher Faulet021a8e42021-03-29 10:46:38 +02008290 See also : "option http-no-delay", "timeout http-request",
8291 "http-request wait-for-body"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008292
8293
Willy Tarreau0f228a02015-05-01 15:37:53 +02008294option http-ignore-probes
8295no option http-ignore-probes
8296 Enable or disable logging of null connections and request timeouts
8297 May be used in sections : defaults | frontend | listen | backend
8298 yes | yes | yes | no
8299 Arguments : none
8300
8301 Recently some browsers started to implement a "pre-connect" feature
8302 consisting in speculatively connecting to some recently visited web sites
8303 just in case the user would like to visit them. This results in many
8304 connections being established to web sites, which end up in 408 Request
8305 Timeout if the timeout strikes first, or 400 Bad Request when the browser
8306 decides to close them first. These ones pollute the log and feed the error
8307 counters. There was already "option dontlognull" but it's insufficient in
8308 this case. Instead, this option does the following things :
8309 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01008310 was received over a connection before it was closed;
8311 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02008312 - prevent any error counter from being incremented
8313
8314 That way the empty connection is silently ignored. Note that it is better
8315 not to use this unless it is clear that it is needed, because it will hide
8316 real problems. The most common reason for not receiving a request and seeing
8317 a 408 is due to an MTU inconsistency between the client and an intermediary
8318 element such as a VPN, which blocks too large packets. These issues are
8319 generally seen with POST requests as well as GET with large cookies. The logs
8320 are often the only way to detect them.
8321
8322 If this option has been enabled in a "defaults" section, it can be disabled
8323 in a specific instance by prepending the "no" keyword before it.
8324
8325 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
8326
8327
Willy Tarreau16bfb022010-01-16 19:48:41 +01008328option http-keep-alive
8329no option http-keep-alive
8330 Enable or disable HTTP keep-alive from client to server
8331 May be used in sections : defaults | frontend | listen | backend
8332 yes | yes | yes | yes
8333 Arguments : none
8334
Willy Tarreau70dffda2014-01-30 03:07:23 +01008335 By default HAProxy operates in keep-alive mode with regards to persistent
8336 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02008337 leaves the connection idle on both sides between the end of a response and
8338 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008339 as "option http-server-close" or "option httpclose". This option allows to
8340 set back the keep-alive mode, which can be useful when another mode was used
8341 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01008342
8343 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
8344 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01008345 network) and the fastest session reuse on the server side at the expense
8346 of maintaining idle connections to the servers. In general, it is possible
8347 with this option to achieve approximately twice the request rate that the
8348 "http-server-close" option achieves on small objects. There are mainly two
8349 situations where this option may be useful :
8350
8351 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008352 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01008353
8354 - when the cost of establishing the connection to the server is significant
8355 compared to the cost of retrieving the associated object from the server.
8356
8357 This last case can happen when the server is a fast static server of cache.
8358 In this case, the server will need to be properly tuned to support high enough
8359 connection counts because connections will last until the client sends another
8360 request.
8361
8362 If the client request has to go to another backend or another server due to
8363 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01008364 immediately be closed and a new one re-opened. Option "prefer-last-server" is
8365 available to try optimize server selection so that if the server currently
8366 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01008367
Willy Tarreau16bfb022010-01-16 19:48:41 +01008368 At the moment, logs will not indicate whether requests came from the same
8369 session or not. The accept date reported in the logs corresponds to the end
8370 of the previous request, and the request time corresponds to the time spent
8371 waiting for a new request. The keep-alive request time is still bound to the
8372 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
8373 not set.
8374
Christopher Faulet159e6672019-07-16 15:09:52 +02008375 This option disables and replaces any previous "option httpclose" or "option
8376 http-server-close". When backend and frontend options differ, all of these 4
8377 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01008378
Christopher Faulet315b39c2018-09-21 16:26:19 +02008379 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01008380 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01008381 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01008382
8383
Willy Tarreau96e31212011-05-30 18:10:30 +02008384option http-no-delay
8385no option http-no-delay
8386 Instruct the system to favor low interactive delays over performance in HTTP
8387 May be used in sections : defaults | frontend | listen | backend
8388 yes | yes | yes | yes
8389 Arguments : none
8390
8391 In HTTP, each payload is unidirectional and has no notion of interactivity.
8392 Any agent is expected to queue data somewhat for a reasonably low delay.
8393 There are some very rare server-to-server applications that abuse the HTTP
8394 protocol and expect the payload phase to be highly interactive, with many
8395 interleaved data chunks in both directions within a single request. This is
8396 absolutely not supported by the HTTP specification and will not work across
8397 most proxies or servers. When such applications attempt to do this through
8398 haproxy, it works but they will experience high delays due to the network
8399 optimizations which favor performance by instructing the system to wait for
8400 enough data to be available in order to only send full packets. Typical
8401 delays are around 200 ms per round trip. Note that this only happens with
8402 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
8403 affected.
8404
8405 When "option http-no-delay" is present in either the frontend or the backend
8406 used by a connection, all such optimizations will be disabled in order to
8407 make the exchanges as fast as possible. Of course this offers no guarantee on
8408 the functionality, as it may break at any other place. But if it works via
8409 HAProxy, it will work as fast as possible. This option should never be used
8410 by default, and should never be used at all unless such a buggy application
8411 is discovered. The impact of using this option is an increase of bandwidth
8412 usage and CPU usage, which may significantly lower performance in high
8413 latency environments.
8414
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008415 See also : "option http-buffer-request"
8416
Willy Tarreau96e31212011-05-30 18:10:30 +02008417
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008418option http-pretend-keepalive
8419no option http-pretend-keepalive
8420 Define whether haproxy will announce keepalive to the server or not
8421 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02008422 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008423 Arguments : none
8424
Christopher Faulet315b39c2018-09-21 16:26:19 +02008425 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008426 adds a "Connection: close" header to the request forwarded to the server.
8427 Unfortunately, when some servers see this header, they automatically refrain
8428 from using the chunked encoding for responses of unknown length, while this
8429 is totally unrelated. The immediate effect is that this prevents haproxy from
8430 maintaining the client connection alive. A second effect is that a client or
8431 a cache could receive an incomplete response without being aware of it, and
8432 consider the response complete.
8433
8434 By setting "option http-pretend-keepalive", haproxy will make the server
8435 believe it will keep the connection alive. The server will then not fall back
8436 to the abnormal undesired above. When haproxy gets the whole response, it
8437 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02008438 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008439 connection is correctly closed on the server side.
8440
8441 It is recommended not to enable this option by default, because most servers
8442 will more efficiently close the connection themselves after the last packet,
8443 and release its buffers slightly earlier. Also, the added packet on the
8444 network could slightly reduce the overall peak performance. However it is
8445 worth noting that when this option is enabled, haproxy will have slightly
8446 less work to do. So if haproxy is the bottleneck on the whole architecture,
8447 enabling this option might save a few CPU cycles.
8448
Christopher Faulet98db9762018-09-21 10:25:19 +02008449 This option may be set in backend and listen sections. Using it in a frontend
8450 section will be ignored and a warning will be reported during startup. It is
8451 a backend related option, so there is no real reason to set it on a
8452 frontend. This option may be combined with "option httpclose", which will
8453 cause keepalive to be announced to the server and close to be announced to
8454 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008455
8456 If this option has been enabled in a "defaults" section, it can be disabled
8457 in a specific instance by prepending the "no" keyword before it.
8458
Christopher Faulet315b39c2018-09-21 16:26:19 +02008459 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01008460 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008461
Willy Tarreauc27debf2008-01-06 08:57:02 +01008462
Willy Tarreaub608feb2010-01-02 22:47:18 +01008463option http-server-close
8464no option http-server-close
8465 Enable or disable HTTP connection closing on the server side
8466 May be used in sections : defaults | frontend | listen | backend
8467 yes | yes | yes | yes
8468 Arguments : none
8469
Willy Tarreau70dffda2014-01-30 03:07:23 +01008470 By default HAProxy operates in keep-alive mode with regards to persistent
8471 connections: for each connection it processes each request and response, and
8472 leaves the connection idle on both sides between the end of a response and
8473 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008474 as "option http-server-close" or "option httpclose". Setting "option
8475 http-server-close" enables HTTP connection-close mode on the server side
8476 while keeping the ability to support HTTP keep-alive and pipelining on the
8477 client side. This provides the lowest latency on the client side (slow
8478 network) and the fastest session reuse on the server side to save server
8479 resources, similarly to "option httpclose". It also permits non-keepalive
8480 capable servers to be served in keep-alive mode to the clients if they
8481 conform to the requirements of RFC7230. Please note that some servers do not
8482 always conform to those requirements when they see "Connection: close" in the
8483 request. The effect will be that keep-alive will never be used. A workaround
8484 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01008485
8486 At the moment, logs will not indicate whether requests came from the same
8487 session or not. The accept date reported in the logs corresponds to the end
8488 of the previous request, and the request time corresponds to the time spent
8489 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01008490 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
8491 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01008492
8493 This option may be set both in a frontend and in a backend. It is enabled if
8494 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008495 It disables and replaces any previous "option httpclose" or "option
8496 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8497 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01008498
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
Christopher Faulet315b39c2018-09-21 16:26:19 +02008502 See also : "option httpclose", "option http-pretend-keepalive",
8503 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01008504
Willy Tarreau88d349d2010-01-25 12:15:43 +01008505option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01008506no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01008507 Make use of non-standard Proxy-Connection header instead of Connection
8508 May be used in sections : defaults | frontend | listen | backend
8509 yes | yes | yes | no
8510 Arguments : none
8511
Lukas Tribus23953682017-04-28 13:24:30 +00008512 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01008513 Connection header to indicate their wish of persistent or non-persistent
8514 connections, both browsers and proxies ignore this header for proxied
8515 connections and make use of the undocumented, non-standard Proxy-Connection
8516 header instead. The issue begins when trying to put a load balancer between
8517 browsers and such proxies, because there will be a difference between what
8518 haproxy understands and what the client and the proxy agree on.
8519
8520 By setting this option in a frontend, haproxy can automatically switch to use
8521 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008522 defined here as one where the URI begins with neither a '/' nor a '*'. This
8523 is incompatible with the HTTP tunnel mode. Note that this option can only be
8524 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01008525
Willy Tarreau844a7e72010-01-31 21:46:18 +01008526 Also, when this option is set, a request which requires authentication will
8527 automatically switch to use proxy authentication headers if it is itself a
8528 proxied request. That makes it possible to check or enforce authentication in
8529 front of an existing proxy.
8530
Willy Tarreau88d349d2010-01-25 12:15:43 +01008531 This option should normally never be used, except in front of a proxy.
8532
Christopher Faulet315b39c2018-09-21 16:26:19 +02008533 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01008534
Willy Tarreaud63335a2010-02-26 12:56:52 +01008535option httpchk
8536option httpchk <uri>
8537option httpchk <method> <uri>
8538option httpchk <method> <uri> <version>
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008539 Enables HTTP protocol to check on the servers health
Willy Tarreaud63335a2010-02-26 12:56:52 +01008540 May be used in sections : defaults | frontend | listen | backend
8541 yes | no | yes | yes
8542 Arguments :
8543 <method> is the optional HTTP method used with the requests. When not set,
8544 the "OPTIONS" method is used, as it generally requires low server
8545 processing and is easy to filter out from the logs. Any method
8546 may be used, though it is not recommended to invent non-standard
8547 ones.
8548
8549 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
8550 which is accessible by default on almost any server, but may be
8551 changed to any other URI. Query strings are permitted.
8552
8553 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
8554 but some servers might behave incorrectly in HTTP 1.0, so turning
8555 it to HTTP/1.1 may sometimes help. Note that the Host field is
Christopher Faulet8acb1282020-04-09 08:44:06 +02008556 mandatory in HTTP/1.1, use "http-check send" directive to add it.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008557
8558 By default, server health checks only consist in trying to establish a TCP
8559 connection. When "option httpchk" is specified, a complete HTTP request is
8560 sent once the TCP connection is established, and responses 2xx and 3xx are
8561 considered valid, while all other ones indicate a server failure, including
8562 the lack of any response.
8563
Christopher Faulete5870d82020-04-15 11:32:03 +02008564 Combined with "http-check" directives, it is possible to customize the
8565 request sent during the HTTP health checks or the matching rules on the
8566 response. It is also possible to configure a send/expect sequence, just like
8567 with the directive "tcp-check" for TCP health checks.
8568
8569 The server configuration is used by default to open connections to perform
8570 HTTP health checks. By it is also possible to overwrite server parameters
8571 using "http-check connect" rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008572
Christopher Faulete5870d82020-04-15 11:32:03 +02008573 "httpchk" option does not necessarily require an HTTP backend, it also works
8574 with plain TCP backends. This is particularly useful to check simple scripts
Christopher Faulet14cd3162020-04-16 14:50:06 +02008575 bound to some dedicated ports using the inetd daemon. However, it will always
Daniel Corbett67a82712020-07-06 23:01:19 -04008576 internally relies on an HTX multiplexer. Thus, it means the request
Christopher Faulet14cd3162020-04-16 14:50:06 +02008577 formatting and the response parsing will be strict.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008578
Christopher Faulet8acb1282020-04-09 08:44:06 +02008579 Note : For a while, there was no way to add headers or body in the request
8580 used for HTTP health checks. So a workaround was to hide it at the end
8581 of the version string with a "\r\n" after the version. It is now
8582 deprecated. The directive "http-check send" must be used instead.
8583
Willy Tarreaud63335a2010-02-26 12:56:52 +01008584 Examples :
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008585 # Relay HTTPS traffic to Apache instance and check service availability
8586 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
8587 backend https_relay
8588 mode tcp
8589 option httpchk OPTIONS * HTTP/1.1
8590 http-check send hdr Host www
8591 server apache1 192.168.1.1:443 check port 80
Willy Tarreaud63335a2010-02-26 12:56:52 +01008592
Simon Hormanafc47ee2013-11-25 10:46:35 +09008593 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
8594 "option pgsql-check", "http-check" and the "check", "port" and
8595 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008596
8597
Willy Tarreauc27debf2008-01-06 08:57:02 +01008598option httpclose
8599no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02008600 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01008601 May be used in sections : defaults | frontend | listen | backend
8602 yes | yes | yes | yes
8603 Arguments : none
8604
Willy Tarreau70dffda2014-01-30 03:07:23 +01008605 By default HAProxy operates in keep-alive mode with regards to persistent
8606 connections: for each connection it processes each request and response, and
8607 leaves the connection idle on both sides between the end of a response and
8608 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008609 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01008610
Christopher Faulet315b39c2018-09-21 16:26:19 +02008611 If "option httpclose" is set, HAProxy will close connections with the server
8612 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05008613 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02008614 and will add one if missing. Any "Connection" header different from "close"
8615 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008616
Christopher Faulet315b39c2018-09-21 16:26:19 +02008617 This option may also be combined with "option http-pretend-keepalive", which
8618 will disable sending of the "Connection: close" header, but will still cause
8619 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008620
8621 This option may be set both in a frontend and in a backend. It is enabled if
8622 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008623 It disables and replaces any previous "option http-server-close" or "option
8624 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8625 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008626
8627 If this option has been enabled in a "defaults" section, it can be disabled
8628 in a specific instance by prepending the "no" keyword before it.
8629
Christopher Faulet315b39c2018-09-21 16:26:19 +02008630 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01008631
8632
Emeric Brun3a058f32009-06-30 18:26:00 +02008633option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01008634 Enable logging of HTTP request, session state and timers
8635 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01008636 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02008637 Arguments :
8638 clf if the "clf" argument is added, then the output format will be
8639 the CLF format instead of HAProxy's default HTTP format. You can
8640 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01008641 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02008642 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008643
8644 By default, the log output format is very poor, as it only contains the
8645 source and destination addresses, and the instance name. By specifying
8646 "option httplog", each log line turns into a much richer format including,
8647 but not limited to, the HTTP request, the connection timers, the session
8648 status, the connections numbers, the captured headers and cookies, the
8649 frontend, backend and server name, and of course the source address and
8650 ports.
8651
PiBa-NLbd556bf2014-12-11 21:31:54 +01008652 Specifying only "option httplog" will automatically clear the 'clf' mode
8653 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02008654
Guillaume de Lafond29f45602017-03-31 19:52:15 +02008655 "option httplog" overrides any previous "log-format" directive.
8656
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008657 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008658
Willy Tarreau55165fe2009-05-10 12:02:55 +02008659
8660option http_proxy
8661no option http_proxy
8662 Enable or disable plain HTTP proxy mode
8663 May be used in sections : defaults | frontend | listen | backend
8664 yes | yes | yes | yes
8665 Arguments : none
8666
8667 It sometimes happens that people need a pure HTTP proxy which understands
8668 basic proxy requests without caching nor any fancy feature. In this case,
8669 it may be worth setting up an HAProxy instance with the "option http_proxy"
8670 set. In this mode, no server is declared, and the connection is forwarded to
8671 the IP address and port found in the URL after the "http://" scheme.
8672
8673 No host address resolution is performed, so this only works when pure IP
8674 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008675 it will probably be used only by experts who know they need exactly it. This
8676 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02008677
8678 If this option has been enabled in a "defaults" section, it can be disabled
8679 in a specific instance by prepending the "no" keyword before it.
8680
8681 Example :
8682 # this backend understands HTTP proxy requests and forwards them directly.
8683 backend direct_forward
8684 option httpclose
8685 option http_proxy
8686
8687 See also : "option httpclose"
8688
Willy Tarreau211ad242009-10-03 21:45:07 +02008689
Jamie Gloudon801a0a32012-08-25 00:18:33 -04008690option independent-streams
8691no option independent-streams
8692 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008693 May be used in sections : defaults | frontend | listen | backend
8694 yes | yes | yes | yes
8695 Arguments : none
8696
8697 By default, when data is sent over a socket, both the write timeout and the
8698 read timeout for that socket are refreshed, because we consider that there is
8699 activity on that socket, and we have no other means of guessing if we should
8700 receive data or not.
8701
Davor Ocelice9ed2812017-12-25 17:49:28 +01008702 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008703 exists a situation where it is desirable to disable it, and only refresh the
8704 read timeout if there are incoming data. This happens on sessions with large
8705 timeouts and low amounts of exchanged data such as telnet session. If the
8706 server suddenly disappears, the output data accumulates in the system's
8707 socket buffers, both timeouts are correctly refreshed, and there is no way
8708 to know the server does not receive them, so we don't timeout. However, when
8709 the underlying protocol always echoes sent data, it would be enough by itself
8710 to detect the issue using the read timeout. Note that this problem does not
8711 happen with more verbose protocols because data won't accumulate long in the
8712 socket buffers.
8713
8714 When this option is set on the frontend, it will disable read timeout updates
8715 on data sent to the client. There probably is little use of this case. When
8716 the option is set on the backend, it will disable read timeout updates on
8717 data sent to the server. Doing so will typically break large HTTP posts from
8718 slow lines, so use it with caution.
8719
Willy Tarreauce887fd2012-05-12 12:50:00 +02008720 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008721
8722
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02008723option ldap-check
8724 Use LDAPv3 health checks for server testing
8725 May be used in sections : defaults | frontend | listen | backend
8726 yes | no | yes | yes
8727 Arguments : none
8728
8729 It is possible to test that the server correctly talks LDAPv3 instead of just
8730 testing that it accepts the TCP connection. When this option is set, an
8731 LDAPv3 anonymous simple bind message is sent to the server, and the response
8732 is analyzed to find an LDAPv3 bind response message.
8733
8734 The server is considered valid only when the LDAP response contains success
8735 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
8736
8737 Logging of bind requests is server dependent see your documentation how to
8738 configure it.
8739
8740 Example :
8741 option ldap-check
8742
8743 See also : "option httpchk"
8744
8745
Simon Horman98637e52014-06-20 12:30:16 +09008746option external-check
8747 Use external processes for server health checks
8748 May be used in sections : defaults | frontend | listen | backend
8749 yes | no | yes | yes
8750
8751 It is possible to test the health of a server using an external command.
8752 This is achieved by running the executable set using "external-check
8753 command".
8754
8755 Requires the "external-check" global to be set.
8756
8757 See also : "external-check", "external-check command", "external-check path"
8758
8759
Willy Tarreau211ad242009-10-03 21:45:07 +02008760option log-health-checks
8761no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02008762 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02008763 May be used in sections : defaults | frontend | listen | backend
8764 yes | no | yes | yes
8765 Arguments : none
8766
Willy Tarreaubef1b322014-05-13 21:01:39 +02008767 By default, failed health check are logged if server is UP and successful
8768 health checks are logged if server is DOWN, so the amount of additional
8769 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02008770
Willy Tarreaubef1b322014-05-13 21:01:39 +02008771 When this option is enabled, any change of the health check status or to
8772 the server's health will be logged, so that it becomes possible to know
8773 that a server was failing occasional checks before crashing, or exactly when
8774 it failed to respond a valid HTTP status, then when the port started to
8775 reject connections, then when the server stopped responding at all.
8776
Davor Ocelice9ed2812017-12-25 17:49:28 +01008777 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02008778 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02008779
Willy Tarreaubef1b322014-05-13 21:01:39 +02008780 See also: "option httpchk", "option ldap-check", "option mysql-check",
8781 "option pgsql-check", "option redis-check", "option smtpchk",
8782 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02008783
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008784
8785option log-separate-errors
8786no option log-separate-errors
8787 Change log level for non-completely successful connections
8788 May be used in sections : defaults | frontend | listen | backend
8789 yes | yes | yes | no
8790 Arguments : none
8791
8792 Sometimes looking for errors in logs is not easy. This option makes haproxy
8793 raise the level of logs containing potentially interesting information such
8794 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
8795 level changes from "info" to "err". This makes it possible to log them
8796 separately to a different file with most syslog daemons. Be careful not to
8797 remove them from the original file, otherwise you would lose ordering which
8798 provides very important information.
8799
8800 Using this option, large sites dealing with several thousand connections per
8801 second may log normal traffic to a rotating buffer and only archive smaller
8802 error logs.
8803
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008804 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008805 logging.
8806
Willy Tarreauc27debf2008-01-06 08:57:02 +01008807
8808option logasap
8809no option logasap
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008810 Enable or disable early logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008811 May be used in sections : defaults | frontend | listen | backend
8812 yes | yes | yes | no
8813 Arguments : none
8814
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008815 By default, logs are emitted when all the log format variables and sample
8816 fetches used in the definition of the log-format string return a value, or
8817 when the session is terminated. This allows the built in log-format strings
8818 to account for the transfer time, or the number of bytes in log messages.
8819
8820 When handling long lived connections such as large file transfers or RDP,
8821 it may take a while for the request or connection to appear in the logs.
8822 Using "option logasap", the log message is created as soon as the server
8823 connection is established in mode tcp, or as soon as the server sends the
8824 complete headers in mode http. Missing information in the logs will be the
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008825 total number of bytes which will only indicate the amount of data transferred
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008826 before the message was created and the total time which will not take the
8827 remainder of the connection life or transfer time into account. For the case
8828 of HTTP, it is good practice to capture the Content-Length response header
8829 so that the logs at least indicate how many bytes are expected to be
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008830 transferred.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008831
Willy Tarreaucc6c8912009-02-22 10:53:55 +01008832 Examples :
8833 listen http_proxy 0.0.0.0:80
8834 mode http
8835 option httplog
8836 option logasap
8837 log 192.168.2.200 local3
8838
8839 >>> Feb 6 12:14:14 localhost \
8840 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
8841 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
8842 "GET /image.iso HTTP/1.0"
8843
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008844 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01008845 logging.
8846
8847
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008848option mysql-check [ user <username> [ { post-41 | pre-41 } ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008849 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008850 May be used in sections : defaults | frontend | listen | backend
8851 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008852 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008853 <username> This is the username which will be used when connecting to MySQL
8854 server.
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008855 post-41 Send post v4.1 client compatible checks (the default)
8856 pre-41 Send pre v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008857
8858 If you specify a username, the check consists of sending two MySQL packet,
8859 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01008860 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008861 Error packet. It is a basic but useful test which does not produce error nor
8862 aborted connect on the server. However, it requires adding an authorization
8863 in the MySQL table, like this :
8864
8865 USE mysql;
8866 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
8867 FLUSH PRIVILEGES;
8868
8869 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008870 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008871 Error packet, we don't send anything in this mode. It was reported that it
8872 can generate lockout if check is too frequent and/or if there is not enough
8873 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
8874 value as if a connection is established successfully within fewer than MySQL
8875 "max_connect_errors" attempts after a previous connection was interrupted,
8876 the error count for the host is cleared to zero. If HAProxy's server get
8877 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
8878
8879 Remember that this does not check database presence nor database consistency.
8880 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008881
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02008882 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008883
8884 Most often, an incoming MySQL server needs to see the client's IP address for
8885 various purposes, including IP privilege matching and connection logging.
8886 When possible, it is often wise to masquerade the client's IP address when
8887 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02008888 which requires the transparent proxy feature to be compiled in, and the MySQL
8889 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008890
8891 See also: "option httpchk"
8892
8893
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008894option nolinger
8895no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008896 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008897 May be used in sections: defaults | frontend | listen | backend
8898 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008899 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008900
Davor Ocelice9ed2812017-12-25 17:49:28 +01008901 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008902 physically disconnected), the session timeouts triggers and the session is
8903 closed. But it will remain in FIN_WAIT1 state for some time in the system,
8904 using some resources and possibly limiting the ability to establish newer
8905 connections.
8906
8907 When this happens, it is possible to activate "option nolinger" which forces
8908 the system to immediately remove any socket's pending data on close. Thus,
Willy Tarreau4a321032020-10-16 04:55:19 +02008909 a TCP RST is emitted, any pending data are truncated, and the session is
8910 instantly purged from the system's tables. The generally visible effect for
8911 a client is that responses are truncated if the close happens with a last
8912 block of data (e.g. on a redirect or error response). On the server side,
8913 it may help release the source ports immediately when forwarding a client
8914 aborts in tunnels. In both cases, TCP resets are emitted and given that
8915 the session is instantly destroyed, there will be no retransmit. On a lossy
8916 network this can increase problems, especially when there is a firewall on
8917 the lossy side, because the firewall might see and process the reset (hence
8918 purge its session) and block any further traffic for this session,, including
8919 retransmits from the other side. So if the other side doesn't receive it,
8920 it will never receive any RST again, and the firewall might log many blocked
8921 packets.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008922
Willy Tarreau4a321032020-10-16 04:55:19 +02008923 For all these reasons, it is strongly recommended NOT to use this option,
8924 unless absolutely needed as a last resort. In most situations, using the
8925 "client-fin" or "server-fin" timeouts achieves similar results with a more
8926 reliable behavior. On Linux it's also possible to use the "tcp-ut" bind or
8927 server setting.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008928
8929 This option may be used both on frontends and backends, depending on the side
8930 where it is required. Use it on the frontend for clients, and on the backend
Willy Tarreau4a321032020-10-16 04:55:19 +02008931 for servers. While this option is technically supported in "defaults"
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +05008932 sections, it must really not be used there as it risks to accidentally
Willy Tarreau4a321032020-10-16 04:55:19 +02008933 propagate to sections that must no use it and to cause problems there.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008934
8935 If this option has been enabled in a "defaults" section, it can be disabled
8936 in a specific instance by prepending the "no" keyword before it.
8937
Willy Tarreau4a321032020-10-16 04:55:19 +02008938 See also: "timeout client-fin", "timeout server-fin", "tcp-ut" bind or server
8939 keywords.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008940
Willy Tarreau55165fe2009-05-10 12:02:55 +02008941option originalto [ except <network> ] [ header <name> ]
8942 Enable insertion of the X-Original-To header to requests sent to servers
8943 May be used in sections : defaults | frontend | listen | backend
8944 yes | yes | yes | yes
8945 Arguments :
8946 <network> is an optional argument used to disable this option for sources
8947 matching <network>
8948 <name> an optional argument to specify a different "X-Original-To"
8949 header name.
8950
8951 Since HAProxy can work in transparent mode, every request from a client can
8952 be redirected to the proxy and HAProxy itself can proxy every request to a
8953 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
8954 be lost. This is annoying when you want access rules based on destination ip
8955 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
8956 added by HAProxy to all requests sent to the server. This header contains a
8957 value representing the original destination IP address. Since this must be
8958 configured to always use the last occurrence of this header only. Note that
8959 only the last occurrence of the header must be used, since it is really
8960 possible that the client has already brought one.
8961
8962 The keyword "header" may be used to supply a different header name to replace
8963 the default "X-Original-To". This can be useful where you might already
8964 have a "X-Original-To" header from a different application, and you need
8965 preserve it. Also if your backend server doesn't use the "X-Original-To"
8966 header and requires different one.
8967
8968 Sometimes, a same HAProxy instance may be shared between a direct client
8969 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
8970 used to decrypt HTTPS traffic). It is possible to disable the addition of the
Amaury Denoyellef8b42922021-03-04 18:41:14 +01008971 header for a known destination address or network by adding the "except"
8972 keyword followed by the network address. In this case, any destination IP
8973 matching the network will not cause an addition of this header. Most common
8974 uses are with private networks or 127.0.0.1. IPv4 and IPv6 are both
8975 supported.
Willy Tarreau55165fe2009-05-10 12:02:55 +02008976
8977 This option may be specified either in the frontend or in the backend. If at
8978 least one of them uses it, the header will be added. Note that the backend's
8979 setting of the header subargument takes precedence over the frontend's if
8980 both are defined.
8981
Willy Tarreau55165fe2009-05-10 12:02:55 +02008982 Examples :
8983 # Original Destination address
8984 frontend www
8985 mode http
8986 option originalto except 127.0.0.1
8987
8988 # Those servers want the IP Address in X-Client-Dst
8989 backend www
8990 mode http
8991 option originalto header X-Client-Dst
8992
Christopher Faulet315b39c2018-09-21 16:26:19 +02008993 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02008994
8995
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008996option persist
8997no option persist
8998 Enable or disable forced persistence on down servers
8999 May be used in sections: defaults | frontend | listen | backend
9000 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009001 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009002
9003 When an HTTP request reaches a backend with a cookie which references a dead
9004 server, by default it is redispatched to another server. It is possible to
9005 force the request to be sent to the dead server first using "option persist"
9006 if absolutely needed. A common use case is when servers are under extreme
9007 load and spend their time flapping. In this case, the users would still be
9008 directed to the server they opened the session on, in the hope they would be
9009 correctly served. It is recommended to use "option redispatch" in conjunction
9010 with this option so that in the event it would not be possible to connect to
9011 the server at all (server definitely dead), the client would finally be
9012 redirected to another valid server.
9013
9014 If this option has been enabled in a "defaults" section, it can be disabled
9015 in a specific instance by prepending the "no" keyword before it.
9016
Willy Tarreau4de91492010-01-22 19:10:05 +01009017 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009018
9019
Willy Tarreau0c122822013-12-15 18:49:01 +01009020option pgsql-check [ user <username> ]
9021 Use PostgreSQL health checks for server testing
9022 May be used in sections : defaults | frontend | listen | backend
9023 yes | no | yes | yes
9024 Arguments :
9025 <username> This is the username which will be used when connecting to
9026 PostgreSQL server.
9027
9028 The check sends a PostgreSQL StartupMessage and waits for either
9029 Authentication request or ErrorResponse message. It is a basic but useful
9030 test which does not produce error nor aborted connect on the server.
9031 This check is identical with the "mysql-check".
9032
9033 See also: "option httpchk"
9034
9035
Willy Tarreau9420b122013-12-15 18:58:25 +01009036option prefer-last-server
9037no option prefer-last-server
9038 Allow multiple load balanced requests to remain on the same server
9039 May be used in sections: defaults | frontend | listen | backend
9040 yes | no | yes | yes
9041 Arguments : none
9042
9043 When the load balancing algorithm in use is not deterministic, and a previous
9044 request was sent to a server to which haproxy still holds a connection, it is
9045 sometimes desirable that subsequent requests on a same session go to the same
9046 server as much as possible. Note that this is different from persistence, as
9047 we only indicate a preference which haproxy tries to apply without any form
9048 of warranty. The real use is for keep-alive connections sent to servers. When
9049 this option is used, haproxy will try to reuse the same connection that is
9050 attached to the server instead of rebalancing to another server, causing a
9051 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01009052 not make much sense to use this in combination with hashing algorithms. Note,
9053 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02009054 to a proxy which sends a 407 (authentication required), when the load
9055 balancing algorithm is not deterministic. This is mandatory for use with the
9056 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01009057 troubleshooting some faulty applications. Option prefer-last-server might be
9058 desirable in these environments as well, to avoid redistributing the traffic
9059 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01009060
9061 If this option has been enabled in a "defaults" section, it can be disabled
9062 in a specific instance by prepending the "no" keyword before it.
9063
9064 See also: "option http-keep-alive"
9065
9066
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009067option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07009068option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009069no option redispatch
9070 Enable or disable session redistribution in case of connection failure
9071 May be used in sections: defaults | frontend | listen | backend
9072 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07009073 Arguments :
9074 <interval> The optional integer value that controls how often redispatches
9075 occur when retrying connections. Positive value P indicates a
9076 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01009077 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07009078 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009079 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07009080 positive value of 1 would indicate a redispatch on every retry,
9081 and a positive value of 3 would indicate a redispatch on every
9082 third retry. You can disable redispatches with a value of 0.
9083
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009084
9085 In HTTP mode, if a server designated by a cookie is down, clients may
9086 definitely stick to it because they cannot flush the cookie, so they will not
9087 be able to access the service anymore.
9088
Willy Tarreau59884a62019-01-02 14:48:31 +01009089 Specifying "option redispatch" will allow the proxy to break cookie or
9090 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009091
Olivier Carrère6e6f59b2020-04-15 11:30:18 +02009092 Active servers are selected from a subset of the list of available
9093 servers. Active servers that are not down or in maintenance (i.e., whose
9094 health is not checked or that have been checked as "up"), are selected in the
9095 following order:
9096
9097 1. Any active, non-backup server, if any, or,
9098
9099 2. If the "allbackups" option is not set, the first backup server in the
9100 list, or
9101
9102 3. If the "allbackups" option is set, any backup server.
9103
9104 When a retry occurs, HAProxy tries to select another server than the last
9105 one. The new server is selected from the current list of servers.
9106
9107 Sometimes, if the list is updated between retries (e.g., if numerous retries
9108 occur and last longer than the time needed to check that a server is down,
9109 remove it from the list and fall back on the list of backup servers),
9110 connections may be redirected to a backup server, though.
9111
Joseph Lynch726ab712015-05-11 23:25:34 -07009112 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009113 connection failures. Of course, it requires having "retries" set to a nonzero
9114 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01009115
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009116 If this option has been enabled in a "defaults" section, it can be disabled
9117 in a specific instance by prepending the "no" keyword before it.
9118
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02009119 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009120
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009121
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02009122option redis-check
9123 Use redis health checks for server testing
9124 May be used in sections : defaults | frontend | listen | backend
9125 yes | no | yes | yes
9126 Arguments : none
9127
9128 It is possible to test that the server correctly talks REDIS protocol instead
9129 of just testing that it accepts the TCP connection. When this option is set,
9130 a PING redis command is sent to the server, and the response is analyzed to
9131 find the "+PONG" response message.
9132
9133 Example :
9134 option redis-check
9135
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009136 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02009137
9138
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009139option smtpchk
9140option smtpchk <hello> <domain>
9141 Use SMTP health checks for server testing
9142 May be used in sections : defaults | frontend | listen | backend
9143 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01009144 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009145 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02009146 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009147 values will be turned into the default command ("HELO").
9148
9149 <domain> is the domain name to present to the server. It may only be
9150 specified (and is mandatory) if the hello command has been
9151 specified. By default, "localhost" is used.
9152
9153 When "option smtpchk" is set, the health checks will consist in TCP
9154 connections followed by an SMTP command. By default, this command is
9155 "HELO localhost". The server's return code is analyzed and only return codes
9156 starting with a "2" will be considered as valid. All other responses,
9157 including a lack of response will constitute an error and will indicate a
9158 dead server.
9159
9160 This test is meant to be used with SMTP servers or relays. Depending on the
9161 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01009162 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009163 25 is often easier than adjusting the configuration.
9164
9165 Most often, an incoming SMTP server needs to see the client's IP address for
9166 various purposes, including spam filtering, anti-spoofing and logging. When
9167 possible, it is often wise to masquerade the client's IP address when
9168 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02009169 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009170
9171 Example :
9172 option smtpchk HELO mydomain.org
9173
9174 See also : "option httpchk", "source"
9175
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009176
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02009177option socket-stats
9178no option socket-stats
9179
9180 Enable or disable collecting & providing separate statistics for each socket.
9181 May be used in sections : defaults | frontend | listen | backend
9182 yes | yes | yes | no
9183
9184 Arguments : none
9185
9186
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009187option splice-auto
9188no option splice-auto
9189 Enable or disable automatic kernel acceleration on sockets in both directions
9190 May be used in sections : defaults | frontend | listen | backend
9191 yes | yes | yes | yes
9192 Arguments : none
9193
9194 When this option is enabled either on a frontend or on a backend, haproxy
9195 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01009196 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009197 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009198 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009199 are not much aggressive in order to limit excessive use of splicing. This
9200 option requires splicing to be enabled at compile time, and may be globally
9201 disabled with the global option "nosplice". Since splice uses pipes, using it
9202 requires that there are enough spare pipes.
9203
9204 Important note: kernel-based TCP splicing is a Linux-specific feature which
9205 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
9206 transfer data between sockets without copying these data to user-space, thus
9207 providing noticeable performance gains and CPU cycles savings. Since many
9208 early implementations are buggy, corrupt data and/or are inefficient, this
9209 feature is not enabled by default, and it should be used with extreme care.
9210 While it is not possible to detect the correctness of an implementation,
9211 2.6.29 is the first version offering a properly working implementation. In
9212 case of doubt, splicing may be globally disabled using the global "nosplice"
9213 keyword.
9214
9215 Example :
9216 option splice-auto
9217
9218 If this option has been enabled in a "defaults" section, it can be disabled
9219 in a specific instance by prepending the "no" keyword before it.
9220
9221 See also : "option splice-request", "option splice-response", and global
9222 options "nosplice" and "maxpipes"
9223
9224
9225option splice-request
9226no option splice-request
9227 Enable or disable automatic kernel acceleration on sockets for requests
9228 May be used in sections : defaults | frontend | listen | backend
9229 yes | yes | yes | yes
9230 Arguments : none
9231
9232 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009233 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009234 the client to the server. It might still use the recv/send scheme if there
9235 are no spare pipes left. This option requires splicing to be enabled at
9236 compile time, and may be globally disabled with the global option "nosplice".
9237 Since splice uses pipes, using it requires that there are enough spare pipes.
9238
9239 Important note: see "option splice-auto" for usage limitations.
9240
9241 Example :
9242 option splice-request
9243
9244 If this option has been enabled in a "defaults" section, it can be disabled
9245 in a specific instance by prepending the "no" keyword before it.
9246
9247 See also : "option splice-auto", "option splice-response", and global options
9248 "nosplice" and "maxpipes"
9249
9250
9251option splice-response
9252no option splice-response
9253 Enable or disable automatic kernel acceleration on sockets for responses
9254 May be used in sections : defaults | frontend | listen | backend
9255 yes | yes | yes | yes
9256 Arguments : none
9257
9258 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009259 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009260 the server to the client. It might still use the recv/send scheme if there
9261 are no spare pipes left. This option requires splicing to be enabled at
9262 compile time, and may be globally disabled with the global option "nosplice".
9263 Since splice uses pipes, using it requires that there are enough spare pipes.
9264
9265 Important note: see "option splice-auto" for usage limitations.
9266
9267 Example :
9268 option splice-response
9269
9270 If this option has been enabled in a "defaults" section, it can be disabled
9271 in a specific instance by prepending the "no" keyword before it.
9272
9273 See also : "option splice-auto", "option splice-request", and global options
9274 "nosplice" and "maxpipes"
9275
9276
Christopher Fauletba7bc162016-11-07 21:07:38 +01009277option spop-check
9278 Use SPOP health checks for server testing
9279 May be used in sections : defaults | frontend | listen | backend
9280 no | no | no | yes
9281 Arguments : none
9282
9283 It is possible to test that the server correctly talks SPOP protocol instead
9284 of just testing that it accepts the TCP connection. When this option is set,
9285 a HELLO handshake is performed between HAProxy and the server, and the
9286 response is analyzed to check no error is reported.
9287
9288 Example :
9289 option spop-check
9290
9291 See also : "option httpchk"
9292
9293
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009294option srvtcpka
9295no option srvtcpka
9296 Enable or disable the sending of TCP keepalive packets on the server side
9297 May be used in sections : defaults | frontend | listen | backend
9298 yes | no | yes | yes
9299 Arguments : none
9300
9301 When there is a firewall or any session-aware component between a client and
9302 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01009303 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009304 components decides to expire a session which has remained idle for too long.
9305
9306 Enabling socket-level TCP keep-alives makes the system regularly send packets
9307 to the other end of the connection, leaving it active. The delay between
9308 keep-alive probes is controlled by the system only and depends both on the
9309 operating system and its tuning parameters.
9310
9311 It is important to understand that keep-alive packets are neither emitted nor
9312 received at the application level. It is only the network stacks which sees
9313 them. For this reason, even if one side of the proxy already uses keep-alives
9314 to maintain its connection alive, those keep-alive packets will not be
9315 forwarded to the other side of the proxy.
9316
9317 Please note that this has nothing to do with HTTP keep-alive.
9318
9319 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
9320 server side of a connection, which should help when session expirations are
9321 noticed between HAProxy and a server.
9322
9323 If this option has been enabled in a "defaults" section, it can be disabled
9324 in a specific instance by prepending the "no" keyword before it.
9325
9326 See also : "option clitcpka", "option tcpka"
9327
9328
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009329option ssl-hello-chk
9330 Use SSLv3 client hello health checks for server testing
9331 May be used in sections : defaults | frontend | listen | backend
9332 yes | no | yes | yes
9333 Arguments : none
9334
9335 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
9336 possible to test that the server correctly talks SSL instead of just testing
9337 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
9338 SSLv3 client hello messages are sent once the connection is established to
9339 the server, and the response is analyzed to find an SSL server hello message.
9340 The server is considered valid only when the response contains this server
9341 hello message.
9342
9343 All servers tested till there correctly reply to SSLv3 client hello messages,
9344 and most servers tested do not even log the requests containing only hello
9345 messages, which is appreciable.
9346
Willy Tarreau763a95b2012-10-04 23:15:39 +02009347 Note that this check works even when SSL support was not built into haproxy
9348 because it forges the SSL message. When SSL support is available, it is best
9349 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009350
Willy Tarreau763a95b2012-10-04 23:15:39 +02009351 See also: "option httpchk", "check-ssl"
9352
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009353
Willy Tarreaued179852013-12-16 01:07:00 +01009354option tcp-check
9355 Perform health checks using tcp-check send/expect sequences
9356 May be used in sections: defaults | frontend | listen | backend
9357 yes | no | yes | yes
9358
9359 This health check method is intended to be combined with "tcp-check" command
9360 lists in order to support send/expect types of health check sequences.
9361
9362 TCP checks currently support 4 modes of operations :
9363 - no "tcp-check" directive : the health check only consists in a connection
9364 attempt, which remains the default mode.
9365
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009366 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01009367 used to send a string along with a connection opening. With some
9368 protocols, it helps sending a "QUIT" message for example that prevents
9369 the server from logging a connection error for each health check. The
9370 check result will still be based on the ability to open the connection
9371 only.
9372
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009373 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01009374 The connection is opened and haproxy waits for the server to present some
9375 contents which must validate some rules. The check result will be based
9376 on the matching between the contents and the rules. This is suited for
9377 POP, IMAP, SMTP, FTP, SSH, TELNET.
9378
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009379 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009380 used to test a hello-type protocol. HAProxy sends a message, the server
9381 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009382 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01009383 suited for protocols which require a binding or a request/response model.
9384 LDAP, MySQL, Redis and SSL are example of such protocols, though they
9385 already all have their dedicated checks with a deeper understanding of
9386 the respective protocols.
9387 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01009388 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01009389
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009390 A fifth mode can be used to insert comments in different steps of the script.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009391
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009392 For each tcp-check rule you create, you can add a "comment" directive,
9393 followed by a string. This string will be reported in the log and stderr in
9394 debug mode. It is useful to make user-friendly error reporting. The
9395 "comment" is of course optional.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009396
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009397 During the execution of a health check, a variable scope is made available to
9398 store data samples, using the "tcp-check set-var" operation. Freeing those
9399 variable is possible using "tcp-check unset-var".
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +01009400
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009401
Willy Tarreaued179852013-12-16 01:07:00 +01009402 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01009403 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01009404 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009405 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01009406
Davor Ocelice9ed2812017-12-25 17:49:28 +01009407 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01009408 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009409 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01009410
9411 # look for the redis master server after ensuring it speaks well
9412 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009413 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01009414 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009415 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01009416 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009417 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009418 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01009419 tcp-check send info\ replication\r\n
9420 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009421 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01009422 tcp-check send QUIT\r\n
9423 tcp-check expect string +OK
9424
Davor Ocelice9ed2812017-12-25 17:49:28 +01009425 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01009426 (send many headers before analyzing)
9427 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009428 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01009429 tcp-check send HEAD\ /\ HTTP/1.1\r\n
9430 tcp-check send Host:\ www.mydomain.com\r\n
9431 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
9432 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009433 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01009434
9435
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009436 See also : "tcp-check connect", "tcp-check expect" and "tcp-check send".
Willy Tarreaued179852013-12-16 01:07:00 +01009437
9438
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009439option tcp-smart-accept
9440no option tcp-smart-accept
9441 Enable or disable the saving of one ACK packet during the accept sequence
9442 May be used in sections : defaults | frontend | listen | backend
9443 yes | yes | yes | no
9444 Arguments : none
9445
9446 When an HTTP connection request comes in, the system acknowledges it on
9447 behalf of HAProxy, then the client immediately sends its request, and the
9448 system acknowledges it too while it is notifying HAProxy about the new
9449 connection. HAProxy then reads the request and responds. This means that we
9450 have one TCP ACK sent by the system for nothing, because the request could
9451 very well be acknowledged by HAProxy when it sends its response.
9452
9453 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
9454 sending this useless ACK on platforms which support it (currently at least
9455 Linux). It must not cause any problem, because the system will send it anyway
9456 after 40 ms if the response takes more time than expected to come.
9457
9458 During complex network debugging sessions, it may be desirable to disable
9459 this optimization because delayed ACKs can make troubleshooting more complex
9460 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01009461 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009462
9463 It is also possible to force it for non-HTTP proxies by simply specifying
9464 "option tcp-smart-accept". For instance, it can make sense with some services
9465 such as SMTP where the server speaks first.
9466
9467 It is recommended to avoid forcing this option in a defaults section. In case
9468 of doubt, consider setting it back to automatic values by prepending the
9469 "default" keyword before it, or disabling it using the "no" keyword.
9470
Willy Tarreaud88edf22009-06-14 15:48:17 +02009471 See also : "option tcp-smart-connect"
9472
9473
9474option tcp-smart-connect
9475no option tcp-smart-connect
9476 Enable or disable the saving of one ACK packet during the connect sequence
9477 May be used in sections : defaults | frontend | listen | backend
9478 yes | no | yes | yes
9479 Arguments : none
9480
9481 On certain systems (at least Linux), HAProxy can ask the kernel not to
9482 immediately send an empty ACK upon a connection request, but to directly
9483 send the buffer request instead. This saves one packet on the network and
9484 thus boosts performance. It can also be useful for some servers, because they
9485 immediately get the request along with the incoming connection.
9486
9487 This feature is enabled when "option tcp-smart-connect" is set in a backend.
9488 It is not enabled by default because it makes network troubleshooting more
9489 complex.
9490
9491 It only makes sense to enable it with protocols where the client speaks first
9492 such as HTTP. In other situations, if there is no data to send in place of
9493 the ACK, a normal ACK is sent.
9494
9495 If this option has been enabled in a "defaults" section, it can be disabled
9496 in a specific instance by prepending the "no" keyword before it.
9497
9498 See also : "option tcp-smart-accept"
9499
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009500
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009501option tcpka
9502 Enable or disable the sending of TCP keepalive packets on both sides
9503 May be used in sections : defaults | frontend | listen | backend
9504 yes | yes | yes | yes
9505 Arguments : none
9506
9507 When there is a firewall or any session-aware component between a client and
9508 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01009509 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009510 components decides to expire a session which has remained idle for too long.
9511
9512 Enabling socket-level TCP keep-alives makes the system regularly send packets
9513 to the other end of the connection, leaving it active. The delay between
9514 keep-alive probes is controlled by the system only and depends both on the
9515 operating system and its tuning parameters.
9516
9517 It is important to understand that keep-alive packets are neither emitted nor
9518 received at the application level. It is only the network stacks which sees
9519 them. For this reason, even if one side of the proxy already uses keep-alives
9520 to maintain its connection alive, those keep-alive packets will not be
9521 forwarded to the other side of the proxy.
9522
9523 Please note that this has nothing to do with HTTP keep-alive.
9524
9525 Using option "tcpka" enables the emission of TCP keep-alive probes on both
9526 the client and server sides of a connection. Note that this is meaningful
9527 only in "defaults" or "listen" sections. If this option is used in a
9528 frontend, only the client side will get keep-alives, and if this option is
9529 used in a backend, only the server side will get keep-alives. For this
9530 reason, it is strongly recommended to explicitly use "option clitcpka" and
9531 "option srvtcpka" when the configuration is split between frontends and
9532 backends.
9533
9534 See also : "option clitcpka", "option srvtcpka"
9535
Willy Tarreau844e3c52008-01-11 16:28:18 +01009536
9537option tcplog
9538 Enable advanced logging of TCP connections with session state and timers
9539 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01009540 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01009541 Arguments : none
9542
9543 By default, the log output format is very poor, as it only contains the
9544 source and destination addresses, and the instance name. By specifying
9545 "option tcplog", each log line turns into a much richer format including, but
9546 not limited to, the connection timers, the session status, the connections
9547 numbers, the frontend, backend and server name, and of course the source
9548 address and ports. This option is useful for pure TCP proxies in order to
9549 find which of the client or server disconnects or times out. For normal HTTP
9550 proxies, it's better to use "option httplog" which is even more complete.
9551
Guillaume de Lafond29f45602017-03-31 19:52:15 +02009552 "option tcplog" overrides any previous "log-format" directive.
9553
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009554 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009555
9556
Willy Tarreau844e3c52008-01-11 16:28:18 +01009557option transparent
9558no option transparent
9559 Enable client-side transparent proxying
9560 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01009561 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01009562 Arguments : none
9563
9564 This option was introduced in order to provide layer 7 persistence to layer 3
9565 load balancers. The idea is to use the OS's ability to redirect an incoming
9566 connection for a remote address to a local process (here HAProxy), and let
9567 this process know what address was initially requested. When this option is
9568 used, sessions without cookies will be forwarded to the original destination
9569 IP address of the incoming request (which should match that of another
9570 equipment), while requests with cookies will still be forwarded to the
9571 appropriate server.
9572
9573 Note that contrary to a common belief, this option does NOT make HAProxy
9574 present the client's IP to the server when establishing the connection.
9575
Willy Tarreaua1146052011-03-01 09:51:54 +01009576 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009577 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009578
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009579
Simon Horman98637e52014-06-20 12:30:16 +09009580external-check command <command>
9581 Executable to run when performing an external-check
9582 May be used in sections : defaults | frontend | listen | backend
9583 yes | no | yes | yes
9584
9585 Arguments :
9586 <command> is the external command to run
9587
Simon Horman98637e52014-06-20 12:30:16 +09009588 The arguments passed to the to the command are:
9589
Cyril Bonté777be862014-12-02 21:21:35 +01009590 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09009591
Cyril Bonté777be862014-12-02 21:21:35 +01009592 The <proxy_address> and <proxy_port> are derived from the first listener
9593 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
9594 listener the proxy_address will be the path of the socket and the
9595 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
9596 possible to determine a listener, and both <proxy_address> and <proxy_port>
9597 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09009598
Cyril Bonté72cda2a2014-12-27 22:28:39 +01009599 Some values are also provided through environment variables.
9600
9601 Environment variables :
9602 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
9603 applicable, for example in a "backend" section).
9604
9605 HAPROXY_PROXY_ID The backend id.
9606
9607 HAPROXY_PROXY_NAME The backend name.
9608
9609 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
9610 applicable, for example in a "backend" section or
9611 for a UNIX socket).
9612
9613 HAPROXY_SERVER_ADDR The server address.
9614
9615 HAPROXY_SERVER_CURCONN The current number of connections on the server.
9616
9617 HAPROXY_SERVER_ID The server id.
9618
9619 HAPROXY_SERVER_MAXCONN The server max connections.
9620
9621 HAPROXY_SERVER_NAME The server name.
9622
9623 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
9624 socket).
9625
9626 PATH The PATH environment variable used when executing
9627 the command may be set using "external-check path".
9628
William Lallemand4d03e432019-06-14 15:35:37 +02009629 See also "2.3. Environment variables" for other variables.
9630
Simon Horman98637e52014-06-20 12:30:16 +09009631 If the command executed and exits with a zero status then the check is
9632 considered to have passed, otherwise the check is considered to have
9633 failed.
9634
9635 Example :
9636 external-check command /bin/true
9637
9638 See also : "external-check", "option external-check", "external-check path"
9639
9640
9641external-check path <path>
9642 The value of the PATH environment variable used when running an external-check
9643 May be used in sections : defaults | frontend | listen | backend
9644 yes | no | yes | yes
9645
9646 Arguments :
9647 <path> is the path used when executing external command to run
9648
9649 The default path is "".
9650
9651 Example :
9652 external-check path "/usr/bin:/bin"
9653
9654 See also : "external-check", "option external-check",
9655 "external-check command"
9656
9657
Emeric Brun647caf12009-06-30 17:57:00 +02009658persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02009659persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02009660 Enable RDP cookie-based persistence
9661 May be used in sections : defaults | frontend | listen | backend
9662 yes | no | yes | yes
9663 Arguments :
9664 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02009665 default cookie name "msts" will be used. There currently is no
9666 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02009667
9668 This statement enables persistence based on an RDP cookie. The RDP cookie
9669 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01009670 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02009671 and if an RDP cookie is found, it is decoded. If it matches a known server
9672 which is still UP (or if "option persist" is set), then the connection is
9673 forwarded to this server.
9674
9675 Note that this only makes sense in a TCP backend, but for this to work, the
9676 frontend must have waited long enough to ensure that an RDP cookie is present
9677 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009678 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02009679 a single "listen" section.
9680
Willy Tarreau61e28f22010-05-16 22:31:05 +02009681 Also, it is important to understand that the terminal server will emit this
9682 RDP cookie only if it is configured for "token redirection mode", which means
9683 that the "IP address redirection" option is disabled.
9684
Emeric Brun647caf12009-06-30 17:57:00 +02009685 Example :
9686 listen tse-farm
9687 bind :3389
9688 # wait up to 5s for an RDP cookie in the request
9689 tcp-request inspect-delay 5s
9690 tcp-request content accept if RDP_COOKIE
9691 # apply RDP cookie persistence
9692 persist rdp-cookie
9693 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02009694 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02009695 balance rdp-cookie
9696 server srv1 1.1.1.1:3389
9697 server srv2 1.1.1.2:3389
9698
Simon Hormanab814e02011-06-24 14:50:20 +09009699 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
9700 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02009701
9702
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009703rate-limit sessions <rate>
9704 Set a limit on the number of new sessions accepted per second on a frontend
9705 May be used in sections : defaults | frontend | listen | backend
9706 yes | yes | yes | no
9707 Arguments :
9708 <rate> The <rate> parameter is an integer designating the maximum number
9709 of new sessions per second to accept on the frontend.
9710
9711 When the frontend reaches the specified number of new sessions per second, it
9712 stops accepting new connections until the rate drops below the limit again.
9713 During this time, the pending sessions will be kept in the socket's backlog
9714 (in system buffers) and haproxy will not even be aware that sessions are
9715 pending. When applying very low limit on a highly loaded service, it may make
9716 sense to increase the socket's backlog using the "backlog" keyword.
9717
9718 This feature is particularly efficient at blocking connection-based attacks
9719 or service abuse on fragile servers. Since the session rate is measured every
9720 millisecond, it is extremely accurate. Also, the limit applies immediately,
9721 no delay is needed at all to detect the threshold.
9722
9723 Example : limit the connection rate on SMTP to 10 per second max
9724 listen smtp
9725 mode tcp
9726 bind :25
9727 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02009728 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009729
Willy Tarreaua17c2d92011-07-25 08:16:20 +02009730 Note : when the maximum rate is reached, the frontend's status is not changed
9731 but its sockets appear as "WAITING" in the statistics if the
9732 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009733
9734 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
9735
9736
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009737redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
9738redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
9739redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009740 Return an HTTP redirection if/unless a condition is matched
9741 May be used in sections : defaults | frontend | listen | backend
9742 no | yes | yes | yes
9743
9744 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01009745 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009746
Willy Tarreau0140f252008-11-19 21:07:09 +01009747 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009748 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009749 the HTTP "Location" header. When used in an "http-request" rule,
9750 <loc> value follows the log-format rules and can include some
9751 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009752
9753 <pfx> With "redirect prefix", the "Location" header is built from the
9754 concatenation of <pfx> and the complete URI path, including the
9755 query string, unless the "drop-query" option is specified (see
9756 below). As a special case, if <pfx> equals exactly "/", then
9757 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009758 redirect to the same URL (for instance, to insert a cookie). When
9759 used in an "http-request" rule, <pfx> value follows the log-format
9760 rules and can include some dynamic values (see Custom Log Format
9761 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009762
9763 <sch> With "redirect scheme", then the "Location" header is built by
9764 concatenating <sch> with "://" then the first occurrence of the
9765 "Host" header, and then the URI path, including the query string
9766 unless the "drop-query" option is specified (see below). If no
9767 path is found or if the path is "*", then "/" is used instead. If
9768 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009769 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009770 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009771 HTTPS. When used in an "http-request" rule, <sch> value follows
9772 the log-format rules and can include some dynamic values (see
9773 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01009774
9775 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009776 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
9777 with 302 used by default if no code is specified. 301 means
9778 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02009779 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009780 cache the redirection. 303 is equivalent to 302 except that the
9781 browser will fetch the location with a GET method. 307 is just
9782 like 302 but makes it clear that the same method must be reused.
9783 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01009784
9785 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009786 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01009787
9788 - "drop-query"
9789 When this keyword is used in a prefix-based redirection, then the
9790 location will be set without any possible query-string, which is useful
9791 for directing users to a non-secure page for instance. It has no effect
9792 with a location-type redirect.
9793
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009794 - "append-slash"
9795 This keyword may be used in conjunction with "drop-query" to redirect
9796 users who use a URL not ending with a '/' to the same one with the '/'.
9797 It can be useful to ensure that search engines will only see one URL.
9798 For this, a return code 301 is preferred.
9799
Willy Tarreau0140f252008-11-19 21:07:09 +01009800 - "set-cookie NAME[=value]"
9801 A "Set-Cookie" header will be added with NAME (and optionally "=value")
9802 to the response. This is sometimes used to indicate that a user has
9803 been seen, for instance to protect against some types of DoS. No other
9804 cookie option is added, so the cookie will be a session cookie. Note
9805 that for a browser, a sole cookie name without an equal sign is
9806 different from a cookie with an equal sign.
9807
9808 - "clear-cookie NAME[=]"
9809 A "Set-Cookie" header will be added with NAME (and optionally "="), but
9810 with the "Max-Age" attribute set to zero. This will tell the browser to
9811 delete this cookie. It is useful for instance on logout pages. It is
9812 important to note that clearing the cookie "NAME" will not remove a
9813 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
9814 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009815
9816 Example: move the login URL only to HTTPS.
9817 acl clear dst_port 80
9818 acl secure dst_port 8080
9819 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01009820 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01009821 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01009822 acl cookie_set hdr_sub(cookie) SEEN=1
9823
9824 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01009825 redirect prefix https://mysite.com if login_page !secure
9826 redirect prefix http://mysite.com drop-query if login_page !uid_given
9827 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01009828 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009829
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009830 Example: send redirects for request for articles without a '/'.
9831 acl missing_slash path_reg ^/article/[^/]*$
9832 redirect code 301 prefix / drop-query append-slash if missing_slash
9833
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009834 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01009835 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009836
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009837 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02009838 http-request redirect code 301 location \
9839 http://www.%[hdr(host)]%[capture.req.uri] \
9840 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009841
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009842 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009843
Willy Tarreau303c0352008-01-17 19:01:39 +01009844
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009845retries <value>
9846 Set the number of retries to perform on a server after a connection failure
9847 May be used in sections: defaults | frontend | listen | backend
9848 yes | no | yes | yes
9849 Arguments :
9850 <value> is the number of times a connection attempt should be retried on
9851 a server when a connection either is refused or times out. The
9852 default value is 3.
9853
9854 It is important to understand that this value applies to the number of
9855 connection attempts, not full requests. When a connection has effectively
9856 been established to a server, there will be no more retry.
9857
9858 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07009859 a turn-around timer of min("timeout connect", one second) is applied before
9860 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009861
9862 When "option redispatch" is set, the last retry may be performed on another
9863 server even if a cookie references a different server.
9864
9865 See also : "option redispatch"
9866
9867
Olivier Houcharda254a372019-04-05 15:30:12 +02009868retry-on [list of keywords]
Jerome Magnin5ce3c142020-05-13 20:09:57 +02009869 Specify when to attempt to automatically retry a failed request.
9870 This setting is only valid when "mode" is set to http and is silently ignored
9871 otherwise.
Olivier Houcharda254a372019-04-05 15:30:12 +02009872 May be used in sections: defaults | frontend | listen | backend
9873 yes | no | yes | yes
9874 Arguments :
9875 <keywords> is a list of keywords or HTTP status codes, each representing a
9876 type of failure event on which an attempt to retry the request
9877 is desired. Please read the notes at the bottom before changing
9878 this setting. The following keywords are supported :
9879
9880 none never retry
9881
9882 conn-failure retry when the connection or the SSL handshake failed
9883 and the request could not be sent. This is the default.
9884
9885 empty-response retry when the server connection was closed after part
9886 of the request was sent, and nothing was received from
9887 the server. This type of failure may be caused by the
9888 request timeout on the server side, poor network
9889 condition, or a server crash or restart while
9890 processing the request.
9891
Olivier Houcharde3249a92019-05-03 23:01:47 +02009892 junk-response retry when the server returned something not looking
9893 like a complete HTTP response. This includes partial
9894 responses headers as well as non-HTTP contents. It
9895 usually is a bad idea to retry on such events, which
9896 may be caused a configuration issue (wrong server port)
9897 or by the request being harmful to the server (buffer
9898 overflow attack for example).
9899
Olivier Houcharda254a372019-04-05 15:30:12 +02009900 response-timeout the server timeout stroke while waiting for the server
9901 to respond to the request. This may be caused by poor
9902 network condition, the reuse of an idle connection
9903 which has expired on the path, or by the request being
9904 extremely expensive to process. It generally is a bad
9905 idea to retry on such events on servers dealing with
9906 heavy database processing (full scans, etc) as it may
9907 amplify denial of service attacks.
9908
Olivier Houchard865d8392019-05-03 22:46:27 +02009909 0rtt-rejected retry requests which were sent over early data and were
9910 rejected by the server. These requests are generally
9911 considered to be safe to retry.
9912
Julien Pivotto2de240a2020-11-12 11:14:05 +01009913 <status> any HTTP status code among "401" (Unauthorized), "403"
9914 (Forbidden), "404" (Not Found), "408" (Request Timeout),
9915 "425" (Too Early), "500" (Server Error), "501" (Not
9916 Implemented), "502" (Bad Gateway), "503" (Service
9917 Unavailable), "504" (Gateway Timeout).
Olivier Houcharda254a372019-04-05 15:30:12 +02009918
Olivier Houchardddf0e032019-05-10 18:05:40 +02009919 all-retryable-errors
9920 retry request for any error that are considered
9921 retryable. This currently activates "conn-failure",
9922 "empty-response", "junk-response", "response-timeout",
9923 "0rtt-rejected", "500", "502", "503", and "504".
9924
Olivier Houcharda254a372019-04-05 15:30:12 +02009925 Using this directive replaces any previous settings with the new ones; it is
9926 not cumulative.
9927
9928 Please note that using anything other than "none" and "conn-failure" requires
9929 to allocate a buffer and copy the whole request into it, so it has memory and
9930 performance impacts. Requests not fitting in a single buffer will never be
9931 retried (see the global tune.bufsize setting).
9932
9933 You have to make sure the application has a replay protection mechanism built
9934 in such as a unique transaction IDs passed in requests, or that replaying the
9935 same request has no consequence, or it is very dangerous to use any retry-on
9936 value beside "conn-failure" and "none". Static file servers and caches are
9937 generally considered safe against any type of retry. Using a status code can
9938 be useful to quickly leave a server showing an abnormal behavior (out of
9939 memory, file system issues, etc), but in this case it may be a good idea to
9940 immediately redispatch the connection to another server (please see "option
9941 redispatch" for this). Last, it is important to understand that most causes
9942 of failures are the requests themselves and that retrying a request causing a
9943 server to misbehave will often make the situation even worse for this server,
9944 or for the whole service in case of redispatch.
9945
9946 Unless you know exactly how the application deals with replayed requests, you
9947 should not use this directive.
9948
9949 The default is "conn-failure".
9950
9951 See also: "retries", "option redispatch", "tune.bufsize"
9952
David du Colombier486df472011-03-17 10:40:26 +01009953server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009954 Declare a server in a backend
9955 May be used in sections : defaults | frontend | listen | backend
9956 no | no | yes | yes
9957 Arguments :
9958 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009959 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05009960 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009961
David du Colombier486df472011-03-17 10:40:26 +01009962 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
9963 resolvable hostname is supported, but this name will be resolved
9964 during start-up. Address "0.0.0.0" or "*" has a special meaning.
9965 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02009966 address as the one from the client connection. This is useful in
9967 transparent proxy architectures where the client's connection is
9968 intercepted and haproxy must forward to the original destination
9969 address. This is more or less what the "transparent" keyword does
9970 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01009971 to report statistics. Optionally, an address family prefix may be
9972 used before the address to force the family regardless of the
9973 address format, which can be useful to specify a path to a unix
9974 socket with no slash ('/'). Currently supported prefixes are :
9975 - 'ipv4@' -> address is always IPv4
9976 - 'ipv6@' -> address is always IPv6
9977 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02009978 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02009979 - 'sockpair@' -> address is the FD of a connected unix
9980 socket or of a socketpair. During a connection, the
9981 backend creates a pair of connected sockets, and passes
9982 one of them over the FD. The bind part will use the
9983 received socket as the client FD. Should be used
9984 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02009985 You may want to reference some environment variables in the
9986 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01009987 variables. The "init-addr" setting can be used to modify the way
9988 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009989
Willy Tarreaub6205fd2012-09-24 12:27:33 +02009990 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009991 be sent to this port. If unset, the same port the client
9992 connected to will be used. The port may also be prefixed by a "+"
9993 or a "-". In this case, the server's port will be determined by
9994 adding this value to the client's port.
9995
9996 <param*> is a list of parameters for this server. The "server" keywords
9997 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009998 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009999
10000 Examples :
10001 server first 10.1.1.1:1080 cookie first check inter 1000
10002 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +010010003 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +020010004 server backup "${SRV_BACKUP}:1080" backup
10005 server www1_dc1 "${LAN_DC1}.101:80"
10006 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010007
Willy Tarreau55dcaf62015-09-27 15:03:15 +020010008 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
10009 sun_path length is used for the address length. Some other programs
10010 such as socat use the string length only by default. Pass the option
10011 ",unix-tightsocklen=0" to any abstract socket definition in socat to
10012 make it compatible with HAProxy's.
10013
Mark Lamourinec2247f02012-01-04 13:02:01 -050010014 See also: "default-server", "http-send-name-header" and section 5 about
10015 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010016
Christopher Faulet583b6de2021-02-12 09:27:10 +010010017server-state-file-name [ { use-backend-name | <file> } ]
Baptiste Assmann01c6cc32015-08-23 11:45:29 +020010018 Set the server state file to read, load and apply to servers available in
Christopher Faulet583b6de2021-02-12 09:27:10 +010010019 this backend.
10020 May be used in sections: defaults | frontend | listen | backend
10021 no | no | yes | yes
10022
10023 It only applies when the directive "load-server-state-from-file" is set to
10024 "local". When <file> is not provided, if "use-backend-name" is used or if
10025 this directive is not set, then backend name is used. If <file> starts with a
10026 slash '/', then it is considered as an absolute path. Otherwise, <file> is
10027 concatenated to the global directive "server-state-base".
Baptiste Assmann01c6cc32015-08-23 11:45:29 +020010028
10029 Example: the minimal configuration below would make HAProxy look for the
10030 state server file '/etc/haproxy/states/bk':
10031
10032 global
10033 server-state-file-base /etc/haproxy/states
10034
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010010035 backend bk
Baptiste Assmann01c6cc32015-08-23 11:45:29 +020010036 load-server-state-from-file
10037
Christopher Faulet583b6de2021-02-12 09:27:10 +010010038 See also: "server-state-base", "load-server-state-from-file", and
Baptiste Assmann01c6cc32015-08-23 11:45:29 +020010039 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010040
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +020010041server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
10042 Set a template to initialize servers with shared parameters.
10043 The names of these servers are built from <prefix> and <num | range> parameters.
10044 May be used in sections : defaults | frontend | listen | backend
10045 no | no | yes | yes
10046
10047 Arguments:
10048 <prefix> A prefix for the server names to be built.
10049
10050 <num | range>
10051 If <num> is provided, this template initializes <num> servers
10052 with 1 up to <num> as server name suffixes. A range of numbers
10053 <num_low>-<num_high> may also be used to use <num_low> up to
10054 <num_high> as server name suffixes.
10055
10056 <fqdn> A FQDN for all the servers this template initializes.
10057
10058 <port> Same meaning as "server" <port> argument (see "server" keyword).
10059
10060 <params*>
10061 Remaining server parameters among all those supported by "server"
10062 keyword.
10063
10064 Examples:
10065 # Initializes 3 servers with srv1, srv2 and srv3 as names,
10066 # google.com as FQDN, and health-check enabled.
10067 server-template srv 1-3 google.com:80 check
10068
10069 # or
10070 server-template srv 3 google.com:80 check
10071
10072 # would be equivalent to:
10073 server srv1 google.com:80 check
10074 server srv2 google.com:80 check
10075 server srv3 google.com:80 check
10076
10077
10078
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010079source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020010080source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +010010081source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010082 Set the source address for outgoing connections
10083 May be used in sections : defaults | frontend | listen | backend
10084 yes | no | yes | yes
10085 Arguments :
10086 <addr> is the IPv4 address HAProxy will bind to before connecting to a
10087 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +010010088
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010089 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +010010090 the most appropriate address to reach its destination. Optionally
10091 an address family prefix may be used before the address to force
10092 the family regardless of the address format, which can be useful
10093 to specify a path to a unix socket with no slash ('/'). Currently
10094 supported prefixes are :
10095 - 'ipv4@' -> address is always IPv4
10096 - 'ipv6@' -> address is always IPv6
10097 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +020010098 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +020010099 You may want to reference some environment variables in the
10100 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010101
10102 <port> is an optional port. It is normally not needed but may be useful
10103 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020010104 the system will select a free port. Note that port ranges are not
10105 supported in the backend. If you want to force port ranges, you
10106 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010107
10108 <addr2> is the IP address to present to the server when connections are
10109 forwarded in full transparent proxy mode. This is currently only
10110 supported on some patched Linux kernels. When this address is
10111 specified, clients connecting to the server will be presented
10112 with this address, while health checks will still use the address
10113 <addr>.
10114
10115 <port2> is the optional port to present to the server when connections
10116 are forwarded in full transparent proxy mode (see <addr2> above).
10117 The default value of zero means the system will select a free
10118 port.
10119
Willy Tarreaubce70882009-09-07 11:51:47 +020010120 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
10121 This is the name of a comma-separated header list which can
10122 contain multiple IP addresses. By default, the last occurrence is
10123 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +010010124 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +020010125 by previous proxy, typically Stunnel. In order to use another
10126 occurrence from the last one, please see the <occ> parameter
10127 below. When the header (or occurrence) is not found, no binding
10128 is performed so that the proxy's default IP address is used. Also
10129 keep in mind that the header name is case insensitive, as for any
10130 HTTP header.
10131
10132 <occ> is the occurrence number of a value to be used in a multi-value
10133 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010134 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +020010135 address. Positive values indicate a position from the first
10136 occurrence, 1 being the first one. Negative values indicate
10137 positions relative to the last one, -1 being the last one. This
10138 is helpful for situations where an X-Forwarded-For header is set
10139 at the entry point of an infrastructure and must be used several
10140 proxy layers away. When this value is not specified, -1 is
10141 assumed. Passing a zero here disables the feature.
10142
Willy Tarreaud53f96b2009-02-04 18:46:54 +010010143 <name> is an optional interface name to which to bind to for outgoing
10144 traffic. On systems supporting this features (currently, only
10145 Linux), this allows one to bind all traffic to the server to
10146 this interface even if it is not the one the system would select
10147 based on routing tables. This should be used with extreme care.
10148 Note that using this option requires root privileges.
10149
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010150 The "source" keyword is useful in complex environments where a specific
10151 address only is allowed to connect to the servers. It may be needed when a
10152 private address must be used through a public gateway for instance, and it is
10153 known that the system cannot determine the adequate source address by itself.
10154
10155 An extension which is available on certain patched Linux kernels may be used
10156 through the "usesrc" optional keyword. It makes it possible to connect to the
10157 servers with an IP address which does not belong to the system itself. This
10158 is called "full transparent proxy mode". For this to work, the destination
10159 servers have to route their traffic back to this address through the machine
10160 running HAProxy, and IP forwarding must generally be enabled on this machine.
10161
10162 In this "full transparent proxy" mode, it is possible to force a specific IP
10163 address to be presented to the servers. This is not much used in fact. A more
10164 common use is to tell HAProxy to present the client's IP address. For this,
10165 there are two methods :
10166
10167 - present the client's IP and port addresses. This is the most transparent
10168 mode, but it can cause problems when IP connection tracking is enabled on
10169 the machine, because a same connection may be seen twice with different
10170 states. However, this solution presents the huge advantage of not
10171 limiting the system to the 64k outgoing address+port couples, because all
10172 of the client ranges may be used.
10173
10174 - present only the client's IP address and select a spare port. This
10175 solution is still quite elegant but slightly less transparent (downstream
10176 firewalls logs will not match upstream's). It also presents the downside
10177 of limiting the number of concurrent connections to the usual 64k ports.
10178 However, since the upstream and downstream ports are different, local IP
10179 connection tracking on the machine will not be upset by the reuse of the
10180 same session.
10181
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010182 This option sets the default source for all servers in the backend. It may
10183 also be specified in a "defaults" section. Finer source address specification
10184 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010185 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010186
Baptiste Assmann91bd3372015-07-17 21:59:42 +020010187 In order to work, "usesrc" requires root privileges.
10188
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010189 Examples :
10190 backend private
10191 # Connect to the servers using our 192.168.1.200 source address
10192 source 192.168.1.200
10193
10194 backend transparent_ssl1
10195 # Connect to the SSL farm from the client's source address
10196 source 192.168.1.200 usesrc clientip
10197
10198 backend transparent_ssl2
10199 # Connect to the SSL farm from the client's source address and port
10200 # not recommended if IP conntrack is present on the local machine.
10201 source 192.168.1.200 usesrc client
10202
10203 backend transparent_ssl3
10204 # Connect to the SSL farm from the client's source address. It
10205 # is more conntrack-friendly.
10206 source 192.168.1.200 usesrc clientip
10207
10208 backend transparent_smtp
10209 # Connect to the SMTP farm from the client's source address/port
10210 # with Tproxy version 4.
10211 source 0.0.0.0 usesrc clientip
10212
Willy Tarreaubce70882009-09-07 11:51:47 +020010213 backend transparent_http
10214 # Connect to the servers using the client's IP as seen by previous
10215 # proxy.
10216 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
10217
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010218 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010219 the Linux kernel on www.balabit.com, the "bind" keyword.
10220
Willy Tarreau844e3c52008-01-11 16:28:18 +010010221
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010222srvtcpka-cnt <count>
10223 Sets the maximum number of keepalive probes TCP should send before dropping
10224 the connection on the server side.
10225 May be used in sections : defaults | frontend | listen | backend
10226 yes | no | yes | yes
10227 Arguments :
10228 <count> is the maximum number of keepalive probes.
10229
10230 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
10231 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +020010232 The availability of this setting depends on the operating system. It is
10233 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010234
10235 See also : "option srvtcpka", "srvtcpka-idle", "srvtcpka-intvl".
10236
10237
10238srvtcpka-idle <timeout>
10239 Sets the time the connection needs to remain idle before TCP starts sending
10240 keepalive probes, if enabled the sending of TCP keepalive packets on the
10241 server side.
10242 May be used in sections : defaults | frontend | listen | backend
10243 yes | no | yes | yes
10244 Arguments :
10245 <timeout> is the time the connection needs to remain idle before TCP starts
10246 sending keepalive probes. It is specified in seconds by default,
10247 but can be in any other unit if the number is suffixed by the
10248 unit, as explained at the top of this document.
10249
10250 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
10251 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +020010252 The availability of this setting depends on the operating system. It is
10253 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010254
10255 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-intvl".
10256
10257
10258srvtcpka-intvl <timeout>
10259 Sets the time between individual keepalive probes on the server side.
10260 May be used in sections : defaults | frontend | listen | backend
10261 yes | no | yes | yes
10262 Arguments :
10263 <timeout> is the time between individual keepalive probes. It is specified
10264 in seconds by default, but can be in any other unit if the number
10265 is suffixed by the unit, as explained at the top of this
10266 document.
10267
10268 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
10269 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +020010270 The availability of this setting depends on the operating system. It is
10271 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010272
10273 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-idle".
10274
10275
Cyril Bonté66c327d2010-10-12 00:14:37 +020010276stats admin { if | unless } <cond>
10277 Enable statistics admin level if/unless a condition is matched
10278 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010279 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +020010280
10281 This statement enables the statistics admin level if/unless a condition is
10282 matched.
10283
10284 The admin level allows to enable/disable servers from the web interface. By
10285 default, statistics page is read-only for security reasons.
10286
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010287 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10288 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010289 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010290
Cyril Bonté23b39d92011-02-10 22:54:44 +010010291 Currently, the POST request is limited to the buffer size minus the reserved
10292 buffer space, which means that if the list of servers is too long, the
10293 request won't be processed. It is recommended to alter few servers at a
10294 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +020010295
10296 Example :
10297 # statistics admin level only for localhost
10298 backend stats_localhost
10299 stats enable
10300 stats admin if LOCALHOST
10301
10302 Example :
10303 # statistics admin level always enabled because of the authentication
10304 backend stats_auth
10305 stats enable
10306 stats auth admin:AdMiN123
10307 stats admin if TRUE
10308
10309 Example :
10310 # statistics admin level depends on the authenticated user
10311 userlist stats-auth
10312 group admin users admin
10313 user admin insecure-password AdMiN123
10314 group readonly users haproxy
10315 user haproxy insecure-password haproxy
10316
10317 backend stats_auth
10318 stats enable
10319 acl AUTH http_auth(stats-auth)
10320 acl AUTH_ADMIN http_auth_group(stats-auth) admin
10321 stats http-request auth unless AUTH
10322 stats admin if AUTH_ADMIN
10323
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010324 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
10325 "bind-process", section 3.4 about userlists and section 7 about
10326 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +020010327
10328
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010329stats auth <user>:<passwd>
10330 Enable statistics with authentication and grant access to an account
10331 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010332 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010333 Arguments :
10334 <user> is a user name to grant access to
10335
10336 <passwd> is the cleartext password associated to this user
10337
10338 This statement enables statistics with default settings, and restricts access
10339 to declared users only. It may be repeated as many times as necessary to
10340 allow as many users as desired. When a user tries to access the statistics
10341 without a valid account, a "401 Forbidden" response will be returned so that
10342 the browser asks the user to provide a valid user and password. The real
10343 which will be returned to the browser is configurable using "stats realm".
10344
10345 Since the authentication method is HTTP Basic Authentication, the passwords
10346 circulate in cleartext on the network. Thus, it was decided that the
10347 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020010348 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010349
10350 It is also possible to reduce the scope of the proxies which appear in the
10351 report using "stats scope".
10352
10353 Though this statement alone is enough to enable statistics reporting, it is
10354 recommended to set all other settings in order to avoid relying on default
10355 unobvious parameters.
10356
10357 Example :
10358 # public access (limited to this backend only)
10359 backend public_www
10360 server srv1 192.168.0.1:80
10361 stats enable
10362 stats hide-version
10363 stats scope .
10364 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010365 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010366 stats auth admin1:AdMiN123
10367 stats auth admin2:AdMiN321
10368
10369 # internal monitoring access (unlimited)
10370 backend private_monitoring
10371 stats enable
10372 stats uri /admin?stats
10373 stats refresh 5s
10374
10375 See also : "stats enable", "stats realm", "stats scope", "stats uri"
10376
10377
10378stats enable
10379 Enable statistics reporting with default settings
10380 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010381 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010382 Arguments : none
10383
10384 This statement enables statistics reporting with default settings defined
10385 at build time. Unless stated otherwise, these settings are used :
10386 - stats uri : /haproxy?stats
10387 - stats realm : "HAProxy Statistics"
10388 - stats auth : no authentication
10389 - stats scope : no restriction
10390
10391 Though this statement alone is enough to enable statistics reporting, it is
10392 recommended to set all other settings in order to avoid relying on default
10393 unobvious parameters.
10394
10395 Example :
10396 # public access (limited to this backend only)
10397 backend public_www
10398 server srv1 192.168.0.1:80
10399 stats enable
10400 stats hide-version
10401 stats scope .
10402 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010403 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010404 stats auth admin1:AdMiN123
10405 stats auth admin2:AdMiN321
10406
10407 # internal monitoring access (unlimited)
10408 backend private_monitoring
10409 stats enable
10410 stats uri /admin?stats
10411 stats refresh 5s
10412
10413 See also : "stats auth", "stats realm", "stats uri"
10414
10415
Willy Tarreaud63335a2010-02-26 12:56:52 +010010416stats hide-version
10417 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010418 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010419 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010420 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010421
Willy Tarreaud63335a2010-02-26 12:56:52 +010010422 By default, the stats page reports some useful status information along with
10423 the statistics. Among them is HAProxy's version. However, it is generally
10424 considered dangerous to report precise version to anyone, as it can help them
10425 target known weaknesses with specific attacks. The "stats hide-version"
10426 statement removes the version from the statistics report. This is recommended
10427 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010428
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +020010429 Though this statement alone is enough to enable statistics reporting, it is
10430 recommended to set all other settings in order to avoid relying on default
10431 unobvious parameters.
10432
Willy Tarreaud63335a2010-02-26 12:56:52 +010010433 Example :
10434 # public access (limited to this backend only)
10435 backend public_www
10436 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +020010437 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +010010438 stats hide-version
10439 stats scope .
10440 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010441 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +010010442 stats auth admin1:AdMiN123
10443 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010444
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010445 # internal monitoring access (unlimited)
10446 backend private_monitoring
10447 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +010010448 stats uri /admin?stats
10449 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +010010450
Willy Tarreaud63335a2010-02-26 12:56:52 +010010451 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010452
Willy Tarreau983e01e2010-01-11 18:42:06 +010010453
Cyril Bonté2be1b3f2010-09-30 23:46:30 +020010454stats http-request { allow | deny | auth [realm <realm>] }
10455 [ { if | unless } <condition> ]
10456 Access control for statistics
10457
10458 May be used in sections: defaults | frontend | listen | backend
10459 no | no | yes | yes
10460
10461 As "http-request", these set of options allow to fine control access to
10462 statistics. Each option may be followed by if/unless and acl.
10463 First option with matched condition (or option without condition) is final.
10464 For "deny" a 403 error will be returned, for "allow" normal processing is
10465 performed, for "auth" a 401/407 error code is returned so the client
10466 should be asked to enter a username and password.
10467
10468 There is no fixed limit to the number of http-request statements per
10469 instance.
10470
10471 See also : "http-request", section 3.4 about userlists and section 7
10472 about ACL usage.
10473
10474
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010475stats realm <realm>
10476 Enable statistics and set authentication realm
10477 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010478 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010479 Arguments :
10480 <realm> is the name of the HTTP Basic Authentication realm reported to
10481 the browser. The browser uses it to display it in the pop-up
10482 inviting the user to enter a valid username and password.
10483
10484 The realm is read as a single word, so any spaces in it should be escaped
10485 using a backslash ('\').
10486
10487 This statement is useful only in conjunction with "stats auth" since it is
10488 only related to authentication.
10489
10490 Though this statement alone is enough to enable statistics reporting, it is
10491 recommended to set all other settings in order to avoid relying on default
10492 unobvious parameters.
10493
10494 Example :
10495 # public access (limited to this backend only)
10496 backend public_www
10497 server srv1 192.168.0.1:80
10498 stats enable
10499 stats hide-version
10500 stats scope .
10501 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010502 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010503 stats auth admin1:AdMiN123
10504 stats auth admin2:AdMiN321
10505
10506 # internal monitoring access (unlimited)
10507 backend private_monitoring
10508 stats enable
10509 stats uri /admin?stats
10510 stats refresh 5s
10511
10512 See also : "stats auth", "stats enable", "stats uri"
10513
10514
10515stats refresh <delay>
10516 Enable statistics with automatic refresh
10517 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010518 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010519 Arguments :
10520 <delay> is the suggested refresh delay, specified in seconds, which will
10521 be returned to the browser consulting the report page. While the
10522 browser is free to apply any delay, it will generally respect it
10523 and refresh the page this every seconds. The refresh interval may
10524 be specified in any other non-default time unit, by suffixing the
10525 unit after the value, as explained at the top of this document.
10526
10527 This statement is useful on monitoring displays with a permanent page
10528 reporting the load balancer's activity. When set, the HTML report page will
10529 include a link "refresh"/"stop refresh" so that the user can select whether
Jackie Tapia749f74c2020-07-22 18:59:40 -050010530 they want automatic refresh of the page or not.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010531
10532 Though this statement alone is enough to enable statistics reporting, it is
10533 recommended to set all other settings in order to avoid relying on default
10534 unobvious parameters.
10535
10536 Example :
10537 # public access (limited to this backend only)
10538 backend public_www
10539 server srv1 192.168.0.1:80
10540 stats enable
10541 stats hide-version
10542 stats scope .
10543 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010544 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010545 stats auth admin1:AdMiN123
10546 stats auth admin2:AdMiN321
10547
10548 # internal monitoring access (unlimited)
10549 backend private_monitoring
10550 stats enable
10551 stats uri /admin?stats
10552 stats refresh 5s
10553
10554 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10555
10556
10557stats scope { <name> | "." }
10558 Enable statistics and limit access scope
10559 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010560 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010561 Arguments :
10562 <name> is the name of a listen, frontend or backend section to be
10563 reported. The special name "." (a single dot) designates the
10564 section in which the statement appears.
10565
10566 When this statement is specified, only the sections enumerated with this
10567 statement will appear in the report. All other ones will be hidden. This
10568 statement may appear as many times as needed if multiple sections need to be
10569 reported. Please note that the name checking is performed as simple string
10570 comparisons, and that it is never checked that a give section name really
10571 exists.
10572
10573 Though this statement alone is enough to enable statistics reporting, it is
10574 recommended to set all other settings in order to avoid relying on default
10575 unobvious parameters.
10576
10577 Example :
10578 # public access (limited to this backend only)
10579 backend public_www
10580 server srv1 192.168.0.1:80
10581 stats enable
10582 stats hide-version
10583 stats scope .
10584 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010585 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010586 stats auth admin1:AdMiN123
10587 stats auth admin2:AdMiN321
10588
10589 # internal monitoring access (unlimited)
10590 backend private_monitoring
10591 stats enable
10592 stats uri /admin?stats
10593 stats refresh 5s
10594
10595 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10596
Willy Tarreaud63335a2010-02-26 12:56:52 +010010597
Willy Tarreauc9705a12010-07-27 20:05:50 +020010598stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +010010599 Enable reporting of a description on the statistics page.
10600 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010601 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010602
Willy Tarreauc9705a12010-07-27 20:05:50 +020010603 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +010010604 description from global section is automatically used instead.
10605
10606 This statement is useful for users that offer shared services to their
10607 customers, where node or description should be different for each customer.
10608
10609 Though this statement alone is enough to enable statistics reporting, it is
10610 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010611 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010612
10613 Example :
10614 # internal monitoring access (unlimited)
10615 backend private_monitoring
10616 stats enable
10617 stats show-desc Master node for Europe, Asia, Africa
10618 stats uri /admin?stats
10619 stats refresh 5s
10620
10621 See also: "show-node", "stats enable", "stats uri" and "description" in
10622 global section.
10623
10624
10625stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +020010626 Enable reporting additional information on the statistics page
10627 May be used in sections : defaults | frontend | listen | backend
10628 yes | yes | yes | yes
10629 Arguments : none
10630
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010631 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +010010632 - cap: capabilities (proxy)
10633 - mode: one of tcp, http or health (proxy)
10634 - id: SNMP ID (proxy, socket, server)
10635 - IP (socket, server)
10636 - cookie (backend, server)
10637
10638 Though this statement alone is enough to enable statistics reporting, it is
10639 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010640 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010641
10642 See also: "stats enable", "stats uri".
10643
10644
Amaury Denoyelle0b70a8a2020-10-05 11:49:45 +020010645stats show-modules
10646 Enable display of extra statistics module on the statistics page
10647 May be used in sections : defaults | frontend | listen | backend
10648 yes | yes | yes | yes
10649 Arguments : none
10650
10651 New columns are added at the end of the line containing the extra statistics
10652 values as a tooltip.
10653
10654 Though this statement alone is enough to enable statistics reporting, it is
10655 recommended to set all other settings in order to avoid relying on default
10656 unobvious parameters. Default behavior is not to show this information.
10657
10658 See also: "stats enable", "stats uri".
10659
10660
Willy Tarreaud63335a2010-02-26 12:56:52 +010010661stats show-node [ <name> ]
10662 Enable reporting of a host name on the statistics page.
10663 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010664 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010665 Arguments:
10666 <name> is an optional name to be reported. If unspecified, the
10667 node name from global section is automatically used instead.
10668
10669 This statement is useful for users that offer shared services to their
10670 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +010010671 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010672
10673 Though this statement alone is enough to enable statistics reporting, it is
10674 recommended to set all other settings in order to avoid relying on default
10675 unobvious parameters.
10676
10677 Example:
10678 # internal monitoring access (unlimited)
10679 backend private_monitoring
10680 stats enable
10681 stats show-node Europe-1
10682 stats uri /admin?stats
10683 stats refresh 5s
10684
10685 See also: "show-desc", "stats enable", "stats uri", and "node" in global
10686 section.
10687
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010688
10689stats uri <prefix>
10690 Enable statistics and define the URI prefix to access them
10691 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010692 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010693 Arguments :
10694 <prefix> is the prefix of any URI which will be redirected to stats. This
10695 prefix may contain a question mark ('?') to indicate part of a
10696 query string.
10697
10698 The statistics URI is intercepted on the relayed traffic, so it appears as a
10699 page within the normal application. It is strongly advised to ensure that the
10700 selected URI will never appear in the application, otherwise it will never be
10701 possible to reach it in the application.
10702
10703 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010704 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010705 It is generally a good idea to include a question mark in the URI so that
10706 intermediate proxies refrain from caching the results. Also, since any string
10707 beginning with the prefix will be accepted as a stats request, the question
10708 mark helps ensuring that no valid URI will begin with the same words.
10709
10710 It is sometimes very convenient to use "/" as the URI prefix, and put that
10711 statement in a "listen" instance of its own. That makes it easy to dedicate
10712 an address or a port to statistics only.
10713
10714 Though this statement alone is enough to enable statistics reporting, it is
10715 recommended to set all other settings in order to avoid relying on default
10716 unobvious parameters.
10717
10718 Example :
10719 # public access (limited to this backend only)
10720 backend public_www
10721 server srv1 192.168.0.1:80
10722 stats enable
10723 stats hide-version
10724 stats scope .
10725 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010726 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010727 stats auth admin1:AdMiN123
10728 stats auth admin2:AdMiN321
10729
10730 # internal monitoring access (unlimited)
10731 backend private_monitoring
10732 stats enable
10733 stats uri /admin?stats
10734 stats refresh 5s
10735
10736 See also : "stats auth", "stats enable", "stats realm"
10737
10738
Willy Tarreaud63335a2010-02-26 12:56:52 +010010739stick match <pattern> [table <table>] [{if | unless} <cond>]
10740 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010741 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +010010742 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010743
10744 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010745 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010746 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010747 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010748 stickiness table. This rule is mandatory.
10749
10750 <table> is an optional stickiness table name. If unspecified, the same
10751 backend's table is used. A stickiness table is declared using
10752 the "stick-table" statement.
10753
10754 <cond> is an optional matching condition. It makes it possible to match
10755 on a certain criterion only when other conditions are met (or
10756 not met). For instance, it could be used to match on a source IP
10757 address except when a request passes through a known proxy, in
10758 which case we'd match on a header containing that IP address.
10759
10760 Some protocols or applications require complex stickiness rules and cannot
10761 always simply rely on cookies nor hashing. The "stick match" statement
10762 describes a rule to extract the stickiness criterion from an incoming request
10763 or connection. See section 7 for a complete list of possible patterns and
10764 transformation rules.
10765
10766 The table has to be declared using the "stick-table" statement. It must be of
10767 a type compatible with the pattern. By default it is the one which is present
10768 in the same backend. It is possible to share a table with other backends by
10769 referencing it using the "table" keyword. If another table is referenced,
10770 the server's ID inside the backends are used. By default, all server IDs
10771 start at 1 in each backend, so the server ordering is enough. But in case of
10772 doubt, it is highly recommended to force server IDs using their "id" setting.
10773
10774 It is possible to restrict the conditions where a "stick match" statement
10775 will apply, using "if" or "unless" followed by a condition. See section 7 for
10776 ACL based conditions.
10777
10778 There is no limit on the number of "stick match" statements. The first that
10779 applies and matches will cause the request to be directed to the same server
10780 as was used for the request which created the entry. That way, multiple
10781 matches can be used as fallbacks.
10782
10783 The stick rules are checked after the persistence cookies, so they will not
10784 affect stickiness if a cookie has already been used to select a server. That
10785 way, it becomes very easy to insert cookies and match on IP addresses in
10786 order to maintain stickiness between HTTP and HTTPS.
10787
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010788 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10789 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010790 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010791
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010792 Example :
10793 # forward SMTP users to the same server they just used for POP in the
10794 # last 30 minutes
10795 backend pop
10796 mode tcp
10797 balance roundrobin
10798 stick store-request src
10799 stick-table type ip size 200k expire 30m
10800 server s1 192.168.1.1:110
10801 server s2 192.168.1.1:110
10802
10803 backend smtp
10804 mode tcp
10805 balance roundrobin
10806 stick match src table pop
10807 server s1 192.168.1.1:25
10808 server s2 192.168.1.1:25
10809
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010810 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010811 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010812
10813
10814stick on <pattern> [table <table>] [{if | unless} <condition>]
10815 Define a request pattern to associate a user to a server
10816 May be used in sections : defaults | frontend | listen | backend
10817 no | no | yes | yes
10818
10819 Note : This form is exactly equivalent to "stick match" followed by
10820 "stick store-request", all with the same arguments. Please refer
10821 to both keywords for details. It is only provided as a convenience
10822 for writing more maintainable configurations.
10823
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010824 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10825 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010826 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010827
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010828 Examples :
10829 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +010010830 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010831
10832 # ...is strictly equivalent to this one :
10833 stick match src table pop if !localhost
10834 stick store-request src table pop if !localhost
10835
10836
10837 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
10838 # well as HTTP without cookie. Share the same table between both accesses.
10839 backend http
10840 mode http
10841 balance roundrobin
10842 stick on src table https
10843 cookie SRV insert indirect nocache
10844 server s1 192.168.1.1:80 cookie s1
10845 server s2 192.168.1.1:80 cookie s2
10846
10847 backend https
10848 mode tcp
10849 balance roundrobin
10850 stick-table type ip size 200k expire 30m
10851 stick on src
10852 server s1 192.168.1.1:443
10853 server s2 192.168.1.1:443
10854
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010855 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010856
10857
10858stick store-request <pattern> [table <table>] [{if | unless} <condition>]
10859 Define a request pattern used to create an entry in a stickiness table
10860 May be used in sections : defaults | frontend | listen | backend
10861 no | no | yes | yes
10862
10863 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010864 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010865 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010866 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010867 server is selected.
10868
10869 <table> is an optional stickiness table name. If unspecified, the same
10870 backend's table is used. A stickiness table is declared using
10871 the "stick-table" statement.
10872
10873 <cond> is an optional storage condition. It makes it possible to store
10874 certain criteria only when some conditions are met (or not met).
10875 For instance, it could be used to store the source IP address
10876 except when the request passes through a known proxy, in which
10877 case we'd store a converted form of a header containing that IP
10878 address.
10879
10880 Some protocols or applications require complex stickiness rules and cannot
10881 always simply rely on cookies nor hashing. The "stick store-request" statement
10882 describes a rule to decide what to extract from the request and when to do
10883 it, in order to store it into a stickiness table for further requests to
10884 match it using the "stick match" statement. Obviously the extracted part must
10885 make sense and have a chance to be matched in a further request. Storing a
10886 client's IP address for instance often makes sense. Storing an ID found in a
10887 URL parameter also makes sense. Storing a source port will almost never make
10888 any sense because it will be randomly matched. See section 7 for a complete
10889 list of possible patterns and transformation rules.
10890
10891 The table has to be declared using the "stick-table" statement. It must be of
10892 a type compatible with the pattern. By default it is the one which is present
10893 in the same backend. It is possible to share a table with other backends by
10894 referencing it using the "table" keyword. If another table is referenced,
10895 the server's ID inside the backends are used. By default, all server IDs
10896 start at 1 in each backend, so the server ordering is enough. But in case of
10897 doubt, it is highly recommended to force server IDs using their "id" setting.
10898
10899 It is possible to restrict the conditions where a "stick store-request"
10900 statement will apply, using "if" or "unless" followed by a condition. This
10901 condition will be evaluated while parsing the request, so any criteria can be
10902 used. See section 7 for ACL based conditions.
10903
10904 There is no limit on the number of "stick store-request" statements, but
10905 there is a limit of 8 simultaneous stores per request or response. This
10906 makes it possible to store up to 8 criteria, all extracted from either the
10907 request or the response, regardless of the number of rules. Only the 8 first
10908 ones which match will be kept. Using this, it is possible to feed multiple
10909 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010010910 another protocol or access method. Using multiple store-request rules with
10911 the same table is possible and may be used to find the best criterion to rely
10912 on, by arranging the rules by decreasing preference order. Only the first
10913 extracted criterion for a given table will be stored. All subsequent store-
10914 request rules referencing the same table will be skipped and their ACLs will
10915 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010916
10917 The "store-request" rules are evaluated once the server connection has been
10918 established, so that the table will contain the real server that processed
10919 the request.
10920
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010921 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10922 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010923 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010924
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010925 Example :
10926 # forward SMTP users to the same server they just used for POP in the
10927 # last 30 minutes
10928 backend pop
10929 mode tcp
10930 balance roundrobin
10931 stick store-request src
10932 stick-table type ip size 200k expire 30m
10933 server s1 192.168.1.1:110
10934 server s2 192.168.1.1:110
10935
10936 backend smtp
10937 mode tcp
10938 balance roundrobin
10939 stick match src table pop
10940 server s1 192.168.1.1:25
10941 server s2 192.168.1.1:25
10942
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010943 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010944 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010945
10946
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010947stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Thayne McCombs92149f92020-11-20 01:28:26 -070010948 size <size> [expire <expire>] [nopurge] [peers <peersect>] [srvkey <srvkey>]
Emeric Brunf099e792010-09-27 12:05:28 +020010949 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +080010950 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010951 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +020010952 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010953
10954 Arguments :
10955 ip a table declared with "type ip" will only store IPv4 addresses.
10956 This form is very compact (about 50 bytes per entry) and allows
10957 very fast entry lookup and stores with almost no overhead. This
10958 is mainly used to store client source IP addresses.
10959
David du Colombier9a6d3c92011-03-17 10:40:24 +010010960 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
10961 This form is very compact (about 60 bytes per entry) and allows
10962 very fast entry lookup and stores with almost no overhead. This
10963 is mainly used to store client source IP addresses.
10964
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010965 integer a table declared with "type integer" will store 32bit integers
10966 which can represent a client identifier found in a request for
10967 instance.
10968
10969 string a table declared with "type string" will store substrings of up
10970 to <len> characters. If the string provided by the pattern
10971 extractor is larger than <len>, it will be truncated before
10972 being stored. During matching, at most <len> characters will be
10973 compared between the string in the table and the extracted
10974 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010975 to 32 characters.
10976
10977 binary a table declared with "type binary" will store binary blocks
10978 of <len> bytes. If the block provided by the pattern
10979 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +020010980 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010981 is shorter than <len>, it will be padded by 0. When not
10982 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010983
10984 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010985 "string" type table (See type "string" above). Or the number
10986 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010987 changing this parameter as memory usage will proportionally
10988 increase.
10989
10990 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +010010991 value directly impacts memory usage. Count approximately
10992 50 bytes per entry, plus the size of a string if any. The size
10993 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010994
10995 [nopurge] indicates that we refuse to purge older entries when the table
10996 is full. When not specified and the table is full when haproxy
10997 wants to store an entry in it, it will flush a few of the oldest
10998 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010999 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011000 be desirable to refuse new entries instead of purging the older
11001 ones. That may be the case when the amount of data to store is
11002 far above the hardware limits and we prefer not to offer access
11003 to new clients than to reject the ones already connected. When
11004 using this parameter, be sure to properly set the "expire"
11005 parameter (see below).
11006
Emeric Brunf099e792010-09-27 12:05:28 +020011007 <peersect> is the name of the peers section to use for replication. Entries
11008 which associate keys to server IDs are kept synchronized with
11009 the remote peers declared in this section. All entries are also
11010 automatically learned from the local peer (old process) during a
11011 soft restart.
11012
Willy Tarreau1abc6732015-05-01 19:21:02 +020011013 NOTE : each peers section may be referenced only by tables
11014 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010011015
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011016 <expire> defines the maximum duration of an entry in the table since it
11017 was last created, refreshed or matched. The expiration delay is
11018 defined using the standard time format, similarly as the various
11019 timeouts. The maximum duration is slightly above 24 days. See
Willy Tarreau4b103022021-02-12 17:59:10 +010011020 section 2.5 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +020011021 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011022 be removed once full. Be sure not to use the "nopurge" parameter
11023 if not expiration delay is specified.
11024
Thayne McCombs92149f92020-11-20 01:28:26 -070011025 <srvkey> specifies how each server is identified for the purposes of the
11026 stick table. The valid values are "name" and "addr". If "name" is
11027 given, then <name> argument for the server (may be generated by
11028 a template). If "addr" is given, then the server is identified
11029 by its current network address, including the port. "addr" is
11030 especially useful if you are using service discovery to generate
11031 the addresses for servers with peered stick-tables and want
11032 to consistently use the same host across peers for a stickiness
11033 token.
11034
Willy Tarreau08d5f982010-06-06 13:34:54 +020011035 <data_type> is used to store additional information in the stick-table. This
11036 may be used by ACLs in order to control various criteria related
11037 to the activity of the client matching the stick-table. For each
11038 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +020011039 that the additional data can fit. Several data types may be
11040 stored with an entry. Multiple data types may be specified after
11041 the "store" keyword, as a comma-separated list. Alternatively,
11042 it is possible to repeat the "store" keyword followed by one or
11043 several data types. Except for the "server_id" type which is
11044 automatically detected and enabled, all data types must be
11045 explicitly declared to be stored. If an ACL references a data
11046 type which is not stored, the ACL will simply not match. Some
11047 data types require an argument which must be passed just after
11048 the type between parenthesis. See below for the supported data
11049 types and their arguments.
11050
11051 The data types that can be stored with an entry are the following :
11052 - server_id : this is an integer which holds the numeric ID of the server a
11053 request was assigned to. It is used by the "stick match", "stick store",
11054 and "stick on" rules. It is automatically enabled when referenced.
11055
11056 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
11057 integer which may be used for anything. Most of the time it will be used
11058 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +010011059 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +020011060
Willy Tarreauba2ffd12013-05-29 15:54:14 +020011061 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
11062 over a period. It is a positive 32-bit integer integer which may be used
11063 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +010011064 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +020011065 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +010011066 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +020011067
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011068 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
11069 integer which may be used for anything. Most of the time it will be used
11070 to put a special tag on some entries, for instance to note that a
11071 specific behavior was detected and must be known for future matches.
11072
11073 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
11074 over a period. It is a positive 32-bit integer integer which may be used
11075 for anything. Just like <gpc1>, it counts events, but instead of keeping
11076 a cumulative number, it maintains the rate at which the counter is
11077 incremented. Most of the time it will be used to measure the frequency of
11078 occurrence of certain events (e.g. requests to a specific URL).
11079
Willy Tarreauc9705a12010-07-27 20:05:50 +020011080 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
11081 the absolute number of connections received from clients which matched
11082 this entry. It does not mean the connections were accepted, just that
11083 they were received.
11084
11085 - conn_cur : Current Connections. It is a positive 32-bit integer which
11086 stores the concurrent connection counts for the entry. It is incremented
11087 once an incoming connection matches the entry, and decremented once the
11088 connection leaves. That way it is possible to know at any time the exact
11089 number of concurrent connections for an entry.
11090
11091 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11092 integer parameter <period> which indicates in milliseconds the length
11093 of the period over which the average is measured. It reports the average
11094 incoming connection rate over that period, in connections per period. The
11095 result is an integer which can be matched using ACLs.
11096
11097 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
11098 the absolute number of sessions received from clients which matched this
11099 entry. A session is a connection that was accepted by the layer 4 rules.
11100
11101 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11102 integer parameter <period> which indicates in milliseconds the length
11103 of the period over which the average is measured. It reports the average
11104 incoming session rate over that period, in sessions per period. The
11105 result is an integer which can be matched using ACLs.
11106
11107 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
11108 counts the absolute number of HTTP requests received from clients which
11109 matched this entry. It does not matter whether they are valid requests or
11110 not. Note that this is different from sessions when keep-alive is used on
11111 the client side.
11112
11113 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11114 integer parameter <period> which indicates in milliseconds the length
11115 of the period over which the average is measured. It reports the average
11116 HTTP request rate over that period, in requests per period. The result is
11117 an integer which can be matched using ACLs. It does not matter whether
11118 they are valid requests or not. Note that this is different from sessions
11119 when keep-alive is used on the client side.
11120
11121 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
11122 counts the absolute number of HTTP requests errors induced by clients
11123 which matched this entry. Errors are counted on invalid and truncated
11124 requests, as well as on denied or tarpitted requests, and on failed
11125 authentications. If the server responds with 4xx, then the request is
11126 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +010011127 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +020011128
11129 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11130 integer parameter <period> which indicates in milliseconds the length
11131 of the period over which the average is measured. It reports the average
11132 HTTP request error rate over that period, in requests per period (see
11133 http_err_cnt above for what is accounted as an error). The result is an
11134 integer which can be matched using ACLs.
11135
Willy Tarreau826f3ab2021-02-10 12:07:15 +010011136 - http_fail_cnt : HTTP Failure Count. It is a positive 32-bit integer which
11137 counts the absolute number of HTTP response failures induced by servers
11138 which matched this entry. Errors are counted on invalid and truncated
11139 responses, as well as any 5xx response other than 501 or 505. It aims at
11140 being used combined with path or URI to detect service failures.
11141
11142 - http_fail_rate(<period>) : frequency counter (takes 12 bytes). It takes
11143 an integer parameter <period> which indicates in milliseconds the length
11144 of the period over which the average is measured. It reports the average
11145 HTTP response failure rate over that period, in requests per period (see
11146 http_fail_cnt above for what is accounted as a failure). The result is an
11147 integer which can be matched using ACLs.
11148
Willy Tarreauc9705a12010-07-27 20:05:50 +020011149 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010011150 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +020011151 which matched this entry. Headers are included in the count. This may be
11152 used to limit abuse of upload features on photo or video servers.
11153
11154 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11155 integer parameter <period> which indicates in milliseconds the length
11156 of the period over which the average is measured. It reports the average
11157 incoming bytes rate over that period, in bytes per period. It may be used
11158 to detect users which upload too much and too fast. Warning: with large
11159 uploads, it is possible that the amount of uploaded data will be counted
11160 once upon termination, thus causing spikes in the average transfer speed
11161 instead of having a smooth one. This may partially be smoothed with
11162 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
11163 recommended for better fairness.
11164
11165 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010011166 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +020011167 matched this entry. Headers are included in the count. This may be used
11168 to limit abuse of bots sucking the whole site.
11169
11170 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
11171 an integer parameter <period> which indicates in milliseconds the length
11172 of the period over which the average is measured. It reports the average
11173 outgoing bytes rate over that period, in bytes per period. It may be used
11174 to detect users which download too much and too fast. Warning: with large
11175 transfers, it is possible that the amount of transferred data will be
11176 counted once upon termination, thus causing spikes in the average
11177 transfer speed instead of having a smooth one. This may partially be
11178 smoothed with "option contstats" though this is not perfect yet. Use of
11179 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +020011180
Willy Tarreauc00cdc22010-06-06 16:48:26 +020011181 There is only one stick-table per proxy. At the moment of writing this doc,
11182 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011183 to be required, simply create a dummy backend with a stick-table in it and
11184 reference it.
11185
11186 It is important to understand that stickiness based on learning information
11187 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +010011188 lost upon restart unless peers are properly configured to transfer such
11189 information upon restart (recommended). In general it can be good as a
11190 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011191
Willy Tarreauc9705a12010-07-27 20:05:50 +020011192 Last, memory requirements may be important when storing many data types.
11193 Indeed, storing all indicators above at once in each entry requires 116 bytes
11194 per entry, or 116 MB for a 1-million entries table. This is definitely not
11195 something that can be ignored.
11196
11197 Example:
11198 # Keep track of counters of up to 1 million IP addresses over 5 minutes
11199 # and store a general purpose counter and the average connection rate
11200 # computed over a sliding window of 30 seconds.
11201 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
11202
Willy Tarreau4b103022021-02-12 17:59:10 +010011203 See also : "stick match", "stick on", "stick store-request", section 2.5
David du Colombiera13d1b92011-03-17 10:40:22 +010011204 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011205
11206
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011207stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +010011208 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011209 May be used in sections : defaults | frontend | listen | backend
11210 no | no | yes | yes
11211
11212 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011213 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011214 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +010011215 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011216 server is selected.
11217
11218 <table> is an optional stickiness table name. If unspecified, the same
11219 backend's table is used. A stickiness table is declared using
11220 the "stick-table" statement.
11221
11222 <cond> is an optional storage condition. It makes it possible to store
11223 certain criteria only when some conditions are met (or not met).
11224 For instance, it could be used to store the SSL session ID only
11225 when the response is a SSL server hello.
11226
11227 Some protocols or applications require complex stickiness rules and cannot
11228 always simply rely on cookies nor hashing. The "stick store-response"
11229 statement describes a rule to decide what to extract from the response and
11230 when to do it, in order to store it into a stickiness table for further
11231 requests to match it using the "stick match" statement. Obviously the
11232 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011233 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011234 See section 7 for a complete list of possible patterns and transformation
11235 rules.
11236
11237 The table has to be declared using the "stick-table" statement. It must be of
11238 a type compatible with the pattern. By default it is the one which is present
11239 in the same backend. It is possible to share a table with other backends by
11240 referencing it using the "table" keyword. If another table is referenced,
11241 the server's ID inside the backends are used. By default, all server IDs
11242 start at 1 in each backend, so the server ordering is enough. But in case of
11243 doubt, it is highly recommended to force server IDs using their "id" setting.
11244
11245 It is possible to restrict the conditions where a "stick store-response"
11246 statement will apply, using "if" or "unless" followed by a condition. This
11247 condition will be evaluated while parsing the response, so any criteria can
11248 be used. See section 7 for ACL based conditions.
11249
11250 There is no limit on the number of "stick store-response" statements, but
11251 there is a limit of 8 simultaneous stores per request or response. This
11252 makes it possible to store up to 8 criteria, all extracted from either the
11253 request or the response, regardless of the number of rules. Only the 8 first
11254 ones which match will be kept. Using this, it is possible to feed multiple
11255 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010011256 another protocol or access method. Using multiple store-response rules with
11257 the same table is possible and may be used to find the best criterion to rely
11258 on, by arranging the rules by decreasing preference order. Only the first
11259 extracted criterion for a given table will be stored. All subsequent store-
11260 response rules referencing the same table will be skipped and their ACLs will
11261 not be evaluated. However, even if a store-request rule references a table, a
11262 store-response rule may also use the same table. This means that each table
11263 may learn exactly one element from the request and one element from the
11264 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011265
11266 The table will contain the real server that processed the request.
11267
11268 Example :
11269 # Learn SSL session ID from both request and response and create affinity.
11270 backend https
11271 mode tcp
11272 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +020011273 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011274 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011275
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011276 acl clienthello req_ssl_hello_type 1
11277 acl serverhello rep_ssl_hello_type 2
11278
11279 # use tcp content accepts to detects ssl client and server hello.
11280 tcp-request inspect-delay 5s
11281 tcp-request content accept if clienthello
11282
11283 # no timeout on response inspect delay by default.
11284 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011285
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011286 # SSL session ID (SSLID) may be present on a client or server hello.
11287 # Its length is coded on 1 byte at offset 43 and its value starts
11288 # at offset 44.
11289
11290 # Match and learn on request if client hello.
11291 stick on payload_lv(43,1) if clienthello
11292
11293 # Learn on response if server hello.
11294 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +020011295
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011296 server s1 192.168.1.1:443
11297 server s2 192.168.1.1:443
11298
11299 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
11300 extraction.
11301
11302
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011303tcp-check comment <string>
11304 Defines a comment for the following the tcp-check rule, reported in logs if
11305 it fails.
11306 May be used in sections : defaults | frontend | listen | backend
11307 yes | no | yes | yes
11308
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011309 Arguments :
11310 <string> is the comment message to add in logs if the following tcp-check
11311 rule fails.
11312
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011313 It only works for connect, send and expect rules. It is useful to make
11314 user-friendly error reporting.
11315
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011316 See also : "option tcp-check", "tcp-check connect", "tcp-check send" and
11317 "tcp-check expect".
11318
11319
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011320tcp-check connect [default] [port <expr>] [addr <ip>] [send-proxy] [via-socks4]
11321 [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +020011322 [proto <name>] [comment <msg>]
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011323 Opens a new connection
11324 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011325 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011326
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011327 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011328 comment <msg> defines a message to report if the rule evaluation fails.
11329
Christopher Faulet4dce5922020-03-30 13:54:42 +020011330 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -040011331 checks. The server options are used only if not redefined.
Christopher Faulet4dce5922020-03-30 13:54:42 +020011332
Christopher Fauletb7d30092020-03-30 15:19:03 +020011333 port <expr> if not set, check port or server port is used.
Christopher Faulet5c288742020-03-31 08:15:58 +020011334 It tells HAProxy where to open the connection to.
11335 <port> must be a valid TCP port source integer, from 1 to
Christopher Fauletb7d30092020-03-30 15:19:03 +020011336 65535 or an sample-fetch expression.
Christopher Faulet5c288742020-03-31 08:15:58 +020011337
11338 addr <ip> defines the IP address to do the health check.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011339
11340 send-proxy send a PROXY protocol string
11341
Christopher Faulet085426a2020-03-30 13:07:02 +020011342 via-socks4 enables outgoing health checks using upstream socks4 proxy.
11343
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011344 ssl opens a ciphered connection
11345
Christopher Faulet79b31d42020-03-30 13:00:05 +020011346 sni <sni> specifies the SNI to use to do health checks over SSL.
11347
Christopher Faulet98572322020-03-30 13:16:44 +020011348 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
11349 list consists in a comma-delimited list of protocol names,
11350 for instance: "http/1.1,http/1.0" (without quotes).
11351 If it is not set, the server ALPN is used.
11352
Christopher Fauletedc6ed92020-04-23 16:27:59 +020011353 proto <name> forces the multiplexer's protocol to use for this connection.
11354 It must be a TCP mux protocol and it must be usable on the
11355 backend side. The list of available protocols is reported in
11356 haproxy -vv.
11357
Christopher Faulet5c288742020-03-31 08:15:58 +020011358 linger cleanly close the connection instead of using a single RST.
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010011359
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011360 When an application lies on more than a single TCP port or when HAProxy
11361 load-balance many services in a single backend, it makes sense to probe all
11362 the services individually before considering a server as operational.
11363
11364 When there are no TCP port configured on the server line neither server port
11365 directive, then the 'tcp-check connect port <port>' must be the first step
11366 of the sequence.
11367
11368 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
11369 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
11370 do.
11371
11372 When a connect must start the ruleset, if may still be preceded by set-var,
11373 unset-var or comment rules.
11374
11375 Examples :
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011376 # check HTTP and HTTPs services on a server.
11377 # first open port 80 thanks to server line port directive, then
11378 # tcp-check opens port 443, ciphered and run a request on it:
11379 option tcp-check
11380 tcp-check connect
11381 tcp-check send GET\ /\ HTTP/1.0\r\n
11382 tcp-check send Host:\ haproxy.1wt.eu\r\n
11383 tcp-check send \r\n
11384 tcp-check expect rstring (2..|3..)
11385 tcp-check connect port 443 ssl
11386 tcp-check send GET\ /\ HTTP/1.0\r\n
11387 tcp-check send Host:\ haproxy.1wt.eu\r\n
11388 tcp-check send \r\n
11389 tcp-check expect rstring (2..|3..)
11390 server www 10.0.0.1 check port 80
11391
11392 # check both POP and IMAP from a single server:
11393 option tcp-check
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010011394 tcp-check connect port 110 linger
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011395 tcp-check expect string +OK\ POP3\ ready
11396 tcp-check connect port 143
11397 tcp-check expect string *\ OK\ IMAP4\ ready
11398 server mail 10.0.0.1 check
11399
11400 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
11401
11402
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011403tcp-check expect [min-recv <int>] [comment <msg>]
Christopher Fauletec07e382020-04-07 14:56:26 +020011404 [ok-status <st>] [error-status <st>] [tout-status <st>]
Christopher Faulet98cc57c2020-04-01 20:52:31 +020011405 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011406 [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +010011407 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011408 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011409 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011410
11411 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011412 comment <msg> defines a message to report if the rule evaluation fails.
11413
Gaetan Rivet1afd8262020-02-07 15:37:17 +010011414 min-recv is optional and can define the minimum amount of data required to
11415 evaluate the current expect rule. If the number of received bytes
11416 is under this limit, the check will wait for more data. This
11417 option can be used to resolve some ambiguous matching rules or to
11418 avoid executing costly regex matches on content known to be still
11419 incomplete. If an exact string (string or binary) is used, the
11420 minimum between the string length and this parameter is used.
11421 This parameter is ignored if it is set to -1. If the expect rule
11422 does not match, the check will wait for more data. If set to 0,
11423 the evaluation result is always conclusive.
11424
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011425 <match> is a keyword indicating how to look for a specific pattern in the
Gaetan Rivetefab6c62020-02-07 15:37:17 +010011426 response. The keyword may be one of "string", "rstring", "binary" or
11427 "rbinary".
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011428 The keyword may be preceded by an exclamation mark ("!") to negate
11429 the match. Spaces are allowed between the exclamation mark and the
11430 keyword. See below for more details on the supported keywords.
11431
Christopher Fauletec07e382020-04-07 14:56:26 +020011432 ok-status <st> is optional and can be used to set the check status if
11433 the expect rule is successfully evaluated and if it is
11434 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011435 "L6OK" and "L4OK" are supported :
11436 - L7OK : check passed on layer 7
Christopher Faulet83662b52020-11-20 17:47:47 +010011437 - L7OKC : check conditionally passed on layer 7, set
11438 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011439 - L6OK : check passed on layer 6
11440 - L4OK : check passed on layer 4
Christopher Fauletec07e382020-04-07 14:56:26 +020011441 By default "L7OK" is used.
11442
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011443 error-status <st> is optional and can be used to set the check status if
11444 an error occurred during the expect rule evaluation.
Christopher Faulet83662b52020-11-20 17:47:47 +010011445 "L7OKC", "L7RSP", "L7STS", "L6RSP" and "L4CON" are
11446 supported :
11447 - L7OKC : check conditionally passed on layer 7, set
11448 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011449 - L7RSP : layer 7 invalid response - protocol error
11450 - L7STS : layer 7 response error, for example HTTP 5xx
11451 - L6RSP : layer 6 invalid response - protocol error
11452 - L4CON : layer 1-4 connection problem
11453 By default "L7RSP" is used.
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011454
Christopher Fauletec07e382020-04-07 14:56:26 +020011455 tout-status <st> is optional and can be used to set the check status if
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011456 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011457 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
11458 - L7TOUT : layer 7 (HTTP/SMTP) timeout
11459 - L6TOUT : layer 6 (SSL) timeout
11460 - L4TOUT : layer 1-4 timeout
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011461 By default "L7TOUT" is used.
11462
Christopher Fauletbe52b4d2020-04-01 16:30:22 +020011463 on-success <fmt> is optional and can be used to customize the
11464 informational message reported in logs if the expect
11465 rule is successfully evaluated and if it is the last rule
11466 in the tcp-check ruleset. <fmt> is a log-format string.
11467
11468 on-error <fmt> is optional and can be used to customize the
11469 informational message reported in logs if an error
11470 occurred during the expect rule evaluation. <fmt> is a
11471 log-format string.
11472
Christopher Faulet98cc57c2020-04-01 20:52:31 +020011473 status-code <expr> is optional and can be used to set the check status code
11474 reported in logs, on success or on error. <expr> is a
11475 standard HAProxy expression formed by a sample-fetch
11476 followed by some converters.
11477
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011478 <pattern> is the pattern to look for. It may be a string or a regular
11479 expression. If the pattern contains spaces, they must be escaped
11480 with the usual backslash ('\').
11481 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +010011482 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011483 two digits will represent a byte. The hexadecimal digits may be
11484 used upper or lower case.
11485
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011486 The available matches are intentionally similar to their http-check cousins :
11487
11488 string <string> : test the exact string matches in the response buffer.
11489 A health check response will be considered valid if the
11490 response's buffer contains this exact string. If the
11491 "string" keyword is prefixed with "!", then the response
11492 will be considered invalid if the body contains this
11493 string. This can be used to look for a mandatory pattern
11494 in a protocol response, or to detect a failure when a
11495 specific error appears in a protocol banner.
11496
11497 rstring <regex> : test a regular expression on the response buffer.
11498 A health check response will be considered valid if the
11499 response's buffer matches this expression. If the
11500 "rstring" keyword is prefixed with "!", then the response
11501 will be considered invalid if the body matches the
11502 expression.
11503
Christopher Fauletaaab0832020-05-05 15:54:22 +020011504 string-lf <fmt> : test a log-format string match in the response's buffer.
11505 A health check response will be considered valid if the
11506 response's buffer contains the string resulting of the
11507 evaluation of <fmt>, which follows the log-format rules.
11508 If prefixed with "!", then the response will be
11509 considered invalid if the buffer contains the string.
11510
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011511 binary <hexstring> : test the exact string in its hexadecimal form matches
11512 in the response buffer. A health check response will
11513 be considered valid if the response's buffer contains
11514 this exact hexadecimal string.
11515 Purpose is to match data on binary protocols.
11516
Gaetan Rivetefab6c62020-02-07 15:37:17 +010011517 rbinary <regex> : test a regular expression on the response buffer, like
11518 "rstring". However, the response buffer is transformed
11519 into its hexadecimal form, including NUL-bytes. This
11520 allows using all regex engines to match any binary
11521 content. The hexadecimal transformation takes twice the
11522 size of the original response. As such, the expected
11523 pattern should work on at-most half the response buffer
11524 size.
11525
Christopher Fauletaaab0832020-05-05 15:54:22 +020011526 binary-lf <hexfmt> : test a log-format string in its hexadecimal form
11527 match in the response's buffer. A health check response
11528 will be considered valid if the response's buffer
11529 contains the hexadecimal string resulting of the
11530 evaluation of <fmt>, which follows the log-format
11531 rules. If prefixed with "!", then the response will be
11532 considered invalid if the buffer contains the
11533 hexadecimal string. The hexadecimal string is converted
11534 in a binary string before matching the response's
11535 buffer.
11536
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011537 It is important to note that the responses will be limited to a certain size
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011538 defined by the global "tune.bufsize" option, which defaults to 16384 bytes.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011539 Thus, too large responses may not contain the mandatory pattern when using
11540 "string", "rstring" or binary. If a large response is absolutely required, it
11541 is possible to change the default max size by setting the global variable.
11542 However, it is worth keeping in mind that parsing very large responses can
11543 waste some CPU cycles, especially when regular expressions are used, and that
11544 it is always better to focus the checks on smaller resources. Also, in its
11545 current state, the check will not find any string nor regex past a null
11546 character in the response. Similarly it is not possible to request matching
11547 the null character.
11548
11549 Examples :
11550 # perform a POP check
11551 option tcp-check
11552 tcp-check expect string +OK\ POP3\ ready
11553
11554 # perform an IMAP check
11555 option tcp-check
11556 tcp-check expect string *\ OK\ IMAP4\ ready
11557
11558 # look for the redis master server
11559 option tcp-check
11560 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +020011561 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011562 tcp-check send info\ replication\r\n
11563 tcp-check expect string role:master
11564 tcp-check send QUIT\r\n
11565 tcp-check expect string +OK
11566
11567
11568 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011569 "tcp-check send-binary", "http-check expect", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011570
11571
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011572tcp-check send <data> [comment <msg>]
11573tcp-check send-lf <fmt> [comment <msg>]
11574 Specify a string or a log-format string to be sent as a question during a
11575 generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011576 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011577 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011578
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011579 Arguments :
11580 comment <msg> defines a message to report if the rule evaluation fails.
11581
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011582 <data> is the string that will be sent during a generic health
11583 check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020011584
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011585 <fmt> is the log-format string that will be sent, once evaluated,
11586 during a generic health check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011587
11588 Examples :
11589 # look for the redis master server
11590 option tcp-check
11591 tcp-check send info\ replication\r\n
11592 tcp-check expect string role:master
11593
11594 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011595 "tcp-check send-binary", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011596
11597
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011598tcp-check send-binary <hexstring> [comment <msg>]
11599tcp-check send-binary-lf <hexfmt> [comment <msg>]
11600 Specify an hex digits string or an hex digits log-format string to be sent as
11601 a binary question during a raw tcp health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011602 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011603 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011604
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011605 Arguments :
11606 comment <msg> defines a message to report if the rule evaluation fails.
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011607
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011608 <hexstring> is the hexadecimal string that will be send, once converted
11609 to binary, during a generic health check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020011610
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011611 <hexfmt> is the hexadecimal log-format string that will be send, once
11612 evaluated and converted to binary, during a generic health
11613 check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011614
11615 Examples :
11616 # redis check in binary
11617 option tcp-check
11618 tcp-check send-binary 50494e470d0a # PING\r\n
11619 tcp-check expect binary 2b504F4e47 # +PONG
11620
11621
11622 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011623 "tcp-check send", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011624
11625
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011626tcp-check set-var(<var-name>) <expr>
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011627 This operation sets the content of a variable. The variable is declared inline.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011628 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011629 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011630
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011631 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011632 <var-name> The name of the variable starts with an indication about its
11633 scope. The scopes allowed for tcp-check are:
11634 "proc" : the variable is shared with the whole process.
11635 "sess" : the variable is shared with the tcp-check session.
11636 "check": the variable is declared for the lifetime of the tcp-check.
11637 This prefix is followed by a name. The separator is a '.'.
11638 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11639 and '-'.
11640
11641 <expr> Is a sample-fetch expression potentially followed by converters.
11642
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011643 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011644 tcp-check set-var(check.port) int(1234)
11645
11646
11647tcp-check unset-var(<var-name>)
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011648 Free a reference to a variable within its scope.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011649 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011650 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011651
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011652 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011653 <var-name> The name of the variable starts with an indication about its
11654 scope. The scopes allowed for tcp-check are:
11655 "proc" : the variable is shared with the whole process.
11656 "sess" : the variable is shared with the tcp-check session.
11657 "check": the variable is declared for the lifetime of the tcp-check.
11658 This prefix is followed by a name. The separator is a '.'.
11659 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11660 and '-'.
11661
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011662 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011663 tcp-check unset-var(check.port)
11664
11665
Willy Tarreaue9656522010-08-17 15:40:09 +020011666tcp-request connection <action> [{if | unless} <condition>]
11667 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +020011668 May be used in sections : defaults | frontend | listen | backend
11669 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +020011670 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011671 <action> defines the action to perform if the condition applies. See
11672 below.
Willy Tarreau1a687942010-05-23 22:40:30 +020011673
Willy Tarreaue9656522010-08-17 15:40:09 +020011674 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011675
11676 Immediately after acceptance of a new incoming connection, it is possible to
11677 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +020011678 or dropped or have its counters tracked. Those conditions cannot make use of
11679 any data contents because the connection has not been read from yet, and the
11680 buffers are not yet allocated. This is used to selectively and very quickly
11681 accept or drop connections from various sources with a very low overhead. If
11682 some contents need to be inspected in order to take the decision, the
11683 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011684
Willy Tarreaue9656522010-08-17 15:40:09 +020011685 The "tcp-request connection" rules are evaluated in their exact declaration
11686 order. If no rule matches or if there is no rule, the default action is to
11687 accept the incoming connection. There is no specific limit to the number of
11688 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011689
Willy Tarreaua9083d02015-05-08 15:27:59 +020011690 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +020011691 - accept :
11692 accepts the connection if the condition is true (when used with "if")
11693 or false (when used with "unless"). The first such rule executed ends
11694 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011695
Willy Tarreaue9656522010-08-17 15:40:09 +020011696 - reject :
11697 rejects the connection if the condition is true (when used with "if")
11698 or false (when used with "unless"). The first such rule executed ends
11699 the rules evaluation. Rejected connections do not even become a
11700 session, which is why they are accounted separately for in the stats,
11701 as "denied connections". They are not considered for the session
11702 rate-limit and are not logged either. The reason is that these rules
11703 should only be used to filter extremely high connection rates such as
11704 the ones encountered during a massive DDoS attack. Under these extreme
11705 conditions, the simple action of logging each event would make the
11706 system collapse and would considerably lower the filtering capacity. If
11707 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +020011708 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011709
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011710 - expect-proxy layer4 :
11711 configures the client-facing connection to receive a PROXY protocol
11712 header before any byte is read from the socket. This is equivalent to
11713 having the "accept-proxy" keyword on the "bind" line, except that using
11714 the TCP rule allows the PROXY protocol to be accepted only for certain
11715 IP address ranges using an ACL. This is convenient when multiple layers
11716 of load balancers are passed through by traffic coming from public
11717 hosts.
11718
Bertrand Jacquin90759682016-06-06 15:35:39 +010011719 - expect-netscaler-cip layer4 :
11720 configures the client-facing connection to receive a NetScaler Client
11721 IP insertion protocol header before any byte is read from the socket.
11722 This is equivalent to having the "accept-netscaler-cip" keyword on the
11723 "bind" line, except that using the TCP rule allows the PROXY protocol
11724 to be accepted only for certain IP address ranges using an ACL. This
11725 is convenient when multiple layers of load balancers are passed
11726 through by traffic coming from public hosts.
11727
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011728 - capture <sample> len <length> :
11729 This only applies to "tcp-request content" rules. It captures sample
11730 expression <sample> from the request buffer, and converts it to a
11731 string of at most <len> characters. The resulting string is stored into
11732 the next request "capture" slot, so it will possibly appear next to
11733 some captured HTTP headers. It will then automatically appear in the
11734 logs, and it will be possible to extract it using sample fetch rules to
11735 feed it into headers or anything. The length should be limited given
11736 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +020011737 session life. Please check section 7.3 (Fetching samples) and "capture
11738 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011739
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011740 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +020011741 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011742 rules do not stop evaluation and do not change default action. The
11743 number of counters that may be simultaneously tracked by the same
11744 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -050011745 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Matteo Contrini1857b8c2020-10-16 17:35:54 +020011746 and (MAX_SESS_STKCTR-1). The first "track-sc0" rule executed enables
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011747 tracking of the counters of the specified table as the first set. The
11748 first "track-sc1" rule executed enables tracking of the counters of the
11749 specified table as the second set. The first "track-sc2" rule executed
11750 enables tracking of the counters of the specified table as the third
11751 set. It is a recommended practice to use the first set of counters for
11752 the per-frontend counters and the second set for the per-backend ones.
11753 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011754
Willy Tarreaue9656522010-08-17 15:40:09 +020011755 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011756 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +020011757 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +010011758 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011759 and used to select which table entry to update the counters.
11760 Note that "tcp-request connection" cannot use content-based
11761 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011762
Willy Tarreaue9656522010-08-17 15:40:09 +020011763 <table> is an optional table to be used instead of the default one,
11764 which is the stick-table declared in the current proxy. All
11765 the counters for the matches and updates for the key will
11766 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011767
Willy Tarreaue9656522010-08-17 15:40:09 +020011768 Once a "track-sc*" rule is executed, the key is looked up in the table
11769 and if it is not found, an entry is allocated for it. Then a pointer to
11770 that entry is kept during all the session's life, and this entry's
11771 counters are updated as often as possible, every time the session's
11772 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011773 Counters are only updated for events that happen after the tracking has
11774 been started. For example, connection counters will not be updated when
11775 tracking layer 7 information, since the connection event happens before
11776 layer7 information is extracted.
11777
Willy Tarreaue9656522010-08-17 15:40:09 +020011778 If the entry tracks concurrent connection counters, one connection is
11779 counted for as long as the entry is tracked, and the entry will not
11780 expire during that time. Tracking counters also provides a performance
11781 advantage over just checking the keys, because only one table lookup is
11782 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011783
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011784 - sc-inc-gpc0(<sc-id>):
11785 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
11786 counter designated by <sc-id>. If an error occurs, this action silently
11787 fails and the actions evaluation continues.
11788
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011789 - sc-inc-gpc1(<sc-id>):
11790 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
11791 counter designated by <sc-id>. If an error occurs, this action silently
11792 fails and the actions evaluation continues.
11793
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011794 - sc-set-gpt0(<sc-id>) { <int> | <expr> }:
11795 This action sets the 32-bit unsigned GPT0 tag according to the sticky
11796 counter designated by <sc-id> and the value of <int>/<expr>. The
11797 expected result is a boolean. If an error occurs, this action silently
11798 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011799
William Lallemand2e785f22016-05-25 01:48:42 +020011800 - set-src <expr> :
11801 Is used to set the source IP address to the value of specified
11802 expression. Useful if you want to mask source IP for privacy.
11803 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011804 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +020011805
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011806 Arguments:
11807 <expr> Is a standard HAProxy expression formed by a sample-fetch
11808 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +020011809
11810 Example:
William Lallemand2e785f22016-05-25 01:48:42 +020011811 tcp-request connection set-src src,ipmask(24)
11812
Willy Tarreau0c630532016-10-21 17:52:58 +020011813 When possible, set-src preserves the original source port as long as the
11814 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +020011815
William Lallemand44be6402016-05-25 01:51:35 +020011816 - set-src-port <expr> :
11817 Is used to set the source port address to the value of specified
11818 expression.
11819
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011820 Arguments:
11821 <expr> Is a standard HAProxy expression formed by a sample-fetch
11822 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +020011823
11824 Example:
William Lallemand44be6402016-05-25 01:51:35 +020011825 tcp-request connection set-src-port int(4000)
11826
Willy Tarreau0c630532016-10-21 17:52:58 +020011827 When possible, set-src-port preserves the original source address as long
11828 as the address family supports a port, otherwise it forces the source
11829 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +020011830
William Lallemand13e9b0c2016-05-25 02:34:07 +020011831 - set-dst <expr> :
11832 Is used to set the destination IP address to the value of specified
11833 expression. Useful if you want to mask IP for privacy in log.
11834 If you want to provide an IP from a HTTP header use "http-request
11835 set-dst". If you want to connect to the new address/port, use
11836 '0.0.0.0:0' as a server address in the backend.
11837
11838 <expr> Is a standard HAProxy expression formed by a sample-fetch
11839 followed by some converters.
11840
11841 Example:
11842
11843 tcp-request connection set-dst dst,ipmask(24)
11844 tcp-request connection set-dst ipv4(10.0.0.1)
11845
Willy Tarreau0c630532016-10-21 17:52:58 +020011846 When possible, set-dst preserves the original destination port as long as
11847 the address family allows it, otherwise the destination port is set to 0.
11848
William Lallemand13e9b0c2016-05-25 02:34:07 +020011849 - set-dst-port <expr> :
11850 Is used to set the destination port address to the value of specified
11851 expression. If you want to connect to the new address/port, use
11852 '0.0.0.0:0' as a server address in the backend.
11853
11854
11855 <expr> Is a standard HAProxy expression formed by a sample-fetch
11856 followed by some converters.
11857
11858 Example:
11859
11860 tcp-request connection set-dst-port int(4000)
11861
Willy Tarreau0c630532016-10-21 17:52:58 +020011862 When possible, set-dst-port preserves the original destination address as
11863 long as the address family supports a port, otherwise it forces the
11864 destination address to IPv4 "0.0.0.0" before rewriting the port.
11865
Willy Tarreau2d392c22015-08-24 01:43:45 +020011866 - "silent-drop" :
11867 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011868 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020011869 to prevent the client from being notified. The effect it then that the
11870 client still sees an established connection while there's none on
11871 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
11872 except that it doesn't use any local resource at all on the machine
11873 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010011874 slow down stronger attackers. It is important to understand the impact
11875 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011876 client and HAProxy (firewalls, proxies, load balancers) will also keep
11877 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011878 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011879 TCP_REPAIR socket option is used to block the emission of a TCP
11880 reset. On other systems, the socket's TTL is reduced to 1 so that the
11881 TCP reset doesn't pass the first router, though it's still delivered to
11882 local networks. Do not use it unless you fully understand how it works.
11883
Willy Tarreaue9656522010-08-17 15:40:09 +020011884 Note that the "if/unless" condition is optional. If no condition is set on
11885 the action, it is simply performed unconditionally. That can be useful for
11886 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011887
Willy Tarreaue9656522010-08-17 15:40:09 +020011888 Example: accept all connections from white-listed hosts, reject too fast
11889 connection without counting them, and track accepted connections.
11890 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011891
Willy Tarreaue9656522010-08-17 15:40:09 +020011892 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011893 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011894 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011895
Willy Tarreaue9656522010-08-17 15:40:09 +020011896 Example: accept all connections from white-listed hosts, count all other
11897 connections and reject too fast ones. This results in abusive ones
11898 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011899
Willy Tarreaue9656522010-08-17 15:40:09 +020011900 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011901 tcp-request connection track-sc0 src
11902 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011903
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011904 Example: enable the PROXY protocol for traffic coming from all known proxies.
11905
11906 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
11907
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011908 See section 7 about ACL usage.
11909
Willy Tarreau4f614292016-10-21 17:49:36 +020011910 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011911
11912
Willy Tarreaue9656522010-08-17 15:40:09 +020011913tcp-request content <action> [{if | unless} <condition>]
11914 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +020011915 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020011916 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +020011917 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011918 <action> defines the action to perform if the condition applies. See
11919 below.
Willy Tarreau62644772008-07-16 18:36:06 +020011920
Willy Tarreaue9656522010-08-17 15:40:09 +020011921 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +020011922
Davor Ocelice9ed2812017-12-25 17:49:28 +010011923 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +020011924 called "TCP content inspection". During this stage, ACL-based rules are
11925 evaluated every time the request contents are updated, until either an
Christopher Fauletae863c62021-03-15 12:03:44 +010011926 "accept", a "reject" or a "switch-mode" rule matches, or the TCP request
11927 inspection delay expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +020011928
Willy Tarreaue9656522010-08-17 15:40:09 +020011929 The first difference between these rules and "tcp-request connection" rules
11930 is that "tcp-request content" rules can make use of contents to take a
11931 decision. Most often, these decisions will consider a protocol recognition or
11932 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +010011933 both frontends and backends. In case of HTTP keep-alive with the client, all
11934 tcp-request content rules are evaluated again, so haproxy keeps a record of
11935 what sticky counters were assigned by a "tcp-request connection" versus a
11936 "tcp-request content" rule, and flushes all the content-related ones after
11937 processing an HTTP request, so that they may be evaluated again by the rules
11938 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011939 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +010011940 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +020011941
Willy Tarreaue9656522010-08-17 15:40:09 +020011942 Content-based rules are evaluated in their exact declaration order. If no
11943 rule matches or if there is no rule, the default action is to accept the
11944 contents. There is no specific limit to the number of rules which may be
11945 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +020011946
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011947 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011948 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +010011949 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011950 - reject : the request is rejected and the connection is closed
11951 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -040011952 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011953 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011954 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011955 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011956 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011957 - set-dst <expr>
11958 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011959 - set-var(<var-name>) <expr>
Christopher Fauletae863c62021-03-15 12:03:44 +010011960 - switch-mode http [ proto <name> ]
Christopher Faulet85d79c92016-11-09 16:54:56 +010011961 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +020011962 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +010011963 - send-spoe-group <engine-name> <group-name>
Christopher Faulet579d83b2019-11-22 15:34:17 +010011964 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +020011965
Willy Tarreaue9656522010-08-17 15:40:09 +020011966 They have the same meaning as their counter-parts in "tcp-request connection"
11967 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +010011968 For "do-resolve" action, please check the "http-request do-resolve"
11969 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +020011970
Willy Tarreauf3338342014-01-28 21:40:28 +010011971 While there is nothing mandatory about it, it is recommended to use the
11972 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
11973 content" rules in the frontend, and track-sc2 for "tcp-request content"
11974 rules in the backend, because that makes the configuration more readable
11975 and easier to troubleshoot, but this is just a guideline and all counters
11976 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +020011977
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011978 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +020011979 the action, it is simply performed unconditionally. That can be useful for
11980 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +020011981
Christopher Faulet2079a4a2020-10-02 11:48:57 +020011982 Note also that it is recommended to use a "tcp-request session" rule to track
11983 information that does *not* depend on Layer 7 contents, especially for HTTP
11984 frontends. Some HTTP processing are performed at the session level and may
11985 lead to an early rejection of the requests. Thus, the tracking at the content
11986 level may be disturbed in such case. A warning is emitted during startup to
11987 prevent, as far as possible, such unreliable usage.
11988
Willy Tarreaue9656522010-08-17 15:40:09 +020011989 It is perfectly possible to match layer 7 contents with "tcp-request content"
Christopher Faulet7ea509e2020-10-02 11:38:46 +020011990 rules from a TCP proxy, since HTTP-specific ACL matches are able to
11991 preliminarily parse the contents of a buffer before extracting the required
11992 data. If the buffered contents do not parse as a valid HTTP message, then the
11993 ACL does not match. The parser which is involved there is exactly the same
11994 as for all other HTTP processing, so there is no risk of parsing something
11995 differently. In an HTTP frontend or an HTTP backend, it is guaranteed that
11996 HTTP contents will always be immediately present when the rule is evaluated
11997 first because the HTTP parsing is performed in the early stages of the
11998 connection processing, at the session level. But for such proxies, using
11999 "http-request" rules is much more natural and recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020012000
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012001 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020012002 are present when the rule is processed. The rule processing engine is able to
12003 wait until the inspect delay expires when the data to be tracked is not yet
12004 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012005
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020012006 The "set-dst" and "set-dst-port" are used to set respectively the destination
12007 IP and port. More information on how to use it at "http-request set-dst".
12008
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012009 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +020012010 declared inline. For "tcp-request session" rules, only session-level
12011 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012012
Daniel Schneller0b547052016-03-21 20:46:57 +010012013 <var-name> The name of the variable starts with an indication about
12014 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012015 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012016 "sess" : the variable is shared with the whole session
12017 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012018 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010012019 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012020 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010012021 "res" : the variable is shared only during response
12022 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012023 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010012024 The name may only contain characters 'a-z', 'A-Z', '0-9',
12025 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012026
12027 <expr> Is a standard HAProxy expression formed by a sample-fetch
12028 followed by some converters.
12029
Christopher Fauletae863c62021-03-15 12:03:44 +010012030 The "switch-mode" is used to perform a conntection upgrade. Only HTTP
12031 upgrades are supported for now. The protocol may optionally be
12032 specified. This action is only available for a proxy with the frontend
12033 capability. The connection upgrade is immediately performed, following
12034 "tcp-request content" rules are not evaluated. This upgrade method should be
12035 preferred to the implicit one consisting to rely on the backend mode. When
12036 used, it is possible to set HTTP directives in a frontend without any
12037 warning. These directives will be conditionnaly evaluated if the HTTP upgrade
12038 is performed. However, an HTTP backend must still be selected. It remains
12039 unsupported to route an HTTP connection (upgraded or not) to a TCP server.
12040
Christopher Faulet4d37e532021-03-26 14:44:00 +010012041 See section 4 about Proxies for more details on HTTP upgrades.
12042
Christopher Faulet85d79c92016-11-09 16:54:56 +010012043 The "unset-var" is used to unset a variable. See above for details about
12044 <var-name>.
12045
Patrick Hemmer268a7072018-05-11 12:52:31 -040012046 The "set-priority-class" is used to set the queue priority class of the
12047 current request. The value must be a sample expression which converts to an
12048 integer in the range -2047..2047. Results outside this range will be
12049 truncated. The priority class determines the order in which queued requests
12050 are processed. Lower values have higher priority.
12051
12052 The "set-priority-offset" is used to set the queue priority timestamp offset
12053 of the current request. The value must be a sample expression which converts
12054 to an integer in the range -524287..524287. Results outside this range will be
12055 truncated. When a request is queued, it is ordered first by the priority
12056 class, then by the current timestamp adjusted by the given offset in
12057 milliseconds. Lower values have higher priority.
12058 Note that the resulting timestamp is is only tracked with enough precision for
12059 524,287ms (8m44s287ms). If the request is queued long enough to where the
12060 adjusted timestamp exceeds this value, it will be misidentified as highest
12061 priority. Thus it is important to set "timeout queue" to a value, where when
12062 combined with the offset, does not exceed this limit.
12063
Christopher Faulet76c09ef2017-09-21 11:03:52 +020012064 The "send-spoe-group" is used to trigger sending of a group of SPOE
12065 messages. To do so, the SPOE engine used to send messages must be defined, as
12066 well as the SPOE group to send. Of course, the SPOE engine must refer to an
12067 existing SPOE filter. If not engine name is provided on the SPOE filter line,
12068 the SPOE agent name must be used.
12069
12070 <engine-name> The SPOE engine name.
12071
12072 <group-name> The SPOE group name as specified in the engine configuration.
12073
Christopher Faulet579d83b2019-11-22 15:34:17 +010012074 The "use-service" is used to executes a TCP service which will reply to the
12075 request and stop the evaluation of the rules. This service may choose to
12076 reply by sending any valid response or it may immediately close the
12077 connection without sending anything. Outside natives services, it is possible
12078 to write your own services in Lua. No further "tcp-request" rules are
12079 evaluated.
12080
12081 Example:
12082 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
12083
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012084 Example:
12085
12086 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +010012087 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012088
Willy Tarreau62644772008-07-16 18:36:06 +020012089 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +020012090 # Accept HTTP requests containing a Host header saying "example.com"
Christopher Fauletae863c62021-03-15 12:03:44 +010012091 # and reject everything else. (Only works for HTTP/1 connections)
Willy Tarreaue9656522010-08-17 15:40:09 +020012092 acl is_host_com hdr(Host) -i example.com
12093 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +020012094 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +020012095 tcp-request content reject
12096
Christopher Fauletae863c62021-03-15 12:03:44 +010012097 # Accept HTTP requests containing a Host header saying "example.com"
12098 # and reject everything else. (works for HTTP/1 and HTTP/2 connections)
12099 acl is_host_com hdr(Host) -i example.com
12100 tcp-request inspect-delay 5s
12101 tcp-request switch-mode http if HTTP
12102 tcp-request reject # non-HTTP traffic is implicit here
12103 ...
12104 http-request reject unless is_host_com
12105
Willy Tarreaue9656522010-08-17 15:40:09 +020012106 Example:
Willy Tarreau62644772008-07-16 18:36:06 +020012107 # reject SMTP connection if client speaks first
12108 tcp-request inspect-delay 30s
12109 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012110 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +020012111
12112 # Forward HTTPS connection only if client speaks
12113 tcp-request inspect-delay 30s
12114 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012115 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +020012116 tcp-request content reject
12117
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012118 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012119 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012120 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020012121 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012122 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
12123 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012124
12125 Example:
12126 # track request counts per "base" (concatenation of Host+URL)
12127 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020012128 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012129
Willy Tarreaue9656522010-08-17 15:40:09 +020012130 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012131 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +020012132
12133 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +010012134 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +020012135 # protecting all our sites
12136 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012137 tcp-request connection track-sc0 src
12138 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +020012139 ...
12140 use_backend http_dynamic if { path_end .php }
12141
12142 backend http_dynamic
12143 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012144 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +020012145 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012146 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012147 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012148 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +020012149 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +020012150
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012151 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +020012152
Jarno Huuskonen95b012b2017-04-06 13:59:14 +030012153 See also : "tcp-request connection", "tcp-request session",
12154 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +020012155
12156
12157tcp-request inspect-delay <timeout>
12158 Set the maximum allowed time to wait for data during content inspection
12159 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020012160 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +020012161 Arguments :
12162 <timeout> is the timeout value specified in milliseconds by default, but
12163 can be in any other unit if the number is suffixed by the unit,
12164 as explained at the top of this document.
12165
12166 People using haproxy primarily as a TCP relay are often worried about the
12167 risk of passing any type of protocol to a server without any analysis. In
12168 order to be able to analyze the request contents, we must first withhold
12169 the data then analyze them. This statement simply enables withholding of
12170 data for at most the specified amount of time.
12171
Willy Tarreaufb356202010-08-03 14:02:05 +020012172 TCP content inspection applies very early when a connection reaches a
12173 frontend, then very early when the connection is forwarded to a backend. This
12174 means that a connection may experience a first delay in the frontend and a
12175 second delay in the backend if both have tcp-request rules.
12176
Willy Tarreau62644772008-07-16 18:36:06 +020012177 Note that when performing content inspection, haproxy will evaluate the whole
12178 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012179 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +020012180 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +010012181 contents are definitive. If no delay is set, haproxy will not wait at all
12182 and will immediately apply a verdict based on the available information.
12183 Obviously this is unlikely to be very useful and might even be racy, so such
12184 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020012185
12186 As soon as a rule matches, the request is released and continues as usual. If
12187 the timeout is reached and no rule matches, the default policy will be to let
12188 it pass through unaffected.
12189
12190 For most protocols, it is enough to set it to a few seconds, as most clients
12191 send the full request immediately upon connection. Add 3 or more seconds to
12192 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +010012193 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +010012194 before the server (e.g. SMTP), or to wait for a client to talk before passing
12195 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +020012196 least the inspection delay, otherwise it will expire first. If the client
12197 closes the connection or if the buffer is full, the delay immediately expires
12198 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +020012199
Willy Tarreau55165fe2009-05-10 12:02:55 +020012200 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020012201 "timeout client".
12202
12203
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012204tcp-response content <action> [{if | unless} <condition>]
12205 Perform an action on a session response depending on a layer 4-7 condition
12206 May be used in sections : defaults | frontend | listen | backend
12207 no | no | yes | yes
12208 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020012209 <action> defines the action to perform if the condition applies. See
12210 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012211
12212 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
12213
Davor Ocelice9ed2812017-12-25 17:49:28 +010012214 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012215 called "TCP content inspection". During this stage, ACL-based rules are
12216 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020012217 "accept", "close" or a "reject" rule matches, or a TCP response inspection
12218 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012219
12220 Most often, these decisions will consider a protocol recognition or validity.
12221
12222 Content-based rules are evaluated in their exact declaration order. If no
12223 rule matches or if there is no rule, the default action is to accept the
12224 contents. There is no specific limit to the number of rules which may be
12225 inserted.
12226
Thierry FOURNIER236657b2015-08-19 08:25:14 +020012227 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012228 - accept :
12229 accepts the response if the condition is true (when used with "if")
12230 or false (when used with "unless"). The first such rule executed ends
12231 the rules evaluation.
12232
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020012233 - close :
12234 immediately closes the connection with the server if the condition is
12235 true (when used with "if"), or false (when used with "unless"). The
12236 first such rule executed ends the rules evaluation. The main purpose of
12237 this action is to force a connection to be finished between a client
12238 and a server after an exchange when the application protocol expects
12239 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012240 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020012241 protocols.
12242
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012243 - reject :
12244 rejects the response if the condition is true (when used with "if")
12245 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012246 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012247
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012248 - set-var(<var-name>) <expr>
12249 Sets a variable.
12250
Christopher Faulet85d79c92016-11-09 16:54:56 +010012251 - unset-var(<var-name>)
12252 Unsets a variable.
12253
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020012254 - sc-inc-gpc0(<sc-id>):
12255 This action increments the GPC0 counter according to the sticky
12256 counter designated by <sc-id>. If an error occurs, this action fails
12257 silently and the actions evaluation continues.
12258
Frédéric Lécaille6778b272018-01-29 15:22:53 +010012259 - sc-inc-gpc1(<sc-id>):
12260 This action increments the GPC1 counter according to the sticky
12261 counter designated by <sc-id>. If an error occurs, this action fails
12262 silently and the actions evaluation continues.
12263
Cédric Dufour0d7712d2019-11-06 18:38:53 +010012264 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
12265 This action sets the 32-bit unsigned GPT0 tag according to the sticky
12266 counter designated by <sc-id> and the value of <int>/<expr>. The
12267 expected result is a boolean. If an error occurs, this action silently
12268 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020012269
Willy Tarreau2d392c22015-08-24 01:43:45 +020012270 - "silent-drop" :
12271 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012272 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020012273 to prevent the client from being notified. The effect it then that the
12274 client still sees an established connection while there's none on
12275 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
12276 except that it doesn't use any local resource at all on the machine
12277 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010012278 slow down stronger attackers. It is important to understand the impact
12279 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020012280 client and HAProxy (firewalls, proxies, load balancers) will also keep
12281 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012282 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020012283 TCP_REPAIR socket option is used to block the emission of a TCP
12284 reset. On other systems, the socket's TTL is reduced to 1 so that the
12285 TCP reset doesn't pass the first router, though it's still delivered to
12286 local networks. Do not use it unless you fully understand how it works.
12287
Christopher Faulet76c09ef2017-09-21 11:03:52 +020012288 - send-spoe-group <engine-name> <group-name>
12289 Send a group of SPOE messages.
12290
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012291 Note that the "if/unless" condition is optional. If no condition is set on
12292 the action, it is simply performed unconditionally. That can be useful for
12293 for changing the default action to a reject.
12294
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012295 It is perfectly possible to match layer 7 contents with "tcp-response
12296 content" rules, but then it is important to ensure that a full response has
12297 been buffered, otherwise no contents will match. In order to achieve this,
12298 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012299 period.
12300
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012301 The "set-var" is used to set the content of a variable. The variable is
12302 declared inline.
12303
Daniel Schneller0b547052016-03-21 20:46:57 +010012304 <var-name> The name of the variable starts with an indication about
12305 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012306 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012307 "sess" : the variable is shared with the whole session
12308 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012309 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010012310 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012311 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010012312 "res" : the variable is shared only during response
12313 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012314 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010012315 The name may only contain characters 'a-z', 'A-Z', '0-9',
12316 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012317
12318 <expr> Is a standard HAProxy expression formed by a sample-fetch
12319 followed by some converters.
12320
12321 Example:
12322
12323 tcp-request content set-var(sess.my_var) src
12324
Christopher Faulet85d79c92016-11-09 16:54:56 +010012325 The "unset-var" is used to unset a variable. See above for details about
12326 <var-name>.
12327
12328 Example:
12329
12330 tcp-request content unset-var(sess.my_var)
12331
Christopher Faulet76c09ef2017-09-21 11:03:52 +020012332 The "send-spoe-group" is used to trigger sending of a group of SPOE
12333 messages. To do so, the SPOE engine used to send messages must be defined, as
12334 well as the SPOE group to send. Of course, the SPOE engine must refer to an
12335 existing SPOE filter. If not engine name is provided on the SPOE filter line,
12336 the SPOE agent name must be used.
12337
12338 <engine-name> The SPOE engine name.
12339
12340 <group-name> The SPOE group name as specified in the engine configuration.
12341
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012342 See section 7 about ACL usage.
12343
12344 See also : "tcp-request content", "tcp-response inspect-delay"
12345
12346
Willy Tarreau4f614292016-10-21 17:49:36 +020012347tcp-request session <action> [{if | unless} <condition>]
12348 Perform an action on a validated session depending on a layer 5 condition
12349 May be used in sections : defaults | frontend | listen | backend
12350 no | yes | yes | no
12351 Arguments :
12352 <action> defines the action to perform if the condition applies. See
12353 below.
12354
12355 <condition> is a standard layer5-only ACL-based condition (see section 7).
12356
Davor Ocelice9ed2812017-12-25 17:49:28 +010012357 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020012358 it is possible to evaluate some conditions to decide whether this session
12359 must be accepted or dropped or have its counters tracked. Those conditions
12360 cannot make use of any data contents because no buffers are allocated yet and
12361 the processing cannot wait at this stage. The main use case it to copy some
12362 early information into variables (since variables are accessible in the
12363 session), or to keep track of some information collected after the handshake,
12364 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010012365 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020012366 extracted information can thus be copied to a variable or tracked using
12367 "track-sc" rules. Of course it is also possible to decide to accept/reject as
12368 with other rulesets. Most operations performed here could also be performed
12369 in "tcp-request content" rules, except that in HTTP these rules are evaluated
12370 for each new request, and that might not always be acceptable. For example a
12371 rule might increment a counter on each evaluation. It would also be possible
12372 that a country is resolved by geolocation from the source IP address,
12373 assigned to a session-wide variable, then the source address rewritten from
12374 an HTTP header for all requests. If some contents need to be inspected in
12375 order to take the decision, the "tcp-request content" statements must be used
12376 instead.
12377
12378 The "tcp-request session" rules are evaluated in their exact declaration
12379 order. If no rule matches or if there is no rule, the default action is to
12380 accept the incoming session. There is no specific limit to the number of
12381 rules which may be inserted.
12382
12383 Several types of actions are supported :
12384 - accept : the request is accepted
12385 - reject : the request is rejected and the connection is closed
12386 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
12387 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010012388 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010012389 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Willy Tarreau4f614292016-10-21 17:49:36 +020012390 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010012391 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020012392 - silent-drop
12393
12394 These actions have the same meaning as their respective counter-parts in
12395 "tcp-request connection" and "tcp-request content", so please refer to these
12396 sections for a complete description.
12397
12398 Note that the "if/unless" condition is optional. If no condition is set on
12399 the action, it is simply performed unconditionally. That can be useful for
12400 "track-sc*" actions as well as for changing the default action to a reject.
12401
12402 Example: track the original source address by default, or the one advertised
12403 in the PROXY protocol header for connection coming from the local
12404 proxies. The first connection-level rule enables receipt of the
12405 PROXY protocol for these ones, the second rule tracks whatever
12406 address we decide to keep after optional decoding.
12407
12408 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
12409 tcp-request session track-sc0 src
12410
12411 Example: accept all sessions from white-listed hosts, reject too fast
12412 sessions without counting them, and track accepted sessions.
12413 This results in session rate being capped from abusive sources.
12414
12415 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
12416 tcp-request session reject if { src_sess_rate gt 10 }
12417 tcp-request session track-sc0 src
12418
12419 Example: accept all sessions from white-listed hosts, count all other
12420 sessions and reject too fast ones. This results in abusive ones
12421 being blocked as long as they don't slow down.
12422
12423 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
12424 tcp-request session track-sc0 src
12425 tcp-request session reject if { sc0_sess_rate gt 10 }
12426
12427 See section 7 about ACL usage.
12428
12429 See also : "tcp-request connection", "tcp-request content", "stick-table"
12430
12431
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012432tcp-response inspect-delay <timeout>
12433 Set the maximum allowed time to wait for a response during content inspection
12434 May be used in sections : defaults | frontend | listen | backend
12435 no | no | yes | yes
12436 Arguments :
12437 <timeout> is the timeout value specified in milliseconds by default, but
12438 can be in any other unit if the number is suffixed by the unit,
12439 as explained at the top of this document.
12440
12441 See also : "tcp-response content", "tcp-request inspect-delay".
12442
12443
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012444timeout check <timeout>
12445 Set additional check timeout, but only after a connection has been already
12446 established.
12447
12448 May be used in sections: defaults | frontend | listen | backend
12449 yes | no | yes | yes
12450 Arguments:
12451 <timeout> is the timeout value specified in milliseconds by default, but
12452 can be in any other unit if the number is suffixed by the unit,
12453 as explained at the top of this document.
12454
12455 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
12456 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010012457 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012458 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010012459 (Please also note that there is no valid reason to have such long connect
12460 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
12461 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012462
12463 If "timeout check" is not set haproxy uses "inter" for complete check
12464 timeout (connect + read) exactly like all <1.3.15 version.
12465
12466 In most cases check request is much simpler and faster to handle than normal
12467 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010012468 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012469
12470 This parameter is specific to backends, but can be specified once for all in
12471 "defaults" sections. This is in fact one of the easiest solutions not to
12472 forget about it.
12473
Willy Tarreau41a340d2008-01-22 12:25:31 +010012474 See also: "timeout connect", "timeout queue", "timeout server",
12475 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012476
12477
Willy Tarreau0ba27502007-12-24 16:55:16 +010012478timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010012479 Set the maximum inactivity time on the client side.
12480 May be used in sections : defaults | frontend | listen | backend
12481 yes | yes | yes | no
12482 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012483 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010012484 can be in any other unit if the number is suffixed by the unit,
12485 as explained at the top of this document.
12486
12487 The inactivity timeout applies when the client is expected to acknowledge or
12488 send data. In HTTP mode, this timeout is particularly important to consider
12489 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010012490 response while it is reading data sent by the server. That said, for the
12491 first phase, it is preferable to set the "timeout http-request" to better
12492 protect HAProxy from Slowloris like attacks. The value is specified in
12493 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010012494 suffixed by the unit, as specified at the top of this document. In TCP mode
12495 (and to a lesser extent, in HTTP mode), it is highly recommended that the
12496 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012497 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010012498 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010012499 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
12500 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020012501 which overrides "timeout client" and "timeout server" for tunnels, as well as
12502 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012503
12504 This parameter is specific to frontends, but can be specified once for all in
12505 "defaults" sections. This is in fact one of the easiest solutions not to
12506 forget about it. An unspecified timeout results in an infinite timeout, which
12507 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012508 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010012509 the system if the system's timeouts are not configured either.
12510
Willy Tarreau95c4e142017-11-26 12:18:55 +010012511 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010012512
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012513 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012514
Willy Tarreau0ba27502007-12-24 16:55:16 +010012515
Willy Tarreau05cdd962014-05-10 14:30:07 +020012516timeout client-fin <timeout>
12517 Set the inactivity timeout on the client side for half-closed connections.
12518 May be used in sections : defaults | frontend | listen | backend
12519 yes | yes | yes | no
12520 Arguments :
12521 <timeout> is the timeout value specified in milliseconds by default, but
12522 can be in any other unit if the number is suffixed by the unit,
12523 as explained at the top of this document.
12524
12525 The inactivity timeout applies when the client is expected to acknowledge or
12526 send data while one direction is already shut down. This timeout is different
12527 from "timeout client" in that it only applies to connections which are closed
12528 in one direction. This is particularly useful to avoid keeping connections in
12529 FIN_WAIT state for too long when clients do not disconnect cleanly. This
12530 problem is particularly common long connections such as RDP or WebSocket.
12531 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010012532 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
12533 frame was sent, often indicating an expectation that the connection quickly
12534 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020012535
12536 This parameter is specific to frontends, but can be specified once for all in
12537 "defaults" sections. By default it is not set, so half-closed connections
12538 will use the other timeouts (timeout.client or timeout.tunnel).
12539
12540 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
12541
12542
Willy Tarreau0ba27502007-12-24 16:55:16 +010012543timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010012544 Set the maximum time to wait for a connection attempt to a server to succeed.
12545 May be used in sections : defaults | frontend | listen | backend
12546 yes | no | yes | yes
12547 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012548 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010012549 can be in any other unit if the number is suffixed by the unit,
12550 as explained at the top of this document.
12551
12552 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012553 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010012554 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010012555 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012556 connect timeout also presets both queue and tarpit timeouts to the same value
12557 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012558
12559 This parameter is specific to backends, but can be specified once for all in
12560 "defaults" sections. This is in fact one of the easiest solutions not to
12561 forget about it. An unspecified timeout results in an infinite timeout, which
12562 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012563 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010012564 the system if the system's timeouts are not configured either.
12565
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012566 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012567
Willy Tarreau0ba27502007-12-24 16:55:16 +010012568
Willy Tarreaub16a5742010-01-10 14:46:16 +010012569timeout http-keep-alive <timeout>
12570 Set the maximum allowed time to wait for a new HTTP request to appear
12571 May be used in sections : defaults | frontend | listen | backend
12572 yes | yes | yes | yes
12573 Arguments :
12574 <timeout> is the timeout value specified in milliseconds by default, but
12575 can be in any other unit if the number is suffixed by the unit,
12576 as explained at the top of this document.
12577
12578 By default, the time to wait for a new request in case of keep-alive is set
12579 by "timeout http-request". However this is not always convenient because some
12580 people want very short keep-alive timeouts in order to release connections
12581 faster, and others prefer to have larger ones but still have short timeouts
12582 once the request has started to present itself.
12583
12584 The "http-keep-alive" timeout covers these needs. It will define how long to
12585 wait for a new HTTP request to start coming after a response was sent. Once
12586 the first byte of request has been seen, the "http-request" timeout is used
12587 to wait for the complete request to come. Note that empty lines prior to a
12588 new request do not refresh the timeout and are not counted as a new request.
12589
12590 There is also another difference between the two timeouts : when a connection
12591 expires during timeout http-keep-alive, no error is returned, the connection
12592 just closes. If the connection expires in "http-request" while waiting for a
12593 connection to complete, a HTTP 408 error is returned.
12594
12595 In general it is optimal to set this value to a few tens to hundreds of
12596 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010012597 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010012598 1 millisecond) it will probably only accept pipelined requests but not the
12599 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020012600 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010012601
12602 If this parameter is not set, the "http-request" timeout applies, and if both
12603 are not set, "timeout client" still applies at the lower level. It should be
12604 set in the frontend to take effect, unless the frontend is in TCP mode, in
12605 which case the HTTP backend's timeout will be used.
12606
Willy Tarreau95c4e142017-11-26 12:18:55 +010012607 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
12608 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010012609 (where we only have one connection per client and a connection setup).
12610
Willy Tarreaub16a5742010-01-10 14:46:16 +010012611 See also : "timeout http-request", "timeout client".
12612
12613
Willy Tarreau036fae02008-01-06 13:24:40 +010012614timeout http-request <timeout>
12615 Set the maximum allowed time to wait for a complete HTTP request
12616 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012617 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010012618 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012619 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010012620 can be in any other unit if the number is suffixed by the unit,
12621 as explained at the top of this document.
12622
12623 In order to offer DoS protection, it may be required to lower the maximum
12624 accepted time to receive a complete HTTP request without affecting the client
12625 timeout. This helps protecting against established connections on which
12626 nothing is sent. The client timeout cannot offer a good protection against
12627 this abuse because it is an inactivity timeout, which means that if the
12628 attacker sends one character every now and then, the timeout will not
12629 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020012630 types, the request will be aborted if it does not complete in time. When the
12631 timeout expires, an HTTP 408 response is sent to the client to inform it
12632 about the problem, and the connection is closed. The logs will report
12633 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012634 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020012635 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
12636 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010012637
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012638 By default, this timeout only applies to the header part of the request,
12639 and not to any data. As soon as the empty line is received, this timeout is
12640 not used anymore. When combined with "option http-buffer-request", this
12641 timeout also applies to the body of the request..
12642 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010012643 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010012644
12645 Generally it is enough to set it to a few seconds, as most clients send the
12646 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010012647 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010012648 generally work on local networks as long as there are no packet losses. This
12649 will prevent people from sending bare HTTP requests using telnet.
12650
12651 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012652 chunk of the incoming request. It should be set in the frontend to take
12653 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
12654 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010012655
Willy Tarreau0f228a02015-05-01 15:37:53 +020012656 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012657 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010012658
Willy Tarreau844e3c52008-01-11 16:28:18 +010012659
12660timeout queue <timeout>
12661 Set the maximum time to wait in the queue for a connection slot to be free
12662 May be used in sections : defaults | frontend | listen | backend
12663 yes | no | yes | yes
12664 Arguments :
12665 <timeout> is the timeout value specified in milliseconds by default, but
12666 can be in any other unit if the number is suffixed by the unit,
12667 as explained at the top of this document.
12668
12669 When a server's maxconn is reached, connections are left pending in a queue
12670 which may be server-specific or global to the backend. In order not to wait
12671 indefinitely, a timeout is applied to requests pending in the queue. If the
12672 timeout is reached, it is considered that the request will almost never be
12673 served, so it is dropped and a 503 error is returned to the client.
12674
12675 The "timeout queue" statement allows to fix the maximum time for a request to
12676 be left pending in a queue. If unspecified, the same value as the backend's
12677 connection timeout ("timeout connect") is used, for backwards compatibility
12678 with older versions with no "timeout queue" parameter.
12679
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012680 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012681
12682
12683timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010012684 Set the maximum inactivity time on the server side.
12685 May be used in sections : defaults | frontend | listen | backend
12686 yes | no | yes | yes
12687 Arguments :
12688 <timeout> is the timeout value specified in milliseconds by default, but
12689 can be in any other unit if the number is suffixed by the unit,
12690 as explained at the top of this document.
12691
12692 The inactivity timeout applies when the server is expected to acknowledge or
12693 send data. In HTTP mode, this timeout is particularly important to consider
12694 during the first phase of the server's response, when it has to send the
12695 headers, as it directly represents the server's processing time for the
12696 request. To find out what value to put there, it's often good to start with
12697 what would be considered as unacceptable response times, then check the logs
12698 to observe the response time distribution, and adjust the value accordingly.
12699
12700 The value is specified in milliseconds by default, but can be in any other
12701 unit if the number is suffixed by the unit, as specified at the top of this
12702 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
12703 recommended that the client timeout remains equal to the server timeout in
12704 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012705 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010012706 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010012707 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
12708 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020012709 "timeout tunnel", which overrides "timeout client" and "timeout server" for
12710 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012711
12712 This parameter is specific to backends, but can be specified once for all in
12713 "defaults" sections. This is in fact one of the easiest solutions not to
12714 forget about it. An unspecified timeout results in an infinite timeout, which
12715 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012716 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010012717 the system if the system's timeouts are not configured either.
12718
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012719 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012720
Willy Tarreau05cdd962014-05-10 14:30:07 +020012721
12722timeout server-fin <timeout>
12723 Set the inactivity timeout on the server side for half-closed connections.
12724 May be used in sections : defaults | frontend | listen | backend
12725 yes | no | yes | yes
12726 Arguments :
12727 <timeout> is the timeout value specified in milliseconds by default, but
12728 can be in any other unit if the number is suffixed by the unit,
12729 as explained at the top of this document.
12730
12731 The inactivity timeout applies when the server is expected to acknowledge or
12732 send data while one direction is already shut down. This timeout is different
12733 from "timeout server" in that it only applies to connections which are closed
12734 in one direction. This is particularly useful to avoid keeping connections in
12735 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
12736 This problem is particularly common long connections such as RDP or WebSocket.
12737 Note that this timeout can override "timeout tunnel" when a connection shuts
12738 down in one direction. This setting was provided for completeness, but in most
12739 situations, it should not be needed.
12740
12741 This parameter is specific to backends, but can be specified once for all in
12742 "defaults" sections. By default it is not set, so half-closed connections
12743 will use the other timeouts (timeout.server or timeout.tunnel).
12744
12745 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
12746
Willy Tarreau844e3c52008-01-11 16:28:18 +010012747
12748timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010012749 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010012750 May be used in sections : defaults | frontend | listen | backend
12751 yes | yes | yes | yes
12752 Arguments :
12753 <timeout> is the tarpit duration specified in milliseconds by default, but
12754 can be in any other unit if the number is suffixed by the unit,
12755 as explained at the top of this document.
12756
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020012757 When a connection is tarpitted using "http-request tarpit", it is maintained
12758 open with no activity for a certain amount of time, then closed. "timeout
12759 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012760
12761 The value is specified in milliseconds by default, but can be in any other
12762 unit if the number is suffixed by the unit, as specified at the top of this
12763 document. If unspecified, the same value as the backend's connection timeout
12764 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010012765 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012766
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012767 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012768
12769
Willy Tarreauce887fd2012-05-12 12:50:00 +020012770timeout tunnel <timeout>
12771 Set the maximum inactivity time on the client and server side for tunnels.
12772 May be used in sections : defaults | frontend | listen | backend
12773 yes | no | yes | yes
12774 Arguments :
12775 <timeout> is the timeout value specified in milliseconds by default, but
12776 can be in any other unit if the number is suffixed by the unit,
12777 as explained at the top of this document.
12778
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012779 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020012780 between a client and a server, and the connection remains inactive in both
12781 directions. This timeout supersedes both the client and server timeouts once
12782 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012783 analyzer remains attached to either connection (e.g. tcp content rules are
12784 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020012785 when switching to the WebSocket protocol, or forwarding a CONNECT request
12786 to a proxy), or after the first response when no keepalive/close option is
12787 specified.
12788
Willy Tarreau05cdd962014-05-10 14:30:07 +020012789 Since this timeout is usually used in conjunction with long-lived connections,
12790 it usually is a good idea to also set "timeout client-fin" to handle the
12791 situation where a client suddenly disappears from the net and does not
12792 acknowledge a close, or sends a shutdown and does not acknowledge pending
12793 data anymore. This can happen in lossy networks where firewalls are present,
12794 and is detected by the presence of large amounts of sessions in a FIN_WAIT
12795 state.
12796
Willy Tarreauce887fd2012-05-12 12:50:00 +020012797 The value is specified in milliseconds by default, but can be in any other
12798 unit if the number is suffixed by the unit, as specified at the top of this
12799 document. Whatever the expected normal idle time, it is a good practice to
12800 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010012801 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020012802
12803 This parameter is specific to backends, but can be specified once for all in
12804 "defaults" sections. This is in fact one of the easiest solutions not to
12805 forget about it.
12806
12807 Example :
12808 defaults http
12809 option http-server-close
12810 timeout connect 5s
12811 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020012812 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020012813 timeout server 30s
12814 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
12815
Willy Tarreau05cdd962014-05-10 14:30:07 +020012816 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020012817
12818
Willy Tarreau844e3c52008-01-11 16:28:18 +010012819transparent (deprecated)
12820 Enable client-side transparent proxying
12821 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010012822 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010012823 Arguments : none
12824
12825 This keyword was introduced in order to provide layer 7 persistence to layer
12826 3 load balancers. The idea is to use the OS's ability to redirect an incoming
12827 connection for a remote address to a local process (here HAProxy), and let
12828 this process know what address was initially requested. When this option is
12829 used, sessions without cookies will be forwarded to the original destination
12830 IP address of the incoming request (which should match that of another
12831 equipment), while requests with cookies will still be forwarded to the
12832 appropriate server.
12833
12834 The "transparent" keyword is deprecated, use "option transparent" instead.
12835
12836 Note that contrary to a common belief, this option does NOT make HAProxy
12837 present the client's IP to the server when establishing the connection.
12838
Willy Tarreau844e3c52008-01-11 16:28:18 +010012839 See also: "option transparent"
12840
William Lallemanda73203e2012-03-12 12:48:57 +010012841unique-id-format <string>
12842 Generate a unique ID for each request.
12843 May be used in sections : defaults | frontend | listen | backend
12844 yes | yes | yes | no
12845 Arguments :
12846 <string> is a log-format string.
12847
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012848 This keyword creates a ID for each request using the custom log format. A
12849 unique ID is useful to trace a request passing through many components of
12850 a complex infrastructure. The newly created ID may also be logged using the
12851 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010012852
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012853 The format should be composed from elements that are guaranteed to be
12854 unique when combined together. For instance, if multiple haproxy instances
12855 are involved, it might be important to include the node name. It is often
12856 needed to log the incoming connection's source and destination addresses
12857 and ports. Note that since multiple requests may be performed over the same
12858 connection, including a request counter may help differentiate them.
12859 Similarly, a timestamp may protect against a rollover of the counter.
12860 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010012861
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012862 It is recommended to use hexadecimal notation for many fields since it
12863 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010012864
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012865 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012866
Julien Vehentf21be322014-03-07 08:27:34 -050012867 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012868
12869 will generate:
12870
12871 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12872
12873 See also: "unique-id-header"
12874
12875unique-id-header <name>
12876 Add a unique ID header in the HTTP request.
12877 May be used in sections : defaults | frontend | listen | backend
12878 yes | yes | yes | no
12879 Arguments :
12880 <name> is the name of the header.
12881
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012882 Add a unique-id header in the HTTP request sent to the server, using the
12883 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010012884
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012885 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012886
Julien Vehentf21be322014-03-07 08:27:34 -050012887 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012888 unique-id-header X-Unique-ID
12889
12890 will generate:
12891
12892 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12893
12894 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010012895
Willy Tarreauf51658d2014-04-23 01:21:56 +020012896use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012897 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012898 May be used in sections : defaults | frontend | listen | backend
12899 no | yes | yes | no
12900 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012901 <backend> is the name of a valid backend or "listen" section, or a
12902 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012903
Willy Tarreauf51658d2014-04-23 01:21:56 +020012904 <condition> is a condition composed of ACLs, as described in section 7. If
12905 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012906
12907 When doing content-switching, connections arrive on a frontend and are then
12908 dispatched to various backends depending on a number of conditions. The
12909 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012910 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010012911 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012912 source address validation) or combined with a "tcp-request" rule to wait for
12913 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012914
12915 There may be as many "use_backend" rules as desired. All of these rules are
12916 evaluated in their declaration order, and the first one which matches will
12917 assign the backend.
12918
12919 In the first form, the backend will be used if the condition is met. In the
12920 second form, the backend will be used if the condition is not met. If no
12921 condition is valid, the backend defined with "default_backend" will be used.
12922 If no default backend is defined, either the servers in the same section are
12923 used (in case of a "listen" section) or, in case of a frontend, no server is
12924 used and a 503 service unavailable response is returned.
12925
Willy Tarreau51aecc72009-07-12 09:47:04 +020012926 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012927 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020012928 and backend processing will immediately follow, or the backend will wait for
12929 a complete HTTP request to get in. This feature is useful when a frontend
12930 must decode several protocols on a unique port, one of them being HTTP.
12931
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012932 When <backend> is a simple name, it is resolved at configuration time, and an
12933 error is reported if the specified backend does not exist. If <backend> is
12934 a log-format string instead, no check may be done at configuration time, so
12935 the backend name is resolved dynamically at run time. If the resulting
12936 backend name does not correspond to any valid backend, no other rule is
12937 evaluated, and the default_backend directive is applied instead. Note that
12938 when using dynamic backend names, it is highly recommended to use a prefix
12939 that no other backend uses in order to ensure that an unauthorized backend
12940 cannot be forced from the request.
12941
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012942 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012943 used to detect the association between frontends and backends to compute the
12944 backend's "fullconn" setting. This cannot be done for dynamic names.
12945
12946 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
12947 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010012948
Christopher Fauletb30b3102019-09-12 23:03:09 +020012949use-fcgi-app <name>
12950 Defines the FastCGI application to use for the backend.
12951 May be used in sections : defaults | frontend | listen | backend
12952 no | no | yes | yes
12953 Arguments :
12954 <name> is the name of the FastCGI application to use.
12955
12956 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010012957
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012958use-server <server> if <condition>
12959use-server <server> unless <condition>
12960 Only use a specific server if/unless an ACL-based condition is matched.
12961 May be used in sections : defaults | frontend | listen | backend
12962 no | no | yes | yes
12963 Arguments :
Jerome Magnin824186b2020-03-29 09:37:12 +020012964 <server> is the name of a valid server in the same backend section
12965 or a "log-format" string resolving to a server name.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012966
12967 <condition> is a condition composed of ACLs, as described in section 7.
12968
12969 By default, connections which arrive to a backend are load-balanced across
12970 the available servers according to the configured algorithm, unless a
12971 persistence mechanism such as a cookie is used and found in the request.
12972
12973 Sometimes it is desirable to forward a particular request to a specific
12974 server without having to declare a dedicated backend for this server. This
12975 can be achieved using the "use-server" rules. These rules are evaluated after
12976 the "redirect" rules and before evaluating cookies, and they have precedence
12977 on them. There may be as many "use-server" rules as desired. All of these
12978 rules are evaluated in their declaration order, and the first one which
12979 matches will assign the server.
12980
12981 If a rule designates a server which is down, and "option persist" is not used
12982 and no force-persist rule was validated, it is ignored and evaluation goes on
12983 with the next rules until one matches.
12984
12985 In the first form, the server will be used if the condition is met. In the
12986 second form, the server will be used if the condition is not met. If no
12987 condition is valid, the processing continues and the server will be assigned
12988 according to other persistence mechanisms.
12989
12990 Note that even if a rule is matched, cookie processing is still performed but
12991 does not assign the server. This allows prefixed cookies to have their prefix
12992 stripped.
12993
12994 The "use-server" statement works both in HTTP and TCP mode. This makes it
12995 suitable for use with content-based inspection. For instance, a server could
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012996 be selected in a farm according to the TLS SNI field when using protocols with
12997 implicit TLS (also see "req_ssl_sni"). And if these servers have their weight
12998 set to zero, they will not be used for other traffic.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012999
13000 Example :
13001 # intercept incoming TLS requests based on the SNI field
13002 use-server www if { req_ssl_sni -i www.example.com }
13003 server www 192.168.0.1:443 weight 0
13004 use-server mail if { req_ssl_sni -i mail.example.com }
Lukas Tribusa267b5d2020-07-19 00:25:06 +020013005 server mail 192.168.0.1:465 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013006 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000013007 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013008 # all the rest is forwarded to this server
13009 server default 192.168.0.2:443 check
13010
Jerome Magnin824186b2020-03-29 09:37:12 +020013011 When <server> is a simple name, it is checked against existing servers in the
13012 configuration and an error is reported if the specified server does not exist.
13013 If it is a log-format, no check is performed when parsing the configuration,
13014 and if we can't resolve a valid server name at runtime but the use-server rule
Ilya Shipitsin11057a32020-06-21 21:18:27 +050013015 was conditioned by an ACL returning true, no other use-server rule is applied
Jerome Magnin824186b2020-03-29 09:37:12 +020013016 and we fall back to load balancing.
13017
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013018 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013019
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013020
Davor Ocelice9ed2812017-12-25 17:49:28 +0100130215. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013022--------------------------
13023
13024The "bind", "server" and "default-server" keywords support a number of settings
13025depending on some build options and on the system HAProxy was built on. These
13026settings generally each consist in one word sometimes followed by a value,
13027written on the same line as the "bind" or "server" line. All these options are
13028described in this section.
13029
13030
130315.1. Bind options
13032-----------------
13033
13034The "bind" keyword supports a certain number of settings which are all passed
13035as arguments on the same line. The order in which those arguments appear makes
13036no importance, provided that they appear after the bind address. All of these
13037parameters are optional. Some of them consist in a single words (booleans),
13038while other ones expect a value after them. In this case, the value must be
13039provided immediately after the setting name.
13040
13041The currently supported settings are the following ones.
13042
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010013043accept-netscaler-cip <magic number>
13044 Enforces the use of the NetScaler Client IP insertion protocol over any
13045 connection accepted by any of the TCP sockets declared on the same line. The
13046 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
13047 the incoming connection to be used everywhere an address is used, with the
13048 only exception of "tcp-request connection" rules which will only see the
13049 real connection address. Logs will reflect the addresses indicated in the
13050 protocol, unless it is violated, in which case the real address will still
13051 be used. This keyword combined with support from external components can be
13052 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010013053 mechanism which is not always reliable and not even always usable. See also
13054 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
13055 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010013056
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013057accept-proxy
13058 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020013059 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
13060 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013061 3/4 addresses of the incoming connection to be used everywhere an address is
13062 used, with the only exception of "tcp-request connection" rules which will
13063 only see the real connection address. Logs will reflect the addresses
13064 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010013065 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013066 components can be used as an efficient and reliable alternative to the
13067 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020013068 usable. See also "tcp-request connection expect-proxy" for a finer-grained
13069 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013070
Olivier Houchardc2aae742017-09-22 18:26:28 +020013071allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010013072 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010013073 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050013074 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010013075 that are idempotent. You can use the "wait-for-handshake" action for any
13076 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020013077
Willy Tarreauab861d32013-04-02 02:30:41 +020013078alpn <protocols>
13079 This enables the TLS ALPN extension and advertises the specified protocol
13080 list as supported on top of ALPN. The protocol list consists in a comma-
13081 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050013082 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020013083 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010013084 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
13085 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
13086 now obsolete NPN extension. At the time of writing this, most browsers still
13087 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
13088 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
13089 are expected to be supported, both versions can be advertised, in order of
13090 preference, like below :
13091
13092 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020013093
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013094backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010013095 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013096 backlog is used instead, which generally defaults to the maxconn value.
13097
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010013098curves <curves>
13099 This setting is only available when support for OpenSSL was built in. It sets
13100 the string describing the list of elliptic curves algorithms ("curve suite")
13101 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
13102 string is a colon-delimited list of curve name.
13103 Example: "X25519:P-256" (without quote)
13104 When "curves" is set, "ecdhe" parameter is ignored.
13105
Emeric Brun7fb34422012-09-28 15:26:15 +020013106ecdhe <named curve>
13107 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010013108 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
13109 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020013110
Emeric Brunfd33a262012-10-11 16:28:27 +020013111ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020013112 This setting is only available when support for OpenSSL was built in. It
13113 designates a PEM file from which to load CA certificates used to verify
13114 client's certificate.
13115
Emeric Brunb6dc9342012-09-28 17:55:37 +020013116ca-ignore-err [all|<errorID>,...]
13117 This setting is only available when support for OpenSSL was built in.
13118 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
13119 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
13120 error is ignored.
13121
Christopher Faulet31af49d2015-06-09 17:29:50 +020013122ca-sign-file <cafile>
13123 This setting is only available when support for OpenSSL was built in. It
13124 designates a PEM file containing both the CA certificate and the CA private
13125 key used to create and sign server's certificates. This is a mandatory
13126 setting when the dynamic generation of certificates is enabled. See
13127 'generate-certificates' for details.
13128
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000013129ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020013130 This setting is only available when support for OpenSSL was built in. It is
13131 the CA private key passphrase. This setting is optional and used only when
13132 the dynamic generation of certificates is enabled. See
13133 'generate-certificates' for details.
13134
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010013135ca-verify-file <cafile>
13136 This setting designates a PEM file from which to load CA certificates used to
13137 verify client's certificate. It designates CA certificates which must not be
13138 included in CA names sent in server hello message. Typically, "ca-file" must
13139 be defined with intermediate certificates, and "ca-verify-file" with
13140 certificates to ending the chain, like root CA.
13141
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013142ciphers <ciphers>
13143 This setting is only available when support for OpenSSL was built in. It sets
13144 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000013145 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013146 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013147 information and recommendations see e.g.
13148 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
13149 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
13150 cipher configuration, please check the "ciphersuites" keyword.
13151
13152ciphersuites <ciphersuites>
13153 This setting is only available when support for OpenSSL was built in and
13154 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
13155 the list of cipher algorithms ("cipher suite") that are negotiated during the
13156 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013157 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
13158 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013159
Emeric Brunfd33a262012-10-11 16:28:27 +020013160crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020013161 This setting is only available when support for OpenSSL was built in. It
13162 designates a PEM file from which to load certificate revocation list used
13163 to verify client's certificate.
13164
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013165crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000013166 This setting is only available when support for OpenSSL was built in. It
13167 designates a PEM file containing both the required certificates and any
13168 associated private keys. This file can be built by concatenating multiple
13169 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
13170 requires an intermediate certificate, this can also be concatenated into this
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +010013171 file. Intermediate certificate can also be shared in a directory via
13172 "issuers-chain-path" directive.
Alex Davies0fbf0162013-03-02 16:04:50 +000013173
William Lallemand4c5adbf2020-02-24 14:23:22 +010013174 If the file does not contain a private key, HAProxy will try to load
13175 the key at the same path suffixed by a ".key".
13176
Alex Davies0fbf0162013-03-02 16:04:50 +000013177 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
13178 are loaded.
13179
13180 If a directory name is used instead of a PEM file, then all files found in
William Lallemand3f25ae32020-02-24 16:30:12 +010013181 that directory will be loaded in alphabetic order unless their name ends
13182 with '.key', '.issuer', '.ocsp' or '.sctl' (reserved extensions). This
13183 directive may be specified multiple times in order to load certificates from
13184 multiple files or directories. The certificates will be presented to clients
13185 who provide a valid TLS Server Name Indication field matching one of their
13186 CN or alt subjects. Wildcards are supported, where a wildcard character '*'
13187 is used instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010013188 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000013189
13190 If no SNI is provided by the client or if the SSL library does not support
13191 TLS extensions, or if the client provides an SNI hostname which does not
13192 match any certificate, then the first loaded certificate will be presented.
13193 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010013194 recommended to load the default one first as a file or to ensure that it will
13195 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000013196
Emeric Brune032bfa2012-09-28 13:01:45 +020013197 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013198
Davor Ocelice9ed2812017-12-25 17:49:28 +010013199 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000013200 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010013201 choose a web server that the CA believes requires an intermediate CA (for
13202 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000013203 others, e.g. nginx, result in a wrong bundle that will not work for some
13204 clients).
13205
Emeric Brun4147b2e2014-06-16 18:36:30 +020013206 For each PEM file, haproxy checks for the presence of file at the same path
13207 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
13208 Status Request extension (also known as "OCSP stapling") is automatically
13209 enabled. The content of this file is optional. If not empty, it must contain
13210 a valid OCSP Response in DER format. In order to be valid an OCSP Response
13211 must comply with the following rules: it has to indicate a good status,
13212 it has to be a single response for the certificate of the PEM file, and it
13213 has to be valid at the moment of addition. If these rules are not respected
13214 the OCSP Response is ignored and a warning is emitted. In order to identify
13215 which certificate an OCSP Response applies to, the issuer's certificate is
13216 necessary. If the issuer's certificate is not found in the PEM file, it will
13217 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
13218 if it exists otherwise it will fail with an error.
13219
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010013220 For each PEM file, haproxy also checks for the presence of file at the same
13221 path suffixed by ".sctl". If such file is found, support for Certificate
13222 Transparency (RFC6962) TLS extension is enabled. The file must contain a
13223 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
13224 to check basic syntax, but no signatures are verified.
13225
yanbzhu6c25e9e2016-01-05 12:52:02 -050013226 There are cases where it is desirable to support multiple key types, e.g. RSA
13227 and ECDSA in the cipher suites offered to the clients. This allows clients
13228 that support EC certificates to be able to use EC ciphers, while
13229 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050013230
William Lallemandf9ff3ec2020-10-02 17:57:44 +020013231 To achieve this, OpenSSL 1.1.1 is required, you can configure this behavior
13232 by providing one crt entry per certificate type, or by configuring a "cert
13233 bundle" like it was required before HAProxy 1.8. See "ssl-load-extra-files".
yanbzhud19630c2015-12-14 15:10:25 -050013234
Emeric Brunb6dc9342012-09-28 17:55:37 +020013235crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000013236 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010013237 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013238 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000013239 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020013240
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013241crt-list <file>
13242 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013243 designates a list of PEM file with an optional ssl configuration and a SNI
13244 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013245
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013246 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
13247
William Lallemand5d036392020-06-30 16:11:36 +020013248 sslbindconf supports "allow-0rtt", "alpn", "ca-file", "ca-verify-file",
13249 "ciphers", "ciphersuites", "crl-file", "curves", "ecdhe", "no-ca-names",
13250 "npn", "verify" configuration. With BoringSSL and Openssl >= 1.1.1
13251 "ssl-min-ver" and "ssl-max-ver" are also supported. It overrides the
13252 configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013253
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020013254 Wildcards are supported in the SNI filter. Negative filter are also supported,
Joao Moraise51fab02020-11-21 07:42:20 -030013255 useful in combination with a wildcard filter to exclude a particular SNI, or
13256 after the first certificate to exclude a pattern from its CN or Subject Alt
13257 Name (SAN). The certificates will be presented to clients who provide a valid
13258 TLS Server Name Indication field matching one of the SNI filters. If no SNI
13259 filter is specified, the CN and SAN are used. This directive may be specified
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020013260 multiple times. See the "crt" option for more information. The default
13261 certificate is still needed to meet OpenSSL expectations. If it is not used,
13262 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013263
William Lallemandf9ff3ec2020-10-02 17:57:44 +020013264 Multi-cert bundling (see "ssl-load-extra-files") is supported with crt-list,
13265 as long as only the base name is given in the crt-list. SNI filter will do
13266 the same work on all bundled certificates.
yanbzhud19630c2015-12-14 15:10:25 -050013267
William Lallemand7c26ed72020-06-03 17:34:48 +020013268 Empty lines as well as lines beginning with a hash ('#') will be ignored.
13269
Joao Moraisaa8fcc42020-11-24 08:24:30 -030013270 The first declared certificate of a bind line is used as the default
13271 certificate, either from crt or crt-list option, which haproxy should use in
13272 the TLS handshake if no other certificate matches. This certificate will also
13273 be used if the provided SNI matches its CN or SAN, even if a matching SNI
13274 filter is found on any crt-list. The SNI filter !* can be used after the first
13275 declared certificate to not include its CN and SAN in the SNI tree, so it will
13276 never match except if no other certificate matches. This way the first
13277 declared certificate act as a fallback.
Joao Moraise51fab02020-11-21 07:42:20 -030013278
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013279 crt-list file example:
Joao Moraise51fab02020-11-21 07:42:20 -030013280 cert1.pem !*
William Lallemand7c26ed72020-06-03 17:34:48 +020013281 # comment
Emmanuel Hocdet05942112017-02-20 16:11:50 +010013282 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013283 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010013284 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013285
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013286defer-accept
13287 Is an optional keyword which is supported only on certain Linux kernels. It
13288 states that a connection will only be accepted once some data arrive on it,
13289 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010013290 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013291 performance by ensuring that most of the request is already available when
13292 the connection is accepted. On the other hand, it will not be able to detect
13293 connections which don't talk. It is important to note that this option is
13294 broken in all kernels up to 2.6.31, as the connection is never accepted until
13295 the client talks. This can cause issues with front firewalls which would see
13296 an established connection while the proxy will only see it in SYN_RECV. This
13297 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
13298
William Lallemandf6975e92017-05-26 17:42:10 +020013299expose-fd listeners
13300 This option is only usable with the stats socket. It gives your stats socket
13301 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020013302 During a reload with the master-worker mode, the process is automatically
13303 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013304 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020013305
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013306force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013307 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013308 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013309 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013310 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013311
13312force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013313 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013314 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013315 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013316
13317force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013318 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013319 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013320 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013321
13322force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013323 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013324 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013325 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013326
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013327force-tlsv13
13328 This option enforces use of TLSv1.3 only on SSL connections instantiated from
13329 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013330 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013331
Christopher Faulet31af49d2015-06-09 17:29:50 +020013332generate-certificates
13333 This setting is only available when support for OpenSSL was built in. It
13334 enables the dynamic SSL certificates generation. A CA certificate and its
13335 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
13336 a transparent forward proxy, SSL requests generate errors because of a common
13337 name mismatch on the certificate presented to the client. With this option
13338 enabled, HAProxy will try to forge a certificate using the SNI hostname
13339 indicated by the client. This is done only if no certificate matches the SNI
13340 hostname (see 'crt-list'). If an error occurs, the default certificate is
13341 used, else the 'strict-sni' option is set.
13342 It can also be used when HAProxy is configured as a reverse proxy to ease the
13343 deployment of an architecture with many backends.
13344
13345 Creating a SSL certificate is an expensive operation, so a LRU cache is used
13346 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010013347 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020013348 certificate is used many times.
13349
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013350gid <gid>
13351 Sets the group of the UNIX sockets to the designated system gid. It can also
13352 be set by default in the global section's "unix-bind" statement. Note that
13353 some platforms simply ignore this. This setting is equivalent to the "group"
13354 setting except that the group ID is used instead of its name. This setting is
13355 ignored by non UNIX sockets.
13356
13357group <group>
13358 Sets the group of the UNIX sockets to the designated system group. It can
13359 also be set by default in the global section's "unix-bind" statement. Note
13360 that some platforms simply ignore this. This setting is equivalent to the
13361 "gid" setting except that the group name is used instead of its gid. This
13362 setting is ignored by non UNIX sockets.
13363
13364id <id>
13365 Fixes the socket ID. By default, socket IDs are automatically assigned, but
13366 sometimes it is more convenient to fix them to ease monitoring. This value
13367 must be strictly positive and unique within the listener/frontend. This
13368 option can only be used when defining only a single socket.
13369
13370interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010013371 Restricts the socket to a specific interface. When specified, only packets
13372 received from that particular interface are processed by the socket. This is
13373 currently only supported on Linux. The interface must be a primary system
13374 interface, not an aliased interface. It is also possible to bind multiple
13375 frontends to the same address if they are bound to different interfaces. Note
13376 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010013377 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
13378 uses the same interface as inbound traffic, and its associated routing table,
13379 even if there are explicit routes through different interfaces configured.
13380 This can prove useful to address asymmetric routing issues when the same
13381 client IP addresses need to be able to reach frontends hosted on different
13382 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013383
Willy Tarreauabb175f2012-09-24 12:43:26 +020013384level <level>
13385 This setting is used with the stats sockets only to restrict the nature of
13386 the commands that can be issued on the socket. It is ignored by other
13387 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013388 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020013389 read, and no change is allowed. It would make sense on systems where it
13390 is not easy to restrict access to the socket.
13391 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010013392 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020013393 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010013394 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020013395 all counters).
13396
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020013397severity-output <format>
13398 This setting is used with the stats sockets only to configure severity
13399 level output prepended to informational feedback messages. Severity
13400 level of messages can range between 0 and 7, conforming to syslog
13401 rfc5424. Valid and successful socket commands requesting data
13402 (i.e. "show map", "get acl foo" etc.) will never have a severity level
13403 prepended. It is ignored by other sockets. <format> can be one of :
13404 - "none" (default) no severity level is prepended to feedback messages.
13405 - "number" severity level is prepended as a number.
13406 - "string" severity level is prepended as a string following the
13407 rfc5424 convention.
13408
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013409maxconn <maxconn>
13410 Limits the sockets to this number of concurrent connections. Extraneous
13411 connections will remain in the system's backlog until a connection is
13412 released. If unspecified, the limit will be the same as the frontend's
13413 maxconn. Note that in case of port ranges or multiple addresses, the same
13414 value will be applied to each socket. This setting enables different
13415 limitations on expensive sockets, for instance SSL entries which may easily
13416 eat all memory.
13417
13418mode <mode>
13419 Sets the octal mode used to define access permissions on the UNIX socket. It
13420 can also be set by default in the global section's "unix-bind" statement.
13421 Note that some platforms simply ignore this. This setting is ignored by non
13422 UNIX sockets.
13423
13424mss <maxseg>
13425 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
13426 connections. This can be used to force a lower MSS for certain specific
13427 ports, for instance for connections passing through a VPN. Note that this
13428 relies on a kernel feature which is theoretically supported under Linux but
13429 was buggy in all versions prior to 2.6.28. It may or may not work on other
13430 operating systems. It may also not change the advertised value but change the
13431 effective size of outgoing segments. The commonly advertised value for TCPv4
13432 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
13433 positive, it will be used as the advertised MSS. If it is negative, it will
13434 indicate by how much to reduce the incoming connection's advertised MSS for
13435 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
13436
13437name <name>
13438 Sets an optional name for these sockets, which will be reported on the stats
13439 page.
13440
Willy Tarreaud72f0f32015-10-13 14:50:22 +020013441namespace <name>
13442 On Linux, it is possible to specify which network namespace a socket will
13443 belong to. This directive makes it possible to explicitly bind a listener to
13444 a namespace different from the default one. Please refer to your operating
13445 system's documentation to find more details about network namespaces.
13446
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013447nice <nice>
13448 Sets the 'niceness' of connections initiated from the socket. Value must be
13449 in the range -1024..1024 inclusive, and defaults to zero. Positive values
13450 means that such connections are more friendly to others and easily offer
13451 their place in the scheduler. On the opposite, negative values mean that
13452 connections want to run with a higher priority than others. The difference
13453 only happens under high loads when the system is close to saturation.
13454 Negative values are appropriate for low-latency or administration services,
13455 and high values are generally recommended for CPU intensive tasks such as SSL
13456 processing or bulk transfers which are less sensible to latency. For example,
13457 it may make sense to use a positive value for an SMTP socket and a negative
13458 one for an RDP socket.
13459
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020013460no-ca-names
13461 This setting is only available when support for OpenSSL was built in. It
13462 prevents from send CA names in server hello message when ca-file is used.
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010013463 Use "ca-verify-file" instead of "ca-file" with "no-ca-names".
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020013464
Emeric Brun9b3009b2012-10-05 11:55:06 +020013465no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013466 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013467 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013468 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013469 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013470 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
13471 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013472
Emeric Brun90ad8722012-10-02 14:00:59 +020013473no-tls-tickets
13474 This setting is only available when support for OpenSSL was built in. It
13475 disables the stateless session resumption (RFC 5077 TLS Ticket
13476 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013477 session resumption is more expensive in CPU usage. This option is also
13478 available on global statement "ssl-default-bind-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010013479 The TLS ticket mechanism is only used up to TLS 1.2.
13480 Forward Secrecy is compromised with TLS tickets, unless ticket keys
13481 are periodically rotated (via reload or by using "tls-ticket-keys").
Emeric Brun90ad8722012-10-02 14:00:59 +020013482
Emeric Brun9b3009b2012-10-05 11:55:06 +020013483no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013484 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013485 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013486 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013487 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013488 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13489 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013490
Emeric Brun9b3009b2012-10-05 11:55:06 +020013491no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020013492 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013493 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013494 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013495 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013496 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13497 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020013498
Emeric Brun9b3009b2012-10-05 11:55:06 +020013499no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020013500 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013501 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013502 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013503 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013504 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13505 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020013506
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013507no-tlsv13
13508 This setting is only available when support for OpenSSL was built in. It
13509 disables support for TLSv1.3 on any sockets instantiated from the listener
13510 when SSL is supported. Note that SSLv2 is forced disabled in the code and
13511 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013512 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13513 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013514
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020013515npn <protocols>
13516 This enables the NPN TLS extension and advertises the specified protocol list
13517 as supported on top of NPN. The protocol list consists in a comma-delimited
13518 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050013519 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020013520 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010013521 replaced with the ALPN extension (see the "alpn" keyword), though this one is
13522 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
13523 version of OpenSSL, NPN might still be used as most clients still support it
13524 at the time of writing this. It is possible to enable both NPN and ALPN
13525 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020013526
Lukas Tribus53ae85c2017-05-04 15:45:40 +000013527prefer-client-ciphers
13528 Use the client's preference when selecting the cipher suite, by default
13529 the server's preference is enforced. This option is also available on
13530 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020013531 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
13532 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
13533 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000013534
Christopher Fauletc644fa92017-11-23 22:44:11 +010013535process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010013536 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010013537 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010013538 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010013539 between the two is applied. If in the end the listener is not allowed to run
13540 on any remaining process, a warning is emitted, and the listener will either
13541 run on the first process of the listener if a single process was specified,
13542 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010013543 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010013544 connections for this listener, for the the process set. If multiple processes
13545 and threads are configured, a warning is emitted, as it either results from a
13546 configuration error or a misunderstanding of these models. For the unlikely
13547 case where several ranges are needed, this directive may be repeated.
13548 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010013549
13550 all | odd | even | number[-[number]]
13551
13552 Ranges can be partially defined. The higher bound can be omitted. In such
13553 case, it is replaced by the corresponding maximum value. The main purpose of
13554 this directive is to be used with the stats sockets and have one different
13555 socket per process. The second purpose is to have multiple bind lines sharing
13556 the same IP:port but not the same process in a listener, so that the system
13557 can distribute the incoming connections into multiple queues and allow a
13558 smoother inter-process load balancing. Currently Linux 3.9 and above is known
13559 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020013560
Christopher Fauleta717b992018-04-10 14:43:00 +020013561proto <name>
13562 Forces the multiplexer's protocol to use for the incoming connections. It
13563 must be compatible with the mode of the frontend (TCP or HTTP). It must also
13564 be usable on the frontend side. The list of available protocols is reported
Christopher Faulet982e17d2021-03-26 14:44:18 +010013565 in haproxy -vv. The protocols properties are reported : the mode (TCP/HTTP),
13566 the side (FE/BE), the mux name and its flags.
13567
13568 Some protocols report errors on aborts (flag=CLEAN_ABRT). Some others are
13569 subject to the head-of-line blocking on server side (flag=HOL_RISK). Finally
13570 some protocols don't support upgrades (flag=NO_UPG). The HTX compatibility is
13571 also reported (flag=HTX).
13572
13573 Here are the protocols that may be used as argument to a "proto" directive on
13574 a bind line :
13575
13576 h2 : mode=HTTP side=FE|BE mux=H2 flags=HTX|CLEAN_ABRT|HOL_RISK|NO_UPG
13577 h1 : mode=HTTP side=FE|BE mux=H1 flags=HTX|NO_UPG
13578 none : mode=TCP side=FE|BE mux=PASS flags=NO_UPG
13579
Daniel Corbett67a82712020-07-06 23:01:19 -040013580 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauleta717b992018-04-10 14:43:00 +020013581 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080013582 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020013583 h2" on the bind line.
13584
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013585ssl
13586 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013587 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013588 certificate is necessary (see "crt" above). All contents in the buffers will
13589 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020013590 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
13591 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013592
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013593ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
13594 This option enforces use of <version> or lower on SSL connections instantiated
William Lallemand50df1cb2020-06-02 10:52:24 +020013595 from this listener. Using this setting without "ssl-min-ver" can be
13596 ambiguous because the default ssl-min-ver value could change in future HAProxy
13597 versions. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013598 "ssl-default-bind-options". See also "ssl-min-ver".
13599
13600ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
William Lallemand50df1cb2020-06-02 10:52:24 +020013601 This option enforces use of <version> or upper on SSL connections
13602 instantiated from this listener. The default value is "TLSv1.2". This option
13603 is also available on global statement "ssl-default-bind-options".
13604 See also "ssl-max-ver".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013605
Emmanuel Hocdet65623372013-01-24 17:17:15 +010013606strict-sni
13607 This setting is only available when support for OpenSSL was built in. The
13608 SSL/TLS negotiation is allow only if the client provided an SNI which match
13609 a certificate. The default certificate is not used.
13610 See the "crt" option for more information.
13611
Willy Tarreau2af207a2015-02-04 00:45:58 +010013612tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010013613 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010013614 listening socket. This option is available on Linux since version 2.6.37. It
13615 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010013616 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010013617 useful on long-lived connections experiencing long idle periods such as
13618 remote terminals or database connection pools, where the client and server
13619 timeouts must remain high to allow a long period of idle, but where it is
13620 important to detect that the client has disappeared in order to release all
13621 resources associated with its connection (and the server's session). The
13622 argument is a delay expressed in milliseconds by default. This only works
13623 for regular TCP connections, and is ignored for other protocols.
13624
Willy Tarreau1c862c52012-10-05 16:21:00 +020013625tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010013626 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020013627 enables TCP Fast Open on the listening socket, which means that clients which
13628 support this feature will be able to send a request and receive a response
13629 during the 3-way handshake starting from second connection, thus saving one
13630 round-trip after the first connection. This only makes sense with protocols
13631 that use high connection rates and where each round trip matters. This can
13632 possibly cause issues with many firewalls which do not accept data on SYN
13633 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020013634 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
13635 need to build HAProxy with USE_TFO=1 if your libc doesn't define
13636 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020013637
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010013638tls-ticket-keys <keyfile>
13639 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010013640 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
13641 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
13642 The first key determines the key length used for next keys: you can't mix
13643 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
13644 build option (default 3) and at least as many keys need to be present in
13645 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
13646 penultimate one for encryption. This enables easy key rotation by just
13647 appending new key to the file and reloading the process. Keys must be
13648 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
13649 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010013650 storage such as hard drives (hint: use tmpfs and don't swap those files).
13651 Lifetime hint can be changed using tune.ssl.timeout.
13652
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013653transparent
13654 Is an optional keyword which is supported only on certain Linux kernels. It
13655 indicates that the addresses will be bound even if they do not belong to the
13656 local machine, and that packets targeting any of these addresses will be
13657 intercepted just as if the addresses were locally configured. This normally
13658 requires that IP forwarding is enabled. Caution! do not use this with the
13659 default address '*', as it would redirect any traffic for the specified port.
13660 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
13661 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
13662 kernel version. Some distribution kernels include backports of the feature,
13663 so check for support with your vendor.
13664
Willy Tarreau77e3af92012-11-24 15:07:23 +010013665v4v6
13666 Is an optional keyword which is supported only on most recent systems
13667 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
13668 and IPv6 when it uses the default address. Doing so is sometimes necessary
13669 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013670 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010013671
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013672v6only
13673 Is an optional keyword which is supported only on most recent systems
13674 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
13675 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010013676 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
13677 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013678
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013679uid <uid>
13680 Sets the owner of the UNIX sockets to the designated system uid. It can also
13681 be set by default in the global section's "unix-bind" statement. Note that
13682 some platforms simply ignore this. This setting is equivalent to the "user"
13683 setting except that the user numeric ID is used instead of its name. This
13684 setting is ignored by non UNIX sockets.
13685
13686user <user>
13687 Sets the owner of the UNIX sockets to the designated system user. It can also
13688 be set by default in the global section's "unix-bind" statement. Note that
13689 some platforms simply ignore this. This setting is equivalent to the "uid"
13690 setting except that the user name is used instead of its uid. This setting is
13691 ignored by non UNIX sockets.
13692
Emeric Brun1a073b42012-09-28 17:07:34 +020013693verify [none|optional|required]
13694 This setting is only available when support for OpenSSL was built in. If set
13695 to 'none', client certificate is not requested. This is the default. In other
13696 cases, a client certificate is requested. If the client does not provide a
13697 certificate after the request and if 'verify' is set to 'required', then the
13698 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020013699 certificate provided by the client is always verified using CAs from
13700 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
13701 is aborted, regardless of the 'verify' option, unless the error code exactly
13702 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013703
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200137045.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010013705------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013706
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013707The "server" and "default-server" keywords support a certain number of settings
13708which are all passed as arguments on the server line. The order in which those
13709arguments appear does not count, and they are all optional. Some of those
13710settings are single words (booleans) while others expect one or several values
13711after them. In this case, the values must immediately follow the setting name.
13712Except default-server, all those settings must be specified after the server's
13713address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013714
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013715 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013716 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020013717
Frédéric Lécailled2376272017-03-21 18:52:12 +010013718Note that all these settings are supported both by "server" and "default-server"
13719keywords, except "id" which is only supported by "server".
13720
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013721The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013722
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013723addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013724 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010013725 to send health-checks or to probe the agent-check. On some servers, it may be
13726 desirable to dedicate an IP address to specific component able to perform
13727 complex tests which are more suitable to health-checks than the application.
13728 This parameter is ignored if the "check" parameter is not set. See also the
13729 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013730
Simon Hormand60d6912013-11-25 10:46:36 +090013731agent-check
13732 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010013733 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010013734 to the port set by the "agent-port" parameter and reading an ASCII string
13735 terminated by the first '\r' or '\n' met. The string is made of a series of
13736 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090013737
Willy Tarreau81f5d942013-12-09 20:51:51 +010013738 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090013739 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020013740 weight of a server as configured when haproxy starts. Note that a zero
13741 weight is reported on the stats page as "DRAIN" since it has the same
13742 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090013743
Davor Ocelice9ed2812017-12-25 17:49:28 +010013744 - The string "maxconn:" followed by an integer (no space between). Values
13745 in this format will set the maxconn of a server. The maximum number of
13746 connections advertised needs to be multiplied by the number of load
13747 balancers and different backends that use this health check to get the
13748 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020013749
Willy Tarreau81f5d942013-12-09 20:51:51 +010013750 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013751 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090013752
Willy Tarreau81f5d942013-12-09 20:51:51 +010013753 - The word "drain". This will turn the server's administrative state to the
13754 DRAIN mode, thus it will not accept any new connections other than those
13755 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090013756
Willy Tarreau81f5d942013-12-09 20:51:51 +010013757 - The word "maint". This will turn the server's administrative state to the
13758 MAINT mode, thus it will not accept any new connections at all, and health
13759 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090013760
William Dauchyf8e795c2020-09-26 13:35:51 +020013761 - The words "down", "fail", or "stopped", optionally followed by a
Willy Tarreau81f5d942013-12-09 20:51:51 +010013762 description string after a sharp ('#'). All of these mark the server's
13763 operating state as DOWN, but since the word itself is reported on the stats
13764 page, the difference allows an administrator to know if the situation was
13765 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010013766 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010013767 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090013768
Willy Tarreau81f5d942013-12-09 20:51:51 +010013769 - The word "up" sets back the server's operating state as UP if health checks
13770 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090013771
Willy Tarreau81f5d942013-12-09 20:51:51 +010013772 Parameters which are not advertised by the agent are not changed. For
13773 example, an agent might be designed to monitor CPU usage and only report a
13774 relative weight and never interact with the operating status. Similarly, an
13775 agent could be designed as an end-user interface with 3 radio buttons
13776 allowing an administrator to change only the administrative state. However,
13777 it is important to consider that only the agent may revert its own actions,
13778 so if a server is set to DRAIN mode or to DOWN state using the agent, the
13779 agent must implement the other equivalent actions to bring the service into
13780 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090013781
Simon Horman2f1f9552013-11-25 10:46:37 +090013782 Failure to connect to the agent is not considered an error as connectivity
13783 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010013784 parameter. Warning though, it is not a good idea to stop an agent after it
13785 reports "down", since only an agent reporting "up" will be able to turn the
13786 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010013787 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090013788
Willy Tarreau81f5d942013-12-09 20:51:51 +010013789 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010013790 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090013791
James Brown55f9ff12015-10-21 18:19:05 -070013792agent-send <string>
13793 If this option is specified, haproxy will send the given string (verbatim)
13794 to the agent server upon connection. You could, for example, encode
13795 the backend name into this string, which would enable your agent to send
13796 different responses based on the backend. Make sure to include a '\n' if
13797 you want to terminate your request with a newline.
13798
Simon Hormand60d6912013-11-25 10:46:36 +090013799agent-inter <delay>
13800 The "agent-inter" parameter sets the interval between two agent checks
13801 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
13802
13803 Just as with every other time-based parameter, it may be entered in any
13804 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
13805 parameter also serves as a timeout for agent checks "timeout check" is
13806 not set. In order to reduce "resonance" effects when multiple servers are
13807 hosted on the same hardware, the agent and health checks of all servers
13808 are started with a small time offset between them. It is also possible to
13809 add some random noise in the agent and health checks interval using the
13810 global "spread-checks" keyword. This makes sense for instance when a lot
13811 of backends use the same servers.
13812
13813 See also the "agent-check" and "agent-port" parameters.
13814
Misiek768d8602017-01-09 09:52:43 +010013815agent-addr <addr>
13816 The "agent-addr" parameter sets address for agent check.
13817
13818 You can offload agent-check to another target, so you can make single place
13819 managing status and weights of servers defined in haproxy in case you can't
13820 make self-aware and self-managing services. You can specify both IP or
13821 hostname, it will be resolved.
13822
Simon Hormand60d6912013-11-25 10:46:36 +090013823agent-port <port>
13824 The "agent-port" parameter sets the TCP port used for agent checks.
13825
13826 See also the "agent-check" and "agent-inter" parameters.
13827
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013828allow-0rtt
13829 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020013830 Note that early data will be sent only if the client used early data, or
13831 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013832
Olivier Houchardc7566002018-11-20 23:33:50 +010013833alpn <protocols>
13834 This enables the TLS ALPN extension and advertises the specified protocol
13835 list as supported on top of ALPN. The protocol list consists in a comma-
13836 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050013837 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010013838 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
13839 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
13840 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
13841 now obsolete NPN extension.
13842 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
13843 be advertised, in order of preference, like below :
13844
13845 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
13846
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013847backup
13848 When "backup" is present on a server line, the server is only used in load
13849 balancing when all other non-backup servers are unavailable. Requests coming
13850 with a persistence cookie referencing the server will always be served
13851 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010013852 the "allbackups" option is set in the backend. See also the "no-backup" and
13853 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013854
Emeric Brunef42d922012-10-11 16:11:36 +020013855ca-file <cafile>
13856 This setting is only available when support for OpenSSL was built in. It
13857 designates a PEM file from which to load CA certificates used to verify
13858 server's certificate.
13859
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013860check
Jerome Magnin90702bc2020-04-26 14:23:04 +020013861 This option enables health checks on a server:
13862 - when not set, no health checking is performed, and the server is always
13863 considered available.
13864 - when set and no other check method is configured, the server is considered
13865 available when a connection can be established at the highest configured
13866 transport layer. This means TCP by default, or SSL/TLS when "ssl" or
13867 "check-ssl" are set, both possibly combined with connection prefixes such
13868 as a PROXY protocol header when "send-proxy" or "check-send-proxy" are
13869 set.
13870 - when set and an application-level health check is defined, the
13871 application-level exchanges are performed on top of the configured
13872 transport layer and the server is considered available if all of the
13873 exchanges succeed.
13874
13875 By default, health checks are performed on the same address and port as
13876 configured on the server, using the same encapsulation parameters (SSL/TLS,
13877 proxy-protocol header, etc... ). It is possible to change the destination
13878 address using "addr" and the port using "port". When done, it is assumed the
13879 server isn't checked on the service port, and configured encapsulation
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +050013880 parameters are not reused. One must explicitly set "check-send-proxy" to send
Jerome Magnin90702bc2020-04-26 14:23:04 +020013881 connection headers, "check-ssl" to use SSL/TLS.
13882
13883 When "sni" or "alpn" are set on the server line, their value is not used for
13884 health checks and one must use "check-sni" or "check-alpn".
13885
13886 The default source address for health check traffic is the same as the one
13887 defined in the backend. It can be changed with the "source" keyword.
13888
13889 The interval between checks can be set using the "inter" keyword, and the
13890 "rise" and "fall" keywords can be used to define how many successful or
13891 failed health checks are required to flag a server available or not
13892 available.
13893
13894 Optional application-level health checks can be configured with "option
13895 httpchk", "option mysql-check" "option smtpchk", "option pgsql-check",
13896 "option ldap-check", or "option redis-check".
13897
13898 Example:
13899 # simple tcp check
13900 backend foo
13901 server s1 192.168.0.1:80 check
13902 # this does a tcp connect + tls handshake
13903 backend foo
13904 server s1 192.168.0.1:443 ssl check
13905 # simple tcp check is enough for check success
13906 backend foo
13907 option tcp-check
13908 tcp-check connect
13909 server s1 192.168.0.1:443 ssl check
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013910
Willy Tarreau6c16adc2012-10-05 00:04:16 +020013911check-send-proxy
13912 This option forces emission of a PROXY protocol line with outgoing health
13913 checks, regardless of whether the server uses send-proxy or not for the
13914 normal traffic. By default, the PROXY protocol is enabled for health checks
13915 if it is already enabled for normal traffic and if no "port" nor "addr"
13916 directive is present. However, if such a directive is present, the
13917 "check-send-proxy" option needs to be used to force the use of the
13918 protocol. See also the "send-proxy" option for more information.
13919
Olivier Houchard92150142018-12-21 19:47:01 +010013920check-alpn <protocols>
13921 Defines which protocols to advertise with ALPN. The protocol list consists in
13922 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
13923 (without quotes). If it is not set, the server ALPN is used.
13924
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013925check-proto <name>
13926 Forces the multiplexer's protocol to use for the server's health-check
13927 connections. It must be compatible with the health-check type (TCP or
13928 HTTP). It must also be usable on the backend side. The list of available
Christopher Faulet982e17d2021-03-26 14:44:18 +010013929 protocols is reported in haproxy -vv. The protocols properties are
13930 reported : the mode (TCP/HTTP), the side (FE/BE), the mux name and its flags.
13931
13932 Some protocols report errors on aborts (flag=CLEAN_ABRT). Some others are
13933 subject to the head-of-line blocking on server side (flag=HOL_RISK). Finally
13934 some protocols don't support upgrades (flag=NO_UPG). The HTX compatibility is
13935 also reported (flag=HTX).
13936
13937 Here are the protocols that may be used as argument to a "check-proto"
13938 directive on a server line:
13939
13940 h2 : mode=HTTP side=FE|BE mux=H2 flags=HTX|CLEAN_ABRT|HOL_RISK|NO_UPG
13941 fcgi : mode=HTTP side=BE mux=FCGI flags=HTX|HOL_RISK|NO_UPG
13942 h1 : mode=HTTP side=FE|BE mux=H1 flags=HTX|NO_UPG
13943 none : mode=TCP side=FE|BE mux=PASS flags=NO_UPG
13944
Daniel Corbett67a82712020-07-06 23:01:19 -040013945 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013946 protocol for health-check connections established to this server.
13947 If not defined, the server one will be used, if set.
13948
Jérôme Magninae9bb762018-12-09 16:08:26 +010013949check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020013950 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010013951 over SSL. It is only possible to use a string to set <sni>. If you want to
13952 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020013953
Willy Tarreau763a95b2012-10-04 23:15:39 +020013954check-ssl
13955 This option forces encryption of all health checks over SSL, regardless of
13956 whether the server uses SSL or not for the normal traffic. This is generally
13957 used when an explicit "port" or "addr" directive is specified and SSL health
13958 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013959 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020013960 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
13961 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013962 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010013963 See the "ssl" option for more information and "no-check-ssl" to disable
13964 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013965
Alexander Liu2a54bb72019-05-22 19:44:48 +080013966check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050013967 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080013968 default, the health checks won't go through socks tunnel even it was enabled
13969 for normal traffic.
13970
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013971ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013972 This setting is only available when support for OpenSSL was built in. This
13973 option sets the string describing the list of cipher algorithms that is
13974 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013975 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
13976 information and recommendations see e.g.
13977 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
13978 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
13979 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013980
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013981ciphersuites <ciphersuites>
13982 This setting is only available when support for OpenSSL was built in and
13983 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
13984 describing the list of cipher algorithms that is negotiated during the TLS
13985 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013986 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
13987 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
13988 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013989
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013990cookie <value>
13991 The "cookie" parameter sets the cookie value assigned to the server to
13992 <value>. This value will be checked in incoming requests, and the first
13993 operational server possessing the same value will be selected. In return, in
13994 cookie insertion or rewrite modes, this value will be assigned to the cookie
13995 sent to the client. There is nothing wrong in having several servers sharing
13996 the same cookie value, and it is in fact somewhat common between normal and
13997 backup servers. See also the "cookie" keyword in backend section.
13998
Emeric Brunef42d922012-10-11 16:11:36 +020013999crl-file <crlfile>
14000 This setting is only available when support for OpenSSL was built in. It
14001 designates a PEM file from which to load certificate revocation list used
14002 to verify server's certificate.
14003
Emeric Bruna7aa3092012-10-26 12:58:00 +020014004crt <cert>
14005 This setting is only available when support for OpenSSL was built in.
14006 It designates a PEM file from which to load both a certificate and the
14007 associated private key. This file can be built by concatenating both PEM
14008 files into one. This certificate will be sent if the server send a client
14009 certificate request.
14010
Willy Tarreau96839092010-03-29 10:02:24 +020014011disabled
14012 The "disabled" keyword starts the server in the "disabled" state. That means
14013 that it is marked down in maintenance mode, and no connection other than the
14014 ones allowed by persist mode will reach it. It is very well suited to setup
14015 new servers, because normal traffic will never reach them, while it is still
14016 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014017 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020014018
Frédéric Lécailled2376272017-03-21 18:52:12 +010014019enabled
14020 This option may be used as 'server' setting to reset any 'disabled'
14021 setting which would have been inherited from 'default-server' directive as
14022 default value.
14023 It may also be used as 'default-server' setting to reset any previous
14024 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020014025
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014026error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010014027 If health observing is enabled, the "error-limit" parameter specifies the
14028 number of consecutive errors that triggers event selected by the "on-error"
14029 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014030
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014031 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014032
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014033fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014034 The "fall" parameter states that a server will be considered as dead after
14035 <count> consecutive unsuccessful health checks. This value defaults to 3 if
14036 unspecified. See also the "check", "inter" and "rise" parameters.
14037
Emeric Brun8694b9a2012-10-05 14:39:07 +020014038force-sslv3
14039 This option enforces use of SSLv3 only when SSL is used to communicate with
14040 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014041 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014042 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020014043
14044force-tlsv10
14045 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014046 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014047 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020014048
14049force-tlsv11
14050 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014051 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014052 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020014053
14054force-tlsv12
14055 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014056 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014057 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020014058
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020014059force-tlsv13
14060 This option enforces use of TLSv1.3 only when SSL is used to communicate with
14061 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014062 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020014063
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014064id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020014065 Set a persistent ID for the server. This ID must be positive and unique for
14066 the proxy. An unused ID will automatically be assigned if unset. The first
14067 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014068
Willy Tarreau6a031d12016-11-07 19:42:35 +010014069init-addr {last | libc | none | <ip>},[...]*
14070 Indicate in what order the server's address should be resolved upon startup
14071 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014072 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010014073 method which succeeds is used. If the end of the list is reached without
14074 finding a working method, an error is thrown. Method "last" suggests to pick
14075 the address which appears in the state file (see "server-state-file"). Method
14076 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
14077 depending on the operating system and build options). Method "none"
14078 specifically indicates that the server should start without any valid IP
14079 address in a down state. It can be useful to ignore some DNS issues upon
14080 startup, waiting for the situation to get fixed later. Finally, an IP address
14081 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014082 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010014083 server used to catch old sessions and present them with a decent error
14084 message for example. When the "first" load balancing algorithm is used, this
14085 IP address could point to a fake server used to trigger the creation of new
14086 instances on the fly. This option defaults to "last,libc" indicating that the
14087 previous address found in the state file (if any) is used first, otherwise
14088 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014089 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010014090
14091 Example:
14092 defaults
14093 # never fail on address resolution
14094 default-server init-addr last,libc,none
14095
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014096inter <delay>
14097fastinter <delay>
14098downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014099 The "inter" parameter sets the interval between two consecutive health checks
14100 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
14101 It is also possible to use "fastinter" and "downinter" to optimize delays
14102 between checks depending on the server state :
14103
Pieter Baauw44fc9df2015-09-17 21:30:46 +020014104 Server state | Interval used
14105 ----------------------------------------+----------------------------------
14106 UP 100% (non-transitional) | "inter"
14107 ----------------------------------------+----------------------------------
14108 Transitionally UP (going down "fall"), | "fastinter" if set,
14109 Transitionally DOWN (going up "rise"), | "inter" otherwise.
14110 or yet unchecked. |
14111 ----------------------------------------+----------------------------------
14112 DOWN 100% (non-transitional) | "downinter" if set,
14113 | "inter" otherwise.
14114 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010014115
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014116 Just as with every other time-based parameter, they can be entered in any
14117 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
14118 serves as a timeout for health checks sent to servers if "timeout check" is
14119 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090014120 hosted on the same hardware, the agent and health checks of all servers
14121 are started with a small time offset between them. It is also possible to
14122 add some random noise in the agent and health checks interval using the
14123 global "spread-checks" keyword. This makes sense for instance when a lot
14124 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014125
Emeric Brun97556472020-05-30 01:42:45 +020014126log-proto <logproto>
14127 The "log-proto" specifies the protocol used to forward event messages to
14128 a server configured in a ring section. Possible values are "legacy"
14129 and "octet-count" corresponding respectively to "Non-transparent-framing"
14130 and "Octet counting" in rfc6587. "legacy" is the default.
14131
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014132maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014133 The "maxconn" parameter specifies the maximal number of concurrent
14134 connections that will be sent to this server. If the number of incoming
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010014135 concurrent connections goes higher than this value, they will be queued,
14136 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014137 save fragile servers from going down under extreme loads. If a "minconn"
14138 parameter is specified, the limit becomes dynamic. The default value is "0"
14139 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
14140 the backend's "fullconn" keyword.
14141
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010014142 In HTTP mode this parameter limits the number of concurrent requests instead
14143 of the number of connections. Multiple requests might be multiplexed over a
14144 single TCP connection to the server. As an example if you specify a maxconn
14145 of 50 you might see between 1 and 50 actual server connections, but no more
14146 than 50 concurrent requests.
14147
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014148maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014149 The "maxqueue" parameter specifies the maximal number of connections which
14150 will wait in the queue for this server. If this limit is reached, next
14151 requests will be redispatched to other servers instead of indefinitely
14152 waiting to be served. This will break persistence but may allow people to
Willy Tarreau8ae8c482020-10-22 17:19:07 +020014153 quickly re-log in when the server they try to connect to is dying. Some load
14154 balancing algorithms such as leastconn take this into account and accept to
14155 add requests into a server's queue up to this value if it is explicitly set
14156 to a value greater than zero, which often allows to better smooth the load
14157 when dealing with single-digit maxconn values. The default value is "0" which
14158 means the queue is unlimited. See also the "maxconn" and "minconn" parameters
14159 and "balance leastconn".
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014160
Willy Tarreau9c538e02019-01-23 10:21:49 +010014161max-reuse <count>
14162 The "max-reuse" argument indicates the HTTP connection processors that they
14163 should not reuse a server connection more than this number of times to send
14164 new requests. Permitted values are -1 (the default), which disables this
14165 limit, or any positive value. Value zero will effectively disable keep-alive.
14166 This is only used to work around certain server bugs which cause them to leak
14167 resources over time. The argument is not necessarily respected by the lower
14168 layers as there might be technical limitations making it impossible to
14169 enforce. At least HTTP/2 connections to servers will respect it.
14170
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014171minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014172 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
14173 limit following the backend's load. The server will always accept at least
14174 <minconn> connections, never more than <maxconn>, and the limit will be on
14175 the ramp between both values when the backend has less than <fullconn>
14176 concurrent connections. This makes it possible to limit the load on the
14177 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010014178 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014179 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014180
Willy Tarreaud72f0f32015-10-13 14:50:22 +020014181namespace <name>
14182 On Linux, it is possible to specify which network namespace a socket will
14183 belong to. This directive makes it possible to explicitly bind a server to
14184 a namespace different from the default one. Please refer to your operating
14185 system's documentation to find more details about network namespaces.
14186
Frédéric Lécailled2376272017-03-21 18:52:12 +010014187no-agent-check
14188 This option may be used as "server" setting to reset any "agent-check"
14189 setting which would have been inherited from "default-server" directive as
14190 default value.
14191 It may also be used as "default-server" setting to reset any previous
14192 "default-server" "agent-check" setting.
14193
14194no-backup
14195 This option may be used as "server" setting to reset any "backup"
14196 setting which would have been inherited from "default-server" directive as
14197 default value.
14198 It may also be used as "default-server" setting to reset any previous
14199 "default-server" "backup" setting.
14200
14201no-check
14202 This option may be used as "server" setting to reset any "check"
14203 setting which would have been inherited from "default-server" directive as
14204 default value.
14205 It may also be used as "default-server" setting to reset any previous
14206 "default-server" "check" setting.
14207
14208no-check-ssl
14209 This option may be used as "server" setting to reset any "check-ssl"
14210 setting which would have been inherited from "default-server" directive as
14211 default value.
14212 It may also be used as "default-server" setting to reset any previous
14213 "default-server" "check-ssl" setting.
14214
Frédéric Lécailled2376272017-03-21 18:52:12 +010014215no-send-proxy
14216 This option may be used as "server" setting to reset any "send-proxy"
14217 setting which would have been inherited from "default-server" directive as
14218 default value.
14219 It may also be used as "default-server" setting to reset any previous
14220 "default-server" "send-proxy" setting.
14221
14222no-send-proxy-v2
14223 This option may be used as "server" setting to reset any "send-proxy-v2"
14224 setting which would have been inherited from "default-server" directive as
14225 default value.
14226 It may also be used as "default-server" setting to reset any previous
14227 "default-server" "send-proxy-v2" setting.
14228
14229no-send-proxy-v2-ssl
14230 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
14231 setting which would have been inherited from "default-server" directive as
14232 default value.
14233 It may also be used as "default-server" setting to reset any previous
14234 "default-server" "send-proxy-v2-ssl" setting.
14235
14236no-send-proxy-v2-ssl-cn
14237 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
14238 setting which would have been inherited from "default-server" directive as
14239 default value.
14240 It may also be used as "default-server" setting to reset any previous
14241 "default-server" "send-proxy-v2-ssl-cn" setting.
14242
14243no-ssl
14244 This option may be used as "server" setting to reset any "ssl"
14245 setting which would have been inherited from "default-server" directive as
14246 default value.
14247 It may also be used as "default-server" setting to reset any previous
14248 "default-server" "ssl" setting.
14249
William Dauchyf6370442020-11-14 19:25:33 +010014250 Note that using `default-server ssl` setting and `no-ssl` on server will
14251 however init SSL connection, so it can be later be enabled through the
14252 runtime API: see `set server` commands in management doc.
14253
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010014254no-ssl-reuse
14255 This option disables SSL session reuse when SSL is used to communicate with
14256 the server. It will force the server to perform a full handshake for every
14257 new connection. It's probably only useful for benchmarking, troubleshooting,
14258 and for paranoid users.
14259
Emeric Brun9b3009b2012-10-05 11:55:06 +020014260no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014261 This option disables support for SSLv3 when SSL is used to communicate with
14262 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014263 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014264
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014265 Supported in default-server: No
14266
Emeric Brunf9c5c472012-10-11 15:28:34 +020014267no-tls-tickets
14268 This setting is only available when support for OpenSSL was built in. It
14269 disables the stateless session resumption (RFC 5077 TLS Ticket
14270 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014271 session resumption is more expensive in CPU usage for servers. This option
14272 is also available on global statement "ssl-default-server-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010014273 The TLS ticket mechanism is only used up to TLS 1.2.
14274 Forward Secrecy is compromised with TLS tickets, unless ticket keys
14275 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010014276 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020014277
Emeric Brun9b3009b2012-10-05 11:55:06 +020014278no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020014279 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020014280 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14281 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014282 often makes sense to disable it when communicating with local servers. This
14283 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014284 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014285
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014286 Supported in default-server: No
14287
Emeric Brun9b3009b2012-10-05 11:55:06 +020014288no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020014289 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020014290 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14291 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014292 often makes sense to disable it when communicating with local servers. This
14293 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014294 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014295
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014296 Supported in default-server: No
14297
Emeric Brun9b3009b2012-10-05 11:55:06 +020014298no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020014299 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014300 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14301 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014302 often makes sense to disable it when communicating with local servers. This
14303 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014304 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020014305
14306 Supported in default-server: No
14307
14308no-tlsv13
14309 This option disables support for TLSv1.3 when SSL is used to communicate with
14310 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14311 using any configuration option. TLSv1 is more expensive than SSLv3 so it
14312 often makes sense to disable it when communicating with local servers. This
14313 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014314 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014315
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014316 Supported in default-server: No
14317
Frédéric Lécailled2376272017-03-21 18:52:12 +010014318no-verifyhost
14319 This option may be used as "server" setting to reset any "verifyhost"
14320 setting which would have been inherited from "default-server" directive as
14321 default value.
14322 It may also be used as "default-server" setting to reset any previous
14323 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014324
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020014325no-tfo
14326 This option may be used as "server" setting to reset any "tfo"
14327 setting which would have been inherited from "default-server" directive as
14328 default value.
14329 It may also be used as "default-server" setting to reset any previous
14330 "default-server" "tfo" setting.
14331
Simon Hormanfa461682011-06-25 09:39:49 +090014332non-stick
14333 Never add connections allocated to this sever to a stick-table.
14334 This may be used in conjunction with backup to ensure that
14335 stick-table persistence is disabled for backup servers.
14336
Olivier Houchardc7566002018-11-20 23:33:50 +010014337npn <protocols>
14338 This enables the NPN TLS extension and advertises the specified protocol list
14339 as supported on top of NPN. The protocol list consists in a comma-delimited
14340 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050014341 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010014342 enabled (check with haproxy -vv). Note that the NPN extension has been
14343 replaced with the ALPN extension (see the "alpn" keyword), though this one is
14344 only available starting with OpenSSL 1.0.2.
14345
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014346observe <mode>
14347 This option enables health adjusting based on observing communication with
14348 the server. By default this functionality is disabled and enabling it also
14349 requires to enable health checks. There are two supported modes: "layer4" and
14350 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
14351 significant. In layer7, which is only allowed for http proxies, responses
14352 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010014353 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014354
14355 See also the "check", "on-error" and "error-limit".
14356
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014357on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014358 Select what should happen when enough consecutive errors are detected.
14359 Currently, four modes are available:
14360 - fastinter: force fastinter
14361 - fail-check: simulate a failed check, also forces fastinter (default)
14362 - sudden-death: simulate a pre-fatal failed health check, one more failed
14363 check will mark a server down, forces fastinter
14364 - mark-down: mark the server immediately down and force fastinter
14365
14366 See also the "check", "observe" and "error-limit".
14367
Simon Hormane0d1bfb2011-06-21 14:34:58 +090014368on-marked-down <action>
14369 Modify what occurs when a server is marked down.
14370 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014371 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
14372 all connections to the server are immediately terminated when the server
14373 goes down. It might be used if the health check detects more complex cases
14374 than a simple connection status, and long timeouts would cause the service
14375 to remain unresponsive for too long a time. For instance, a health check
14376 might detect that a database is stuck and that there's no chance to reuse
14377 existing connections anymore. Connections killed this way are logged with
14378 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090014379
14380 Actions are disabled by default
14381
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014382on-marked-up <action>
14383 Modify what occurs when a server is marked up.
14384 Currently one action is available:
14385 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
14386 done only if the server is not in backup state and if it is not disabled
14387 (it must have an effective weight > 0). This can be used sometimes to force
14388 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014389 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
14390 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014391 with extreme care. Sessions killed because a server comes up are logged
14392 with an 'U' termination code (for "Up").
14393
14394 Actions are disabled by default
14395
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020014396pool-low-conn <max>
14397 Set a low threshold on the number of idling connections for a server, below
14398 which a thread will not try to steal a connection from another thread. This
14399 can be useful to improve CPU usage patterns in scenarios involving many very
14400 fast servers, in order to ensure all threads will keep a few idle connections
14401 all the time instead of letting them accumulate over one thread and migrating
14402 them from thread to thread. Typical values of twice the number of threads
14403 seem to show very good performance already with sub-millisecond response
14404 times. The default is zero, indicating that any idle connection can be used
14405 at any time. It is the recommended setting for normal use. This only applies
14406 to connections that can be shared according to the same principles as those
Willy Tarreau0784db82021-02-19 11:45:22 +010014407 applying to "http-reuse". In case connection sharing between threads would
14408 be disabled via "tune.idle-pool.shared", it can become very important to use
14409 this setting to make sure each thread always has a few connections, or the
14410 connection reuse rate will decrease as thread count increases.
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020014411
Olivier Houchard006e3102018-12-10 18:30:32 +010014412pool-max-conn <max>
14413 Set the maximum number of idling connections for a server. -1 means unlimited
14414 connections, 0 means no idle connections. The default is -1. When idle
14415 connections are enabled, orphaned idle connections which do not belong to any
14416 client session anymore are moved to a dedicated pool so that they remain
14417 usable by future clients. This only applies to connections that can be shared
14418 according to the same principles as those applying to "http-reuse".
14419
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010014420pool-purge-delay <delay>
14421 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010014422 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020014423 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010014424
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014425port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014426 Using the "port" parameter, it becomes possible to use a different port to
William Dauchy4858fb22021-02-03 22:30:09 +010014427 send health-checks or to probe the agent-check. On some servers, it may be
14428 desirable to dedicate a port to a specific component able to perform complex
14429 tests which are more suitable to health-checks than the application. It is
14430 common to run a simple script in inetd for instance. This parameter is
14431 ignored if the "check" parameter is not set. See also the "addr" parameter.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014432
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014433proto <name>
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014434 Forces the multiplexer's protocol to use for the outgoing connections to this
14435 server. It must be compatible with the mode of the backend (TCP or HTTP). It
14436 must also be usable on the backend side. The list of available protocols is
Christopher Faulet982e17d2021-03-26 14:44:18 +010014437 reported in haproxy -vv.The protocols properties are reported : the mode
14438 (TCP/HTTP), the side (FE/BE), the mux name and its flags.
14439
14440 Some protocols report errors on aborts (flag=CLEAN_ABRT). Some others are
14441 subject to the head-of-line blocking on server side (flag=HOL_RISK). Finally
14442 some protocols don't support upgrades (flag=NO_UPG). The HTX compatibility is
14443 also reported (flag=HTX).
14444
14445 Here are the protocols that may be used as argument to a "proto" directive on
14446 a server line :
14447
14448 h2 : mode=HTTP side=FE|BE mux=H2 flags=HTX|CLEAN_ABRT|HOL_RISK|NO_UPG
14449 fcgi : mode=HTTP side=BE mux=FCGI flags=HTX|HOL_RISK|NO_UPG
14450 h1 : mode=HTTP side=FE|BE mux=H1 flags=HTX|NO_UPG
14451 none : mode=TCP side=FE|BE mux=PASS flags=NO_UPG
14452
Daniel Corbett67a82712020-07-06 23:01:19 -040014453 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014454 protocol for all connections established to this server.
14455
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014456redir <prefix>
14457 The "redir" parameter enables the redirection mode for all GET and HEAD
14458 requests addressing this server. This means that instead of having HAProxy
14459 forward the request to the server, it will send an "HTTP 302" response with
14460 the "Location" header composed of this prefix immediately followed by the
14461 requested URI beginning at the leading '/' of the path component. That means
14462 that no trailing slash should be used after <prefix>. All invalid requests
14463 will be rejected, and all non-GET or HEAD requests will be normally served by
14464 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010014465 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014466 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014467 users to a remote location in case of local disaster. Main use consists in
14468 increasing bandwidth for static servers by having the clients directly
14469 connect to them. Note: never use a relative location here, it would cause a
14470 loop between the client and HAProxy!
14471
14472 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
14473
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014474rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014475 The "rise" parameter states that a server will be considered as operational
14476 after <count> consecutive successful health checks. This value defaults to 2
14477 if unspecified. See also the "check", "inter" and "fall" parameters.
14478
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020014479resolve-opts <option>,<option>,...
14480 Comma separated list of options to apply to DNS resolution linked to this
14481 server.
14482
14483 Available options:
14484
14485 * allow-dup-ip
14486 By default, HAProxy prevents IP address duplication in a backend when DNS
14487 resolution at runtime is in operation.
14488 That said, for some cases, it makes sense that two servers (in the same
14489 backend, being resolved by the same FQDN) have the same IP address.
14490 For such case, simply enable this option.
14491 This is the opposite of prevent-dup-ip.
14492
Daniel Corbettf8716912019-11-17 09:48:56 -050014493 * ignore-weight
14494 Ignore any weight that is set within an SRV record. This is useful when
14495 you would like to control the weights using an alternate method, such as
14496 using an "agent-check" or through the runtime api.
14497
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020014498 * prevent-dup-ip
14499 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
14500 an IP address already set to a server in the same backend and sharing the
14501 same fqdn.
14502 This is the opposite of allow-dup-ip.
14503
14504 Example:
14505 backend b_myapp
14506 default-server init-addr none resolvers dns
14507 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
14508 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
14509
14510 With the option allow-dup-ip set:
14511 * if the nameserver returns a single IP address, then both servers will use
14512 it
14513 * If the nameserver returns 2 IP addresses, then each server will pick up a
14514 different address
14515
14516 Default value: not set
14517
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014518resolve-prefer <family>
14519 When DNS resolution is enabled for a server and multiple IP addresses from
14520 different families are returned, HAProxy will prefer using an IP address
14521 from the family mentioned in the "resolve-prefer" parameter.
14522 Available families: "ipv4" and "ipv6"
14523
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020014524 Default value: ipv6
14525
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014526 Example:
14527
14528 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014529
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014530resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050014531 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014532 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010014533 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010014534 different datacenters. The latency between datacenter is not negligible, so
14535 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014536 configured network, another address is selected.
14537
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014538 Example:
14539
14540 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014541
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014542resolvers <id>
14543 Points to an existing "resolvers" section to resolve current server's
14544 hostname.
14545
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014546 Example:
14547
14548 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014549
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014550 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014551
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010014552send-proxy
14553 The "send-proxy" parameter enforces use of the PROXY protocol over any
14554 connection established to this server. The PROXY protocol informs the other
14555 end about the layer 3/4 addresses of the incoming connection, so that it can
14556 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014557 upper layer protocol. For connections accepted by an "accept-proxy" or
14558 "accept-netscaler-cip" listener, the advertised address will be used. Only
14559 TCPv4 and TCPv6 address families are supported. Other families such as
14560 Unix sockets, will report an UNKNOWN family. Servers using this option can
14561 fully be chained to another instance of haproxy listening with an
14562 "accept-proxy" setting. This setting must not be used if the server isn't
14563 aware of the protocol. When health checks are sent to the server, the PROXY
14564 protocol is automatically used when this option is set, unless there is an
14565 explicit "port" or "addr" directive, in which case an explicit
14566 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014567 See also the "no-send-proxy" option of this section and "accept-proxy" and
14568 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010014569
David Safb76832014-05-08 23:42:08 -040014570send-proxy-v2
14571 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
14572 over any connection established to this server. The PROXY protocol informs
14573 the other end about the layer 3/4 addresses of the incoming connection, so
14574 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020014575 whatever the upper layer protocol. It also send ALPN information if an alpn
14576 have been negotiated. This setting must not be used if the server isn't aware
14577 of this version of the protocol. See also the "no-send-proxy-v2" option of
14578 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014579
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010014580proxy-v2-options <option>[,<option>]*
Tim Duesterhuscf6e0c82020-03-13 12:34:24 +010014581 The "proxy-v2-options" parameter add options to send in PROXY protocol
14582 version 2 when "send-proxy-v2" is used. Options available are:
14583
14584 - ssl : See also "send-proxy-v2-ssl".
14585 - cert-cn : See also "send-proxy-v2-ssl-cn".
14586 - ssl-cipher: Name of the used cipher.
14587 - cert-sig : Signature algorithm of the used certificate.
14588 - cert-key : Key algorithm of the used certificate
14589 - authority : Host name value passed by the client (only SNI from a TLS
14590 connection is supported).
14591 - crc32c : Checksum of the PROXYv2 header.
14592 - unique-id : Send a unique ID generated using the frontend's
14593 "unique-id-format" within the PROXYv2 header.
14594 This unique-id is primarily meant for "mode tcp". It can
14595 lead to unexpected results in "mode http", because the
14596 generated unique ID is also used for the first HTTP request
14597 within a Keep-Alive connection.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010014598
David Safb76832014-05-08 23:42:08 -040014599send-proxy-v2-ssl
14600 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
14601 2 over any connection established to this server. The PROXY protocol informs
14602 the other end about the layer 3/4 addresses of the incoming connection, so
14603 that it can know the client's address or the public address it accessed to,
14604 whatever the upper layer protocol. In addition, the SSL information extension
14605 of the PROXY protocol is added to the PROXY protocol header. This setting
14606 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014607 See also the "no-send-proxy-v2-ssl" option of this section and the
14608 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014609
14610send-proxy-v2-ssl-cn
14611 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
14612 2 over any connection established to this server. The PROXY protocol informs
14613 the other end about the layer 3/4 addresses of the incoming connection, so
14614 that it can know the client's address or the public address it accessed to,
14615 whatever the upper layer protocol. In addition, the SSL information extension
14616 of the PROXY protocol, along along with the Common Name from the subject of
14617 the client certificate (if any), is added to the PROXY protocol header. This
14618 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014619 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
14620 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014621
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014622slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014623 The "slowstart" parameter for a server accepts a value in milliseconds which
14624 indicates after how long a server which has just come back up will run at
14625 full speed. Just as with every other time-based parameter, it can be entered
14626 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
14627 linearly from 0 to 100% during this time. The limitation applies to two
14628 parameters :
14629
14630 - maxconn: the number of connections accepted by the server will grow from 1
14631 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
14632
14633 - weight: when the backend uses a dynamic weighted algorithm, the weight
14634 grows linearly from 1 to 100%. In this case, the weight is updated at every
14635 health-check. For this reason, it is important that the "inter" parameter
14636 is smaller than the "slowstart", in order to maximize the number of steps.
14637
14638 The slowstart never applies when haproxy starts, otherwise it would cause
14639 trouble to running servers. It only applies when a server has been previously
14640 seen as failed.
14641
Willy Tarreau732eac42015-07-09 11:40:25 +020014642sni <expression>
14643 The "sni" parameter evaluates the sample fetch expression, converts it to a
14644 string and uses the result as the host name sent in the SNI TLS extension to
14645 the server. A typical use case is to send the SNI received from the client in
14646 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020014647 expression, though alternatives such as req.hdr(host) can also make sense. If
14648 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014649 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010014650 "verify" directive for more details. If you want to set a SNI for health
14651 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020014652
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014653source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020014654source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014655source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014656 The "source" parameter sets the source address which will be used when
14657 connecting to the server. It follows the exact same parameters and principle
14658 as the backend "source" keyword, except that it only applies to the server
14659 referencing it. Please consult the "source" keyword for details.
14660
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014661 Additionally, the "source" statement on a server line allows one to specify a
14662 source port range by indicating the lower and higher bounds delimited by a
14663 dash ('-'). Some operating systems might require a valid IP address when a
14664 source port range is specified. It is permitted to have the same IP/range for
14665 several servers. Doing so makes it possible to bypass the maximum of 64k
14666 total concurrent connections. The limit will then reach 64k connections per
14667 server.
14668
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000014669 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
14670 specifying the source address without port(s).
14671
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014672ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020014673 This option enables SSL ciphering on outgoing connections to the server. It
14674 is critical to verify server certificates using "verify" when using SSL to
14675 connect to servers, otherwise the communication is prone to trivial man in
14676 the-middle attacks rendering SSL useless. When this option is used, health
14677 checks are automatically sent in SSL too unless there is a "port" or an
14678 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014679 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
14680 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014681
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014682ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
14683 This option enforces use of <version> or lower when SSL is used to communicate
14684 with the server. This option is also available on global statement
14685 "ssl-default-server-options". See also "ssl-min-ver".
14686
14687ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
14688 This option enforces use of <version> or upper when SSL is used to communicate
14689 with the server. This option is also available on global statement
14690 "ssl-default-server-options". See also "ssl-max-ver".
14691
Frédéric Lécailled2376272017-03-21 18:52:12 +010014692ssl-reuse
14693 This option may be used as "server" setting to reset any "no-ssl-reuse"
14694 setting which would have been inherited from "default-server" directive as
14695 default value.
14696 It may also be used as "default-server" setting to reset any previous
14697 "default-server" "no-ssl-reuse" setting.
14698
14699stick
14700 This option may be used as "server" setting to reset any "non-stick"
14701 setting which would have been inherited from "default-server" directive as
14702 default value.
14703 It may also be used as "default-server" setting to reset any previous
14704 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014705
Alexander Liu2a54bb72019-05-22 19:44:48 +080014706socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050014707 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080014708 server. Using this option won't force the health check to go via socks4 by
14709 default. You will have to use the keyword "check-via-socks4" to enable it.
14710
Willy Tarreau163d4622015-10-13 16:16:41 +020014711tcp-ut <delay>
14712 Sets the TCP User Timeout for all outgoing connections to this server. This
14713 option is available on Linux since version 2.6.37. It allows haproxy to
14714 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010014715 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020014716 long-lived connections experiencing long idle periods such as remote
14717 terminals or database connection pools, where the client and server timeouts
14718 must remain high to allow a long period of idle, but where it is important to
14719 detect that the server has disappeared in order to release all resources
14720 associated with its connection (and the client's session). One typical use
14721 case is also to force dead server connections to die when health checks are
14722 too slow or during a soft reload since health checks are then disabled. The
14723 argument is a delay expressed in milliseconds by default. This only works for
14724 regular TCP connections, and is ignored for other protocols.
14725
Willy Tarreau034c88c2017-01-23 23:36:45 +010014726tfo
14727 This option enables using TCP fast open when connecting to servers, on
14728 systems that support it (currently only the Linux kernel >= 4.11).
14729 See the "tfo" bind option for more information about TCP fast open.
14730 Please note that when using tfo, you should also use the "conn-failure",
14731 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020014732 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010014733
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014734track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020014735 This option enables ability to set the current state of the server by tracking
14736 another one. It is possible to track a server which itself tracks another
14737 server, provided that at the end of the chain, a server has health checks
14738 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014739 used, it has to be enabled on both proxies.
14740
Frédéric Lécailled2376272017-03-21 18:52:12 +010014741tls-tickets
14742 This option may be used as "server" setting to reset any "no-tls-tickets"
14743 setting which would have been inherited from "default-server" directive as
14744 default value.
Lukas Tribusbdb386d2020-03-10 00:56:09 +010014745 The TLS ticket mechanism is only used up to TLS 1.2.
14746 Forward Secrecy is compromised with TLS tickets, unless ticket keys
14747 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010014748 It may also be used as "default-server" setting to reset any previous
Bjoern Jacke5ab7eb62020-02-13 14:16:16 +010014749 "default-server" "no-tls-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014750
Emeric Brunef42d922012-10-11 16:11:36 +020014751verify [none|required]
14752 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010014753 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014754 certificate provided by the server is verified using CAs from 'ca-file' and
14755 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014756 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014757 the name passed using the "sni" directive, or if not provided, the static
14758 host name passed using the "verifyhost" directive. When no name is found, the
14759 certificate's names are ignored. For this reason, without SNI it's important
14760 to use "verifyhost". On verification failure the handshake is aborted. It is
14761 critically important to verify server certificates when using SSL to connect
14762 to servers, otherwise the communication is prone to trivial man-in-the-middle
14763 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
14764 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020014765
Evan Broderbe554312013-06-27 00:05:25 -070014766verifyhost <hostname>
14767 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014768 only takes effect if 'verify required' is also specified. This directive sets
14769 a default static hostname to check the server's certificate against when no
14770 SNI was used to connect to the server. If SNI is not used, this is the only
14771 way to enable hostname verification. This static hostname, when set, will
14772 also be used for health checks (which cannot provide an SNI value). If none
14773 of the hostnames in the certificate match the specified hostname, the
14774 handshake is aborted. The hostnames in the server-provided certificate may
14775 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070014776
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014777weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014778 The "weight" parameter is used to adjust the server's weight relative to
14779 other servers. All servers will receive a load proportional to their weight
14780 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020014781 load. The default weight is 1, and the maximal value is 256. A value of 0
14782 means the server will not participate in load-balancing but will still accept
14783 persistent connections. If this parameter is used to distribute the load
14784 according to server's capacity, it is recommended to start with values which
14785 can both grow and shrink, for instance between 10 and 100 to leave enough
14786 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014787
14788
Cyril Bonté46175dd2015-07-02 22:45:32 +0200147895.3. Server IP address resolution using DNS
14790-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014791
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014792HAProxy allows using a host name on the server line to retrieve its IP address
14793using name servers. By default, HAProxy resolves the name when parsing the
Thayne McCombscdbcca92021-01-07 21:24:41 -070014794configuration file, at startup and cache the result for the process's life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014795This is not sufficient in some cases, such as in Amazon where a server's IP
14796can change after a reboot or an ELB Virtual IP can change based on current
14797workload.
14798This chapter describes how HAProxy can be configured to process server's name
14799resolution at run time.
14800Whether run time server name resolution has been enable or not, HAProxy will
14801carry on doing the first resolution when parsing the configuration.
14802
14803
Cyril Bonté46175dd2015-07-02 22:45:32 +0200148045.3.1. Global overview
14805----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014806
14807As we've seen in introduction, name resolution in HAProxy occurs at two
14808different steps of the process life:
14809
14810 1. when starting up, HAProxy parses the server line definition and matches a
14811 host name. It uses libc functions to get the host name resolved. This
14812 resolution relies on /etc/resolv.conf file.
14813
Christopher Faulet67957bd2017-09-27 11:00:59 +020014814 2. at run time, HAProxy performs periodically name resolutions for servers
14815 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014816
14817A few other events can trigger a name resolution at run time:
14818 - when a server's health check ends up in a connection timeout: this may be
14819 because the server has a new IP address. So we need to trigger a name
14820 resolution to know this new IP.
14821
Christopher Faulet67957bd2017-09-27 11:00:59 +020014822When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010014823HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020014824SRV label is specified, then the corresponding SRV records will be retrieved
14825from the DNS server, and the provided hostnames will be used. The SRV label
14826will be checked periodically, and if any server are added or removed, haproxy
14827will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020014828
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014829A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050014830 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014831 first valid response.
14832
14833 - a resolution is considered as invalid (NX, timeout, refused), when all the
14834 servers return an error.
14835
14836
Cyril Bonté46175dd2015-07-02 22:45:32 +0200148375.3.2. The resolvers section
14838----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014839
14840This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020014841HAProxy. There can be as many as resolvers section as needed. Each section can
14842contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014843
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014844When multiple name servers are configured in a resolvers section, then HAProxy
14845uses the first valid response. In case of invalid responses, only the last one
14846is treated. Purpose is to give the chance to a slow server to deliver a valid
14847answer after a fast faulty or outdated server.
14848
14849When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020014850used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014851
Christopher Faulet67957bd2017-09-27 11:00:59 +020014852 1. HAProxy retries the same DNS query with a new query type. The A queries are
14853 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
14854 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014855
Christopher Faulet67957bd2017-09-27 11:00:59 +020014856 2. When the fallback on the query type was done (or not applicable), HAProxy
14857 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014858
Christopher Faulet67957bd2017-09-27 11:00:59 +020014859 3. HAProxy retries previous steps <resolve_retires> times. If no valid
14860 response is received after that, it stops the DNS resolution and reports
14861 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014862
Christopher Faulet67957bd2017-09-27 11:00:59 +020014863For example, with 2 name servers configured in a resolvers section, the
14864following scenarios are possible:
14865
14866 - First response is valid and is applied directly, second response is
14867 ignored
14868
14869 - First response is invalid and second one is valid, then second response is
14870 applied
14871
14872 - First response is a NX domain and second one a truncated response, then
14873 HAProxy retries the query with a new type
14874
14875 - First response is a NX domain and second one is a timeout, then HAProxy
14876 retries the query with a new type
14877
14878 - Query timed out for both name servers, then HAProxy retries it with the
14879 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014880
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014881As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
14882a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020014883<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014884
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014885
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014886resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014887 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014888
14889A resolvers section accept the following parameters:
14890
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014891accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014892 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020014893 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014894 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
14895 by RFC 6891)
14896
Emeric Brun4c751952021-03-08 16:41:29 +010014897 Note: the maximum allowed value is 65535. Recommended value for UDP is
14898 4096 and it is not recommended to exceed 8192 except if you are sure
14899 that your system and network can handle this (over 65507 makes no sense
14900 since is the maximum UDP payload size). If you are using only TCP
14901 nameservers to handle huge DNS responses, you should put this value
14902 to the max: 65535.
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020014903
Emeric Brunc8f3e452021-04-07 16:04:54 +020014904nameserver <name> <address>[:port] [param*]
14905 Used to configure a nameserver. <name> of the nameserver should ne unique.
14906 By default the <address> is considered of type datagram. This means if an
14907 IPv4 or IPv6 is configured without special address prefixes (paragraph 11.)
14908 the UDP protocol will be used. If an stream protocol address prefix is used,
14909 the nameserver will be considered as a stream server (TCP for instance) and
14910 "server" parameters found in 5.2 paragraph which are relevant for DNS
14911 resolving will be considered. Note: currently, in TCP mode, 4 queries are
14912 pipelined on the same connections. A batch of idle connections are removed
14913 every 5 seconds. "maxconn" can be configured to limit the amount of those
Emeric Brun56fc5d92021-02-12 20:05:45 +010014914 concurrent connections and TLS should also usable if the server supports.
14915
Ben Draut44e609b2018-05-29 15:40:08 -060014916parse-resolv-conf
14917 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
14918 list. Ordered as if each nameserver in /etc/resolv.conf was individually
14919 placed in the resolvers section in place of this directive.
14920
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014921hold <status> <period>
14922 Defines <period> during which the last name resolution should be kept based
14923 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014924 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014925 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014926 <period> : interval between two successive name resolution when the last
14927 answer was in <status>. It follows the HAProxy time format.
14928 <period> is in milliseconds by default.
14929
Baptiste Assmann686408b2017-08-18 10:15:42 +020014930 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014931
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014932resolve_retries <nb>
14933 Defines the number <nb> of queries to send to resolve a server name before
14934 giving up.
14935 Default value: 3
14936
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014937 A retry occurs on name server timeout or when the full sequence of DNS query
14938 type failover is over and we need to start up from the default ANY query
14939 type.
14940
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014941timeout <event> <time>
14942 Defines timeouts related to name resolution
14943 <event> : the event on which the <time> timeout period applies to.
14944 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014945 - resolve : default time to trigger name resolutions when no
14946 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014947 Default value: 1s
14948 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014949 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014950 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014951 <time> : time related to the event. It follows the HAProxy time format.
14952 <time> is expressed in milliseconds.
14953
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014954 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014955
14956 resolvers mydns
14957 nameserver dns1 10.0.0.1:53
14958 nameserver dns2 10.0.0.2:53
Emeric Brunc8f3e452021-04-07 16:04:54 +020014959 nameserver dns3 tcp@10.0.0.3:53
Ben Draut44e609b2018-05-29 15:40:08 -060014960 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014961 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020014962 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014963 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014964 hold other 30s
14965 hold refused 30s
14966 hold nx 30s
14967 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014968 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014969 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014970
14971
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200149726. Cache
14973---------
14974
14975HAProxy provides a cache, which was designed to perform cache on small objects
14976(favicon, css...). This is a minimalist low-maintenance cache which runs in
14977RAM.
14978
14979The cache is based on a memory which is shared between processes and threads,
14980this memory is split in blocks of 1k.
14981
14982If an object is not used anymore, it can be deleted to store a new object
14983independently of its expiration date. The oldest objects are deleted first
14984when we try to allocate a new one.
14985
14986The cache uses a hash of the host header and the URI as the key.
14987
14988It's possible to view the status of a cache using the Unix socket command
14989"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
14990for more details.
14991
14992When an object is delivered from the cache, the server name in the log is
14993replaced by "<CACHE>".
14994
14995
149966.1. Limitation
14997----------------
14998
14999The cache won't store and won't deliver objects in these cases:
15000
15001- If the response is not a 200
Remi Tricot-Le Breton4f730832020-11-26 15:51:50 +010015002- If the response contains a Vary header and either the process-vary option is
15003 disabled, or a currently unmanaged header is specified in the Vary value (only
15004 accept-encoding and referer are managed for now)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020015005- If the Content-Length + the headers size is greater than "max-object-size"
15006- If the response is not cacheable
Remi Tricot-Le Bretond493bc82020-11-26 15:51:29 +010015007- If the response does not have an explicit expiration time (s-maxage or max-age
15008 Cache-Control directives or Expires header) or a validator (ETag or Last-Modified
15009 headers)
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010015010- If the process-vary option is enabled and there are already max-secondary-entries
15011 entries with the same primary key as the current response
Remi Tricot-Le Breton6ca89162021-01-07 14:50:51 +010015012- If the process-vary option is enabled and the response has an unknown encoding (not
15013 mentioned in https://www.iana.org/assignments/http-parameters/http-parameters.xhtml)
15014 while varying on the accept-encoding client header
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020015015
15016- If the request is not a GET
15017- If the HTTP version of the request is smaller than 1.1
15018- If the request contains an Authorization header
15019
15020
150216.2. Setup
15022-----------
15023
15024To setup a cache, you must define a cache section and use it in a proxy with
15025the corresponding http-request and response actions.
15026
15027
150286.2.1. Cache section
15029---------------------
15030
15031cache <name>
15032 Declare a cache section, allocate a shared cache memory named <name>, the
15033 size of cache is mandatory.
15034
15035total-max-size <megabytes>
15036 Define the size in RAM of the cache in megabytes. This size is split in
15037 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
15038
15039max-object-size <bytes>
15040 Define the maximum size of the objects to be cached. Must not be greater than
15041 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
15042 All objects with sizes larger than "max-object-size" will not be cached.
15043
15044max-age <seconds>
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010015045 Define the maximum expiration duration. The expiration is set as the lowest
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020015046 value between the s-maxage or max-age (in this order) directive in the
15047 Cache-Control response header and this value. The default value is 60
15048 seconds, which means that you can't cache an object more than 60 seconds by
15049 default.
15050
Remi Tricot-Le Bretone6cc5b52020-12-23 18:13:53 +010015051process-vary <on/off>
15052 Enable or disable the processing of the Vary header. When disabled, a response
Remi Tricot-Le Breton754b2422020-11-16 15:56:10 +010015053 containing such a header will never be cached. When enabled, we need to calculate
15054 a preliminary hash for a subset of request headers on all the incoming requests
15055 (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 +010015056 for a given request (see RFC 7234#4.1). The default value is off (disabled).
Remi Tricot-Le Breton754b2422020-11-16 15:56:10 +010015057
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010015058max-secondary-entries <number>
15059 Define the maximum number of simultaneous secondary entries with the same primary
15060 key in the cache. This needs the vary support to be enabled. Its default value is 10
15061 and should be passed a strictly positive integer.
15062
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020015063
150646.2.2. Proxy section
15065---------------------
15066
15067http-request cache-use <name> [ { if | unless } <condition> ]
15068 Try to deliver a cached object from the cache <name>. This directive is also
15069 mandatory to store the cache as it calculates the cache hash. If you want to
15070 use a condition for both storage and delivering that's a good idea to put it
15071 after this one.
15072
15073http-response cache-store <name> [ { if | unless } <condition> ]
15074 Store an http-response within the cache. The storage of the response headers
15075 is done at this step, which means you can use others http-response actions
15076 to modify headers before or after the storage of the response. This action
15077 is responsible for the setup of the cache storage filter.
15078
15079
15080Example:
15081
15082 backend bck1
15083 mode http
15084
15085 http-request cache-use foobar
15086 http-response cache-store foobar
15087 server srv1 127.0.0.1:80
15088
15089 cache foobar
15090 total-max-size 4
15091 max-age 240
15092
15093
Willy Tarreau74ca5042013-06-11 23:12:07 +0200150947. Using ACLs and fetching samples
15095----------------------------------
15096
Davor Ocelice9ed2812017-12-25 17:49:28 +010015097HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020015098client or server information, from tables, environmental information etc...
15099The action of extracting such data is called fetching a sample. Once retrieved,
15100these samples may be used for various purposes such as a key to a stick-table,
15101but most common usages consist in matching them against predefined constant
15102data called patterns.
15103
15104
151057.1. ACL basics
15106---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015107
15108The use of Access Control Lists (ACL) provides a flexible solution to perform
15109content switching and generally to take decisions based on content extracted
15110from the request, the response or any environmental status. The principle is
15111simple :
15112
Willy Tarreau74ca5042013-06-11 23:12:07 +020015113 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015114 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020015115 - apply one or multiple pattern matching methods on this sample
15116 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015117
Willy Tarreau74ca5042013-06-11 23:12:07 +020015118The actions generally consist in blocking a request, selecting a backend, or
15119adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015120
15121In order to define a test, the "acl" keyword is used. The syntax is :
15122
Willy Tarreau74ca5042013-06-11 23:12:07 +020015123 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015124
15125This creates a new ACL <aclname> or completes an existing one with new tests.
15126Those tests apply to the portion of request/response specified in <criterion>
15127and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015128an operator which may be specified before the set of values. Optionally some
15129conversion operators may be applied to the sample, and they will be specified
15130as a comma-delimited list of keywords just after the first keyword. The values
15131are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015132
15133ACL names must be formed from upper and lower case letters, digits, '-' (dash),
15134'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
15135which means that "my_acl" and "My_Acl" are two different ACLs.
15136
15137There is no enforced limit to the number of ACLs. The unused ones do not affect
15138performance, they just consume a small amount of memory.
15139
Willy Tarreau74ca5042013-06-11 23:12:07 +020015140The criterion generally is the name of a sample fetch method, or one of its ACL
15141specific declinations. The default test method is implied by the output type of
15142this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015143methods of a same sample fetch method. The sample fetch methods are the only
15144ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015145
15146Sample fetch methods return data which can be of the following types :
15147 - boolean
15148 - integer (signed or unsigned)
15149 - IPv4 or IPv6 address
15150 - string
15151 - data block
15152
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015153Converters transform any of these data into any of these. For example, some
15154converters might convert a string to a lower-case string while other ones
15155would turn a string to an IPv4 address, or apply a netmask to an IP address.
15156The resulting sample is of the type of the last converter applied to the list,
15157which defaults to the type of the sample fetch method.
15158
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015159Each sample or converter returns data of a specific type, specified with its
15160keyword in this documentation. When an ACL is declared using a standard sample
15161fetch method, certain types automatically involved a default matching method
15162which are summarized in the table below :
15163
15164 +---------------------+-----------------+
15165 | Sample or converter | Default |
15166 | output type | matching method |
15167 +---------------------+-----------------+
15168 | boolean | bool |
15169 +---------------------+-----------------+
15170 | integer | int |
15171 +---------------------+-----------------+
15172 | ip | ip |
15173 +---------------------+-----------------+
15174 | string | str |
15175 +---------------------+-----------------+
15176 | binary | none, use "-m" |
15177 +---------------------+-----------------+
15178
15179Note that in order to match a binary samples, it is mandatory to specify a
15180matching method, see below.
15181
Willy Tarreau74ca5042013-06-11 23:12:07 +020015182The ACL engine can match these types against patterns of the following types :
15183 - boolean
15184 - integer or integer range
15185 - IP address / network
15186 - string (exact, substring, suffix, prefix, subdir, domain)
15187 - regular expression
15188 - hex block
15189
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015190The following ACL flags are currently supported :
15191
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015192 -i : ignore case during matching of all subsequent patterns.
15193 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015194 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010015195 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010015196 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010015197 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020015198 -- : force end of flags. Useful when a string looks like one of the flags.
15199
Willy Tarreau74ca5042013-06-11 23:12:07 +020015200The "-f" flag is followed by the name of a file from which all lines will be
15201read as individual values. It is even possible to pass multiple "-f" arguments
15202if the patterns are to be loaded from multiple files. Empty lines as well as
15203lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
15204will be stripped. If it is absolutely necessary to insert a valid pattern
15205beginning with a sharp, just prefix it with a space so that it is not taken for
15206a comment. Depending on the data type and match method, haproxy may load the
15207lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
15208exact string matching. In this case, duplicates will automatically be removed.
15209
Thierry FOURNIER9860c412014-01-29 14:23:29 +010015210The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
15211parsed as two column file. The first column contains the patterns used by the
15212ACL, and the second column contain the samples. The sample can be used later by
15213a map. This can be useful in some rare cases where an ACL would just be used to
15214check for the existence of a pattern in a map before a mapping is applied.
15215
Thierry FOURNIER3534d882014-01-20 17:01:44 +010015216The "-u" flag forces the unique id of the ACL. This unique id is used with the
15217socket interface to identify ACL and dynamically change its values. Note that a
15218file is always identified by its name even if an id is set.
15219
Willy Tarreau74ca5042013-06-11 23:12:07 +020015220Also, note that the "-i" flag applies to subsequent entries and not to entries
15221loaded from files preceding it. For instance :
15222
15223 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
15224
15225In this example, each line of "exact-ua.lst" will be exactly matched against
15226the "user-agent" header of the request. Then each line of "generic-ua" will be
15227case-insensitively matched. Then the word "test" will be insensitively matched
15228as well.
15229
15230The "-m" flag is used to select a specific pattern matching method on the input
15231sample. All ACL-specific criteria imply a pattern matching method and generally
15232do not need this flag. However, this flag is useful with generic sample fetch
15233methods to describe how they're going to be matched against the patterns. This
15234is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010015235obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020015236followed by a pattern matching method name, this method is used instead of the
15237default one for the criterion. This makes it possible to match contents in ways
15238that were not initially planned, or with sample fetch methods which return a
15239string. The matching method also affects the way the patterns are parsed.
15240
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010015241The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
15242By default, if the parser cannot parse ip address it considers that the parsed
15243string is maybe a domain name and try dns resolution. The flag "-n" disable this
15244resolution. It is useful for detecting malformed ip lists. Note that if the DNS
15245server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050015246waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010015247flag "-n" disable this behavior. Note also that during the runtime, this
15248function is disabled for the dynamic acl modifications.
15249
Willy Tarreau74ca5042013-06-11 23:12:07 +020015250There are some restrictions however. Not all methods can be used with all
15251sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
15252be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020015253
15254 - "found" : only check if the requested sample could be found in the stream,
15255 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020015256 to pass any pattern to avoid confusion. This matching method is
15257 particularly useful to detect presence of certain contents such
15258 as headers, cookies, etc... even if they are empty and without
15259 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015260
15261 - "bool" : check the value as a boolean. It can only be applied to fetches
15262 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015263 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015264
15265 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020015266 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015267
15268 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020015269 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015270
Davor Ocelice9ed2812017-12-25 17:49:28 +010015271 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020015272 binary sequence. This may be used with binary or string samples.
15273
15274 - "len" : match the sample's length as an integer. This may be used with
15275 binary or string samples.
15276
Willy Tarreau74ca5042013-06-11 23:12:07 +020015277 - "str" : exact match : match the contents against a string. This may be
15278 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015279
Willy Tarreau74ca5042013-06-11 23:12:07 +020015280 - "sub" : substring match : check that the contents contain at least one of
15281 the provided string patterns. This may be used with binary or
15282 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015283
Willy Tarreau74ca5042013-06-11 23:12:07 +020015284 - "reg" : regex match : match the contents against a list of regular
15285 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015286
Willy Tarreau74ca5042013-06-11 23:12:07 +020015287 - "beg" : prefix match : check that the contents begin like the provided
15288 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015289
Willy Tarreau74ca5042013-06-11 23:12:07 +020015290 - "end" : suffix match : check that the contents end like the provided
15291 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015292
Willy Tarreau74ca5042013-06-11 23:12:07 +020015293 - "dir" : subdir match : check that a slash-delimited portion of the
15294 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015295 This may be used with binary or string samples.
15296
Willy Tarreau74ca5042013-06-11 23:12:07 +020015297 - "dom" : domain match : check that a dot-delimited portion of the contents
15298 exactly match one of the provided string patterns. This may be
15299 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015300
15301For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
15302request, it is possible to do :
15303
15304 acl jsess_present cook(JSESSIONID) -m found
15305
15306In order to apply a regular expression on the 500 first bytes of data in the
15307buffer, one would use the following acl :
15308
15309 acl script_tag payload(0,500) -m reg -i <script>
15310
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015311On systems where the regex library is much slower when using "-i", it is
15312possible to convert the sample to lowercase before matching, like this :
15313
15314 acl script_tag payload(0,500),lower -m reg <script>
15315
Willy Tarreau74ca5042013-06-11 23:12:07 +020015316All ACL-specific criteria imply a default matching method. Most often, these
15317criteria are composed by concatenating the name of the original sample fetch
15318method and the matching method. For example, "hdr_beg" applies the "beg" match
15319to samples retrieved using the "hdr" fetch method. Since all ACL-specific
15320criteria rely on a sample fetch method, it is always possible instead to use
15321the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015322
Willy Tarreau74ca5042013-06-11 23:12:07 +020015323If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015324the matching method is simply applied to the underlying sample fetch method.
15325For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015326
Willy Tarreau74ca5042013-06-11 23:12:07 +020015327 acl short_form hdr_beg(host) www.
15328 acl alternate1 hdr_beg(host) -m beg www.
15329 acl alternate2 hdr_dom(host) -m beg www.
15330 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015331
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015332
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015333The table below summarizes the compatibility matrix between sample or converter
15334types and the pattern types to fetch against. It indicates for each compatible
15335combination the name of the matching method to be used, surrounded with angle
15336brackets ">" and "<" when the method is the default one and will work by
15337default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015338
Willy Tarreau74ca5042013-06-11 23:12:07 +020015339 +-------------------------------------------------+
15340 | Input sample type |
15341 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015342 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015343 +----------------------+---------+---------+---------+---------+---------+
15344 | none (presence only) | found | found | found | found | found |
15345 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015346 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015347 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015348 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015349 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015350 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015351 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015352 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015353 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015354 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015355 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015356 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015357 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015358 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015359 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015360 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015361 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015362 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015363 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015364 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015365 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015366 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015367 +----------------------+---------+---------+---------+---------+---------+
15368 | hex block | | | | bin | bin |
15369 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020015370
15371
Willy Tarreau74ca5042013-06-11 23:12:07 +0200153727.1.1. Matching booleans
15373------------------------
15374
15375In order to match a boolean, no value is needed and all values are ignored.
15376Boolean matching is used by default for all fetch methods of type "boolean".
15377When boolean matching is used, the fetched value is returned as-is, which means
15378that a boolean "true" will always match and a boolean "false" will never match.
15379
15380Boolean matching may also be enforced using "-m bool" on fetch methods which
15381return an integer value. Then, integer value 0 is converted to the boolean
15382"false" and all other values are converted to "true".
15383
Willy Tarreau6a06a402007-07-15 20:15:28 +020015384
Willy Tarreau74ca5042013-06-11 23:12:07 +0200153857.1.2. Matching integers
15386------------------------
15387
15388Integer matching applies by default to integer fetch methods. It can also be
15389enforced on boolean fetches using "-m int". In this case, "false" is converted
15390to the integer 0, and "true" is converted to the integer 1.
15391
15392Integer matching also supports integer ranges and operators. Note that integer
15393matching only applies to positive values. A range is a value expressed with a
15394lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015395
15396For instance, "1024:65535" is a valid range to represent a range of
15397unprivileged ports, and "1024:" would also work. "0:1023" is a valid
15398representation of privileged ports, and ":1023" would also work.
15399
Willy Tarreau62644772008-07-16 18:36:06 +020015400As a special case, some ACL functions support decimal numbers which are in fact
15401two integers separated by a dot. This is used with some version checks for
15402instance. All integer properties apply to those decimal numbers, including
15403ranges and operators.
15404
Willy Tarreau6a06a402007-07-15 20:15:28 +020015405For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010015406operators with ranges does not make much sense and is strongly discouraged.
15407Similarly, it does not make much sense to perform order comparisons with a set
15408of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015409
Willy Tarreau0ba27502007-12-24 16:55:16 +010015410Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020015411
15412 eq : true if the tested value equals at least one value
15413 ge : true if the tested value is greater than or equal to at least one value
15414 gt : true if the tested value is greater than at least one value
15415 le : true if the tested value is less than or equal to at least one value
15416 lt : true if the tested value is less than at least one value
15417
Willy Tarreau0ba27502007-12-24 16:55:16 +010015418For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020015419
15420 acl negative-length hdr_val(content-length) lt 0
15421
Willy Tarreau62644772008-07-16 18:36:06 +020015422This one matches SSL versions between 3.0 and 3.1 (inclusive) :
15423
15424 acl sslv3 req_ssl_ver 3:3.1
15425
Willy Tarreau6a06a402007-07-15 20:15:28 +020015426
Willy Tarreau74ca5042013-06-11 23:12:07 +0200154277.1.3. Matching strings
15428-----------------------
15429
15430String matching applies to string or binary fetch methods, and exists in 6
15431different forms :
15432
15433 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015434 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020015435
15436 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015437 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020015438
15439 - prefix match (-m beg) : the patterns are compared with the beginning of
15440 the extracted string, and the ACL matches if any of them matches.
15441
15442 - suffix match (-m end) : the patterns are compared with the end of the
15443 extracted string, and the ACL matches if any of them matches.
15444
Baptiste Assmann33db6002016-03-06 23:32:10 +010015445 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020015446 string, delimited with slashes ("/"), and the ACL matches if any of them
15447 matches.
15448
15449 - domain match (-m dom) : the patterns are looked up inside the extracted
15450 string, delimited with dots ("."), and the ACL matches if any of them
15451 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015452
15453String matching applies to verbatim strings as they are passed, with the
15454exception of the backslash ("\") which makes it possible to escape some
15455characters such as the space. If the "-i" flag is passed before the first
15456string, then the matching will be performed ignoring the case. In order
15457to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010015458before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020015459
Mathias Weiersmuellercb250fc2019-12-02 09:43:40 +010015460Do not use string matches for binary fetches which might contain null bytes
15461(0x00), as the comparison stops at the occurrence of the first null byte.
15462Instead, convert the binary fetch to a hex string with the hex converter first.
15463
15464Example:
15465 # matches if the string <tag> is present in the binary sample
15466 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
15467
Willy Tarreau6a06a402007-07-15 20:15:28 +020015468
Willy Tarreau74ca5042013-06-11 23:12:07 +0200154697.1.4. Matching regular expressions (regexes)
15470---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020015471
15472Just like with string matching, regex matching applies to verbatim strings as
15473they are passed, with the exception of the backslash ("\") which makes it
15474possible to escape some characters such as the space. If the "-i" flag is
15475passed before the first regex, then the matching will be performed ignoring
15476the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010015477the "--" flag before the first string. Same principle applies of course to
15478match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020015479
15480
Willy Tarreau74ca5042013-06-11 23:12:07 +0200154817.1.5. Matching arbitrary data blocks
15482-------------------------------------
15483
15484It is possible to match some extracted samples against a binary block which may
15485not safely be represented as a string. For this, the patterns must be passed as
15486a series of hexadecimal digits in an even number, when the match method is set
15487to binary. Each sequence of two digits will represent a byte. The hexadecimal
15488digits may be used upper or lower case.
15489
15490Example :
15491 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
15492 acl hello payload(0,6) -m bin 48656c6c6f0a
15493
15494
154957.1.6. Matching IPv4 and IPv6 addresses
15496---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020015497
15498IPv4 addresses values can be specified either as plain addresses or with a
15499netmask appended, in which case the IPv4 address matches whenever it is
15500within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010015501host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010015502difficult to read and debug configurations. If hostnames are used, you should
15503at least ensure that they are present in /etc/hosts so that the configuration
15504does not depend on any random DNS match at the moment the configuration is
15505parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015506
Daniel Schnellereba56342016-04-13 00:26:52 +020015507The dotted IPv4 address notation is supported in both regular as well as the
15508abbreviated form with all-0-octets omitted:
15509
15510 +------------------+------------------+------------------+
15511 | Example 1 | Example 2 | Example 3 |
15512 +------------------+------------------+------------------+
15513 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
15514 | 192.168.1 | 10.12 | 127.1 |
15515 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
15516 | 192.168.1/22 | 10.12/8 | 127.1/8 |
15517 +------------------+------------------+------------------+
15518
15519Notice that this is different from RFC 4632 CIDR address notation in which
15520192.168.42/24 would be equivalent to 192.168.42.0/24.
15521
Willy Tarreauceb4ac92012-04-28 00:41:46 +020015522IPv6 may be entered in their usual form, with or without a netmask appended.
15523Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
15524trouble with randomly resolved IP addresses, host names are never allowed in
15525IPv6 patterns.
15526
15527HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
15528following situations :
15529 - tested address is IPv4, pattern address is IPv4, the match applies
15530 in IPv4 using the supplied mask if any.
15531 - tested address is IPv6, pattern address is IPv6, the match applies
15532 in IPv6 using the supplied mask if any.
15533 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
15534 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
15535 ::IPV4 or ::ffff:IPV4, otherwise it fails.
15536 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
15537 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
15538 applied in IPv6 using the supplied IPv6 mask.
15539
Willy Tarreau74ca5042013-06-11 23:12:07 +020015540
155417.2. Using ACLs to form conditions
15542----------------------------------
15543
15544Some actions are only performed upon a valid condition. A condition is a
15545combination of ACLs with operators. 3 operators are supported :
15546
15547 - AND (implicit)
15548 - OR (explicit with the "or" keyword or the "||" operator)
15549 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020015550
Willy Tarreau74ca5042013-06-11 23:12:07 +020015551A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020015552
Willy Tarreau74ca5042013-06-11 23:12:07 +020015553 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020015554
Willy Tarreau74ca5042013-06-11 23:12:07 +020015555Such conditions are generally used after an "if" or "unless" statement,
15556indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020015557
Willy Tarreau74ca5042013-06-11 23:12:07 +020015558For instance, to block HTTP requests to the "*" URL with methods other than
15559"OPTIONS", as well as POST requests without content-length, and GET or HEAD
15560requests with a content-length greater than 0, and finally every request which
15561is not either GET/HEAD/POST/OPTIONS !
15562
15563 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015564 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
15565 http-request deny if METH_GET HTTP_CONTENT
15566 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020015567
15568To select a different backend for requests to static contents on the "www" site
15569and to every request on the "img", "video", "download" and "ftp" hosts :
15570
15571 acl url_static path_beg /static /images /img /css
15572 acl url_static path_end .gif .png .jpg .css .js
15573 acl host_www hdr_beg(host) -i www
15574 acl host_static hdr_beg(host) -i img. video. download. ftp.
15575
Davor Ocelice9ed2812017-12-25 17:49:28 +010015576 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020015577 # of host "www". Use backend "www" for the rest.
15578 use_backend static if host_static or host_www url_static
15579 use_backend www if host_www
15580
15581It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
15582expressions that are built on the fly without needing to be declared. They must
15583be enclosed between braces, with a space before and after each brace (because
15584the braces must be seen as independent words). Example :
15585
15586 The following rule :
15587
15588 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015589 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020015590
15591 Can also be written that way :
15592
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015593 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020015594
15595It is generally not recommended to use this construct because it's a lot easier
15596to leave errors in the configuration when written that way. However, for very
15597simple rules matching only one source IP address for instance, it can make more
15598sense to use them than to declare ACLs with random names. Another example of
15599good use is the following :
15600
15601 With named ACLs :
15602
15603 acl site_dead nbsrv(dynamic) lt 2
15604 acl site_dead nbsrv(static) lt 2
15605 monitor fail if site_dead
15606
15607 With anonymous ACLs :
15608
15609 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
15610
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015611See section 4.2 for detailed help on the "http-request deny" and "use_backend"
15612keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015613
15614
156157.3. Fetching samples
15616---------------------
15617
15618Historically, sample fetch methods were only used to retrieve data to match
15619against patterns using ACLs. With the arrival of stick-tables, a new class of
15620sample fetch methods was created, most often sharing the same syntax as their
15621ACL counterpart. These sample fetch methods are also known as "fetches". As
15622of now, ACLs and fetches have converged. All ACL fetch methods have been made
15623available as fetch methods, and ACLs may use any sample fetch method as well.
15624
15625This section details all available sample fetch methods and their output type.
15626Some sample fetch methods have deprecated aliases that are used to maintain
15627compatibility with existing configurations. They are then explicitly marked as
15628deprecated and should not be used in new setups.
15629
15630The ACL derivatives are also indicated when available, with their respective
15631matching methods. These ones all have a well defined default pattern matching
15632method, so it is never necessary (though allowed) to pass the "-m" option to
15633indicate how the sample will be matched using ACLs.
15634
15635As indicated in the sample type versus matching compatibility matrix above,
15636when using a generic sample fetch method in an ACL, the "-m" option is
15637mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
15638the same keyword exists as an ACL keyword and as a standard fetch method, the
15639ACL engine will automatically pick the ACL-only one by default.
15640
15641Some of these keywords support one or multiple mandatory arguments, and one or
15642multiple optional arguments. These arguments are strongly typed and are checked
15643when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010015644incorrect argument (e.g. an unresolved backend name). Fetch function arguments
15645are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020015646is optional, it will be indicated below between square brackets ('[ ]'). When
15647all arguments are optional, the parenthesis may be omitted.
15648
15649Thus, the syntax of a standard sample fetch method is one of the following :
15650 - name
15651 - name(arg1)
15652 - name(arg1,arg2)
15653
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015654
156557.3.1. Converters
15656-----------------
15657
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015658Sample fetch methods may be combined with transformations to be applied on top
15659of the fetched sample (also called "converters"). These combinations form what
15660is called "sample expressions" and the result is a "sample". Initially this
15661was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010015662has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015663unique-id-format, add-header, ...).
15664
15665These transformations are enumerated as a series of specific keywords after the
15666sample fetch method. These keywords may equally be appended immediately after
15667the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010015668support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015669
Willy Tarreau97707872015-01-27 15:12:13 +010015670A certain category of converters are bitwise and arithmetic operators which
15671support performing basic operations on integers. Some bitwise operations are
15672supported (and, or, xor, cpl) and some arithmetic operations are supported
15673(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
15674bool) which make it possible to report a match without having to write an ACL.
15675
Willy Tarreau74ca5042013-06-11 23:12:07 +020015676The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010015677
Ben Shillitof25e8e52016-12-02 14:25:37 +00001567851d.single(<prop>[,<prop>*])
15679 Returns values for the properties requested as a string, where values are
15680 separated by the delimiter specified with "51degrees-property-separator".
15681 The device is identified using the User-Agent header passed to the
15682 converter. The function can be passed up to five property names, and if a
15683 property name can't be found, the value "NoData" is returned.
15684
15685 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010015686 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
15687 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000015688 # User-Agent passed to the converter.
15689 frontend http-in
15690 bind *:8081
15691 default_backend servers
15692 http-request set-header X-51D-DeviceTypeMobileTablet \
15693 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
15694
Willy Tarreau97707872015-01-27 15:12:13 +010015695add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015696 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015697 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015698 name. The name of the variable starts with an indication about its scope. The
15699 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015700 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015701 "sess" : the variable is shared with the whole session
15702 "txn" : the variable is shared with the transaction (request and response)
15703 "req" : the variable is shared only during request processing
15704 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015705 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015706 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015707
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010015708aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
15709 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
15710 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
15711 need to be base64 encoded and the returned result is in raw byte format.
15712 If the <aead_tag> validation fails, the converter doesn't return any data.
15713 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
15714 converter requires at least OpenSSL 1.0.1.
15715
15716 Example:
15717 http-response set-header X-Decrypted-Text %[var(txn.enc),\
15718 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
15719
Willy Tarreau97707872015-01-27 15:12:13 +010015720and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015721 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015722 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010015723 numeric value or a variable name. The name of the variable starts with an
15724 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015725 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015726 "sess" : the variable is shared with the whole session
15727 "txn" : the variable is shared with the transaction (request and response)
15728 "req" : the variable is shared only during request processing
15729 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015730 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015731 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015732
Holger Just1bfc24b2017-05-06 00:56:53 +020015733b64dec
15734 Converts (decodes) a base64 encoded input string to its binary
15735 representation. It performs the inverse operation of base64().
Moemen MHEDHBI92f7d432021-04-01 20:53:59 +020015736 For base64url("URL and Filename Safe Alphabet" (RFC 4648)) variant
15737 see "ub64dec".
Holger Just1bfc24b2017-05-06 00:56:53 +020015738
Emeric Brun53d1a982014-04-30 18:21:37 +020015739base64
15740 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010015741 transfer binary content in a way that can be reliably transferred (e.g.
Moemen MHEDHBI92f7d432021-04-01 20:53:59 +020015742 an SSL ID can be copied in a header). For base64url("URL and Filename
15743 Safe Alphabet" (RFC 4648)) variant see "ub64enc".
Emeric Brun53d1a982014-04-30 18:21:37 +020015744
Willy Tarreau97707872015-01-27 15:12:13 +010015745bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015746 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010015747 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010015748 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010015749 presence of a flag).
15750
Emeric Brun54c4ac82014-11-03 15:32:43 +010015751bytes(<offset>[,<length>])
15752 Extracts some bytes from an input binary sample. The result is a binary
15753 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010015754 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010015755
Willy Tarreau280f42b2018-02-19 15:34:12 +010015756concat([<start>],[<var>],[<end>])
15757 Concatenates up to 3 fields after the current sample which is then turned to
15758 a string. The first one, <start>, is a constant string, that will be appended
15759 immediately after the existing sample. It may be omitted if not used. The
15760 second one, <var>, is a variable name. The variable will be looked up, its
15761 contents converted to a string, and it will be appended immediately after the
15762 <first> part. If the variable is not found, nothing is appended. It may be
15763 omitted as well. The third field, <end> is a constant string that will be
15764 appended after the variable. It may also be omitted. Together, these elements
15765 allow to concatenate variables with delimiters to an existing set of
15766 variables. This can be used to build new variables made of a succession of
Willy Tarreauef21fac2020-02-14 13:37:20 +010015767 other variables, such as colon-delimited values. If commas or closing
Daniel Corbett67a82712020-07-06 23:01:19 -040015768 parenthesis are needed as delimiters, they must be protected by quotes or
Willy Tarreauef21fac2020-02-14 13:37:20 +010015769 backslashes, themselves protected so that they are not stripped by the first
15770 level parser. See examples below.
Willy Tarreau280f42b2018-02-19 15:34:12 +010015771
15772 Example:
15773 tcp-request session set-var(sess.src) src
15774 tcp-request session set-var(sess.dn) ssl_c_s_dn
15775 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
Willy Tarreauef21fac2020-02-14 13:37:20 +010015776 tcp-request session set-var(txn.ipport) "str(),concat('addr=(',sess.ip),concat(',',sess.port,')')"
Willy Tarreau280f42b2018-02-19 15:34:12 +010015777 http-request set-header x-hap-sig %[var(txn.sig)]
15778
Willy Tarreau97707872015-01-27 15:12:13 +010015779cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015780 Takes the input value of type signed integer, applies a ones-complement
15781 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010015782
Willy Tarreau80599772015-01-20 19:35:24 +010015783crc32([<avalanche>])
15784 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
15785 hash function. Optionally, it is possible to apply a full avalanche hash
15786 function to the output if the optional <avalanche> argument equals 1. This
15787 converter uses the same functions as used by the various hash-based load
15788 balancing algorithms, so it will provide exactly the same results. It is
15789 provided for compatibility with other software which want a CRC32 to be
15790 computed on some input keys, so it follows the most common implementation as
15791 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
15792 but may provide a better or at least less predictable distribution. It must
15793 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015794 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
15795
15796crc32c([<avalanche>])
15797 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
15798 hash function. Optionally, it is possible to apply a full avalanche hash
15799 function to the output if the optional <avalanche> argument equals 1. This
15800 converter uses the same functions as described in RFC4960, Appendix B [8].
15801 It is provided for compatibility with other software which want a CRC32C to be
15802 computed on some input keys. It is slower than the other algorithms and it must
15803 not be used for security purposes as a 32-bit hash is trivial to break. See
15804 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010015805
Christopher Fauletea159d62020-04-01 16:21:44 +020015806cut_crlf
15807 Cuts the string representation of the input sample on the first carriage
15808 return ('\r') or newline ('\n') character found. Only the string length is
15809 updated.
15810
David Carlier29b3ca32015-09-25 14:09:21 +010015811da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020015812 Asks the DeviceAtlas converter to identify the User Agent string passed on
15813 input, and to emit a string made of the concatenation of the properties
15814 enumerated in argument, delimited by the separator defined by the global
15815 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000015816 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020015817 configuration language.
15818
15819 Example:
15820 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015821 bind *:8881
15822 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000015823 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 +020015824
Willy Tarreau0851fd52019-12-17 10:07:25 +010015825debug([<prefix][,<destination>])
15826 This converter is used as debug tool. It takes a capture of the input sample
15827 and sends it to event sink <destination>, which may designate a ring buffer
15828 such as "buf0", as well as "stdout", or "stderr". Available sinks may be
15829 checked at run time by issuing "show events" on the CLI. When not specified,
15830 the output will be "buf0", which may be consulted via the CLI's "show events"
15831 command. An optional prefix <prefix> may be passed to help distinguish
15832 outputs from multiple expressions. It will then appear before the colon in
15833 the output message. The input sample is passed as-is on the output, so that
15834 it is safe to insert the debug converter anywhere in a chain, even with non-
15835 printable sample types.
15836
15837 Example:
15838 tcp-request connection track-sc0 src,debug(track-sc)
Thierry FOURNIER9687c772015-05-07 15:46:29 +020015839
Patrick Gansterer8e366512020-04-22 16:47:57 +020015840digest(<algorithm>)
15841 Converts a binary input sample to a message digest. The result is a binary
15842 sample. The <algorithm> must be an OpenSSL message digest name (e.g. sha256).
15843
15844 Please note that this converter is only available when haproxy has been
15845 compiled with USE_OPENSSL.
15846
Willy Tarreau97707872015-01-27 15:12:13 +010015847div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015848 Divides the input value of type signed integer by <value>, and returns the
15849 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015850 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010015851 variable name. The name of the variable starts with an indication about its
15852 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015853 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015854 "sess" : the variable is shared with the whole session
15855 "txn" : the variable is shared with the transaction (request and response)
15856 "req" : the variable is shared only during request processing
15857 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015858 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015859 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015860
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015861djb2([<avalanche>])
15862 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
15863 hash function. Optionally, it is possible to apply a full avalanche hash
15864 function to the output if the optional <avalanche> argument equals 1. This
15865 converter uses the same functions as used by the various hash-based load
15866 balancing algorithms, so it will provide exactly the same results. It is
15867 mostly intended for debugging, but can be used as a stick-table entry to
15868 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015869 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
15870 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015871
Willy Tarreau97707872015-01-27 15:12:13 +010015872even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015873 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010015874 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
15875
Marcin Deranek9631a282018-04-16 14:30:46 +020015876field(<index>,<delimiters>[,<count>])
15877 Extracts the substring at the given index counting from the beginning
15878 (positive index) or from the end (negative index) considering given delimiters
15879 from an input string. Indexes start at 1 or -1 and delimiters are a string
15880 formatted list of chars. Optionally you can specify <count> of fields to
15881 extract (default: 1). Value of 0 indicates extraction of all remaining
15882 fields.
15883
15884 Example :
15885 str(f1_f2_f3__f5),field(5,_) # f5
15886 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
15887 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
15888 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
15889 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010015890
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015891fix_is_valid
15892 Parses a binary payload and performs sanity checks regarding FIX (Financial
15893 Information eXchange):
15894
15895 - checks that all tag IDs and values are not empty and the tags IDs are well
15896 numeric
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050015897 - checks the BeginString tag is the first tag with a valid FIX version
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015898 - checks the BodyLength tag is the second one with the right body length
Christopher Fauleted4bef72021-03-18 17:40:56 +010015899 - checks the MsgType tag is the third tag.
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015900 - checks that last tag in the message is the CheckSum tag with a valid
15901 checksum
15902
15903 Due to current HAProxy design, only the first message sent by the client and
15904 the server can be parsed.
15905
15906 This converter returns a boolean, true if the payload contains a valid FIX
15907 message, false if not.
15908
15909 See also the fix_tag_value converter.
15910
15911 Example:
15912 tcp-request inspect-delay 10s
15913 tcp-request content reject unless { req.payload(0,0),fix_is_valid }
15914
15915fix_tag_value(<tag>)
15916 Parses a FIX (Financial Information eXchange) message and extracts the value
15917 from the tag <tag>. <tag> can be a string or an integer pointing to the
15918 desired tag. Any integer value is accepted, but only the following strings
15919 are translated into their integer equivalent: BeginString, BodyLength,
Daniel Corbettbefef702021-03-09 23:00:34 -050015920 MsgType, SenderCompID, TargetCompID, CheckSum. More tag names can be easily
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015921 added.
15922
15923 Due to current HAProxy design, only the first message sent by the client and
15924 the server can be parsed. No message validation is performed by this
15925 converter. It is highly recommended to validate the message first using
15926 fix_is_valid converter.
15927
15928 See also the fix_is_valid converter.
15929
15930 Example:
15931 tcp-request inspect-delay 10s
15932 tcp-request content reject unless { req.payload(0,0),fix_is_valid }
15933 # MsgType tag ID is 35, so both lines below will return the same content
15934 tcp-request content set-var(txn.foo) req.payload(0,0),fix_tag_value(35)
15935 tcp-request content set-var(txn.bar) req.payload(0,0),fix_tag_value(MsgType)
15936
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015937hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010015938 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015939 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010015940 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 +020015941 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010015942
Dragan Dosen3f957b22017-10-24 09:27:34 +020015943hex2i
15944 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050015945 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020015946
Christopher Faulet4ccc12f2020-04-01 09:08:32 +020015947htonl
15948 Converts the input integer value to its 32-bit binary representation in the
15949 network byte order. Because sample fetches own signed 64-bit integer, when
15950 this converter is used, the input integer value is first casted to an
15951 unsigned 32-bit integer.
15952
Tim Duesterhusa3082092021-01-21 17:40:49 +010015953hmac(<algorithm>,<key>)
Patrick Gansterer8e366512020-04-22 16:47:57 +020015954 Converts a binary input sample to a message authentication code with the given
15955 key. The result is a binary sample. The <algorithm> must be one of the
15956 registered OpenSSL message digest names (e.g. sha256). The <key> parameter must
15957 be base64 encoded and can either be a string or a variable.
15958
15959 Please note that this converter is only available when haproxy has been
15960 compiled with USE_OPENSSL.
15961
Cyril Bonté6bcd1822019-11-05 23:13:59 +010015962http_date([<offset],[<unit>])
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015963 Converts an integer supposed to contain a date since epoch to a string
15964 representing this date in a format suitable for use in HTTP header fields. If
Damien Claisseae6f1252019-10-30 15:57:28 +000015965 an offset value is specified, then it is added to the date before the
15966 conversion is operated. This is particularly useful to emit Date header fields,
15967 Expires values in responses when combined with a positive offset, or
15968 Last-Modified values when the offset is negative.
15969 If a unit value is specified, then consider the timestamp as either
15970 "s" for seconds (default behavior), "ms" for milliseconds, or "us" for
15971 microseconds since epoch. Offset is assumed to have the same unit as
15972 input timestamp.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015973
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015974iif(<true>,<false>)
15975 Returns the <true> string if the input value is true. Returns the <false>
15976 string otherwise.
15977
15978 Example:
Tim Duesterhus870713b2020-09-11 17:13:12 +020015979 http-request set-header x-forwarded-proto %[ssl_fc,iif(https,http)]
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015980
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015981in_table(<table>)
15982 Uses the string representation of the input sample to perform a look up in
15983 the specified table. If the key is not found in the table, a boolean false
15984 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010015985 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015986 or not a source IP address or an Authorization header was already seen).
15987
Tim Duesterhusa3082092021-01-21 17:40:49 +010015988ipmask(<mask4>,[<mask6>])
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015989 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015990 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015991 table entries and as such use the same server. The mask4 can be passed in
15992 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
15993 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
15994 If no mask6 is given IPv6 addresses will fail to convert for backwards
15995 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015996
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015997json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010015998 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015999 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020016000 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016001 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
16002 of errors:
16003 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
16004 bytes, ...)
16005 - invalid range (the decoded value is within a UTF-8 prohibited range),
16006 - code overlong (the value is encoded with more bytes than necessary).
16007
16008 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
16009 character is greater than 0xffff because the JSON string escape specification
16010 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
16011 in 4 variants designated by a combination of two suffix letters : "p" for
16012 "permissive" and "s" for "silently ignore". The behaviors of the decoders
16013 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010016014 - "ascii" : never fails;
16015 - "utf8" : fails on any detected errors;
16016 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016017 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010016018 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016019 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
16020 characters corresponding to the other errors.
16021
16022 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010016023 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016024
16025 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016026 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020016027 capture request header user-agent len 150
16028 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020016029
16030 Input request from client 127.0.0.1:
16031 GET / HTTP/1.0
16032 User-Agent: Very "Ugly" UA 1/2
16033
16034 Output log:
16035 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
16036
Alex51c8ad42021-04-15 16:45:15 +020016037json_query(<json_path>,[<output_type>])
16038 The json_query converter supports the JSON types string, boolean and
16039 number. Floating point numbers will be returned as a string. By
16040 specifying the output_type 'int' the value will be converted to an
16041 Integer. If conversion is not possible the json_query converter fails.
16042
16043 <json_path> must be a valid JSON Path string as defined in
16044 https://datatracker.ietf.org/doc/draft-ietf-jsonpath-base/
16045
16046 Example:
16047 # get a integer value from the request body
16048 # "{"integer":4}" => 5
16049 http-request set-var(txn.pay_int) req.body,json_query('$.integer','int'),add(1)
16050
16051 # get a key with '.' in the name
16052 # {"my.key":"myvalue"} => myvalue
16053 http-request set-var(txn.pay_mykey) req.body,json_query('$.my\\.key')
16054
16055 # {"boolean-false":false} => 0
16056 http-request set-var(txn.pay_boolean_false) req.body,json_query('$.boolean-false')
16057
16058 # get the value of the key 'iss' from a JWT Bearer token
16059 http-request set-var(txn.token_payload) req.hdr(Authorization),word(2,.),ub64dec,json_query('$.iss')
16060
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016061language(<value>[,<default>])
16062 Returns the value with the highest q-factor from a list as extracted from the
16063 "accept-language" header using "req.fhdr". Values with no q-factor have a
16064 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
16065 belong to the list of semi-colon delimited <values> will be considered. The
16066 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
16067 given list and a default value is provided, it is returned. Note that language
16068 names may have a variant after a dash ('-'). If this variant is present in the
16069 list, it will be matched, but if it is not, only the base language is checked.
16070 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010016071 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016072 ordering of the values in the request counts, as the first value among
16073 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020016074
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016075 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020016076
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016077 # this configuration switches to the backend matching a
16078 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020016079
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016080 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
16081 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
16082 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
16083 use_backend spanish if es
16084 use_backend french if fr
16085 use_backend english if en
16086 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020016087
Willy Tarreau60a2ee72017-12-15 07:13:48 +010016088length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010016089 Get the length of the string. This can only be placed after a string
16090 sample fetch function or after a transformation keyword returning a string
16091 type. The result is of type integer.
16092
Willy Tarreauffcb2e42014-07-10 16:29:08 +020016093lower
16094 Convert a string sample to lower case. This can only be placed after a string
16095 sample fetch function or after a transformation keyword returning a string
16096 type. The result is of type string.
16097
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016098ltime(<format>[,<offset>])
16099 Converts an integer supposed to contain a date since epoch to a string
16100 representing this date in local time using a format defined by the <format>
16101 string using strftime(3). The purpose is to allow any date format to be used
16102 in logs. An optional <offset> in seconds may be applied to the input date
16103 (positive or negative). See the strftime() man page for the format supported
16104 by your operating system. See also the utime converter.
16105
16106 Example :
16107
16108 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010016109 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016110 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
16111
Christopher Faulet51fc9d12020-04-01 17:24:41 +020016112ltrim(<chars>)
16113 Skips any characters from <chars> from the beginning of the string
16114 representation of the input sample.
16115
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016116map(<map_file>[,<default_value>])
16117map_<match_type>(<map_file>[,<default_value>])
16118map_<match_type>_<output_type>(<map_file>[,<default_value>])
16119 Search the input value from <map_file> using the <match_type> matching method,
16120 and return the associated value converted to the type <output_type>. If the
16121 input value cannot be found in the <map_file>, the converter returns the
16122 <default_value>. If the <default_value> is not set, the converter fails and
16123 acts as if no input value could be fetched. If the <match_type> is not set, it
16124 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
16125 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
16126 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016127
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016128 It is important to avoid overlapping between the keys : IP addresses and
16129 strings are stored in trees, so the first of the finest match will be used.
16130 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016131
Tim Düsterhus4896c442016-11-29 02:15:19 +010016132 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016133 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016134
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016135 input type | match method | output type str | output type int | output type ip
16136 -----------+--------------+-----------------+-----------------+---------------
16137 str | str | map_str | map_str_int | map_str_ip
16138 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020016139 str | beg | map_beg | map_beg_int | map_end_ip
16140 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016141 str | sub | map_sub | map_sub_int | map_sub_ip
16142 -----------+--------------+-----------------+-----------------+---------------
16143 str | dir | map_dir | map_dir_int | map_dir_ip
16144 -----------+--------------+-----------------+-----------------+---------------
16145 str | dom | map_dom | map_dom_int | map_dom_ip
16146 -----------+--------------+-----------------+-----------------+---------------
16147 str | end | map_end | map_end_int | map_end_ip
16148 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080016149 str | reg | map_reg | map_reg_int | map_reg_ip
16150 -----------+--------------+-----------------+-----------------+---------------
16151 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016152 -----------+--------------+-----------------+-----------------+---------------
16153 int | int | map_int | map_int_int | map_int_ip
16154 -----------+--------------+-----------------+-----------------+---------------
16155 ip | ip | map_ip | map_ip_int | map_ip_ip
16156 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016157
Thierry Fournier8feaa662016-02-10 22:55:20 +010016158 The special map called "map_regm" expect matching zone in the regular
16159 expression and modify the output replacing back reference (like "\1") by
16160 the corresponding match text.
16161
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016162 The file contains one key + value per line. Lines which start with '#' are
16163 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
16164 is then the first "word" (series of non-space/tabs characters), and the value
16165 is what follows this series of space/tab till the end of the line excluding
16166 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016167
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016168 Example :
16169
16170 # this is a comment and is ignored
16171 2.22.246.0/23 United Kingdom \n
16172 <-><-----------><--><------------><---->
16173 | | | | `- trailing spaces ignored
16174 | | | `---------- value
16175 | | `-------------------- middle spaces ignored
16176 | `---------------------------- key
16177 `------------------------------------ leading spaces ignored
16178
Willy Tarreau97707872015-01-27 15:12:13 +010016179mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016180 Divides the input value of type signed integer by <value>, and returns the
16181 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016182 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016183 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016184 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016185 "sess" : the variable is shared with the whole session
16186 "txn" : the variable is shared with the transaction (request and response)
16187 "req" : the variable is shared only during request processing
16188 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016189 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016190 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016191
Baptiste Assmanne279ca62020-10-27 18:10:06 +010016192mqtt_field_value(<packettype>,<fieldname or property ID>)
16193 Returns value of <fieldname> found in input MQTT payload of type
16194 <packettype>.
16195 <packettype> can be either a string (case insensitive matching) or a numeric
16196 value corresponding to the type of packet we're supposed to extract data
16197 from.
16198 Supported string and integers can be found here:
16199 https://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html#_Toc398718021
16200 https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html#_Toc3901022
16201
16202 <fieldname> depends on <packettype> and can be any of the following below.
16203 (note that <fieldname> matching is case insensitive).
16204 <property id> can only be found in MQTT v5.0 streams. check this table:
16205 https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html#_Toc3901029
16206
16207 - CONNECT (or 1): flags, protocol_name, protocol_version, client_identifier,
16208 will_topic, will_payload, username, password, keepalive
16209 OR any property ID as a numeric value (for MQTT v5.0
16210 packets only):
16211 17: Session Expiry Interval
16212 33: Receive Maximum
16213 39: Maximum Packet Size
16214 34: Topic Alias Maximum
16215 25: Request Response Information
16216 23: Request Problem Information
16217 21: Authentication Method
16218 22: Authentication Data
16219 18: Will Delay Interval
16220 1: Payload Format Indicator
16221 2: Message Expiry Interval
16222 3: Content Type
16223 8: Response Topic
16224 9: Correlation Data
16225 Not supported yet:
16226 38: User Property
16227
16228 - CONNACK (or 2): flags, protocol_version, reason_code
16229 OR any property ID as a numeric value (for MQTT v5.0
16230 packets only):
16231 17: Session Expiry Interval
16232 33: Receive Maximum
16233 36: Maximum QoS
16234 37: Retain Available
16235 39: Maximum Packet Size
16236 18: Assigned Client Identifier
16237 34: Topic Alias Maximum
16238 31: Reason String
16239 40; Wildcard Subscription Available
16240 41: Subscription Identifiers Available
16241 42: Shared Subscription Available
16242 19: Server Keep Alive
16243 26: Response Information
16244 28: Server Reference
16245 21: Authentication Method
16246 22: Authentication Data
16247 Not supported yet:
16248 38: User Property
16249
16250 Due to current HAProxy design, only the first message sent by the client and
16251 the server can be parsed. Thus this converter can extract data only from
16252 CONNECT and CONNACK packet types. CONNECT is the first message sent by the
16253 client and CONNACK is the first response sent by the server.
16254
16255 Example:
16256
16257 acl data_in_buffer req.len ge 4
16258 tcp-request content set-var(txn.username) \
16259 req.payload(0,0),mqtt_field_value(connect,protocol_name) \
16260 if data_in_buffer
16261 # do the same as above
16262 tcp-request content set-var(txn.username) \
16263 req.payload(0,0),mqtt_field_value(1,protocol_name) \
16264 if data_in_buffer
16265
16266mqtt_is_valid
16267 Checks that the binary input is a valid MQTT packet. It returns a boolean.
16268
16269 Due to current HAProxy design, only the first message sent by the client and
16270 the server can be parsed. Thus this converter can extract data only from
16271 CONNECT and CONNACK packet types. CONNECT is the first message sent by the
16272 client and CONNACK is the first response sent by the server.
16273
16274 Example:
16275
16276 acl data_in_buffer req.len ge 4
16277 tcp-request content reject unless req.payload(0,0),mqtt_is_valid
16278
Willy Tarreau97707872015-01-27 15:12:13 +010016279mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016280 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020016281 the product as an signed integer. In case of overflow, the largest possible
16282 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016283 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016284 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016285 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016286 "sess" : the variable is shared with the whole session
16287 "txn" : the variable is shared with the transaction (request and response)
16288 "req" : the variable is shared only during request processing
16289 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016290 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016291 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016292
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010016293nbsrv
16294 Takes an input value of type string, interprets it as a backend name and
16295 returns the number of usable servers in that backend. Can be used in places
16296 where we want to look up a backend from a dynamic name, like a result of a
16297 map lookup.
16298
Willy Tarreau97707872015-01-27 15:12:13 +010016299neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016300 Takes the input value of type signed integer, computes the opposite value,
16301 and returns the remainder as an signed integer. 0 is identity. This operator
16302 is provided for reversed subtracts : in order to subtract the input from a
16303 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010016304
16305not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016306 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010016307 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016308 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010016309 absence of a flag).
16310
16311odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016312 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010016313 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
16314
16315or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016316 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016317 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010016318 numeric value or a variable name. The name of the variable starts with an
16319 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016320 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016321 "sess" : the variable is shared with the whole session
16322 "txn" : the variable is shared with the transaction (request and response)
16323 "req" : the variable is shared only during request processing
16324 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016325 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016326 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016327
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016328protobuf(<field_number>,[<field_type>])
16329 This extracts the protocol buffers message field in raw mode of an input binary
16330 sample representation of a protocol buffer message with <field_number> as field
16331 number (dotted notation) if <field_type> is not present, or as an integer sample
16332 if this field is present (see also "ungrpc" below).
16333 The list of the authorized types is the following one: "int32", "int64", "uint32",
16334 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
16335 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
16336 "float" for the wire type 5. Note that "string" is considered as a length-delimited
16337 type, so it does not require any <field_type> argument to be extracted.
16338 More information may be found here about the protocol buffers message field types:
16339 https://developers.google.com/protocol-buffers/docs/encoding
16340
Willy Tarreauc4dc3502015-01-23 20:39:28 +010016341regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010016342 Applies a regex-based substitution to the input string. It does the same
16343 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
16344 default it will replace in the input string the first occurrence of the
16345 largest part matching the regular expression <regex> with the substitution
16346 string <subst>. It is possible to replace all occurrences instead by adding
16347 the flag "g" in the third argument <flags>. It is also possible to make the
16348 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
16349 string, it is made up from the concatenation of all desired flags. Thus if
16350 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
Willy Tarreauef21fac2020-02-14 13:37:20 +010016351 The first use of this converter is to replace certain characters or sequence
16352 of characters with other ones.
16353
16354 It is highly recommended to enclose the regex part using protected quotes to
16355 improve clarity and never have a closing parenthesis from the regex mixed up
16356 with the parenthesis from the function. Just like in Bourne shell, the first
16357 level of quotes is processed when delimiting word groups on the line, a
16358 second level is usable for argument. It is recommended to use single quotes
16359 outside since these ones do not try to resolve backslashes nor dollar signs.
Willy Tarreau7eda8492015-01-20 19:47:06 +010016360
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010016361 Examples:
Willy Tarreau7eda8492015-01-20 19:47:06 +010016362
16363 # de-duplicate "/" in header "x-path".
16364 # input: x-path: /////a///b/c/xzxyz/
16365 # output: x-path: /a/b/c/xzxyz/
Willy Tarreauef21fac2020-02-14 13:37:20 +010016366 http-request set-header x-path "%[hdr(x-path),regsub('/+','/','g')]"
Willy Tarreau7eda8492015-01-20 19:47:06 +010016367
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010016368 # copy query string to x-query and drop all leading '?', ';' and '&'
16369 http-request set-header x-query "%[query,regsub([?;&]*,'')]"
16370
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010016371 # capture groups and backreferences
16372 # both lines do the same.
Willy Tarreau465dc7d2020-10-08 18:05:56 +020016373 http-request redirect location %[url,'regsub("(foo|bar)([0-9]+)?","\2\1",i)']
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010016374 http-request redirect location %[url,regsub(\"(foo|bar)([0-9]+)?\",\"\2\1\",i)]
16375
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020016376capture-req(<id>)
16377 Capture the string entry in the request slot <id> and returns the entry as
16378 is. If the slot doesn't exist, the capture fails silently.
16379
16380 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020016381 "http-response capture", "capture.req.hdr" and
16382 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020016383
16384capture-res(<id>)
16385 Capture the string entry in the response slot <id> and returns the entry as
16386 is. If the slot doesn't exist, the capture fails silently.
16387
16388 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020016389 "http-response capture", "capture.req.hdr" and
16390 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020016391
Christopher Faulet568415a2020-04-01 17:24:47 +020016392rtrim(<chars>)
16393 Skips any characters from <chars> from the end of the string representation
16394 of the input sample.
16395
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016396sdbm([<avalanche>])
16397 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
16398 hash function. Optionally, it is possible to apply a full avalanche hash
16399 function to the output if the optional <avalanche> argument equals 1. This
16400 converter uses the same functions as used by the various hash-based load
16401 balancing algorithms, so it will provide exactly the same results. It is
16402 mostly intended for debugging, but can be used as a stick-table entry to
16403 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010016404 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
16405 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016406
Tim Duesterhusf38175c2020-06-09 11:48:42 +020016407secure_memcmp(<var>)
16408 Compares the contents of <var> with the input value. Both values are treated
16409 as a binary string. Returns a boolean indicating whether both binary strings
16410 match.
16411
16412 If both binary strings have the same length then the comparison will be
16413 performed in constant time.
16414
16415 Please note that this converter is only available when haproxy has been
16416 compiled with USE_OPENSSL.
16417
16418 Example :
16419
16420 http-request set-var(txn.token) hdr(token)
16421 # Check whether the token sent by the client matches the secret token
16422 # value, without leaking the contents using a timing attack.
16423 acl token_given str(my_secret_token),secure_memcmp(txn.token)
16424
Tim Duesterhusef4e45c2021-01-21 17:40:50 +010016425set-var(<var>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010016426 Sets a variable with the input content and returns the content on the output
16427 as-is. The variable keeps the value and the associated input type. The name of
16428 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016429 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016430 "sess" : the variable is shared with the whole session
16431 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016432 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016433 "req" : the variable is shared only during request processing,
16434 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016435 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016436 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016437
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020016438sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020016439 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020016440 sample with length of 20 bytes.
16441
Tim Duesterhusd4376302019-06-17 12:41:44 +020016442sha2([<bits>])
16443 Converts a binary input sample to a digest in the SHA-2 family. The result
16444 is a binary sample with length of <bits>/8 bytes.
16445
16446 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
16447 SHA-<bits>. The default value is 256.
16448
16449 Please note that this converter is only available when haproxy has been
16450 compiled with USE_OPENSSL.
16451
Nenad Merdanovic177adc92019-08-27 01:58:13 +020016452srv_queue
16453 Takes an input value of type string, either a server name or <backend>/<server>
16454 format and returns the number of queued sessions on that server. Can be used
16455 in places where we want to look up queued sessions from a dynamic name, like a
16456 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
16457 persistence or direct a request elsewhere.
16458
Tim Duesterhusca097c12018-04-27 21:18:45 +020016459strcmp(<var>)
16460 Compares the contents of <var> with the input value of type string. Returns
16461 the result as a signed integer compatible with strcmp(3): 0 if both strings
16462 are identical. A value less than 0 if the left string is lexicographically
16463 smaller than the right string or if the left string is shorter. A value greater
16464 than 0 otherwise (right string greater than left string or the right string is
16465 shorter).
16466
Tim Duesterhusf38175c2020-06-09 11:48:42 +020016467 See also the secure_memcmp converter if you need to compare two binary
16468 strings in constant time.
16469
Tim Duesterhusca097c12018-04-27 21:18:45 +020016470 Example :
16471
16472 http-request set-var(txn.host) hdr(host)
16473 # Check whether the client is attempting domain fronting.
16474 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
16475
16476
Willy Tarreau97707872015-01-27 15:12:13 +010016477sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016478 Subtracts <value> from the input value of type signed integer, and returns
16479 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016480 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010016481 or a variable name. The name of the variable starts with an indication about
16482 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016483 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016484 "sess" : the variable is shared with the whole session
16485 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016486 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016487 "req" : the variable is shared only during request processing,
16488 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016489 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016490 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016491
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016492table_bytes_in_rate(<table>)
16493 Uses the string representation of the input sample to perform a look up in
16494 the specified table. If the key is not found in the table, integer value zero
16495 is returned. Otherwise the converter returns the average client-to-server
16496 bytes rate associated with the input sample in the designated table, measured
16497 in amount of bytes over the period configured in the table. See also the
16498 sc_bytes_in_rate sample fetch keyword.
16499
16500
16501table_bytes_out_rate(<table>)
16502 Uses the string representation of the input sample to perform a look up in
16503 the specified table. If the key is not found in the table, integer value zero
16504 is returned. Otherwise the converter returns the average server-to-client
16505 bytes rate associated with the input sample in the designated table, measured
16506 in amount of bytes over the period configured in the table. See also the
16507 sc_bytes_out_rate sample fetch keyword.
16508
16509table_conn_cnt(<table>)
16510 Uses the string representation of the input sample to perform a look up in
16511 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016512 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016513 connections associated with the input sample in the designated table. See
16514 also the sc_conn_cnt sample fetch keyword.
16515
16516table_conn_cur(<table>)
16517 Uses the string representation of the input sample to perform a look up in
16518 the specified table. If the key is not found in the table, integer value zero
16519 is returned. Otherwise the converter returns the current amount of concurrent
16520 tracked connections associated with the input sample in the designated table.
16521 See also the sc_conn_cur sample fetch keyword.
16522
16523table_conn_rate(<table>)
16524 Uses the string representation of the input sample to perform a look up in
16525 the specified table. If the key is not found in the table, integer value zero
16526 is returned. Otherwise the converter returns the average incoming connection
16527 rate associated with the input sample in the designated table. See also the
16528 sc_conn_rate sample fetch keyword.
16529
Thierry FOURNIER236657b2015-08-19 08:25:14 +020016530table_gpt0(<table>)
16531 Uses the string representation of the input sample to perform a look up in
16532 the specified table. If the key is not found in the table, boolean value zero
16533 is returned. Otherwise the converter returns the current value of the first
16534 general purpose tag associated with the input sample in the designated table.
16535 See also the sc_get_gpt0 sample fetch keyword.
16536
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016537table_gpc0(<table>)
16538 Uses the string representation of the input sample to perform a look up in
16539 the specified table. If the key is not found in the table, integer value zero
16540 is returned. Otherwise the converter returns the current value of the first
16541 general purpose counter associated with the input sample in the designated
16542 table. See also the sc_get_gpc0 sample fetch keyword.
16543
16544table_gpc0_rate(<table>)
16545 Uses the string representation of the input sample to perform a look up in
16546 the specified table. If the key is not found in the table, integer value zero
16547 is returned. Otherwise the converter returns the frequency which the gpc0
16548 counter was incremented over the configured period in the table, associated
16549 with the input sample in the designated table. See also the sc_get_gpc0_rate
16550 sample fetch keyword.
16551
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016552table_gpc1(<table>)
16553 Uses the string representation of the input sample to perform a look up in
16554 the specified table. If the key is not found in the table, integer value zero
16555 is returned. Otherwise the converter returns the current value of the second
16556 general purpose counter associated with the input sample in the designated
16557 table. See also the sc_get_gpc1 sample fetch keyword.
16558
16559table_gpc1_rate(<table>)
16560 Uses the string representation of the input sample to perform a look up in
16561 the specified table. If the key is not found in the table, integer value zero
16562 is returned. Otherwise the converter returns the frequency which the gpc1
16563 counter was incremented over the configured period in the table, associated
16564 with the input sample in the designated table. See also the sc_get_gpc1_rate
16565 sample fetch keyword.
16566
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016567table_http_err_cnt(<table>)
16568 Uses the string representation of the input sample to perform a look up in
16569 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016570 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016571 errors associated with the input sample in the designated table. See also the
16572 sc_http_err_cnt sample fetch keyword.
16573
16574table_http_err_rate(<table>)
16575 Uses the string representation of the input sample to perform a look up in
16576 the specified table. If the key is not found in the table, integer value zero
16577 is returned. Otherwise the average rate of HTTP errors associated with the
16578 input sample in the designated table, measured in amount of errors over the
16579 period configured in the table. See also the sc_http_err_rate sample fetch
16580 keyword.
16581
Willy Tarreau826f3ab2021-02-10 12:07:15 +010016582table_http_fail_cnt(<table>)
16583 Uses the string representation of the input sample to perform a look up in
16584 the specified table. If the key is not found in the table, integer value zero
16585 is returned. Otherwise the converter returns the cumulative number of HTTP
16586 failures associated with the input sample in the designated table. See also
16587 the sc_http_fail_cnt sample fetch keyword.
16588
16589table_http_fail_rate(<table>)
16590 Uses the string representation of the input sample to perform a look up in
16591 the specified table. If the key is not found in the table, integer value zero
16592 is returned. Otherwise the average rate of HTTP failures associated with the
16593 input sample in the designated table, measured in amount of failures over the
16594 period configured in the table. See also the sc_http_fail_rate sample fetch
16595 keyword.
16596
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016597table_http_req_cnt(<table>)
16598 Uses the string representation of the input sample to perform a look up in
16599 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016600 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016601 requests associated with the input sample in the designated table. See also
16602 the sc_http_req_cnt sample fetch keyword.
16603
16604table_http_req_rate(<table>)
16605 Uses the string representation of the input sample to perform a look up in
16606 the specified table. If the key is not found in the table, integer value zero
16607 is returned. Otherwise the average rate of HTTP requests associated with the
16608 input sample in the designated table, measured in amount of requests over the
16609 period configured in the table. See also the sc_http_req_rate sample fetch
16610 keyword.
16611
16612table_kbytes_in(<table>)
16613 Uses the string representation of the input sample to perform a look up in
16614 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016615 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016616 to-server data associated with the input sample in the designated table,
16617 measured in kilobytes. The test is currently performed on 32-bit integers,
16618 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
16619 keyword.
16620
16621table_kbytes_out(<table>)
16622 Uses the string representation of the input sample to perform a look up in
16623 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016624 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016625 to-client data associated with the input sample in the designated table,
16626 measured in kilobytes. The test is currently performed on 32-bit integers,
16627 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
16628 keyword.
16629
16630table_server_id(<table>)
16631 Uses the string representation of the input sample to perform a look up in
16632 the specified table. If the key is not found in the table, integer value zero
16633 is returned. Otherwise the converter returns the server ID associated with
16634 the input sample in the designated table. A server ID is associated to a
16635 sample by a "stick" rule when a connection to a server succeeds. A server ID
16636 zero means that no server is associated with this key.
16637
16638table_sess_cnt(<table>)
16639 Uses the string representation of the input sample to perform a look up in
16640 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016641 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016642 sessions associated with the input sample in the designated table. Note that
16643 a session here refers to an incoming connection being accepted by the
16644 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
16645 keyword.
16646
16647table_sess_rate(<table>)
16648 Uses the string representation of the input sample to perform a look up in
16649 the specified table. If the key is not found in the table, integer value zero
16650 is returned. Otherwise the converter returns the average incoming session
16651 rate associated with the input sample in the designated table. Note that a
16652 session here refers to an incoming connection being accepted by the
16653 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
16654 keyword.
16655
16656table_trackers(<table>)
16657 Uses the string representation of the input sample to perform a look up in
16658 the specified table. If the key is not found in the table, integer value zero
16659 is returned. Otherwise the converter returns the current amount of concurrent
16660 connections tracking the same key as the input sample in the designated
16661 table. It differs from table_conn_cur in that it does not rely on any stored
16662 information but on the table's reference count (the "use" value which is
16663 returned by "show table" on the CLI). This may sometimes be more suited for
16664 layer7 tracking. It can be used to tell a server how many concurrent
16665 connections there are from a given address for example. See also the
16666 sc_trackers sample fetch keyword.
16667
Moemen MHEDHBI92f7d432021-04-01 20:53:59 +020016668ub64dec
16669 This converter is the base64url variant of b64dec converter. base64url
16670 encoding is the "URL and Filename Safe Alphabet" variant of base64 encoding.
16671 It is also the encoding used in JWT (JSON Web Token) standard.
16672
16673 Example:
16674 # Decoding a JWT payload:
16675 http-request set-var(txn.token_payload) req.hdr(Authorization),word(2,.),ub64dec
16676
16677ub64enc
16678 This converter is the base64url variant of base64 converter.
16679
Willy Tarreauffcb2e42014-07-10 16:29:08 +020016680upper
16681 Convert a string sample to upper case. This can only be placed after a string
16682 sample fetch function or after a transformation keyword returning a string
16683 type. The result is of type string.
16684
Willy Tarreau62ba9ba2020-04-23 17:54:47 +020016685url_dec([<in_form>])
16686 Takes an url-encoded string provided as input and returns the decoded version
16687 as output. The input and the output are of type string. If the <in_form>
16688 argument is set to a non-zero integer value, the input string is assumed to
16689 be part of a form or query string and the '+' character will be turned into a
16690 space (' '). Otherwise this will only happen after a question mark indicating
16691 a query string ('?').
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020016692
William Dauchy888b0ae2021-01-06 23:39:50 +010016693url_enc([<enc_type>])
16694 Takes a string provided as input and returns the encoded version as output.
16695 The input and the output are of type string. By default the type of encoding
16696 is meant for `query` type. There is no other type supported for now but the
16697 optional argument is here for future changes.
16698
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016699ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016700 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016701 sample representation of a gRPC message with <field_number> as field number
16702 (dotted notation) if <field_type> is not present, or as an integer sample if this
16703 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016704 The list of the authorized types is the following one: "int32", "int64", "uint32",
16705 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
16706 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
16707 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010016708 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016709 More information may be found here about the protocol buffers message field types:
16710 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016711
16712 Example:
16713 // with such a protocol buffer .proto file content adapted from
16714 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
16715
16716 message Point {
16717 int32 latitude = 1;
16718 int32 longitude = 2;
16719 }
16720
16721 message PPoint {
16722 Point point = 59;
16723 }
16724
16725 message Rectangle {
16726 // One corner of the rectangle.
16727 PPoint lo = 48;
16728 // The other corner of the rectangle.
16729 PPoint hi = 49;
16730 }
16731
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016732 let's say a body request is made of a "Rectangle" object value (two PPoint
16733 protocol buffers messages), the four protocol buffers fields could be
16734 extracted with these "ungrpc" directives:
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016735
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016736 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
16737 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050016738 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016739 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
16740
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016741 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016742
Frédéric Lécaille93d33162019-03-06 09:35:59 +010016743 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016744
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016745 As a gRPC message is always made of a gRPC header followed by protocol buffers
16746 messages, in the previous example the "latitude" of "lo" first PPoint
16747 could be extracted with these equivalent directives:
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016748
16749 req.body,ungrpc(48.59),protobuf(1,int32)
16750 req.body,ungrpc(48),protobuf(59.1,int32)
16751 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
16752
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016753 Note that the first convert must be "ungrpc", the remaining ones must be
16754 "protobuf" and only the last one may have or not a second argument to
16755 interpret the previous binary sample.
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016756
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016757
Tim Duesterhusef4e45c2021-01-21 17:40:50 +010016758unset-var(<var>)
Christopher Faulet85d79c92016-11-09 16:54:56 +010016759 Unsets a variable if the input content is defined. The name of the variable
16760 starts with an indication about its scope. The scopes allowed are:
16761 "proc" : the variable is shared with the whole process
16762 "sess" : the variable is shared with the whole session
16763 "txn" : the variable is shared with the transaction (request and
16764 response),
16765 "req" : the variable is shared only during request processing,
16766 "res" : the variable is shared only during response processing.
16767 This prefix is followed by a name. The separator is a '.'. The name may only
16768 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
16769
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016770utime(<format>[,<offset>])
16771 Converts an integer supposed to contain a date since epoch to a string
16772 representing this date in UTC time using a format defined by the <format>
16773 string using strftime(3). The purpose is to allow any date format to be used
16774 in logs. An optional <offset> in seconds may be applied to the input date
16775 (positive or negative). See the strftime() man page for the format supported
16776 by your operating system. See also the ltime converter.
16777
16778 Example :
16779
16780 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010016781 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016782 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
16783
Marcin Deranek9631a282018-04-16 14:30:46 +020016784word(<index>,<delimiters>[,<count>])
16785 Extracts the nth word counting from the beginning (positive index) or from
16786 the end (negative index) considering given delimiters from an input string.
16787 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
Jerome Magnin88209322020-01-28 13:33:44 +010016788 Delimiters at the beginning or end of the input string are ignored.
Marcin Deranek9631a282018-04-16 14:30:46 +020016789 Optionally you can specify <count> of words to extract (default: 1).
16790 Value of 0 indicates extraction of all remaining words.
16791
16792 Example :
16793 str(f1_f2_f3__f5),word(4,_) # f5
16794 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
16795 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
16796 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
16797 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Jerome Magnin88209322020-01-28 13:33:44 +010016798 str(/f1/f2/f3/f4),word(1,/) # f1
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010016799
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016800wt6([<avalanche>])
16801 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
16802 hash function. Optionally, it is possible to apply a full avalanche hash
16803 function to the output if the optional <avalanche> argument equals 1. This
16804 converter uses the same functions as used by the various hash-based load
16805 balancing algorithms, so it will provide exactly the same results. It is
16806 mostly intended for debugging, but can be used as a stick-table entry to
16807 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010016808 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
16809 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016810
Willy Tarreau97707872015-01-27 15:12:13 +010016811xor(<value>)
16812 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016813 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016814 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016815 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016816 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016817 "sess" : the variable is shared with the whole session
16818 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016819 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016820 "req" : the variable is shared only during request processing,
16821 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016822 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016823 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016824
Dragan Dosen04bf0cc2020-12-22 21:44:33 +010016825xxh3([<seed>])
16826 Hashes a binary input sample into a signed 64-bit quantity using the XXH3
16827 64-bit variant of the XXhash hash function. This hash supports a seed which
16828 defaults to zero but a different value maybe passed as the <seed> argument.
16829 This hash is known to be very good and very fast so it can be used to hash
16830 URLs and/or URL parameters for use as stick-table keys to collect statistics
16831 with a low collision rate, though care must be taken as the algorithm is not
16832 considered as cryptographically secure.
16833
Thierry FOURNIER01e09742016-12-26 11:46:11 +010016834xxh32([<seed>])
16835 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
16836 variant of the XXHash hash function. This hash supports a seed which defaults
16837 to zero but a different value maybe passed as the <seed> argument. This hash
16838 is known to be very good and very fast so it can be used to hash URLs and/or
16839 URL parameters for use as stick-table keys to collect statistics with a low
16840 collision rate, though care must be taken as the algorithm is not considered
16841 as cryptographically secure.
16842
16843xxh64([<seed>])
16844 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
16845 variant of the XXHash hash function. This hash supports a seed which defaults
16846 to zero but a different value maybe passed as the <seed> argument. This hash
16847 is known to be very good and very fast so it can be used to hash URLs and/or
16848 URL parameters for use as stick-table keys to collect statistics with a low
16849 collision rate, though care must be taken as the algorithm is not considered
16850 as cryptographically secure.
16851
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016852
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200168537.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020016854--------------------------------------------
16855
16856A first set of sample fetch methods applies to internal information which does
16857not even relate to any client information. These ones are sometimes used with
16858"monitor-fail" directives to report an internal status to external watchers.
16859The sample fetch methods described in this section are usable anywhere.
16860
16861always_false : boolean
16862 Always returns the boolean "false" value. It may be used with ACLs as a
16863 temporary replacement for another one when adjusting configurations.
16864
16865always_true : boolean
16866 Always returns the boolean "true" value. It may be used with ACLs as a
16867 temporary replacement for another one when adjusting configurations.
16868
16869avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016870 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020016871 divided by the number of active servers. The current backend is used if no
16872 backend is specified. This is very similar to "queue" except that the size of
16873 the farm is considered, in order to give a more accurate measurement of the
16874 time it may take for a new connection to be processed. The main usage is with
16875 ACL to return a sorry page to new users when it becomes certain they will get
16876 a degraded service, or to pass to the backend servers in a header so that
16877 they decide to work in degraded mode or to disable some functions to speed up
16878 the processing a bit. Note that in the event there would not be any active
16879 server anymore, twice the number of queued connections would be considered as
16880 the measured value. This is a fair estimate, as we expect one server to get
16881 back soon anyway, but we still prefer to send new traffic to another backend
16882 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
16883 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010016884
Willy Tarreau74ca5042013-06-11 23:12:07 +020016885be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020016886 Applies to the number of currently established connections on the backend,
16887 possibly including the connection being evaluated. If no backend name is
16888 specified, the current one is used. But it is also possible to check another
16889 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040016890 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
16891
16892be_conn_free([<backend>]) : integer
16893 Returns an integer value corresponding to the number of available connections
16894 across available servers in the backend. Queue slots are not included. Backup
16895 servers are also not included, unless all other servers are down. If no
16896 backend name is specified, the current one is used. But it is also possible
16897 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040016898 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
16899 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040016900
16901 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
16902 (meaning unlimited), then this fetch clearly does not make sense, in which
16903 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016904
Willy Tarreau74ca5042013-06-11 23:12:07 +020016905be_sess_rate([<backend>]) : integer
16906 Returns an integer value corresponding to the sessions creation rate on the
16907 backend, in number of new sessions per second. This is used with ACLs to
16908 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010016909 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020016910 online dictionary). It can also be useful to add this element to logs using a
16911 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016912
16913 Example :
16914 # Redirect to an error page if the dictionary is requested too often
16915 backend dynamic
16916 mode http
16917 acl being_scanned be_sess_rate gt 100
16918 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010016919
Davor Ocelice9ed2812017-12-25 17:49:28 +010016920bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016921 Returns a binary chain. The input is the hexadecimal representation
16922 of the string.
16923
16924bool(<bool>) : bool
16925 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
16926 'false' and '0' are the same. 'true' and '1' are the same.
16927
Willy Tarreau74ca5042013-06-11 23:12:07 +020016928connslots([<backend>]) : integer
16929 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016930 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020016931 connections on all servers and the maximum queue size. This is probably only
16932 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050016933
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016934 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020016935 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016936 usage; see "use_backend" keyword) can be redirected to a different backend.
16937
Willy Tarreau55165fe2009-05-10 12:02:55 +020016938 'connslots' = number of available server connection slots, + number of
16939 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016940
Willy Tarreaua36af912009-10-10 12:02:45 +020016941 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020016942 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020016943 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020016944 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010016945 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020016946 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020016947 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016948
Willy Tarreau55165fe2009-05-10 12:02:55 +020016949 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
16950 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020016951 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020016952 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016953
Willy Tarreau70fe9442018-11-22 16:07:39 +010016954cpu_calls : integer
16955 Returns the number of calls to the task processing the stream or current
16956 request since it was allocated. This number is reset for each new request on
16957 the same connections in case of HTTP keep-alive. This value should usually be
16958 low and stable (around 2 calls for a typically simple request) but may become
16959 high if some processing (compression, caching or analysis) is performed. This
16960 is purely for performance monitoring purposes.
16961
16962cpu_ns_avg : integer
16963 Returns the average number of nanoseconds spent in each call to the task
16964 processing the stream or current request. This number is reset for each new
16965 request on the same connections in case of HTTP keep-alive. This value
16966 indicates the overall cost of processing the request or the connection for
16967 each call. There is no good nor bad value but the time spent in a call
16968 automatically causes latency for other processing (see lat_ns_avg below),
16969 and may affect other connection's apparent response time. Certain operations
16970 like compression, complex regex matching or heavy Lua operations may directly
16971 affect this value, and having it in the logs will make it easier to spot the
16972 faulty processing that needs to be fixed to recover decent performance.
16973 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
16974
16975cpu_ns_tot : integer
16976 Returns the total number of nanoseconds spent in each call to the task
16977 processing the stream or current request. This number is reset for each new
16978 request on the same connections in case of HTTP keep-alive. This value
16979 indicates the overall cost of processing the request or the connection for
16980 each call. There is no good nor bad value but the time spent in a call
16981 automatically causes latency for other processing (see lat_ns_avg below),
16982 induces CPU costs on the machine, and may affect other connection's apparent
16983 response time. Certain operations like compression, complex regex matching or
16984 heavy Lua operations may directly affect this value, and having it in the
16985 logs will make it easier to spot the faulty processing that needs to be fixed
16986 to recover decent performance. The value may be artificially high due to a
16987 high cpu_calls count, for example when processing many HTTP chunks, and for
16988 this reason it is often preferred to log cpu_ns_avg instead.
16989
Cyril Bonté6bcd1822019-11-05 23:13:59 +010016990date([<offset>],[<unit>]) : integer
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016991 Returns the current date as the epoch (number of seconds since 01/01/1970).
Damien Claisseae6f1252019-10-30 15:57:28 +000016992
16993 If an offset value is specified, then it is added to the current date before
16994 returning the value. This is particularly useful to compute relative dates,
16995 as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020016996 It is useful combined with the http_date converter.
16997
Damien Claisseae6f1252019-10-30 15:57:28 +000016998 <unit> is facultative, and can be set to "s" for seconds (default behavior),
16999 "ms" for milliseconds or "us" for microseconds.
17000 If unit is set, return value is an integer reflecting either seconds,
17001 milliseconds or microseconds since epoch, plus offset.
17002 It is useful when a time resolution of less than a second is needed.
17003
Willy Tarreau276fae92013-07-25 14:36:01 +020017004 Example :
17005
17006 # set an expires header to now+1 hour in every response
17007 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020017008
Damien Claisseae6f1252019-10-30 15:57:28 +000017009 # set an expires header to now+1 hour in every response, with
17010 # millisecond granularity
17011 http-response set-header Expires %[date(3600000,ms),http_date(0,ms)]
17012
Etienne Carrierea792a0a2018-01-17 13:43:24 +010017013date_us : integer
17014 Return the microseconds part of the date (the "second" part is returned by
17015 date sample). This sample is coherent with the date sample as it is comes
17016 from the same timeval structure.
17017
Willy Tarreaud716f9b2017-10-13 11:03:15 +020017018distcc_body(<token>[,<occ>]) : binary
17019 Parses a distcc message and returns the body associated to occurrence #<occ>
17020 of the token <token>. Occurrences start at 1, and when unspecified, any may
17021 match though in practice only the first one is checked for now. This can be
17022 used to extract file names or arguments in files built using distcc through
17023 haproxy. Please refer to distcc's protocol documentation for the complete
17024 list of supported tokens.
17025
17026distcc_param(<token>[,<occ>]) : integer
17027 Parses a distcc message and returns the parameter associated to occurrence
17028 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
17029 any may match though in practice only the first one is checked for now. This
17030 can be used to extract certain information such as the protocol version, the
17031 file size or the argument in files built using distcc through haproxy.
17032 Another use case consists in waiting for the start of the preprocessed file
17033 contents before connecting to the server to avoid keeping idle connections.
17034 Please refer to distcc's protocol documentation for the complete list of
17035 supported tokens.
17036
17037 Example :
17038 # wait up to 20s for the pre-processed file to be uploaded
17039 tcp-request inspect-delay 20s
17040 tcp-request content accept if { distcc_param(DOTI) -m found }
17041 # send large files to the big farm
17042 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
17043
Willy Tarreau595ec542013-06-12 21:34:28 +020017044env(<name>) : string
17045 Returns a string containing the value of environment variable <name>. As a
17046 reminder, environment variables are per-process and are sampled when the
17047 process starts. This can be useful to pass some information to a next hop
17048 server, or with ACLs to take specific action when the process is started a
17049 certain way.
17050
17051 Examples :
17052 # Pass the Via header to next hop with the local hostname in it
17053 http-request add-header Via 1.1\ %[env(HOSTNAME)]
17054
17055 # reject cookie-less requests when the STOP environment variable is set
17056 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
17057
Willy Tarreau74ca5042013-06-11 23:12:07 +020017058fe_conn([<frontend>]) : integer
17059 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010017060 possibly including the connection being evaluated. If no frontend name is
17061 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020017062 frontend. It can be used to return a sorry page before hard-blocking, or to
17063 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010017064 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020017065 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
17066 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020017067
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020017068fe_req_rate([<frontend>]) : integer
17069 Returns an integer value corresponding to the number of HTTP requests per
17070 second sent to a frontend. This number can differ from "fe_sess_rate" in
17071 situations where client-side keep-alive is enabled.
17072
Willy Tarreau74ca5042013-06-11 23:12:07 +020017073fe_sess_rate([<frontend>]) : integer
17074 Returns an integer value corresponding to the sessions creation rate on the
17075 frontend, in number of new sessions per second. This is used with ACLs to
17076 limit the incoming session rate to an acceptable range in order to prevent
17077 abuse of service at the earliest moment, for example when combined with other
17078 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
17079 down below the limit. It can also be useful to add this element to logs using
17080 a log-format directive. See also the "rate-limit sessions" directive for use
17081 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010017082
17083 Example :
17084 # This frontend limits incoming mails to 10/s with a max of 100
17085 # concurrent connections. We accept any connection below 10/s, and
17086 # force excess clients to wait for 100 ms. Since clients are limited to
17087 # 100 max, there cannot be more than 10 incoming mails per second.
17088 frontend mail
17089 bind :25
17090 mode tcp
17091 maxconn 100
17092 acl too_fast fe_sess_rate ge 10
17093 tcp-request inspect-delay 100ms
17094 tcp-request content accept if ! too_fast
17095 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010017096
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010017097hostname : string
17098 Returns the system hostname.
17099
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020017100int(<integer>) : signed integer
17101 Returns a signed integer.
17102
Thierry FOURNIERcc103292015-06-06 19:30:17 +020017103ipv4(<ipv4>) : ipv4
17104 Returns an ipv4.
17105
17106ipv6(<ipv6>) : ipv6
17107 Returns an ipv6.
17108
Willy Tarreau70fe9442018-11-22 16:07:39 +010017109lat_ns_avg : integer
17110 Returns the average number of nanoseconds spent between the moment the task
17111 handling the stream is woken up and the moment it is effectively called. This
17112 number is reset for each new request on the same connections in case of HTTP
17113 keep-alive. This value indicates the overall latency inflicted to the current
17114 request by all other requests being processed in parallel, and is a direct
17115 indicator of perceived performance due to noisy neighbours. In order to keep
17116 the value low, it is possible to reduce the scheduler's run queue depth using
17117 "tune.runqueue-depth", to reduce the number of concurrent events processed at
17118 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020017119 the "nice" option on the "bind" lines or in the frontend, to enable low
17120 latency scheduling using "tune.sched.low-latency", or to look for other heavy
17121 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
17122 processing needs to be adjusted or fixed. Compression of large buffers could
17123 be a culprit, like heavy regex or long lists of regex. Note: this value is
17124 exactly lat_ns_tot divided by cpu_calls.
Willy Tarreau70fe9442018-11-22 16:07:39 +010017125
17126lat_ns_tot : integer
17127 Returns the total number of nanoseconds spent between the moment the task
17128 handling the stream is woken up and the moment it is effectively called. This
17129 number is reset for each new request on the same connections in case of HTTP
17130 keep-alive. This value indicates the overall latency inflicted to the current
17131 request by all other requests being processed in parallel, and is a direct
17132 indicator of perceived performance due to noisy neighbours. In order to keep
17133 the value low, it is possible to reduce the scheduler's run queue depth using
17134 "tune.runqueue-depth", to reduce the number of concurrent events processed at
17135 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020017136 the "nice" option on the "bind" lines or in the frontend, to enable low
17137 latency scheduling using "tune.sched.low-latency", or to look for other heavy
17138 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
17139 processing needs to be adjusted or fixed. Compression of large buffers could
17140 be a culprit, like heavy regex or long lists of regex. Note: while it
Willy Tarreau70fe9442018-11-22 16:07:39 +010017141 may intuitively seem that the total latency adds to a transfer time, it is
17142 almost never true because while a task waits for the CPU, network buffers
17143 continue to fill up and the next call will process more at once. The value
17144 may be artificially high due to a high cpu_calls count, for example when
17145 processing many HTTP chunks, and for this reason it is often preferred to log
17146 lat_ns_avg instead, which is a more relevant performance indicator.
17147
Thierry FOURNIERcc103292015-06-06 19:30:17 +020017148meth(<method>) : method
17149 Returns a method.
17150
Willy Tarreau0f30d262014-11-24 16:02:05 +010017151nbproc : integer
17152 Returns an integer value corresponding to the number of processes that were
17153 started (it equals the global "nbproc" setting). This is useful for logging
17154 and debugging purposes.
17155
Willy Tarreau74ca5042013-06-11 23:12:07 +020017156nbsrv([<backend>]) : integer
17157 Returns an integer value corresponding to the number of usable servers of
17158 either the current backend or the named backend. This is mostly used with
17159 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010017160 switch to an alternate backend when the number of servers is too low to
17161 to handle some load. It is useful to report a failure when combined with
17162 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010017163
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040017164prio_class : integer
17165 Returns the priority class of the current session for http mode or connection
17166 for tcp mode. The value will be that set by the last call to "http-request
17167 set-priority-class" or "tcp-request content set-priority-class".
17168
17169prio_offset : integer
17170 Returns the priority offset of the current session for http mode or
17171 connection for tcp mode. The value will be that set by the last call to
17172 "http-request set-priority-offset" or "tcp-request content
17173 set-priority-offset".
17174
Willy Tarreau0f30d262014-11-24 16:02:05 +010017175proc : integer
17176 Returns an integer value corresponding to the position of the process calling
17177 the function, between 1 and global.nbproc. This is useful for logging and
17178 debugging purposes.
17179
Willy Tarreau74ca5042013-06-11 23:12:07 +020017180queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010017181 Returns the total number of queued connections of the designated backend,
17182 including all the connections in server queues. If no backend name is
17183 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020017184 one. This is useful with ACLs or to pass statistics to backend servers. This
17185 can be used to take actions when queuing goes above a known level, generally
17186 indicating a surge of traffic or a massive slowdown on the servers. One
17187 possible action could be to reject new users but still accept old ones. See
17188 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
17189
Willy Tarreau84310e22014-02-14 11:59:04 +010017190rand([<range>]) : integer
17191 Returns a random integer value within a range of <range> possible values,
17192 starting at zero. If the range is not specified, it defaults to 2^32, which
17193 gives numbers between 0 and 4294967295. It can be useful to pass some values
17194 needed to take some routing decisions for example, or just for debugging
17195 purposes. This random must not be used for security purposes.
17196
Luca Schimweg8a694b82019-09-10 15:42:52 +020017197uuid([<version>]) : string
17198 Returns a UUID following the RFC4122 standard. If the version is not
17199 specified, a UUID version 4 (fully random) is returned.
17200 Currently, only version 4 is supported.
17201
Willy Tarreau74ca5042013-06-11 23:12:07 +020017202srv_conn([<backend>/]<server>) : integer
17203 Returns an integer value corresponding to the number of currently established
17204 connections on the designated server, possibly including the connection being
17205 evaluated. If <backend> is omitted, then the server is looked up in the
17206 current backend. It can be used to use a specific farm when one server is
17207 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040017208 connections with it. See also the "fe_conn", "be_conn", "queue", and
17209 "srv_conn_free" fetch methods.
17210
17211srv_conn_free([<backend>/]<server>) : integer
17212 Returns an integer value corresponding to the number of available connections
17213 on the designated server, possibly including the connection being evaluated.
17214 The value does not include queue slots. If <backend> is omitted, then the
17215 server is looked up in the current backend. It can be used to use a specific
17216 farm when one server is full, or to inform the server about our view of the
17217 number of active connections with it. See also the "be_conn_free" and
17218 "srv_conn" fetch methods.
17219
17220 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
17221 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017222
17223srv_is_up([<backend>/]<server>) : boolean
17224 Returns true when the designated server is UP, and false when it is either
17225 DOWN or in maintenance mode. If <backend> is omitted, then the server is
17226 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010017227 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017228 availability). Another possible use which is more of a hack consists in
17229 using dummy servers as boolean variables that can be enabled or disabled from
17230 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
17231
Willy Tarreauff2b7af2017-10-13 11:46:26 +020017232srv_queue([<backend>/]<server>) : integer
17233 Returns an integer value corresponding to the number of connections currently
17234 pending in the designated server's queue. If <backend> is omitted, then the
17235 server is looked up in the current backend. It can sometimes be used together
17236 with the "use-server" directive to force to use a known faster server when it
17237 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
17238 fetch methods.
17239
Willy Tarreau74ca5042013-06-11 23:12:07 +020017240srv_sess_rate([<backend>/]<server>) : integer
17241 Returns an integer corresponding to the sessions creation rate on the
17242 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017243 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020017244 used with ACLs but can make sense with logs too. This is used to switch to an
17245 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010017246 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020017247 overloading servers).
17248
17249 Example :
17250 # Redirect to a separate back
17251 acl srv1_full srv_sess_rate(be1/srv1) gt 50
17252 acl srv2_full srv_sess_rate(be1/srv2) gt 50
17253 use_backend be2 if srv1_full or srv2_full
17254
Christopher Faulet1bea8652020-07-10 16:03:45 +020017255srv_iweight([<backend>/]<server>): integer
17256 Returns an integer corresponding to the server's initial weight. If <backend>
17257 is omitted, then the server is looked up in the current backend. See also
17258 "srv_weight" and "srv_uweight".
17259
17260srv_uweight([<backend>/]<server>): integer
17261 Returns an integer corresponding to the user visible server's weight. If
17262 <backend> is omitted, then the server is looked up in the current
17263 backend. See also "srv_weight" and "srv_iweight".
17264
17265srv_weight([<backend>/]<server>): integer
17266 Returns an integer corresponding to the current (or effective) server's
17267 weight. If <backend> is omitted, then the server is looked up in the current
17268 backend. See also "srv_iweight" and "srv_uweight".
17269
Willy Tarreau0f30d262014-11-24 16:02:05 +010017270stopping : boolean
17271 Returns TRUE if the process calling the function is currently stopping. This
17272 can be useful for logging, or for relaxing certain checks or helping close
17273 certain connections upon graceful shutdown.
17274
Thierry FOURNIERcc103292015-06-06 19:30:17 +020017275str(<string>) : string
17276 Returns a string.
17277
Willy Tarreau74ca5042013-06-11 23:12:07 +020017278table_avl([<table>]) : integer
17279 Returns the total number of available entries in the current proxy's
17280 stick-table or in the designated stick-table. See also table_cnt.
17281
17282table_cnt([<table>]) : integer
17283 Returns the total number of entries currently in use in the current proxy's
17284 stick-table or in the designated stick-table. See also src_conn_cnt and
17285 table_avl for other entry counting methods.
17286
Christopher Faulet34adb2a2017-11-21 21:45:38 +010017287thread : integer
17288 Returns an integer value corresponding to the position of the thread calling
17289 the function, between 0 and (global.nbthread-1). This is useful for logging
17290 and debugging purposes.
17291
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017292var(<var-name>) : undefined
17293 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010017294 sample fetch fails. The name of the variable starts with an indication
17295 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010017296 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010017297 "sess" : the variable is shared with the whole session
17298 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017299 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010017300 "req" : the variable is shared only during request processing,
17301 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017302 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010017303 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017304
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200173057.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020017306----------------------------------
17307
17308The layer 4 usually describes just the transport layer which in haproxy is
17309closest to the connection, where no content is yet made available. The fetch
17310methods described here are usable as low as the "tcp-request connection" rule
17311sets unless they require some future information. Those generally include
17312TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017313the incoming connection. For retrieving a value from a sticky counters, the
17314counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020017315"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
17316used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
17317can be specified as the first integer argument when using the "sc_" prefix.
17318Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
17319table may be specified with the "sc*" form, in which case the currently
17320tracked key will be looked up into this alternate table instead of the table
17321currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017322
Christopher Faulet7d081f02021-04-15 09:38:37 +020017323bc_dst : ip
17324 This is the destination ip address of the connection on the server side,
17325 which is the server address HAProxy connected to. It is of type IP and works
17326 on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 address is mapped to its
17327 IPv6 equivalent, according to RFC 4291.
17328
17329bc_dst_port : integer
17330 Returns an integer value corresponding to the destination TCP port of the
17331 connection on the server side, which is the port HAproxy connected to.
17332
Jérôme Magnin35e53a62019-01-16 14:38:37 +010017333bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010017334 Returns the backend connection's HTTP major version encoding, which may be 1
17335 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
17336 encoding and not the version present in the request header.
17337
Christopher Faulet7d081f02021-04-15 09:38:37 +020017338bc_src : ip
17339 This is the source ip address of the connection on the server side, which is
17340 the server address haproxy connected from. It is of type IP and works on both
17341 IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are mapped to their IPv6
17342 equivalent, according to RFC 4291.
17343
17344bc_src_port : integer
17345 Returns an integer value corresponding to the TCP source port of the
17346 connection on the server side, which is the port HAproxy connected from.
17347
Willy Tarreau74ca5042013-06-11 23:12:07 +020017348be_id : integer
17349 Returns an integer containing the current backend's id. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020017350 frontends with responses to check which backend processed the request. It can
17351 also be used in a tcp-check or an http-check ruleset.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017352
Marcin Deranekd2471c22016-12-12 14:08:05 +010017353be_name : string
17354 Returns a string containing the current backend's name. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020017355 frontends with responses to check which backend processed the request. It can
17356 also be used in a tcp-check or an http-check ruleset.
Marcin Deranekd2471c22016-12-12 14:08:05 +010017357
Amaury Denoyelled91d7792020-12-10 13:43:56 +010017358be_server_timeout : integer
17359 Returns the configuration value in millisecond for the server timeout of the
17360 current backend. This timeout can be overwritten by a "set-timeout" rule. See
17361 also the "cur_server_timeout".
17362
17363be_tunnel_timeout : integer
17364 Returns the configuration value in millisecond for the tunnel timeout of the
17365 current backend. This timeout can be overwritten by a "set-timeout" rule. See
17366 also the "cur_tunnel_timeout".
17367
Amaury Denoyellef7719a22020-12-10 13:43:58 +010017368cur_server_timeout : integer
17369 Returns the currently applied server timeout in millisecond for the stream.
17370 In the default case, this will be equal to be_server_timeout unless a
17371 "set-timeout" rule has been applied. See also "be_server_timeout".
17372
17373cur_tunnel_timeout : integer
17374 Returns the currently applied tunnel timeout in millisecond for the stream.
17375 In the default case, this will be equal to be_tunnel_timeout unless a
17376 "set-timeout" rule has been applied. See also "be_tunnel_timeout".
17377
Willy Tarreau74ca5042013-06-11 23:12:07 +020017378dst : ip
17379 This is the destination IPv4 address of the connection on the client side,
17380 which is the address the client connected to. It can be useful when running
17381 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
17382 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010017383 RFC 4291. When the incoming connection passed through address translation or
17384 redirection involving connection tracking, the original destination address
17385 before the redirection will be reported. On Linux systems, the source and
17386 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
17387 is set, because a late response may reopen a timed out connection and switch
17388 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017389
17390dst_conn : integer
17391 Returns an integer value corresponding to the number of currently established
17392 connections on the same socket including the one being evaluated. It is
17393 normally used with ACLs but can as well be used to pass the information to
17394 servers in an HTTP header or in logs. It can be used to either return a sorry
17395 page before hard-blocking, or to use a specific backend to drain new requests
17396 when the socket is considered saturated. This offers the ability to assign
17397 different limits to different listening ports or addresses. See also the
17398 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017399
Willy Tarreau16e01562016-08-09 16:46:18 +020017400dst_is_local : boolean
17401 Returns true if the destination address of the incoming connection is local
17402 to the system, or false if the address doesn't exist on the system, meaning
17403 that it was intercepted in transparent mode. It can be useful to apply
17404 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010017405 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020017406 be delivered only on this address, or SSH access could be locally redirected.
17407 Please note that the check involves a few system calls, so it's better to do
17408 it only once per connection.
17409
Willy Tarreau74ca5042013-06-11 23:12:07 +020017410dst_port : integer
17411 Returns an integer value corresponding to the destination TCP port of the
17412 connection on the client side, which is the port the client connected to.
17413 This might be used when running in transparent mode, when assigning dynamic
17414 ports to some clients for a whole application session, to stick all users to
17415 a same server, or to pass the destination port information to a server using
17416 an HTTP header.
17417
Willy Tarreau60ca10a2017-08-18 15:26:54 +020017418fc_http_major : integer
17419 Reports the front connection's HTTP major version encoding, which may be 1
17420 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
17421 encoding and not on the version present in the request header.
17422
Geoff Simmons7185b782019-08-27 18:31:16 +020017423fc_pp_authority : string
17424 Returns the authority TLV sent by the client in the PROXY protocol header,
17425 if any.
17426
Tim Duesterhusd1b15b62020-03-13 12:34:23 +010017427fc_pp_unique_id : string
17428 Returns the unique ID TLV sent by the client in the PROXY protocol header,
17429 if any.
17430
Emeric Brun4f603012017-01-05 15:11:44 +010017431fc_rcvd_proxy : boolean
17432 Returns true if the client initiated the connection with a PROXY protocol
17433 header.
17434
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020017435fc_rtt(<unit>) : integer
17436 Returns the Round Trip Time (RTT) measured by the kernel for the client
17437 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
17438 can be set to "ms" for milliseconds or "us" for microseconds. If the server
17439 connection is not established, if the connection is not TCP or if the
17440 operating system does not support TCP_INFO, for example Linux kernels before
17441 2.4, the sample fetch fails.
17442
17443fc_rttvar(<unit>) : integer
17444 Returns the Round Trip Time (RTT) variance measured by the kernel for the
17445 client connection. <unit> is facultative, by default the unit is milliseconds.
17446 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
17447 server connection is not established, if the connection is not TCP or if the
17448 operating system does not support TCP_INFO, for example Linux kernels before
17449 2.4, the sample fetch fails.
17450
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017451fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017452 Returns the unacked counter measured by the kernel for the client connection.
17453 If the server connection is not established, if the connection is not TCP or
17454 if the operating system does not support TCP_INFO, for example Linux kernels
17455 before 2.4, the sample fetch fails.
17456
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017457fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017458 Returns the sacked counter measured by the kernel for the client connection.
17459 If the server connection is not established, if the connection is not TCP or
17460 if the operating system does not support TCP_INFO, for example Linux kernels
17461 before 2.4, the sample fetch fails.
17462
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017463fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017464 Returns the retransmits counter measured by the kernel for the client
17465 connection. If the server connection is not established, if the connection is
17466 not TCP or if the operating system does not support TCP_INFO, for example
17467 Linux kernels before 2.4, the sample fetch fails.
17468
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017469fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017470 Returns the fack counter measured by the kernel for the client
17471 connection. If the server connection is not established, if the connection is
17472 not TCP or if the operating system does not support TCP_INFO, for example
17473 Linux kernels before 2.4, the sample fetch fails.
17474
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017475fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017476 Returns the lost counter measured by the kernel for the client
17477 connection. If the server connection is not established, if the connection is
17478 not TCP or if the operating system does not support TCP_INFO, for example
17479 Linux kernels before 2.4, the sample fetch fails.
17480
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017481fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017482 Returns the reordering counter measured by the kernel for the client
17483 connection. If the server connection is not established, if the connection is
17484 not TCP or if the operating system does not support TCP_INFO, for example
17485 Linux kernels before 2.4, the sample fetch fails.
17486
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020017487fe_defbe : string
17488 Returns a string containing the frontend's default backend name. It can be
17489 used in frontends to check which backend will handle requests by default.
17490
Willy Tarreau74ca5042013-06-11 23:12:07 +020017491fe_id : integer
17492 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010017493 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020017494 coming via a same frontend to the same server.
17495
Marcin Deranekd2471c22016-12-12 14:08:05 +010017496fe_name : string
17497 Returns a string containing the current frontend's name. It can be used in
17498 backends to check from which frontend it was called, or to stick all users
17499 coming via a same frontend to the same server.
17500
Amaury Denoyelleda184d52020-12-10 13:43:55 +010017501fe_client_timeout : integer
17502 Returns the configuration value in millisecond for the client timeout of the
17503 current frontend.
17504
Cyril Bonté62ba8702014-04-22 23:52:25 +020017505sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017506sc0_bytes_in_rate([<table>]) : integer
17507sc1_bytes_in_rate([<table>]) : integer
17508sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017509 Returns the average client-to-server bytes rate from the currently tracked
17510 counters, measured in amount of bytes over the period configured in the
17511 table. See also src_bytes_in_rate.
17512
Cyril Bonté62ba8702014-04-22 23:52:25 +020017513sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017514sc0_bytes_out_rate([<table>]) : integer
17515sc1_bytes_out_rate([<table>]) : integer
17516sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017517 Returns the average server-to-client bytes rate from the currently tracked
17518 counters, measured in amount of bytes over the period configured in the
17519 table. See also src_bytes_out_rate.
17520
Cyril Bonté62ba8702014-04-22 23:52:25 +020017521sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017522sc0_clr_gpc0([<table>]) : integer
17523sc1_clr_gpc0([<table>]) : integer
17524sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020017525 Clears the first General Purpose Counter associated to the currently tracked
17526 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010017527 stored value is zero, so first invocation will always return zero. This is
17528 typically used as a second ACL in an expression in order to mark a connection
17529 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020017530
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017531 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020017532 # block if 5 consecutive requests continue to come faster than 10 sess
17533 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017534 acl abuse sc0_http_req_rate gt 10
17535 acl kill sc0_inc_gpc0 gt 5
17536 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020017537 tcp-request connection accept if !abuse save
17538 tcp-request connection reject if abuse kill
17539
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017540sc_clr_gpc1(<ctr>[,<table>]) : integer
17541sc0_clr_gpc1([<table>]) : integer
17542sc1_clr_gpc1([<table>]) : integer
17543sc2_clr_gpc1([<table>]) : integer
17544 Clears the second General Purpose Counter associated to the currently tracked
17545 counters, and returns its previous value. Before the first invocation, the
17546 stored value is zero, so first invocation will always return zero. This is
17547 typically used as a second ACL in an expression in order to mark a connection
17548 when a first ACL was verified.
17549
Cyril Bonté62ba8702014-04-22 23:52:25 +020017550sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017551sc0_conn_cnt([<table>]) : integer
17552sc1_conn_cnt([<table>]) : integer
17553sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017554 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017555 counters. See also src_conn_cnt.
17556
Cyril Bonté62ba8702014-04-22 23:52:25 +020017557sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017558sc0_conn_cur([<table>]) : integer
17559sc1_conn_cur([<table>]) : integer
17560sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017561 Returns the current amount of concurrent connections tracking the same
17562 tracked counters. This number is automatically incremented when tracking
17563 begins and decremented when tracking stops. See also src_conn_cur.
17564
Cyril Bonté62ba8702014-04-22 23:52:25 +020017565sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017566sc0_conn_rate([<table>]) : integer
17567sc1_conn_rate([<table>]) : integer
17568sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017569 Returns the average connection rate from the currently tracked counters,
17570 measured in amount of connections over the period configured in the table.
17571 See also src_conn_rate.
17572
Cyril Bonté62ba8702014-04-22 23:52:25 +020017573sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017574sc0_get_gpc0([<table>]) : integer
17575sc1_get_gpc0([<table>]) : integer
17576sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017577 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017578 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017579
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017580sc_get_gpc1(<ctr>[,<table>]) : integer
17581sc0_get_gpc1([<table>]) : integer
17582sc1_get_gpc1([<table>]) : integer
17583sc2_get_gpc1([<table>]) : integer
17584 Returns the value of the second General Purpose Counter associated to the
17585 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
17586
Thierry FOURNIER236657b2015-08-19 08:25:14 +020017587sc_get_gpt0(<ctr>[,<table>]) : integer
17588sc0_get_gpt0([<table>]) : integer
17589sc1_get_gpt0([<table>]) : integer
17590sc2_get_gpt0([<table>]) : integer
17591 Returns the value of the first General Purpose Tag associated to the
17592 currently tracked counters. See also src_get_gpt0.
17593
Cyril Bonté62ba8702014-04-22 23:52:25 +020017594sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017595sc0_gpc0_rate([<table>]) : integer
17596sc1_gpc0_rate([<table>]) : integer
17597sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017598 Returns the average increment rate of the first General Purpose Counter
17599 associated to the currently tracked counters. It reports the frequency
17600 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017601 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
17602 that the "gpc0_rate" counter must be stored in the stick-table for a value to
17603 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020017604
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017605sc_gpc1_rate(<ctr>[,<table>]) : integer
17606sc0_gpc1_rate([<table>]) : integer
17607sc1_gpc1_rate([<table>]) : integer
17608sc2_gpc1_rate([<table>]) : integer
17609 Returns the average increment rate of the second General Purpose Counter
17610 associated to the currently tracked counters. It reports the frequency
17611 which the gpc1 counter was incremented over the configured period. See also
17612 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
17613 that the "gpc1_rate" counter must be stored in the stick-table for a value to
17614 be returned, as "gpc1" only holds the event count.
17615
Cyril Bonté62ba8702014-04-22 23:52:25 +020017616sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017617sc0_http_err_cnt([<table>]) : integer
17618sc1_http_err_cnt([<table>]) : integer
17619sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017620 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017621 counters. This includes the both request errors and 4xx error responses.
17622 See also src_http_err_cnt.
17623
Cyril Bonté62ba8702014-04-22 23:52:25 +020017624sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017625sc0_http_err_rate([<table>]) : integer
17626sc1_http_err_rate([<table>]) : integer
17627sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017628 Returns the average rate of HTTP errors from the currently tracked counters,
17629 measured in amount of errors over the period configured in the table. This
17630 includes the both request errors and 4xx error responses. See also
17631 src_http_err_rate.
17632
Willy Tarreau826f3ab2021-02-10 12:07:15 +010017633sc_http_fail_cnt(<ctr>[,<table>]) : integer
17634sc0_http_fail_cnt([<table>]) : integer
17635sc1_http_fail_cnt([<table>]) : integer
17636sc2_http_fail_cnt([<table>]) : integer
17637 Returns the cumulative number of HTTP response failures from the currently
17638 tracked counters. This includes the both response errors and 5xx status codes
17639 other than 501 and 505. See also src_http_fail_cnt.
17640
17641sc_http_fail_rate(<ctr>[,<table>]) : integer
17642sc0_http_fail_rate([<table>]) : integer
17643sc1_http_fail_rate([<table>]) : integer
17644sc2_http_fail_rate([<table>]) : integer
17645 Returns the average rate of HTTP response failures from the currently tracked
17646 counters, measured in amount of failures over the period configured in the
17647 table. This includes the both response errors and 5xx status codes other than
17648 501 and 505. See also src_http_fail_rate.
17649
Cyril Bonté62ba8702014-04-22 23:52:25 +020017650sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017651sc0_http_req_cnt([<table>]) : integer
17652sc1_http_req_cnt([<table>]) : integer
17653sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017654 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017655 counters. This includes every started request, valid or not. See also
17656 src_http_req_cnt.
17657
Cyril Bonté62ba8702014-04-22 23:52:25 +020017658sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017659sc0_http_req_rate([<table>]) : integer
17660sc1_http_req_rate([<table>]) : integer
17661sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017662 Returns the average rate of HTTP requests from the currently tracked
17663 counters, measured in amount of requests over the period configured in
17664 the table. This includes every started request, valid or not. See also
17665 src_http_req_rate.
17666
Cyril Bonté62ba8702014-04-22 23:52:25 +020017667sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017668sc0_inc_gpc0([<table>]) : integer
17669sc1_inc_gpc0([<table>]) : integer
17670sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017671 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010017672 tracked counters, and returns its new value. Before the first invocation,
17673 the stored value is zero, so first invocation will increase it to 1 and will
17674 return 1. This is typically used as a second ACL in an expression in order
17675 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020017676
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017677 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017678 acl abuse sc0_http_req_rate gt 10
17679 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020017680 tcp-request connection reject if abuse kill
17681
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017682sc_inc_gpc1(<ctr>[,<table>]) : integer
17683sc0_inc_gpc1([<table>]) : integer
17684sc1_inc_gpc1([<table>]) : integer
17685sc2_inc_gpc1([<table>]) : integer
17686 Increments the second General Purpose Counter associated to the currently
17687 tracked counters, and returns its new value. Before the first invocation,
17688 the stored value is zero, so first invocation will increase it to 1 and will
17689 return 1. This is typically used as a second ACL in an expression in order
17690 to mark a connection when a first ACL was verified.
17691
Cyril Bonté62ba8702014-04-22 23:52:25 +020017692sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017693sc0_kbytes_in([<table>]) : integer
17694sc1_kbytes_in([<table>]) : integer
17695sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017696 Returns the total amount of client-to-server data from the currently tracked
17697 counters, measured in kilobytes. The test is currently performed on 32-bit
17698 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020017699
Cyril Bonté62ba8702014-04-22 23:52:25 +020017700sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017701sc0_kbytes_out([<table>]) : integer
17702sc1_kbytes_out([<table>]) : integer
17703sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017704 Returns the total amount of server-to-client data from the currently tracked
17705 counters, measured in kilobytes. The test is currently performed on 32-bit
17706 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020017707
Cyril Bonté62ba8702014-04-22 23:52:25 +020017708sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017709sc0_sess_cnt([<table>]) : integer
17710sc1_sess_cnt([<table>]) : integer
17711sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017712 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020017713 into sessions, which means that they were accepted by a "tcp-request
17714 connection" rule, from the currently tracked counters. A backend may count
17715 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017716 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020017717 with the client. See also src_sess_cnt.
17718
Cyril Bonté62ba8702014-04-22 23:52:25 +020017719sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017720sc0_sess_rate([<table>]) : integer
17721sc1_sess_rate([<table>]) : integer
17722sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017723 Returns the average session rate from the currently tracked counters,
17724 measured in amount of sessions over the period configured in the table. A
17725 session is a connection that got past the early "tcp-request connection"
17726 rules. A backend may count more sessions than connections because each
17727 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017728 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020017729
Cyril Bonté62ba8702014-04-22 23:52:25 +020017730sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020017731sc0_tracked([<table>]) : boolean
17732sc1_tracked([<table>]) : boolean
17733sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020017734 Returns true if the designated session counter is currently being tracked by
17735 the current session. This can be useful when deciding whether or not we want
17736 to set some values in a header passed to the server.
17737
Cyril Bonté62ba8702014-04-22 23:52:25 +020017738sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017739sc0_trackers([<table>]) : integer
17740sc1_trackers([<table>]) : integer
17741sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010017742 Returns the current amount of concurrent connections tracking the same
17743 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017744 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010017745 that it does not rely on any stored information but on the table's reference
17746 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020017747 may sometimes be more suited for layer7 tracking. It can be used to tell a
17748 server how many concurrent connections there are from a given address for
17749 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010017750
Willy Tarreau74ca5042013-06-11 23:12:07 +020017751so_id : integer
17752 Returns an integer containing the current listening socket's id. It is useful
17753 in frontends involving many "bind" lines, or to stick all users coming via a
17754 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017755
Jerome Magnineb421b22020-03-27 22:08:40 +010017756so_name : string
17757 Returns a string containing the current listening socket's name, as defined
17758 with name on a "bind" line. It can serve the same purposes as so_id but with
17759 strings instead of integers.
17760
Willy Tarreau74ca5042013-06-11 23:12:07 +020017761src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010017762 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020017763 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
17764 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
17765 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017766 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
17767 directive is used, it can be the address of a client behind another
17768 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010017769 "tcp-request connection" which sees the real address. When the incoming
17770 connection passed through address translation or redirection involving
17771 connection tracking, the original destination address before the redirection
17772 will be reported. On Linux systems, the source and destination may seldom
17773 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
17774 response may reopen a timed out connection and switch what is believed to be
17775 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017776
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010017777 Example:
17778 # add an HTTP header in requests with the originating address' country
17779 http-request set-header X-Country %[src,map_ip(geoip.lst)]
17780
Willy Tarreau74ca5042013-06-11 23:12:07 +020017781src_bytes_in_rate([<table>]) : integer
17782 Returns the average bytes rate from the incoming connection's source address
17783 in the current proxy's stick-table or in the designated stick-table, measured
17784 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017785 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017786
Willy Tarreau74ca5042013-06-11 23:12:07 +020017787src_bytes_out_rate([<table>]) : integer
17788 Returns the average bytes rate to the incoming connection's source address in
17789 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017790 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017791 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017792
Willy Tarreau74ca5042013-06-11 23:12:07 +020017793src_clr_gpc0([<table>]) : integer
17794 Clears the first General Purpose Counter associated to the incoming
17795 connection's source address in the current proxy's stick-table or in the
17796 designated stick-table, and returns its previous value. If the address is not
17797 found, an entry is created and 0 is returned. This is typically used as a
17798 second ACL in an expression in order to mark a connection when a first ACL
17799 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020017800
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017801 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020017802 # block if 5 consecutive requests continue to come faster than 10 sess
17803 # per second, and reset the counter as soon as the traffic slows down.
17804 acl abuse src_http_req_rate gt 10
17805 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010017806 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020017807 tcp-request connection accept if !abuse save
17808 tcp-request connection reject if abuse kill
17809
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017810src_clr_gpc1([<table>]) : integer
17811 Clears the second General Purpose Counter associated to the incoming
17812 connection's source address in the current proxy's stick-table or in the
17813 designated stick-table, and returns its previous value. If the address is not
17814 found, an entry is created and 0 is returned. This is typically used as a
17815 second ACL in an expression in order to mark a connection when a first ACL
17816 was verified.
17817
Willy Tarreau74ca5042013-06-11 23:12:07 +020017818src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017819 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020017820 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017821 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017822 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017823
Willy Tarreau74ca5042013-06-11 23:12:07 +020017824src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020017825 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017826 current incoming connection's source address in the current proxy's
17827 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017828 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017829
Willy Tarreau74ca5042013-06-11 23:12:07 +020017830src_conn_rate([<table>]) : integer
17831 Returns the average connection rate from the incoming connection's source
17832 address in the current proxy's stick-table or in the designated stick-table,
17833 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017834 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017835
Willy Tarreau74ca5042013-06-11 23:12:07 +020017836src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020017837 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017838 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017839 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017840 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017841
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017842src_get_gpc1([<table>]) : integer
17843 Returns the value of the second General Purpose Counter associated to the
17844 incoming connection's source address in the current proxy's stick-table or in
17845 the designated stick-table. If the address is not found, zero is returned.
17846 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
17847
Thierry FOURNIER236657b2015-08-19 08:25:14 +020017848src_get_gpt0([<table>]) : integer
17849 Returns the value of the first General Purpose Tag associated to the
17850 incoming connection's source address in the current proxy's stick-table or in
17851 the designated stick-table. If the address is not found, zero is returned.
17852 See also sc/sc0/sc1/sc2_get_gpt0.
17853
Willy Tarreau74ca5042013-06-11 23:12:07 +020017854src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017855 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020017856 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017857 stick-table or in the designated stick-table. It reports the frequency
17858 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017859 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
17860 that the "gpc0_rate" counter must be stored in the stick-table for a value to
17861 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017862
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017863src_gpc1_rate([<table>]) : integer
17864 Returns the average increment rate of the second General Purpose Counter
17865 associated to the incoming connection's source address in the current proxy's
17866 stick-table or in the designated stick-table. It reports the frequency
17867 which the gpc1 counter was incremented over the configured period. See also
17868 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
17869 that the "gpc1_rate" counter must be stored in the stick-table for a value to
17870 be returned, as "gpc1" only holds the event count.
17871
Willy Tarreau74ca5042013-06-11 23:12:07 +020017872src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017873 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017874 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020017875 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017876 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020017877 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017878
Willy Tarreau74ca5042013-06-11 23:12:07 +020017879src_http_err_rate([<table>]) : integer
17880 Returns the average rate of HTTP errors from the incoming connection's source
17881 address in the current proxy's stick-table or in the designated stick-table,
17882 measured in amount of errors over the period configured in the table. This
17883 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017884 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017885
Willy Tarreau826f3ab2021-02-10 12:07:15 +010017886src_http_fail_cnt([<table>]) : integer
17887 Returns the cumulative number of HTTP response failures triggered by the
17888 incoming connection's source address in the current proxy's stick-table or in
Ilya Shipitsin0de36ad2021-02-20 00:23:36 +050017889 the designated stick-table. This includes the both response errors and 5xx
Willy Tarreau826f3ab2021-02-10 12:07:15 +010017890 status codes other than 501 and 505. See also sc/sc0/sc1/sc2_http_fail_cnt.
17891 If the address is not found, zero is returned.
17892
17893src_http_fail_rate([<table>]) : integer
17894 Returns the average rate of HTTP response failures triggered by the incoming
17895 connection's source address in the current proxy's stick-table or in the
17896 designated stick-table, measured in amount of failures over the period
17897 configured in the table. This includes the both response errors and 5xx
17898 status codes other than 501 and 505. If the address is not found, zero is
17899 returned. See also sc/sc0/sc1/sc2_http_fail_rate.
17900
Willy Tarreau74ca5042013-06-11 23:12:07 +020017901src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017902 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017903 source address in the current proxy's stick-table or in the designated stick-
17904 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017905 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017906
Willy Tarreau74ca5042013-06-11 23:12:07 +020017907src_http_req_rate([<table>]) : integer
17908 Returns the average rate of HTTP requests from the incoming connection's
17909 source address in the current proxy's stick-table or in the designated stick-
17910 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020017911 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017912 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017913
Willy Tarreau74ca5042013-06-11 23:12:07 +020017914src_inc_gpc0([<table>]) : integer
17915 Increments the first General Purpose Counter associated to the incoming
17916 connection's source address in the current proxy's stick-table or in the
17917 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017918 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017919 This is typically used as a second ACL in an expression in order to mark a
17920 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020017921
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017922 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020017923 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010017924 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020017925 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020017926
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017927src_inc_gpc1([<table>]) : integer
17928 Increments the second General Purpose Counter associated to the incoming
17929 connection's source address in the current proxy's stick-table or in the
17930 designated stick-table, and returns its new value. If the address is not
17931 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
17932 This is typically used as a second ACL in an expression in order to mark a
17933 connection when a first ACL was verified.
17934
Willy Tarreau16e01562016-08-09 16:46:18 +020017935src_is_local : boolean
17936 Returns true if the source address of the incoming connection is local to the
17937 system, or false if the address doesn't exist on the system, meaning that it
17938 comes from a remote machine. Note that UNIX addresses are considered local.
17939 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017940 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020017941 note that the check involves a few system calls, so it's better to do it only
17942 once per connection.
17943
Willy Tarreau74ca5042013-06-11 23:12:07 +020017944src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017945 Returns the total amount of data received from the incoming connection's
17946 source address in the current proxy's stick-table or in the designated
17947 stick-table, measured in kilobytes. If the address is not found, zero is
17948 returned. The test is currently performed on 32-bit integers, which limits
17949 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017950
Willy Tarreau74ca5042013-06-11 23:12:07 +020017951src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017952 Returns the total amount of data sent to the incoming connection's source
17953 address in the current proxy's stick-table or in the designated stick-table,
17954 measured in kilobytes. If the address is not found, zero is returned. The
17955 test is currently performed on 32-bit integers, which limits values to 4
17956 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017957
Willy Tarreau74ca5042013-06-11 23:12:07 +020017958src_port : integer
17959 Returns an integer value corresponding to the TCP source port of the
17960 connection on the client side, which is the port the client connected from.
17961 Usage of this function is very limited as modern protocols do not care much
17962 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010017963
Willy Tarreau74ca5042013-06-11 23:12:07 +020017964src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017965 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020017966 connection's source IPv4 address in the current proxy's stick-table or in the
17967 designated stick-table, that were transformed into sessions, which means that
17968 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017969 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017970
Willy Tarreau74ca5042013-06-11 23:12:07 +020017971src_sess_rate([<table>]) : integer
17972 Returns the average session rate from the incoming connection's source
17973 address in the current proxy's stick-table or in the designated stick-table,
17974 measured in amount of sessions over the period configured in the table. A
17975 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017976 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017977
Willy Tarreau74ca5042013-06-11 23:12:07 +020017978src_updt_conn_cnt([<table>]) : integer
17979 Creates or updates the entry associated to the incoming connection's source
17980 address in the current proxy's stick-table or in the designated stick-table.
17981 This table must be configured to store the "conn_cnt" data type, otherwise
17982 the match will be ignored. The current count is incremented by one, and the
17983 expiration timer refreshed. The updated count is returned, so this match
17984 can't return zero. This was used to reject service abusers based on their
17985 source address. Note: it is recommended to use the more complete "track-sc*"
17986 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017987
17988 Example :
17989 # This frontend limits incoming SSH connections to 3 per 10 second for
17990 # each source address, and rejects excess connections until a 10 second
17991 # silence is observed. At most 20 addresses are tracked.
17992 listen ssh
17993 bind :22
17994 mode tcp
17995 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020017996 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020017997 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017998 server local 127.0.0.1:22
17999
Willy Tarreau74ca5042013-06-11 23:12:07 +020018000srv_id : integer
18001 Returns an integer containing the server's id when processing the response.
18002 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020018003 debugging. It can also be used in a tcp-check or an http-check ruleset.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020018004
vkill1dfd1652019-10-30 16:58:14 +080018005srv_name : string
18006 Returns a string containing the server's name when processing the response.
18007 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020018008 debugging. It can also be used in a tcp-check or an http-check ruleset.
vkill1dfd1652019-10-30 16:58:14 +080018009
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200180107.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020018011----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020018012
Willy Tarreau74ca5042013-06-11 23:12:07 +020018013The layer 5 usually describes just the session layer which in haproxy is
18014closest to the session once all the connection handshakes are finished, but
18015when no content is yet made available. The fetch methods described here are
18016usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018017future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020018018
Ben Shillitof25e8e52016-12-02 14:25:37 +00001801951d.all(<prop>[,<prop>*]) : string
18020 Returns values for the properties requested as a string, where values are
18021 separated by the delimiter specified with "51degrees-property-separator".
18022 The device is identified using all the important HTTP headers from the
18023 request. The function can be passed up to five property names, and if a
18024 property name can't be found, the value "NoData" is returned.
18025
18026 Example :
18027 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
18028 # containing the three properties requested using all relevant headers from
18029 # the request.
18030 frontend http-in
18031 bind *:8081
18032 default_backend servers
18033 http-request set-header X-51D-DeviceTypeMobileTablet \
18034 %[51d.all(DeviceType,IsMobile,IsTablet)]
18035
Emeric Brun645ae792014-04-30 14:21:06 +020018036ssl_bc : boolean
18037 Returns true when the back connection was made via an SSL/TLS transport
18038 layer and is locally deciphered. This means the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018039 other a server with the "ssl" option. It can be used in a tcp-check or an
18040 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018041
18042ssl_bc_alg_keysize : integer
18043 Returns the symmetric cipher key size supported in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018044 connection was made over an SSL/TLS transport layer. It can be used in a
18045 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018046
Olivier Houchard6b77f492018-11-22 18:18:29 +010018047ssl_bc_alpn : string
18048 This extracts the Application Layer Protocol Negotiation field from an
18049 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020018050 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010018051 server. The SSL library must have been built with support for TLS
18052 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
18053 not advertised unless the "alpn" keyword on the "server" line specifies a
18054 protocol list. Also, nothing forces the server to pick a protocol from this
18055 list, any other one may be requested. The TLS ALPN extension is meant to
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018056 replace the TLS NPN extension. See also "ssl_bc_npn". It can be used in a
18057 tcp-check or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010018058
Emeric Brun645ae792014-04-30 14:21:06 +020018059ssl_bc_cipher : string
18060 Returns the name of the used cipher when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018061 over an SSL/TLS transport layer. It can be used in a tcp-check or an
18062 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018063
Patrick Hemmer65674662019-06-04 08:13:03 -040018064ssl_bc_client_random : binary
18065 Returns the client random of the back connection when the incoming connection
18066 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
18067 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018068 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040018069
Emeric Brun74f7ffa2018-02-19 16:14:12 +010018070ssl_bc_is_resumed : boolean
18071 Returns true when the back connection was made over an SSL/TLS transport
18072 layer and the newly created SSL session was resumed using a cached
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018073 session or a TLS ticket. It can be used in a tcp-check or an http-check
18074 ruleset.
Emeric Brun74f7ffa2018-02-19 16:14:12 +010018075
Olivier Houchard6b77f492018-11-22 18:18:29 +010018076ssl_bc_npn : string
18077 This extracts the Next Protocol Negotiation field from an outgoing connection
18078 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020018079 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010018080 built with support for TLS extensions enabled (check haproxy -vv). Note that
18081 the TLS NPN extension is not advertised unless the "npn" keyword on the
18082 "server" line specifies a protocol list. Also, nothing forces the server to
18083 pick a protocol from this list, any other one may be used. Please note that
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018084 the TLS NPN extension was replaced with ALPN. It can be used in a tcp-check
18085 or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010018086
Emeric Brun645ae792014-04-30 14:21:06 +020018087ssl_bc_protocol : string
18088 Returns the name of the used protocol when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018089 over an SSL/TLS transport layer. It can be used in a tcp-check or an
18090 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018091
Emeric Brunb73a9b02014-04-30 18:49:19 +020018092ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020018093 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020018094 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018095 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64". It
18096 can be used in a tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018097
Patrick Hemmer65674662019-06-04 08:13:03 -040018098ssl_bc_server_random : binary
18099 Returns the server random of the back connection when the incoming connection
18100 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
18101 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018102 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040018103
Emeric Brun645ae792014-04-30 14:21:06 +020018104ssl_bc_session_id : binary
18105 Returns the SSL ID of the back connection when the outgoing connection was
18106 made over an SSL/TLS transport layer. It is useful to log if we want to know
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018107 if session was reused or not. It can be used in a tcp-check or an http-check
18108 ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018109
Patrick Hemmere0275472018-04-28 19:15:51 -040018110ssl_bc_session_key : binary
18111 Returns the SSL session master key of the back connection when the outgoing
18112 connection was made over an SSL/TLS transport layer. It is useful to decrypt
18113 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018114 BoringSSL. It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmere0275472018-04-28 19:15:51 -040018115
Emeric Brun645ae792014-04-30 14:21:06 +020018116ssl_bc_use_keysize : integer
18117 Returns the symmetric cipher key size used in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020018118 connection was made over an SSL/TLS transport layer. It can be used in a
18119 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020018120
Willy Tarreau74ca5042013-06-11 23:12:07 +020018121ssl_c_ca_err : integer
18122 When the incoming connection was made over an SSL/TLS transport layer,
18123 returns the ID of the first error detected during verification of the client
18124 certificate at depth > 0, or 0 if no error was encountered during this
18125 verification process. Please refer to your SSL library's documentation to
18126 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020018127
Willy Tarreau74ca5042013-06-11 23:12:07 +020018128ssl_c_ca_err_depth : integer
18129 When the incoming connection was made over an SSL/TLS transport layer,
18130 returns the depth in the CA chain of the first error detected during the
18131 verification of the client certificate. If no error is encountered, 0 is
18132 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010018133
Christopher Faulet70d10d12020-11-06 12:10:33 +010018134ssl_c_chain_der : binary
William Dauchya598b502020-08-06 18:11:38 +020018135 Returns the DER formatted chain certificate presented by the client when the
18136 incoming connection was made over an SSL/TLS transport layer. When used for
18137 an ACL, the value(s) to match against can be passed in hexadecimal form. One
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050018138 can parse the result with any lib accepting ASN.1 DER data. It currently
William Dauchya598b502020-08-06 18:11:38 +020018139 does not support resumed sessions.
18140
Christopher Faulet70d10d12020-11-06 12:10:33 +010018141ssl_c_der : binary
18142 Returns the DER formatted certificate presented by the client when the
18143 incoming connection was made over an SSL/TLS transport layer. When used for
18144 an ACL, the value(s) to match against can be passed in hexadecimal form.
18145
Willy Tarreau74ca5042013-06-11 23:12:07 +020018146ssl_c_err : integer
18147 When the incoming connection was made over an SSL/TLS transport layer,
18148 returns the ID of the first error detected during verification at depth 0, or
18149 0 if no error was encountered during this verification process. Please refer
18150 to your SSL library's documentation to find the exhaustive list of error
18151 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020018152
Elliot Otchet71f82972020-01-15 08:12:14 -050018153ssl_c_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018154 When the incoming connection was made over an SSL/TLS transport layer,
18155 returns the full distinguished name of the issuer of the certificate
18156 presented by the client when no <entry> is specified, or the value of the
18157 first given entry found from the beginning of the DN. If a positive/negative
18158 occurrence number is specified as the optional second argument, it returns
18159 the value of the nth given entry value from the beginning/end of the DN.
18160 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
18161 "ssl_c_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018162 The <format> parameter allows you to receive the DN suitable for
18163 consumption by different protocols. Currently supported is rfc2253 for
18164 LDAP v3.
18165 If you'd like to modify the format only you can specify an empty string
18166 and zero for the first two parameters. Example: ssl_c_i_dn(,0,rfc2253)
Willy Tarreau62644772008-07-16 18:36:06 +020018167
Willy Tarreau74ca5042013-06-11 23:12:07 +020018168ssl_c_key_alg : string
18169 Returns the name of the algorithm used to generate the key of the certificate
18170 presented by the client when the incoming connection was made over an SSL/TLS
18171 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020018172
Willy Tarreau74ca5042013-06-11 23:12:07 +020018173ssl_c_notafter : string
18174 Returns the end date presented by the client as a formatted string
18175 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18176 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020018177
Willy Tarreau74ca5042013-06-11 23:12:07 +020018178ssl_c_notbefore : string
18179 Returns the start date presented by the client as a formatted string
18180 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18181 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010018182
Elliot Otchet71f82972020-01-15 08:12:14 -050018183ssl_c_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018184 When the incoming connection was made over an SSL/TLS transport layer,
18185 returns the full distinguished name of the subject of the certificate
18186 presented by the client when no <entry> is specified, or the value of the
18187 first given entry found from the beginning of the DN. If a positive/negative
18188 occurrence number is specified as the optional second argument, it returns
18189 the value of the nth given entry value from the beginning/end of the DN.
18190 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
18191 "ssl_c_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018192 The <format> parameter allows you to receive the DN suitable for
18193 consumption by different protocols. Currently supported is rfc2253 for
18194 LDAP v3.
18195 If you'd like to modify the format only you can specify an empty string
18196 and zero for the first two parameters. Example: ssl_c_s_dn(,0,rfc2253)
Willy Tarreaub6672b52011-12-12 17:23:41 +010018197
Willy Tarreau74ca5042013-06-11 23:12:07 +020018198ssl_c_serial : binary
18199 Returns the serial of the certificate presented by the client when the
18200 incoming connection was made over an SSL/TLS transport layer. When used for
18201 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020018202
Willy Tarreau74ca5042013-06-11 23:12:07 +020018203ssl_c_sha1 : binary
18204 Returns the SHA-1 fingerprint of the certificate presented by the client when
18205 the incoming connection was made over an SSL/TLS transport layer. This can be
18206 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020018207 Note that the output is binary, so if you want to pass that signature to the
18208 server, you need to encode it in hex or base64, such as in the example below:
18209
Jarno Huuskonen676f6222017-03-30 09:19:45 +030018210 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020018211 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020018212
Willy Tarreau74ca5042013-06-11 23:12:07 +020018213ssl_c_sig_alg : string
18214 Returns the name of the algorithm used to sign the certificate presented by
18215 the client when the incoming connection was made over an SSL/TLS transport
18216 layer.
Emeric Brun87855892012-10-17 17:39:35 +020018217
Willy Tarreau74ca5042013-06-11 23:12:07 +020018218ssl_c_used : boolean
18219 Returns true if current SSL session uses a client certificate even if current
18220 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020018221
Willy Tarreau74ca5042013-06-11 23:12:07 +020018222ssl_c_verify : integer
18223 Returns the verify result error ID when the incoming connection was made over
18224 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
18225 refer to your SSL library's documentation for an exhaustive list of error
18226 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020018227
Willy Tarreau74ca5042013-06-11 23:12:07 +020018228ssl_c_version : integer
18229 Returns the version of the certificate presented by the client when the
18230 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020018231
Emeric Brun43e79582014-10-29 19:03:26 +010018232ssl_f_der : binary
18233 Returns the DER formatted certificate presented by the frontend when the
18234 incoming connection was made over an SSL/TLS transport layer. When used for
18235 an ACL, the value(s) to match against can be passed in hexadecimal form.
18236
Elliot Otchet71f82972020-01-15 08:12:14 -050018237ssl_f_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018238 When the incoming connection was made over an SSL/TLS transport layer,
18239 returns the full distinguished name of the issuer of the certificate
18240 presented by the frontend when no <entry> is specified, or the value of the
18241 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020018242 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020018243 the value of the nth given entry value from the beginning/end of the DN.
18244 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
18245 "ssl_f_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018246 The <format> parameter allows you to receive the DN suitable for
18247 consumption by different protocols. Currently supported is rfc2253 for
18248 LDAP v3.
18249 If you'd like to modify the format only you can specify an empty string
18250 and zero for the first two parameters. Example: ssl_f_i_dn(,0,rfc2253)
Emeric Brun87855892012-10-17 17:39:35 +020018251
Willy Tarreau74ca5042013-06-11 23:12:07 +020018252ssl_f_key_alg : string
18253 Returns the name of the algorithm used to generate the key of the certificate
18254 presented by the frontend when the incoming connection was made over an
18255 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020018256
Willy Tarreau74ca5042013-06-11 23:12:07 +020018257ssl_f_notafter : string
18258 Returns the end date presented by the frontend as a formatted string
18259 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18260 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020018261
Willy Tarreau74ca5042013-06-11 23:12:07 +020018262ssl_f_notbefore : string
18263 Returns the start date presented by the frontend as a formatted string
18264 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18265 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020018266
Elliot Otchet71f82972020-01-15 08:12:14 -050018267ssl_f_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018268 When the incoming connection was made over an SSL/TLS transport layer,
18269 returns the full distinguished name of the subject of the certificate
18270 presented by the frontend when no <entry> is specified, or the value of the
18271 first given entry found from the beginning of the DN. If a positive/negative
18272 occurrence number is specified as the optional second argument, it returns
18273 the value of the nth given entry value from the beginning/end of the DN.
18274 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
18275 "ssl_f_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018276 The <format> parameter allows you to receive the DN suitable for
18277 consumption by different protocols. Currently supported is rfc2253 for
18278 LDAP v3.
18279 If you'd like to modify the format only you can specify an empty string
18280 and zero for the first two parameters. Example: ssl_f_s_dn(,0,rfc2253)
Emeric Brunce5ad802012-10-22 14:11:22 +020018281
Willy Tarreau74ca5042013-06-11 23:12:07 +020018282ssl_f_serial : binary
18283 Returns the serial of the certificate presented by the frontend when the
18284 incoming connection was made over an SSL/TLS transport layer. When used for
18285 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020018286
Emeric Brun55f4fa82014-04-30 17:11:25 +020018287ssl_f_sha1 : binary
18288 Returns the SHA-1 fingerprint of the certificate presented by the frontend
18289 when the incoming connection was made over an SSL/TLS transport layer. This
18290 can be used to know which certificate was chosen using SNI.
18291
Willy Tarreau74ca5042013-06-11 23:12:07 +020018292ssl_f_sig_alg : string
18293 Returns the name of the algorithm used to sign the certificate presented by
18294 the frontend when the incoming connection was made over an SSL/TLS transport
18295 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020018296
Willy Tarreau74ca5042013-06-11 23:12:07 +020018297ssl_f_version : integer
18298 Returns the version of the certificate presented by the frontend when the
18299 incoming connection was made over an SSL/TLS transport layer.
18300
18301ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020018302 Returns true when the front connection was made via an SSL/TLS transport
18303 layer and is locally deciphered. This means it has matched a socket declared
18304 with a "bind" line having the "ssl" option.
18305
Willy Tarreau74ca5042013-06-11 23:12:07 +020018306 Example :
18307 # This passes "X-Proto: https" to servers when client connects over SSL
18308 listen http-https
18309 bind :80
18310 bind :443 ssl crt /etc/haproxy.pem
18311 http-request add-header X-Proto https if { ssl_fc }
18312
18313ssl_fc_alg_keysize : integer
18314 Returns the symmetric cipher key size supported in bits when the incoming
18315 connection was made over an SSL/TLS transport layer.
18316
18317ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018318 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020018319 incoming connection made via a TLS transport layer and locally deciphered by
18320 haproxy. The result is a string containing the protocol name advertised by
18321 the client. The SSL library must have been built with support for TLS
18322 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
18323 not advertised unless the "alpn" keyword on the "bind" line specifies a
18324 protocol list. Also, nothing forces the client to pick a protocol from this
18325 list, any other one may be requested. The TLS ALPN extension is meant to
18326 replace the TLS NPN extension. See also "ssl_fc_npn".
18327
Willy Tarreau74ca5042013-06-11 23:12:07 +020018328ssl_fc_cipher : string
18329 Returns the name of the used cipher when the incoming connection was made
18330 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020018331
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018332ssl_fc_cipherlist_bin : binary
18333 Returns the binary form of the client hello cipher list. The maximum returned
18334 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010018335 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018336
18337ssl_fc_cipherlist_hex : string
18338 Returns the binary form of the client hello cipher list encoded as
18339 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010018340 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018341
18342ssl_fc_cipherlist_str : string
18343 Returns the decoded text form of the client hello cipher list. The maximum
18344 number of ciphers returned is according with the value of
18345 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010018346 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020018347 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018348
18349ssl_fc_cipherlist_xxh : integer
18350 Returns a xxh64 of the cipher list. This hash can be return only is the value
18351 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010018352 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018353
Patrick Hemmer65674662019-06-04 08:13:03 -040018354ssl_fc_client_random : binary
18355 Returns the client random of the front connection when the incoming connection
18356 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
18357 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
18358
William Lallemand7d42ef52020-07-06 11:41:30 +020018359ssl_fc_client_early_traffic_secret : string
18360 Return the CLIENT_EARLY_TRAFFIC_SECRET as an hexadecimal string for the
18361 front connection when the incoming connection was made over a TLS 1.3
18362 transport layer.
18363 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18364 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18365 activated with "tune.ssl.keylog on" in the global section. See also
18366 "tune.ssl.keylog"
18367
18368ssl_fc_client_handshake_traffic_secret : string
18369 Return the CLIENT_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
18370 front connection when the incoming connection was made over a TLS 1.3
18371 transport layer.
18372 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18373 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18374 activated with "tune.ssl.keylog on" in the global section. See also
18375 "tune.ssl.keylog"
18376
18377ssl_fc_client_traffic_secret_0 : string
18378 Return the CLIENT_TRAFFIC_SECRET_0 as an hexadecimal string for the
18379 front connection when the incoming connection was made over a TLS 1.3
18380 transport layer.
18381 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18382 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18383 activated with "tune.ssl.keylog on" in the global section. See also
18384 "tune.ssl.keylog"
18385
18386ssl_fc_exporter_secret : string
18387 Return the EXPORTER_SECRET as an hexadecimal string for the
18388 front connection when the incoming connection was made over a TLS 1.3
18389 transport layer.
18390 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18391 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18392 activated with "tune.ssl.keylog on" in the global section. See also
18393 "tune.ssl.keylog"
18394
18395ssl_fc_early_exporter_secret : string
18396 Return the EARLY_EXPORTER_SECRET as an hexadecimal string for the
18397 front connection when the incoming connection was made over an TLS 1.3
18398 transport layer.
18399 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18400 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18401 activated with "tune.ssl.keylog on" in the global section. See also
18402 "tune.ssl.keylog"
18403
Willy Tarreau74ca5042013-06-11 23:12:07 +020018404ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020018405 Returns true if a client certificate is present in an incoming connection over
18406 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010018407 Note: on SSL session resumption with Session ID or TLS ticket, client
18408 certificate is not present in the current connection but may be retrieved
18409 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
18410 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020018411
Olivier Houchardccaa7de2017-10-02 11:51:03 +020018412ssl_fc_has_early : boolean
18413 Returns true if early data were sent, and the handshake didn't happen yet. As
18414 it has security implications, it is useful to be able to refuse those, or
18415 wait until the handshake happened.
18416
Willy Tarreau74ca5042013-06-11 23:12:07 +020018417ssl_fc_has_sni : boolean
18418 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020018419 in an incoming connection was made over an SSL/TLS transport layer. Returns
18420 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050018421 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020018422 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020018423
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020018424ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020018425 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010018426 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
18427 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020018428
Willy Tarreau74ca5042013-06-11 23:12:07 +020018429ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018430 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020018431 made via a TLS transport layer and locally deciphered by haproxy. The result
18432 is a string containing the protocol name advertised by the client. The SSL
18433 library must have been built with support for TLS extensions enabled (check
18434 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
18435 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
18436 forces the client to pick a protocol from this list, any other one may be
18437 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020018438
Willy Tarreau74ca5042013-06-11 23:12:07 +020018439ssl_fc_protocol : string
18440 Returns the name of the used protocol when the incoming connection was made
18441 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020018442
Emeric Brunb73a9b02014-04-30 18:49:19 +020018443ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040018444 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020018445 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
18446 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040018447
William Lallemand7d42ef52020-07-06 11:41:30 +020018448ssl_fc_server_handshake_traffic_secret : string
18449 Return the SERVER_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
18450 front connection when the incoming connection was made over a TLS 1.3
18451 transport layer.
18452 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18453 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18454 activated with "tune.ssl.keylog on" in the global section. See also
18455 "tune.ssl.keylog"
18456
18457ssl_fc_server_traffic_secret_0 : string
18458 Return the SERVER_TRAFFIC_SECRET_0 as an hexadecimal string for the
18459 front connection when the incoming connection was made over an TLS 1.3
18460 transport layer.
18461 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18462 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18463 activated with "tune.ssl.keylog on" in the global section. See also
18464 "tune.ssl.keylog"
18465
Patrick Hemmer65674662019-06-04 08:13:03 -040018466ssl_fc_server_random : binary
18467 Returns the server random of the front connection when the incoming connection
18468 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
18469 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
18470
Willy Tarreau74ca5042013-06-11 23:12:07 +020018471ssl_fc_session_id : binary
18472 Returns the SSL ID of the front connection when the incoming connection was
18473 made over an SSL/TLS transport layer. It is useful to stick a given client to
18474 a server. It is important to note that some browsers refresh their session ID
18475 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020018476
Patrick Hemmere0275472018-04-28 19:15:51 -040018477ssl_fc_session_key : binary
18478 Returns the SSL session master key of the front connection when the incoming
18479 connection was made over an SSL/TLS transport layer. It is useful to decrypt
18480 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
18481 BoringSSL.
18482
18483
Willy Tarreau74ca5042013-06-11 23:12:07 +020018484ssl_fc_sni : string
18485 This extracts the Server Name Indication TLS extension (SNI) field from an
18486 incoming connection made via an SSL/TLS transport layer and locally
18487 deciphered by haproxy. The result (when present) typically is a string
18488 matching the HTTPS host name (253 chars or less). The SSL library must have
18489 been built with support for TLS extensions enabled (check haproxy -vv).
18490
18491 This fetch is different from "req_ssl_sni" above in that it applies to the
18492 connection being deciphered by haproxy and not to SSL contents being blindly
18493 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050018494 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020018495 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020018496
Willy Tarreau74ca5042013-06-11 23:12:07 +020018497 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020018498 ssl_fc_sni_end : suffix match
18499 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020018500
Willy Tarreau74ca5042013-06-11 23:12:07 +020018501ssl_fc_use_keysize : integer
18502 Returns the symmetric cipher key size used in bits when the incoming
18503 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018504
William Lallemandbfa3e812020-06-25 20:07:18 +020018505ssl_s_der : binary
18506 Returns the DER formatted certificate presented by the server when the
18507 outgoing connection was made over an SSL/TLS transport layer. When used for
18508 an ACL, the value(s) to match against can be passed in hexadecimal form.
18509
William Dauchya598b502020-08-06 18:11:38 +020018510ssl_s_chain_der : binary
18511 Returns the DER formatted chain certificate presented by the server when the
18512 outgoing connection was made over an SSL/TLS transport layer. When used for
18513 an ACL, the value(s) to match against can be passed in hexadecimal form. One
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050018514 can parse the result with any lib accepting ASN.1 DER data. It currently
William Dauchya598b502020-08-06 18:11:38 +020018515 does not support resumed sessions.
18516
William Lallemandbfa3e812020-06-25 20:07:18 +020018517ssl_s_key_alg : string
18518 Returns the name of the algorithm used to generate the key of the certificate
18519 presented by the server when the outgoing connection was made over an
18520 SSL/TLS transport layer.
18521
18522ssl_s_notafter : string
18523 Returns the end date presented by the server as a formatted string
18524 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
18525 transport layer.
18526
18527ssl_s_notbefore : string
18528 Returns the start date presented by the server as a formatted string
18529 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
18530 transport layer.
18531
18532ssl_s_i_dn([<entry>[,<occ>[,<format>]]]) : string
18533 When the outgoing connection was made over an SSL/TLS transport layer,
18534 returns the full distinguished name of the issuer of the certificate
18535 presented by the server when no <entry> is specified, or the value of the
18536 first given entry found from the beginning of the DN. If a positive/negative
18537 occurrence number is specified as the optional second argument, it returns
18538 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020018539 For instance, "ssl_s_i_dn(OU,2)" the second organization unit, and
18540 "ssl_s_i_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020018541 The <format> parameter allows you to receive the DN suitable for
18542 consumption by different protocols. Currently supported is rfc2253 for
18543 LDAP v3.
18544 If you'd like to modify the format only you can specify an empty string
18545 and zero for the first two parameters. Example: ssl_s_i_dn(,0,rfc2253)
18546
18547ssl_s_s_dn([<entry>[,<occ>[,<format>]]]) : string
18548 When the outgoing connection was made over an SSL/TLS transport layer,
18549 returns the full distinguished name of the subject of the certificate
18550 presented by the server when no <entry> is specified, or the value of the
18551 first given entry found from the beginning of the DN. If a positive/negative
18552 occurrence number is specified as the optional second argument, it returns
18553 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020018554 For instance, "ssl_s_s_dn(OU,2)" the second organization unit, and
18555 "ssl_s_s_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020018556 The <format> parameter allows you to receive the DN suitable for
18557 consumption by different protocols. Currently supported is rfc2253 for
18558 LDAP v3.
18559 If you'd like to modify the format only you can specify an empty string
18560 and zero for the first two parameters. Example: ssl_s_s_dn(,0,rfc2253)
18561
18562ssl_s_serial : binary
18563 Returns the serial of the certificate presented by the server when the
18564 outgoing connection was made over an SSL/TLS transport layer. When used for
18565 an ACL, the value(s) to match against can be passed in hexadecimal form.
18566
18567ssl_s_sha1 : binary
18568 Returns the SHA-1 fingerprint of the certificate presented by the server
18569 when the outgoing connection was made over an SSL/TLS transport layer. This
18570 can be used to know which certificate was chosen using SNI.
18571
18572ssl_s_sig_alg : string
18573 Returns the name of the algorithm used to sign the certificate presented by
18574 the server when the outgoing connection was made over an SSL/TLS transport
18575 layer.
18576
18577ssl_s_version : integer
18578 Returns the version of the certificate presented by the server when the
18579 outgoing connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018580
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200185817.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020018582------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018583
Willy Tarreau74ca5042013-06-11 23:12:07 +020018584Fetching samples from buffer contents is a bit different from the previous
18585sample fetches above because the sampled data are ephemeral. These data can
18586only be used when they're available and will be lost when they're forwarded.
18587For this reason, samples fetched from buffer contents during a request cannot
18588be used in a response for example. Even while the data are being fetched, they
18589can change. Sometimes it is necessary to set some delays or combine multiple
18590sample fetch methods to ensure that the expected data are complete and usable,
18591for example through TCP request content inspection. Please see the "tcp-request
18592content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020018593
Christopher Fauleta434a002021-03-25 11:58:51 +010018594Warning : Following sample fetches are ignored if used from HTTP proxies. They
18595 only deal with raw contents found in the buffers. On their side,
18596 HTTTP proxies use structured content. Thus raw representation of
18597 these data are meaningless. A warning is emitted if an ACL relies on
18598 one of the following sample fetches. But it is not possible to detect
18599 all invalid usage (for instance inside a log-format string or a
18600 sample expression). So be careful.
18601
Willy Tarreau74ca5042013-06-11 23:12:07 +020018602payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010018603 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018604 "stick on", "stick match"), and for "res.payload" when used in the context of
18605 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010018606
Willy Tarreau74ca5042013-06-11 23:12:07 +020018607payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
18608 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010018609 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020018610 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010018611
Willy Tarreau74ca5042013-06-11 23:12:07 +020018612req.len : integer
18613req_len : integer (deprecated)
18614 Returns an integer value corresponding to the number of bytes present in the
18615 request buffer. This is mostly used in ACL. It is important to understand
18616 that this test does not return false as long as the buffer is changing. This
18617 means that a check with equality to zero will almost always immediately match
18618 at the beginning of the session, while a test for more data will wait for
18619 that data to come in and return false only when haproxy is certain that no
18620 more data will come in. This test was designed to be used with TCP request
18621 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018622
Willy Tarreau74ca5042013-06-11 23:12:07 +020018623req.payload(<offset>,<length>) : binary
18624 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020018625 in the request buffer. As a special case, if the <length> argument is zero,
18626 the the whole buffer from <offset> to the end is extracted. This can be used
18627 with ACLs in order to check for the presence of some content in a buffer at
18628 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018629
Willy Tarreau74ca5042013-06-11 23:12:07 +020018630 ACL alternatives :
18631 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018632
Willy Tarreau74ca5042013-06-11 23:12:07 +020018633req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
18634 This extracts a binary block whose size is specified at <offset1> for <length>
18635 bytes, and which starts at <offset2> if specified or just after the length in
18636 the request buffer. The <offset2> parameter also supports relative offsets if
18637 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018638
Willy Tarreau74ca5042013-06-11 23:12:07 +020018639 ACL alternatives :
18640 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018641
Willy Tarreau74ca5042013-06-11 23:12:07 +020018642 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018643
Willy Tarreau74ca5042013-06-11 23:12:07 +020018644req.proto_http : boolean
18645req_proto_http : boolean (deprecated)
18646 Returns true when data in the request buffer look like HTTP and correctly
18647 parses as such. It is the same parser as the common HTTP request parser which
18648 is used so there should be no surprises. The test does not match until the
18649 request is complete, failed or timed out. This test may be used to report the
18650 protocol in TCP logs, but the biggest use is to block TCP request analysis
18651 until a complete HTTP request is present in the buffer, for example to track
18652 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018653
Willy Tarreau74ca5042013-06-11 23:12:07 +020018654 Example:
18655 # track request counts per "base" (concatenation of Host+URL)
18656 tcp-request inspect-delay 10s
18657 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020018658 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018659
Willy Tarreau74ca5042013-06-11 23:12:07 +020018660req.rdp_cookie([<name>]) : string
18661rdp_cookie([<name>]) : string (deprecated)
18662 When the request buffer looks like the RDP protocol, extracts the RDP cookie
18663 <name>, or any cookie if unspecified. The parser only checks for the first
18664 cookie, as illustrated in the RDP protocol specification. The cookie name is
18665 case insensitive. Generally the "MSTS" cookie name will be used, as it can
18666 contain the user name of the client connecting to the server if properly
18667 configured on the client. The "MSTSHASH" cookie is often used as well for
18668 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018669
Willy Tarreau74ca5042013-06-11 23:12:07 +020018670 This differs from "balance rdp-cookie" in that any balancing algorithm may be
18671 used and thus the distribution of clients to backend servers is not linked to
18672 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
18673 such as "balance roundrobin" or "balance leastconn" will lead to a more even
18674 distribution of clients to backend servers than the hash used by "balance
18675 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018676
Willy Tarreau74ca5042013-06-11 23:12:07 +020018677 ACL derivatives :
18678 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018679
Willy Tarreau74ca5042013-06-11 23:12:07 +020018680 Example :
18681 listen tse-farm
18682 bind 0.0.0.0:3389
18683 # wait up to 5s for an RDP cookie in the request
18684 tcp-request inspect-delay 5s
18685 tcp-request content accept if RDP_COOKIE
18686 # apply RDP cookie persistence
18687 persist rdp-cookie
18688 # Persist based on the mstshash cookie
18689 # This is only useful makes sense if
18690 # balance rdp-cookie is not used
18691 stick-table type string size 204800
18692 stick on req.rdp_cookie(mstshash)
18693 server srv1 1.1.1.1:3389
18694 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018695
Willy Tarreau74ca5042013-06-11 23:12:07 +020018696 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
18697 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018698
Willy Tarreau74ca5042013-06-11 23:12:07 +020018699req.rdp_cookie_cnt([name]) : integer
18700rdp_cookie_cnt([name]) : integer (deprecated)
18701 Tries to parse the request buffer as RDP protocol, then returns an integer
18702 corresponding to the number of RDP cookies found. If an optional cookie name
18703 is passed, only cookies matching this name are considered. This is mostly
18704 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018705
Willy Tarreau74ca5042013-06-11 23:12:07 +020018706 ACL derivatives :
18707 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018708
Alex Zorin4afdd132018-12-30 13:56:28 +110018709req.ssl_alpn : string
18710 Returns a string containing the values of the Application-Layer Protocol
18711 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
18712 ClientHello message. Note that this only applies to raw contents found in the
18713 request buffer and not to the contents deciphered via an SSL data layer, so
18714 this will not work with "bind" lines having the "ssl" option. This is useful
18715 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020018716 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110018717
18718 Examples :
18719 # Wait for a client hello for at most 5 seconds
18720 tcp-request inspect-delay 5s
18721 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020018722 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110018723 default_backend bk_default
18724
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020018725req.ssl_ec_ext : boolean
18726 Returns a boolean identifying if client sent the Supported Elliptic Curves
18727 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020018728 message. This can be used to present ECC compatible clients with EC
18729 certificate and to use RSA for all others, on the same IP address. Note that
18730 this only applies to raw contents found in the request buffer and not to
18731 contents deciphered via an SSL data layer, so this will not work with "bind"
18732 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020018733
Willy Tarreau74ca5042013-06-11 23:12:07 +020018734req.ssl_hello_type : integer
18735req_ssl_hello_type : integer (deprecated)
18736 Returns an integer value containing the type of the SSL hello message found
18737 in the request buffer if the buffer contains data that parse as a complete
18738 SSL (v3 or superior) client hello message. Note that this only applies to raw
18739 contents found in the request buffer and not to contents deciphered via an
18740 SSL data layer, so this will not work with "bind" lines having the "ssl"
18741 option. This is mostly used in ACL to detect presence of an SSL hello message
18742 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018743
Willy Tarreau74ca5042013-06-11 23:12:07 +020018744req.ssl_sni : string
18745req_ssl_sni : string (deprecated)
18746 Returns a string containing the value of the Server Name TLS extension sent
18747 by a client in a TLS stream passing through the request buffer if the buffer
18748 contains data that parse as a complete SSL (v3 or superior) client hello
18749 message. Note that this only applies to raw contents found in the request
18750 buffer and not to contents deciphered via an SSL data layer, so this will not
Lukas Tribusa267b5d2020-07-19 00:25:06 +020018751 work with "bind" lines having the "ssl" option. This will only work for actual
18752 implicit TLS based protocols like HTTPS (443), IMAPS (993), SMTPS (465),
18753 however it will not work for explicit TLS based protocols, like SMTP (25/587)
18754 or IMAP (143). SNI normally contains the name of the host the client tries to
18755 connect to (for recent browsers). SNI is useful for allowing or denying access
18756 to certain hosts when SSL/TLS is used by the client. This test was designed to
18757 be used with TCP request content inspection. If content switching is needed,
18758 it is recommended to first wait for a complete client hello (type 1), like in
18759 the example below. See also "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018760
Willy Tarreau74ca5042013-06-11 23:12:07 +020018761 ACL derivatives :
18762 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018763
Willy Tarreau74ca5042013-06-11 23:12:07 +020018764 Examples :
18765 # Wait for a client hello for at most 5 seconds
18766 tcp-request inspect-delay 5s
18767 tcp-request content accept if { req_ssl_hello_type 1 }
18768 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
18769 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018770
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053018771req.ssl_st_ext : integer
18772 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
18773 Returns 1 if the client sent SessionTicket TLS Extension
18774 Returns 2 if the client also sent non-zero length TLS SessionTicket
18775 Note that this only applies to raw contents found in the request buffer and
18776 not to contents deciphered via an SSL data layer, so this will not work with
18777 "bind" lines having the "ssl" option. This can for example be used to detect
18778 whether the client sent a SessionTicket or not and stick it accordingly, if
18779 no SessionTicket then stick on SessionID or don't stick as there's no server
18780 side state is there when SessionTickets are in use.
18781
Willy Tarreau74ca5042013-06-11 23:12:07 +020018782req.ssl_ver : integer
18783req_ssl_ver : integer (deprecated)
18784 Returns an integer value containing the version of the SSL/TLS protocol of a
18785 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
18786 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
18787 composed of the major version multiplied by 65536, added to the minor
18788 version. Note that this only applies to raw contents found in the request
18789 buffer and not to contents deciphered via an SSL data layer, so this will not
18790 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010018791 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020018792 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018793
Willy Tarreau74ca5042013-06-11 23:12:07 +020018794 ACL derivatives :
18795 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010018796
Willy Tarreau47e8eba2013-09-11 23:28:46 +020018797res.len : integer
18798 Returns an integer value corresponding to the number of bytes present in the
18799 response buffer. This is mostly used in ACL. It is important to understand
18800 that this test does not return false as long as the buffer is changing. This
18801 means that a check with equality to zero will almost always immediately match
18802 at the beginning of the session, while a test for more data will wait for
18803 that data to come in and return false only when haproxy is certain that no
18804 more data will come in. This test was designed to be used with TCP response
Christopher Faulete596d182020-05-05 17:46:34 +020018805 content inspection. But it may also be used in tcp-check based expect rules.
Willy Tarreau47e8eba2013-09-11 23:28:46 +020018806
Willy Tarreau74ca5042013-06-11 23:12:07 +020018807res.payload(<offset>,<length>) : binary
18808 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020018809 in the response buffer. As a special case, if the <length> argument is zero,
Christopher Faulete596d182020-05-05 17:46:34 +020018810 the whole buffer from <offset> to the end is extracted. This can be used
Willy Tarreau00f00842013-08-02 11:07:32 +020018811 with ACLs in order to check for the presence of some content in a buffer at
Christopher Faulete596d182020-05-05 17:46:34 +020018812 any location. It may also be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018813
Willy Tarreau74ca5042013-06-11 23:12:07 +020018814res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
18815 This extracts a binary block whose size is specified at <offset1> for <length>
18816 bytes, and which starts at <offset2> if specified or just after the length in
18817 the response buffer. The <offset2> parameter also supports relative offsets
Christopher Faulete596d182020-05-05 17:46:34 +020018818 if prepended with a '+' or '-' sign. It may also be used in tcp-check based
18819 expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018820
Willy Tarreau74ca5042013-06-11 23:12:07 +020018821 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018822
Willy Tarreau971f7b62015-09-29 14:06:59 +020018823res.ssl_hello_type : integer
18824rep_ssl_hello_type : integer (deprecated)
18825 Returns an integer value containing the type of the SSL hello message found
18826 in the response buffer if the buffer contains data that parses as a complete
18827 SSL (v3 or superior) hello message. Note that this only applies to raw
18828 contents found in the response buffer and not to contents deciphered via an
18829 SSL data layer, so this will not work with "server" lines having the "ssl"
18830 option. This is mostly used in ACL to detect presence of an SSL hello message
18831 that is supposed to contain an SSL session ID usable for stickiness.
18832
Willy Tarreau74ca5042013-06-11 23:12:07 +020018833wait_end : boolean
18834 This fetch either returns true when the inspection period is over, or does
18835 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018836 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020018837 actions, such as a delayed reject for some special addresses. Since it either
18838 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018839 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020018840 "WAIT_END" is always usable without prior declaration. This test was designed
18841 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018842
Willy Tarreau74ca5042013-06-11 23:12:07 +020018843 Examples :
18844 # delay every incoming request by 2 seconds
18845 tcp-request inspect-delay 2s
18846 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010018847
Willy Tarreau74ca5042013-06-11 23:12:07 +020018848 # don't immediately tell bad guys they are rejected
18849 tcp-request inspect-delay 10s
18850 acl goodguys src 10.0.0.0/24
18851 acl badguys src 10.0.1.0/24
18852 tcp-request content accept if goodguys
18853 tcp-request content reject if badguys WAIT_END
18854 tcp-request content reject
18855
18856
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200188577.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020018858--------------------------------------
18859
18860It is possible to fetch samples from HTTP contents, requests and responses.
18861This application layer is also called layer 7. It is only possible to fetch the
18862data in this section when a full HTTP request or response has been parsed from
18863its respective request or response buffer. This is always the case with all
18864HTTP specific rules and for sections running with "mode http". When using TCP
18865content inspection, it may be necessary to support an inspection delay in order
18866to let the request or response come in first. These fetches may require a bit
18867more CPU resources than the layer 4 ones, but not much since the request and
18868response are indexed.
18869
Christopher Faulet4d37e532021-03-26 14:44:00 +010018870Note : Regarding HTTP processing from the tcp-request content rules, everything
18871 will work as expected from an HTTP proxy. However, from a TCP proxy,
18872 without an HTTP upgrade, it will only work for HTTP/1 content. For
18873 HTTP/2 content, only the preface is visible. Thus, it is only possible
18874 to rely to "req.proto_http", "req.ver" and eventually "method" sample
18875 fetches. All other L7 sample fetches will fail. After an HTTP upgrade,
18876 they will work in the same manner than from an HTTP proxy.
18877
Willy Tarreau74ca5042013-06-11 23:12:07 +020018878base : string
18879 This returns the concatenation of the first Host header and the path part of
18880 the request, which starts at the first slash and ends before the question
18881 mark. It can be useful in virtual hosted environments to detect URL abuses as
18882 well as to improve shared caches efficiency. Using this with a limited size
18883 stick table also allows one to collect statistics about most commonly
18884 requested objects by host/path. With ACLs it can allow simple content
18885 switching rules involving the host and the path at the same time, such as
18886 "www.example.com/favicon.ico". See also "path" and "uri".
18887
18888 ACL derivatives :
18889 base : exact string match
18890 base_beg : prefix match
18891 base_dir : subdir match
18892 base_dom : domain match
18893 base_end : suffix match
18894 base_len : length match
18895 base_reg : regex match
18896 base_sub : substring match
18897
18898base32 : integer
18899 This returns a 32-bit hash of the value returned by the "base" fetch method
18900 above. This is useful to track per-URL activity on high traffic sites without
18901 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020018902 memory. The output type is an unsigned integer. The hash function used is
18903 SDBM with full avalanche on the output. Technically, base32 is exactly equal
18904 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020018905
18906base32+src : binary
18907 This returns the concatenation of the base32 fetch above and the src fetch
18908 below. The resulting type is of type binary, with a size of 8 or 20 bytes
18909 depending on the source address family. This can be used to track per-IP,
18910 per-URL counters.
18911
Yves Lafonb4d37082021-02-11 11:01:28 +010018912baseq : string
18913 This returns the concatenation of the first Host header and the path part of
18914 the request with the query-string, which starts at the first slash. Using this
18915 instead of "base" allows one to properly identify the target resource, for
18916 statistics or caching use cases. See also "path", "pathq" and "base".
18917
William Lallemand65ad6e12014-01-31 15:08:02 +010018918capture.req.hdr(<idx>) : string
18919 This extracts the content of the header captured by the "capture request
18920 header", idx is the position of the capture keyword in the configuration.
18921 The first entry is an index of 0. See also: "capture request header".
18922
18923capture.req.method : string
18924 This extracts the METHOD of an HTTP request. It can be used in both request
18925 and response. Unlike "method", it can be used in both request and response
18926 because it's allocated.
18927
18928capture.req.uri : string
18929 This extracts the request's URI, which starts at the first slash and ends
18930 before the first space in the request (without the host part). Unlike "path"
18931 and "url", it can be used in both request and response because it's
18932 allocated.
18933
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020018934capture.req.ver : string
18935 This extracts the request's HTTP version and returns either "HTTP/1.0" or
18936 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
18937 logs because it relies on a persistent flag.
18938
William Lallemand65ad6e12014-01-31 15:08:02 +010018939capture.res.hdr(<idx>) : string
18940 This extracts the content of the header captured by the "capture response
18941 header", idx is the position of the capture keyword in the configuration.
18942 The first entry is an index of 0.
18943 See also: "capture response header"
18944
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020018945capture.res.ver : string
18946 This extracts the response's HTTP version and returns either "HTTP/1.0" or
18947 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
18948 persistent flag.
18949
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018950req.body : binary
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018951 This returns the HTTP request's available body as a block of data. It is
18952 recommended to use "option http-buffer-request" to be sure to wait, as much
18953 as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018954
Thierry FOURNIER9826c772015-05-20 15:50:54 +020018955req.body_param([<name>) : string
18956 This fetch assumes that the body of the POST request is url-encoded. The user
18957 can check if the "content-type" contains the value
18958 "application/x-www-form-urlencoded". This extracts the first occurrence of the
18959 parameter <name> in the body, which ends before '&'. The parameter name is
18960 case-sensitive. If no name is given, any parameter will match, and the first
18961 one will be returned. The result is a string corresponding to the value of the
18962 parameter <name> as presented in the request body (no URL decoding is
18963 performed). Note that the ACL version of this fetch iterates over multiple
18964 parameters and will iteratively report all parameters values if no name is
18965 given.
18966
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018967req.body_len : integer
18968 This returns the length of the HTTP request's available body in bytes. It may
18969 be lower than the advertised length if the body is larger than the buffer. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018970 is recommended to use "option http-buffer-request" to be sure to wait, as
18971 much as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018972
18973req.body_size : integer
18974 This returns the advertised length of the HTTP request's body in bytes. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018975 will represent the advertised Content-Length header, or the size of the
18976 available data in case of chunked encoding.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018977
Willy Tarreau74ca5042013-06-11 23:12:07 +020018978req.cook([<name>]) : string
18979cook([<name>]) : string (deprecated)
18980 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18981 header line from the request, and returns its value as string. If no name is
18982 specified, the first cookie value is returned. When used with ACLs, all
18983 matching cookies are evaluated. Spaces around the name and the value are
18984 ignored as requested by the Cookie header specification (RFC6265). The cookie
18985 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
18986 well return an empty value if it is present. Use the "found" match to detect
18987 presence. Use the res.cook() variant for response cookies sent by the server.
18988
18989 ACL derivatives :
18990 cook([<name>]) : exact string match
18991 cook_beg([<name>]) : prefix match
18992 cook_dir([<name>]) : subdir match
18993 cook_dom([<name>]) : domain match
18994 cook_end([<name>]) : suffix match
18995 cook_len([<name>]) : length match
18996 cook_reg([<name>]) : regex match
18997 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010018998
Willy Tarreau74ca5042013-06-11 23:12:07 +020018999req.cook_cnt([<name>]) : integer
19000cook_cnt([<name>]) : integer (deprecated)
19001 Returns an integer value representing the number of occurrences of the cookie
19002 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019003
Willy Tarreau74ca5042013-06-11 23:12:07 +020019004req.cook_val([<name>]) : integer
19005cook_val([<name>]) : integer (deprecated)
19006 This extracts the last occurrence of the cookie name <name> on a "Cookie"
19007 header line from the request, and converts its value to an integer which is
19008 returned. If no name is specified, the first cookie value is returned. When
19009 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020019010
Willy Tarreau74ca5042013-06-11 23:12:07 +020019011cookie([<name>]) : string (deprecated)
19012 This extracts the last occurrence of the cookie name <name> on a "Cookie"
19013 header line from the request, or a "Set-Cookie" header from the response, and
19014 returns its value as a string. A typical use is to get multiple clients
19015 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020019016 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020019017 multi-peer synchronization and state keeping across restarts. If no name is
19018 specified, the first cookie value is returned. This fetch should not be used
19019 anymore and should be replaced by req.cook() or res.cook() instead as it
19020 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019021
Willy Tarreau74ca5042013-06-11 23:12:07 +020019022hdr([<name>[,<occ>]]) : string
19023 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
19024 used on responses. Please refer to these respective fetches for more details.
19025 In case of doubt about the fetch direction, please use the explicit ones.
19026 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019027 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019028
Willy Tarreau74ca5042013-06-11 23:12:07 +020019029req.fhdr(<name>[,<occ>]) : string
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019030 This returns the full value of the last occurrence of header <name> in an
19031 HTTP request. It differs from req.hdr() in that any commas present in the
19032 value are returned and are not used as delimiters. This is sometimes useful
19033 with headers such as User-Agent.
19034
19035 When used from an ACL, all occurrences are iterated over until a match is
19036 found.
19037
Willy Tarreau74ca5042013-06-11 23:12:07 +020019038 Optionally, a specific occurrence might be specified as a position number.
19039 Positive values indicate a position from the first occurrence, with 1 being
19040 the first one. Negative values indicate positions relative to the last one,
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019041 with -1 being the last one.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019042
Willy Tarreau74ca5042013-06-11 23:12:07 +020019043req.fhdr_cnt([<name>]) : integer
19044 Returns an integer value representing the number of occurrences of request
19045 header field name <name>, or the total number of header fields if <name> is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019046 not specified. Like req.fhdr() it differs from res.hdr_cnt() by not splitting
19047 headers at commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019048
Willy Tarreau74ca5042013-06-11 23:12:07 +020019049req.hdr([<name>[,<occ>]]) : string
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019050 This returns the last comma-separated value of the header <name> in an HTTP
19051 request. The fetch considers any comma as a delimiter for distinct values.
19052 This is useful if you need to process headers that are defined to be a list
19053 of values, such as Accept, or X-Forwarded-For. If full-line headers are
19054 desired instead, use req.fhdr(). Please carefully check RFC 7231 to know how
19055 certain headers are supposed to be parsed. Also, some of them are case
19056 insensitive (e.g. Connection).
19057
19058 When used from an ACL, all occurrences are iterated over until a match is
19059 found.
19060
Willy Tarreau74ca5042013-06-11 23:12:07 +020019061 Optionally, a specific occurrence might be specified as a position number.
19062 Positive values indicate a position from the first occurrence, with 1 being
19063 the first one. Negative values indicate positions relative to the last one,
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019064 with -1 being the last one.
19065
19066 A typical use is with the X-Forwarded-For header once converted to IP,
19067 associated with an IP stick-table.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019068
Willy Tarreau74ca5042013-06-11 23:12:07 +020019069 ACL derivatives :
19070 hdr([<name>[,<occ>]]) : exact string match
19071 hdr_beg([<name>[,<occ>]]) : prefix match
19072 hdr_dir([<name>[,<occ>]]) : subdir match
19073 hdr_dom([<name>[,<occ>]]) : domain match
19074 hdr_end([<name>[,<occ>]]) : suffix match
19075 hdr_len([<name>[,<occ>]]) : length match
19076 hdr_reg([<name>[,<occ>]]) : regex match
19077 hdr_sub([<name>[,<occ>]]) : substring match
19078
19079req.hdr_cnt([<name>]) : integer
19080hdr_cnt([<header>]) : integer (deprecated)
19081 Returns an integer value representing the number of occurrences of request
19082 header field name <name>, or the total number of header field values if
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019083 <name> is not specified. Like req.hdr() it counts each comma separated
19084 part of the header's value. If counting of full-line headers is desired,
19085 then req.fhdr_cnt() should be used instead.
19086
19087 With ACLs, it can be used to detect presence, absence or abuse of a specific
19088 header, as well as to block request smuggling attacks by rejecting requests
19089 which contain more than one of certain headers.
19090
19091 Refer to req.hdr() for more information on header matching.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019092
19093req.hdr_ip([<name>[,<occ>]]) : ip
19094hdr_ip([<name>[,<occ>]]) : ip (deprecated)
19095 This extracts the last occurrence of header <name> in an HTTP request,
19096 converts it to an IPv4 or IPv6 address and returns this address. When used
19097 with ACLs, all occurrences are checked, and if <name> is omitted, every value
Willy Tarreau7b0e00d2021-03-25 14:12:29 +010019098 of every header is checked. The parser strictly adheres to the format
19099 described in RFC7239, with the extension that IPv4 addresses may optionally
19100 be followed by a colon (':') and a valid decimal port number (0 to 65535),
19101 which will be silently dropped. All other forms will not match and will
19102 cause the address to be ignored.
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019103
19104 The <occ> parameter is processed as with req.hdr().
19105
19106 A typical use is with the X-Forwarded-For and X-Client-IP headers.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019107
19108req.hdr_val([<name>[,<occ>]]) : integer
19109hdr_val([<name>[,<occ>]]) : integer (deprecated)
19110 This extracts the last occurrence of header <name> in an HTTP request, and
19111 converts it to an integer value. When used with ACLs, all occurrences are
19112 checked, and if <name> is omitted, every value of every header is checked.
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019113
19114 The <occ> parameter is processed as with req.hdr().
19115
19116 A typical use is with the X-Forwarded-For header.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019117
Christopher Faulet687a68e2020-11-24 17:13:24 +010019118req.hdrs : string
19119 Returns the current request headers as string including the last empty line
19120 separating headers from the request body. The last empty line can be used to
19121 detect a truncated header block. This sample fetch is useful for some SPOE
19122 headers analyzers and for advanced logging.
19123
19124req.hdrs_bin : binary
19125 Returns the current request headers contained in preparsed binary form. This
19126 is useful for offloading some processing with SPOE. Each string is described
19127 by a length followed by the number of bytes indicated in the length. The
19128 length is represented using the variable integer encoding detailed in the
19129 SPOE documentation. The end of the list is marked by a couple of empty header
19130 names and values (length of 0 for both).
19131
19132 *(<str:header-name><str:header-value>)<empty string><empty string>
Frédéric Lécailleec891192019-02-26 15:02:35 +010019133
Christopher Faulet687a68e2020-11-24 17:13:24 +010019134 int: refer to the SPOE documentation for the encoding
19135 str: <int:length><bytes>
Frédéric Lécailleec891192019-02-26 15:02:35 +010019136
Willy Tarreau74ca5042013-06-11 23:12:07 +020019137http_auth(<userlist>) : boolean
19138 Returns a boolean indicating whether the authentication data received from
19139 the client match a username & password stored in the specified userlist. This
19140 fetch function is not really useful outside of ACLs. Currently only http
19141 basic auth is supported.
19142
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010019143http_auth_group(<userlist>) : string
19144 Returns a string corresponding to the user name found in the authentication
19145 data received from the client if both the user name and password are valid
19146 according to the specified userlist. The main purpose is to use it in ACLs
19147 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019148 This fetch function is not really useful outside of ACLs. Currently only http
19149 basic auth is supported.
19150
19151 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010019152 http_auth_group(<userlist>) : group ...
19153 Returns true when the user extracted from the request and whose password is
19154 valid according to the specified userlist belongs to at least one of the
19155 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019156
Christopher Fauleta4063562019-08-02 11:51:37 +020019157http_auth_pass : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010019158 Returns the user's password found in the authentication data received from
19159 the client, as supplied in the Authorization header. Not checks are
19160 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020019161
19162http_auth_type : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010019163 Returns the authentication method found in the authentication data received from
19164 the client, as supplied in the Authorization header. Not checks are
19165 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020019166
19167http_auth_user : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010019168 Returns the user name found in the authentication data received from the
19169 client, as supplied in the Authorization header. Not checks are performed by
19170 this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020019171
Willy Tarreau74ca5042013-06-11 23:12:07 +020019172http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020019173 Returns true when the request being processed is the first one of the
19174 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020019175 from some requests when a request is not the first one, or to help grouping
19176 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020019177
Willy Tarreau74ca5042013-06-11 23:12:07 +020019178method : integer + string
19179 Returns an integer value corresponding to the method in the HTTP request. For
19180 example, "GET" equals 1 (check sources to establish the matching). Value 9
19181 means "other method" and may be converted to a string extracted from the
19182 stream. This should not be used directly as a sample, this is only meant to
19183 be used from ACLs, which transparently convert methods from patterns to these
19184 integer + string values. Some predefined ACL already check for most common
19185 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019186
Willy Tarreau74ca5042013-06-11 23:12:07 +020019187 ACL derivatives :
19188 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020019189
Willy Tarreau74ca5042013-06-11 23:12:07 +020019190 Example :
19191 # only accept GET and HEAD requests
19192 acl valid_method method GET HEAD
19193 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020019194
Willy Tarreau74ca5042013-06-11 23:12:07 +020019195path : string
19196 This extracts the request's URL path, which starts at the first slash and
19197 ends before the question mark (without the host part). A typical use is with
19198 prefetch-capable caches, and with portals which need to aggregate multiple
19199 information from databases and keep them in caches. Note that with outgoing
19200 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010019201 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020019202 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019203
Willy Tarreau74ca5042013-06-11 23:12:07 +020019204 ACL derivatives :
19205 path : exact string match
19206 path_beg : prefix match
19207 path_dir : subdir match
19208 path_dom : domain match
19209 path_end : suffix match
19210 path_len : length match
19211 path_reg : regex match
19212 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020019213
Christopher Faulete720c322020-09-02 17:25:18 +020019214pathq : string
19215 This extracts the request's URL path with the query-string, which starts at
19216 the first slash. This sample fetch is pretty handy to always retrieve a
19217 relative URI, excluding the scheme and the authority part, if any. Indeed,
19218 while it is the common representation for an HTTP/1.1 request target, in
19219 HTTP/2, an absolute URI is often used. This sample fetch will return the same
19220 result in both cases.
19221
Willy Tarreau49ad95c2015-01-19 15:06:26 +010019222query : string
19223 This extracts the request's query string, which starts after the first
19224 question mark. If no question mark is present, this fetch returns nothing. If
19225 a question mark is present but nothing follows, it returns an empty string.
19226 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010019227 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010019228 which stops before the question mark.
19229
Willy Tarreaueb27ec72015-02-20 13:55:29 +010019230req.hdr_names([<delim>]) : string
19231 This builds a string made from the concatenation of all header names as they
19232 appear in the request when the rule is evaluated. The default delimiter is
19233 the comma (',') but it may be overridden as an optional argument <delim>. In
19234 this case, only the first character of <delim> is considered.
19235
Willy Tarreau74ca5042013-06-11 23:12:07 +020019236req.ver : string
19237req_ver : string (deprecated)
19238 Returns the version string from the HTTP request, for example "1.1". This can
19239 be useful for logs, but is mostly there for ACL. Some predefined ACL already
19240 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019241
Willy Tarreau74ca5042013-06-11 23:12:07 +020019242 ACL derivatives :
19243 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020019244
Christopher Faulete596d182020-05-05 17:46:34 +020019245res.body : binary
19246 This returns the HTTP response's available body as a block of data. Unlike
19247 the request side, there is no directive to wait for the response's body. This
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019248 sample fetch is really useful (and usable) in the health-check context.
19249
19250 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019251
19252res.body_len : integer
19253 This returns the length of the HTTP response available body in bytes. Unlike
19254 the request side, there is no directive to wait for the response's body. This
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019255 sample fetch is really useful (and usable) in the health-check context.
19256
19257 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019258
19259res.body_size : integer
19260 This returns the advertised length of the HTTP response body in bytes. It
19261 will represent the advertised Content-Length header, or the size of the
19262 available data in case of chunked encoding. Unlike the request side, there is
19263 no directive to wait for the response body. This sample fetch is really
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019264 useful (and usable) in the health-check context.
19265
19266 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019267
Remi Tricot-Le Bretonbf971212020-10-27 11:55:57 +010019268res.cache_hit : boolean
19269 Returns the boolean "true" value if the response has been built out of an
19270 HTTP cache entry, otherwise returns boolean "false".
19271
19272res.cache_name : string
19273 Returns a string containing the name of the HTTP cache that was used to
19274 build the HTTP response if res.cache_hit is true, otherwise returns an
19275 empty string.
19276
Willy Tarreau74ca5042013-06-11 23:12:07 +020019277res.comp : boolean
19278 Returns the boolean "true" value if the response has been compressed by
19279 HAProxy, otherwise returns boolean "false". This may be used to add
19280 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019281
Willy Tarreau74ca5042013-06-11 23:12:07 +020019282res.comp_algo : string
19283 Returns a string containing the name of the algorithm used if the response
19284 was compressed by HAProxy, for example : "deflate". This may be used to add
19285 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019286
Willy Tarreau74ca5042013-06-11 23:12:07 +020019287res.cook([<name>]) : string
19288scook([<name>]) : string (deprecated)
19289 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
19290 header line from the response, and returns its value as string. If no name is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019291 specified, the first cookie value is returned.
19292
19293 It may be used in tcp-check based expect rules.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020019294
Willy Tarreau74ca5042013-06-11 23:12:07 +020019295 ACL derivatives :
19296 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020019297
Willy Tarreau74ca5042013-06-11 23:12:07 +020019298res.cook_cnt([<name>]) : integer
19299scook_cnt([<name>]) : integer (deprecated)
19300 Returns an integer value representing the number of occurrences of the cookie
19301 <name> in the response, or all cookies if <name> is not specified. This is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019302 mostly useful when combined with ACLs to detect suspicious responses.
19303
19304 It may be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019305
Willy Tarreau74ca5042013-06-11 23:12:07 +020019306res.cook_val([<name>]) : integer
19307scook_val([<name>]) : integer (deprecated)
19308 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
19309 header line from the response, and converts its value to an integer which is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019310 returned. If no name is specified, the first cookie value is returned.
19311
19312 It may be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019313
Willy Tarreau74ca5042013-06-11 23:12:07 +020019314res.fhdr([<name>[,<occ>]]) : string
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019315 This fetch works like the req.fhdr() fetch with the difference that it acts
19316 on the headers within an HTTP response.
19317
19318 Like req.fhdr() the res.fhdr() fetch returns full values. If the header is
19319 defined to be a list you should use res.hdr().
19320
19321 This fetch is sometimes useful with headers such as Date or Expires.
19322
19323 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019324
Willy Tarreau74ca5042013-06-11 23:12:07 +020019325res.fhdr_cnt([<name>]) : integer
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019326 This fetch works like the req.fhdr_cnt() fetch with the difference that it
19327 acts on the headers within an HTTP response.
19328
19329 Like req.fhdr_cnt() the res.fhdr_cnt() fetch acts on full values. If the
19330 header is defined to be a list you should use res.hdr_cnt().
19331
19332 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019333
Willy Tarreau74ca5042013-06-11 23:12:07 +020019334res.hdr([<name>[,<occ>]]) : string
19335shdr([<name>[,<occ>]]) : string (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019336 This fetch works like the req.hdr() fetch with the difference that it acts
19337 on the headers within an HTTP response.
19338
Ilya Shipitsinacf84592021-02-06 22:29:08 +050019339 Like req.hdr() the res.hdr() fetch considers the comma to be a delimiter. If
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019340 this is not desired res.fhdr() should be used.
19341
19342 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019343
Willy Tarreau74ca5042013-06-11 23:12:07 +020019344 ACL derivatives :
19345 shdr([<name>[,<occ>]]) : exact string match
19346 shdr_beg([<name>[,<occ>]]) : prefix match
19347 shdr_dir([<name>[,<occ>]]) : subdir match
19348 shdr_dom([<name>[,<occ>]]) : domain match
19349 shdr_end([<name>[,<occ>]]) : suffix match
19350 shdr_len([<name>[,<occ>]]) : length match
19351 shdr_reg([<name>[,<occ>]]) : regex match
19352 shdr_sub([<name>[,<occ>]]) : substring match
19353
19354res.hdr_cnt([<name>]) : integer
19355shdr_cnt([<name>]) : integer (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019356 This fetch works like the req.hdr_cnt() fetch with the difference that it
19357 acts on the headers within an HTTP response.
19358
19359 Like req.hdr_cnt() the res.hdr_cnt() fetch considers the comma to be a
Ilya Shipitsinacf84592021-02-06 22:29:08 +050019360 delimiter. If this is not desired res.fhdr_cnt() should be used.
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019361
19362 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019363
Willy Tarreau74ca5042013-06-11 23:12:07 +020019364res.hdr_ip([<name>[,<occ>]]) : ip
19365shdr_ip([<name>[,<occ>]]) : ip (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019366 This fetch works like the req.hdr_ip() fetch with the difference that it
19367 acts on the headers within an HTTP response.
19368
19369 This can be useful to learn some data into a stick table.
19370
19371 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019372
Willy Tarreaueb27ec72015-02-20 13:55:29 +010019373res.hdr_names([<delim>]) : string
19374 This builds a string made from the concatenation of all header names as they
19375 appear in the response when the rule is evaluated. The default delimiter is
19376 the comma (',') but it may be overridden as an optional argument <delim>. In
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019377 this case, only the first character of <delim> is considered.
19378
19379 It may be used in tcp-check based expect rules.
Willy Tarreaueb27ec72015-02-20 13:55:29 +010019380
Willy Tarreau74ca5042013-06-11 23:12:07 +020019381res.hdr_val([<name>[,<occ>]]) : integer
19382shdr_val([<name>[,<occ>]]) : integer (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019383 This fetch works like the req.hdr_val() fetch with the difference that it
19384 acts on the headers within an HTTP response.
19385
19386 This can be useful to learn some data into a stick table.
19387
19388 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019389
19390res.hdrs : string
19391 Returns the current response headers as string including the last empty line
19392 separating headers from the request body. The last empty line can be used to
19393 detect a truncated header block. This sample fetch is useful for some SPOE
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019394 headers analyzers and for advanced logging.
19395
19396 It may also be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019397
19398res.hdrs_bin : binary
19399 Returns the current response headers contained in preparsed binary form. This
19400 is useful for offloading some processing with SPOE. It may be used in
19401 tcp-check based expect rules. Each string is described by a length followed
19402 by the number of bytes indicated in the length. The length is represented
19403 using the variable integer encoding detailed in the SPOE documentation. The
19404 end of the list is marked by a couple of empty header names and values
19405 (length of 0 for both).
19406
19407 *(<str:header-name><str:header-value>)<empty string><empty string>
19408
19409 int: refer to the SPOE documentation for the encoding
19410 str: <int:length><bytes>
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010019411
Willy Tarreau74ca5042013-06-11 23:12:07 +020019412res.ver : string
19413resp_ver : string (deprecated)
19414 Returns the version string from the HTTP response, for example "1.1". This
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019415 can be useful for logs, but is mostly there for ACL.
19416
19417 It may be used in tcp-check based expect rules.
Willy Tarreau0e698542011-09-16 08:32:32 +020019418
Willy Tarreau74ca5042013-06-11 23:12:07 +020019419 ACL derivatives :
19420 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010019421
Willy Tarreau74ca5042013-06-11 23:12:07 +020019422set-cookie([<name>]) : string (deprecated)
19423 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
19424 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020019425 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020019426 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010019427
Willy Tarreau74ca5042013-06-11 23:12:07 +020019428 This fetch function is deprecated and has been superseded by the "res.cook"
19429 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010019430
Willy Tarreau74ca5042013-06-11 23:12:07 +020019431status : integer
19432 Returns an integer containing the HTTP status code in the HTTP response, for
19433 example, 302. It is mostly used within ACLs and integer ranges, for example,
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019434 to remove any Location header if the response is not a 3xx.
19435
19436 It may be used in tcp-check based expect rules.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019437
Thierry Fournier0e00dca2016-04-07 15:47:40 +020019438unique-id : string
19439 Returns the unique-id attached to the request. The directive
19440 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
19441 fails. Note that the unique-id is usually used with HTTP requests, however this
19442 sample fetch can be used with other protocols. Obviously, if it is used with
19443 other protocols than HTTP, the unique-id-format directive must not contain
19444 HTTP parts. See: unique-id-format and unique-id-header
19445
Willy Tarreau74ca5042013-06-11 23:12:07 +020019446url : string
19447 This extracts the request's URL as presented in the request. A typical use is
19448 with prefetch-capable caches, and with portals which need to aggregate
19449 multiple information from databases and keep them in caches. With ACLs, using
19450 "path" is preferred over using "url", because clients may send a full URL as
19451 is normally done with proxies. The only real use is to match "*" which does
19452 not match in "path", and for which there is already a predefined ACL. See
19453 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019454
Willy Tarreau74ca5042013-06-11 23:12:07 +020019455 ACL derivatives :
19456 url : exact string match
19457 url_beg : prefix match
19458 url_dir : subdir match
19459 url_dom : domain match
19460 url_end : suffix match
19461 url_len : length match
19462 url_reg : regex match
19463 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019464
Willy Tarreau74ca5042013-06-11 23:12:07 +020019465url_ip : ip
19466 This extracts the IP address from the request's URL when the host part is
19467 presented as an IP address. Its use is very limited. For instance, a
19468 monitoring system might use this field as an alternative for the source IP in
19469 order to test what path a given source address would follow, or to force an
19470 entry in a table for a given source address. With ACLs it can be used to
19471 restrict access to certain systems through a proxy, for example when combined
19472 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019473
Willy Tarreau74ca5042013-06-11 23:12:07 +020019474url_port : integer
19475 This extracts the port part from the request's URL. Note that if the port is
19476 not specified in the request, port 80 is assumed. With ACLs it can be used to
19477 restrict access to certain systems through a proxy, for example when combined
19478 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019479
Willy Tarreau1ede1da2015-05-07 16:06:18 +020019480urlp([<name>[,<delim>]]) : string
19481url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020019482 This extracts the first occurrence of the parameter <name> in the query
19483 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020019484 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
19485 any parameter will match, and the first one will be returned. The result is
19486 a string corresponding to the value of the parameter <name> as presented in
19487 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020019488 stickiness based on a client ID, to extract an application cookie passed as a
19489 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020019490 this fetch iterates over multiple parameters and will iteratively report all
19491 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019492
Willy Tarreau74ca5042013-06-11 23:12:07 +020019493 ACL derivatives :
19494 urlp(<name>[,<delim>]) : exact string match
19495 urlp_beg(<name>[,<delim>]) : prefix match
19496 urlp_dir(<name>[,<delim>]) : subdir match
19497 urlp_dom(<name>[,<delim>]) : domain match
19498 urlp_end(<name>[,<delim>]) : suffix match
19499 urlp_len(<name>[,<delim>]) : length match
19500 urlp_reg(<name>[,<delim>]) : regex match
19501 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019502
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019503
Willy Tarreau74ca5042013-06-11 23:12:07 +020019504 Example :
19505 # match http://example.com/foo?PHPSESSIONID=some_id
19506 stick on urlp(PHPSESSIONID)
19507 # match http://example.com/foo;JSESSIONID=some_id
19508 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019509
Jarno Huuskonen676f6222017-03-30 09:19:45 +030019510urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020019511 See "urlp" above. This one extracts the URL parameter <name> in the request
19512 and converts it to an integer value. This can be used for session stickiness
19513 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020019514
Dragan Dosen0070cd52016-06-16 12:19:49 +020019515url32 : integer
19516 This returns a 32-bit hash of the value obtained by concatenating the first
19517 Host header and the whole URL including parameters (not only the path part of
19518 the request, as in the "base32" fetch above). This is useful to track per-URL
19519 activity. A shorter hash is stored, saving a lot of memory. The output type
19520 is an unsigned integer.
19521
19522url32+src : binary
19523 This returns the concatenation of the "url32" fetch and the "src" fetch. The
19524 resulting type is of type binary, with a size of 8 or 20 bytes depending on
19525 the source address family. This can be used to track per-IP, per-URL counters.
19526
Christopher Faulet16032ab2020-04-30 11:30:00 +020019527
Christopher Faulete596d182020-05-05 17:46:34 +0200195287.3.7. Fetching samples for developers
Christopher Fauletd47941d2020-01-08 14:40:19 +010019529---------------------------------------
19530
19531This set of sample fetch methods is reserved to developers and must never be
19532used on a production environment, except on developer demand, for debugging
19533purposes. Moreover, no special care will be taken on backwards compatibility.
19534There is no warranty the following sample fetches will never change, be renamed
19535or simply removed. So be really careful if you should use one of them. To avoid
19536any ambiguity, these sample fetches are placed in the dedicated scope "internal",
19537for instance "internal.strm.is_htx".
19538
19539internal.htx.data : integer
19540 Returns the size in bytes used by data in the HTX message associated to a
19541 channel. The channel is chosen depending on the sample direction.
19542
19543internal.htx.free : integer
19544 Returns the free space (size - used) in bytes in the HTX message associated
19545 to a channel. The channel is chosen depending on the sample direction.
19546
19547internal.htx.free_data : integer
19548 Returns the free space for the data in bytes in the HTX message associated to
19549 a channel. The channel is chosen depending on the sample direction.
19550
19551internal.htx.has_eom : boolean
Christopher Fauletd1ac2b92020-12-02 19:12:22 +010019552 Returns true if the HTX message associated to a channel contains the
19553 end-of-message flag (EOM). Otherwise, it returns false. The channel is chosen
19554 depending on the sample direction.
Christopher Fauletd47941d2020-01-08 14:40:19 +010019555
19556internal.htx.nbblks : integer
19557 Returns the number of blocks present in the HTX message associated to a
19558 channel. The channel is chosen depending on the sample direction.
19559
19560internal.htx.size : integer
19561 Returns the total size in bytes of the HTX message associated to a
19562 channel. The channel is chosen depending on the sample direction.
19563
19564internal.htx.used : integer
19565 Returns the total size used in bytes (data + metadata) in the HTX message
19566 associated to a channel. The channel is chosen depending on the sample
19567 direction.
19568
19569internal.htx_blk.size(<idx>) : integer
19570 Returns the size of the block at the position <idx> in the HTX message
19571 associated to a channel or 0 if it does not exist. The channel is chosen
19572 depending on the sample direction. <idx> may be any positive integer or one
19573 of the special value :
19574 * head : The oldest inserted block
19575 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019576 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019577
19578internal.htx_blk.type(<idx>) : string
19579 Returns the type of the block at the position <idx> in the HTX message
19580 associated to a channel or "HTX_BLK_UNUSED" if it does not exist. The channel
19581 is chosen depending on the sample direction. <idx> may be any positive
19582 integer or one of the special value :
19583 * head : The oldest inserted block
19584 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019585 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019586
19587internal.htx_blk.data(<idx>) : binary
19588 Returns the value of the DATA block at the position <idx> in the HTX message
19589 associated to a channel or an empty string if it does not exist or if it is
19590 not a DATA block. The channel is chosen depending on the sample direction.
19591 <idx> may be any positive integer or one of the special value :
19592
19593 * head : The oldest inserted block
19594 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019595 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019596
19597internal.htx_blk.hdrname(<idx>) : string
19598 Returns the header name of the HEADER block at the position <idx> in the HTX
19599 message associated to a channel or an empty string if it does not exist or if
19600 it is not an HEADER block. The channel is chosen depending on the sample
19601 direction. <idx> may be any positive integer or one of the special value :
19602
19603 * head : The oldest inserted block
19604 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019605 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019606
19607internal.htx_blk.hdrval(<idx>) : string
19608 Returns the header value of the HEADER block at the position <idx> in the HTX
19609 message associated to a channel or an empty string if it does not exist or if
19610 it is not an HEADER block. The channel is chosen depending on the sample
19611 direction. <idx> may be any positive integer or one of the special value :
19612
19613 * head : The oldest inserted block
19614 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019615 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019616
19617internal.htx_blk.start_line(<idx>) : string
19618 Returns the value of the REQ_SL or RES_SL block at the position <idx> in the
19619 HTX message associated to a channel or an empty string if it does not exist
19620 or if it is not a SL block. The channel is chosen depending on the sample
19621 direction. <idx> may be any positive integer or one of the special value :
19622
19623 * head : The oldest inserted block
19624 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019625 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019626
19627internal.strm.is_htx : boolean
19628 Returns true if the current stream is an HTX stream. It means the data in the
19629 channels buffers are stored using the internal HTX representation. Otherwise,
19630 it returns false.
19631
19632
Willy Tarreau74ca5042013-06-11 23:12:07 +0200196337.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019634---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010019635
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019636Some predefined ACLs are hard-coded so that they do not have to be declared in
19637every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020019638order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010019639
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019640ACL name Equivalent to Usage
Christopher Faulet779184e2021-04-01 17:24:04 +020019641---------------+----------------------------------+------------------------------------------------------
19642FALSE always_false never match
19643HTTP req.proto_http match if request protocol is valid HTTP
19644HTTP_1.0 req.ver 1.0 match if HTTP request version is 1.0
19645HTTP_1.1 req.ver 1.1 match if HTTP request version is 1.1
Christopher Faulet8043e832021-03-26 16:00:54 +010019646HTTP_2.0 req.ver 2.0 match if HTTP request version is 2.0
Christopher Faulet779184e2021-04-01 17:24:04 +020019647HTTP_CONTENT req.hdr_val(content-length) gt 0 match an existing content-length in the HTTP request
19648HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
19649HTTP_URL_SLASH url_beg / match URL beginning with "/"
19650HTTP_URL_STAR url * match URL equal to "*"
19651LOCALHOST src 127.0.0.1/8 match connection from local host
19652METH_CONNECT method CONNECT match HTTP CONNECT method
19653METH_DELETE method DELETE match HTTP DELETE method
19654METH_GET method GET HEAD match HTTP GET or HEAD method
19655METH_HEAD method HEAD match HTTP HEAD method
19656METH_OPTIONS method OPTIONS match HTTP OPTIONS method
19657METH_POST method POST match HTTP POST method
19658METH_PUT method PUT match HTTP PUT method
19659METH_TRACE method TRACE match HTTP TRACE method
19660RDP_COOKIE req.rdp_cookie_cnt gt 0 match presence of an RDP cookie in the request buffer
19661REQ_CONTENT req.len gt 0 match data in the request buffer
19662TRUE always_true always match
19663WAIT_END wait_end wait for end of content analysis
19664---------------+----------------------------------+------------------------------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010019665
Willy Tarreaub937b7e2010-01-12 15:27:54 +010019666
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200196678. Logging
19668----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010019669
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019670One of HAProxy's strong points certainly lies is its precise logs. It probably
19671provides the finest level of information available for such a product, which is
19672very important for troubleshooting complex environments. Standard information
19673provided in logs include client ports, TCP/HTTP state timers, precise session
19674state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010019675to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019676headers.
19677
19678In order to improve administrators reactivity, it offers a great transparency
19679about encountered problems, both internal and external, and it is possible to
19680send logs to different sources at the same time with different level filters :
19681
19682 - global process-level logs (system errors, start/stop, etc..)
19683 - per-instance system and internal errors (lack of resource, bugs, ...)
19684 - per-instance external troubles (servers up/down, max connections)
19685 - per-instance activity (client connections), either at the establishment or
19686 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010019687 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060019688 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019689
19690The ability to distribute different levels of logs to different log servers
19691allow several production teams to interact and to fix their problems as soon
19692as possible. For example, the system team might monitor system-wide errors,
19693while the application team might be monitoring the up/down for their servers in
19694real time, and the security team might analyze the activity logs with one hour
19695delay.
19696
19697
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200196988.1. Log levels
19699---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019700
Simon Hormandf791f52011-05-29 15:01:10 +090019701TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019702source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090019703HTTP request, HTTP return code, number of bytes transmitted, conditions
19704in which the session ended, and even exchanged cookies values. For example
19705track a particular user's problems. All messages may be sent to up to two
19706syslog servers. Check the "log" keyword in section 4.2 for more information
19707about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019708
19709
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200197108.2. Log formats
19711----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019712
William Lallemand48940402012-01-30 16:47:22 +010019713HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090019714and will be detailed in the following sections. A few of them may vary
19715slightly with the configuration, due to indicators specific to certain
19716options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019717
19718 - the default format, which is very basic and very rarely used. It only
19719 provides very basic information about the incoming connection at the moment
19720 it is accepted : source IP:port, destination IP:port, and frontend-name.
19721 This mode will eventually disappear so it will not be described to great
19722 extents.
19723
19724 - the TCP format, which is more advanced. This format is enabled when "option
19725 tcplog" is set on the frontend. HAProxy will then usually wait for the
19726 connection to terminate before logging. This format provides much richer
19727 information, such as timers, connection counts, queue size, etc... This
19728 format is recommended for pure TCP proxies.
19729
19730 - the HTTP format, which is the most advanced for HTTP proxying. This format
19731 is enabled when "option httplog" is set on the frontend. It provides the
19732 same information as the TCP format with some HTTP-specific fields such as
19733 the request, the status code, and captures of headers and cookies. This
19734 format is recommended for HTTP proxies.
19735
Emeric Brun3a058f32009-06-30 18:26:00 +020019736 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
19737 fields arranged in the same order as the CLF format. In this mode, all
19738 timers, captures, flags, etc... appear one per field after the end of the
19739 common fields, in the same order they appear in the standard HTTP format.
19740
William Lallemand48940402012-01-30 16:47:22 +010019741 - the custom log format, allows you to make your own log line.
19742
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019743Next sections will go deeper into details for each of these formats. Format
19744specification will be performed on a "field" basis. Unless stated otherwise, a
19745field is a portion of text delimited by any number of spaces. Since syslog
19746servers are susceptible of inserting fields at the beginning of a line, it is
19747always assumed that the first field is the one containing the process name and
19748identifier.
19749
19750Note : Since log lines may be quite long, the log examples in sections below
19751 might be broken into multiple lines. The example log lines will be
19752 prefixed with 3 closing angle brackets ('>>>') and each time a log is
19753 broken into multiple lines, each non-final line will end with a
19754 backslash ('\') and the next line will start indented by two characters.
19755
19756
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200197578.2.1. Default log format
19758-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019759
19760This format is used when no specific option is set. The log is emitted as soon
19761as the connection is accepted. One should note that this currently is the only
19762format which logs the request's destination IP and ports.
19763
19764 Example :
19765 listen www
19766 mode http
19767 log global
19768 server srv1 127.0.0.1:8000
19769
19770 >>> Feb 6 12:12:09 localhost \
19771 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
19772 (www/HTTP)
19773
19774 Field Format Extract from the example above
19775 1 process_name '[' pid ']:' haproxy[14385]:
19776 2 'Connect from' Connect from
19777 3 source_ip ':' source_port 10.0.1.2:33312
19778 4 'to' to
19779 5 destination_ip ':' destination_port 10.0.3.31:8012
19780 6 '(' frontend_name '/' mode ')' (www/HTTP)
19781
19782Detailed fields description :
19783 - "source_ip" is the IP address of the client which initiated the connection.
19784 - "source_port" is the TCP port of the client which initiated the connection.
19785 - "destination_ip" is the IP address the client connected to.
19786 - "destination_port" is the TCP port the client connected to.
19787 - "frontend_name" is the name of the frontend (or listener) which received
19788 and processed the connection.
19789 - "mode is the mode the frontend is operating (TCP or HTTP).
19790
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019791In case of a UNIX socket, the source and destination addresses are marked as
19792"unix:" and the ports reflect the internal ID of the socket which accepted the
19793connection (the same ID as reported in the stats).
19794
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019795It is advised not to use this deprecated format for newer installations as it
19796will eventually disappear.
19797
19798
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200197998.2.2. TCP log format
19800---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019801
19802The TCP format is used when "option tcplog" is specified in the frontend, and
19803is the recommended format for pure TCP proxies. It provides a lot of precious
19804information for troubleshooting. Since this format includes timers and byte
19805counts, the log is normally emitted at the end of the session. It can be
19806emitted earlier if "option logasap" is specified, which makes sense in most
19807environments with long sessions such as remote terminals. Sessions which match
19808the "monitor" rules are never logged. It is also possible not to emit logs for
19809sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019810specifying "option dontlognull" in the frontend. Successful connections will
19811not be logged if "option dontlog-normal" is specified in the frontend. A few
19812fields may slightly vary depending on some configuration options, those are
19813marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019814
19815 Example :
19816 frontend fnt
19817 mode tcp
19818 option tcplog
19819 log global
19820 default_backend bck
19821
19822 backend bck
19823 server srv1 127.0.0.1:8000
19824
19825 >>> Feb 6 12:12:56 localhost \
19826 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
19827 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
19828
19829 Field Format Extract from the example above
19830 1 process_name '[' pid ']:' haproxy[14387]:
19831 2 client_ip ':' client_port 10.0.1.2:33313
19832 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
19833 4 frontend_name fnt
19834 5 backend_name '/' server_name bck/srv1
19835 6 Tw '/' Tc '/' Tt* 0/0/5007
19836 7 bytes_read* 212
19837 8 termination_state --
19838 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
19839 10 srv_queue '/' backend_queue 0/0
19840
19841Detailed fields description :
19842 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019843 connection to haproxy. If the connection was accepted on a UNIX socket
19844 instead, the IP address would be replaced with the word "unix". Note that
19845 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019846 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010019847 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019848 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019849
19850 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019851 If the connection was accepted on a UNIX socket instead, the port would be
19852 replaced with the ID of the accepting socket, which is also reported in the
19853 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019854
19855 - "accept_date" is the exact date when the connection was received by haproxy
19856 (which might be very slightly different from the date observed on the
19857 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020019858 the same date which may appear in any upstream firewall's log. When used in
19859 HTTP mode, the accept_date field will be reset to the first moment the
19860 connection is ready to receive a new request (end of previous response for
19861 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019862
19863 - "frontend_name" is the name of the frontend (or listener) which received
19864 and processed the connection.
19865
19866 - "backend_name" is the name of the backend (or listener) which was selected
19867 to manage the connection to the server. This will be the same as the
19868 frontend if no switching rule has been applied, which is common for TCP
19869 applications.
19870
19871 - "server_name" is the name of the last server to which the connection was
19872 sent, which might differ from the first one if there were connection errors
19873 and a redispatch occurred. Note that this server belongs to the backend
19874 which processed the request. If the connection was aborted before reaching
19875 a server, "<NOSRV>" is indicated instead of a server name.
19876
19877 - "Tw" is the total time in milliseconds spent waiting in the various queues.
19878 It can be "-1" if the connection was aborted before reaching the queue.
19879 See "Timers" below for more details.
19880
19881 - "Tc" is the total time in milliseconds spent waiting for the connection to
19882 establish to the final server, including retries. It can be "-1" if the
19883 connection was aborted before a connection could be established. See
19884 "Timers" below for more details.
19885
19886 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019887 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019888 "option logasap" was specified, then the time counting stops at the moment
19889 the log is emitted. In this case, a '+' sign is prepended before the value,
19890 indicating that the final one will be larger. See "Timers" below for more
19891 details.
19892
19893 - "bytes_read" is the total number of bytes transmitted from the server to
19894 the client when the log is emitted. If "option logasap" is specified, the
19895 this value will be prefixed with a '+' sign indicating that the final one
19896 may be larger. Please note that this value is a 64-bit counter, so log
19897 analysis tools must be able to handle it without overflowing.
19898
19899 - "termination_state" is the condition the session was in when the session
19900 ended. This indicates the session state, which side caused the end of
19901 session to happen, and for what reason (timeout, error, ...). The normal
19902 flags should be "--", indicating the session was closed by either end with
19903 no data remaining in buffers. See below "Session state at disconnection"
19904 for more details.
19905
19906 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019907 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019908 limits have been reached. For instance, if actconn is close to 512 when
19909 multiple connection errors occur, chances are high that the system limits
19910 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019911 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019912
19913 - "feconn" is the total number of concurrent connections on the frontend when
19914 the session was logged. It is useful to estimate the amount of resource
19915 required to sustain high loads, and to detect when the frontend's "maxconn"
19916 has been reached. Most often when this value increases by huge jumps, it is
19917 because there is congestion on the backend servers, but sometimes it can be
19918 caused by a denial of service attack.
19919
19920 - "beconn" is the total number of concurrent connections handled by the
19921 backend when the session was logged. It includes the total number of
19922 concurrent connections active on servers as well as the number of
19923 connections pending in queues. It is useful to estimate the amount of
19924 additional servers needed to support high loads for a given application.
19925 Most often when this value increases by huge jumps, it is because there is
19926 congestion on the backend servers, but sometimes it can be caused by a
19927 denial of service attack.
19928
19929 - "srv_conn" is the total number of concurrent connections still active on
19930 the server when the session was logged. It can never exceed the server's
19931 configured "maxconn" parameter. If this value is very often close or equal
19932 to the server's "maxconn", it means that traffic regulation is involved a
19933 lot, meaning that either the server's maxconn value is too low, or that
19934 there aren't enough servers to process the load with an optimal response
19935 time. When only one of the server's "srv_conn" is high, it usually means
19936 that this server has some trouble causing the connections to take longer to
19937 be processed than on other servers.
19938
19939 - "retries" is the number of connection retries experienced by this session
19940 when trying to connect to the server. It must normally be zero, unless a
19941 server is being stopped at the same moment the connection was attempted.
19942 Frequent retries generally indicate either a network problem between
19943 haproxy and the server, or a misconfigured system backlog on the server
19944 preventing new connections from being queued. This field may optionally be
19945 prefixed with a '+' sign, indicating that the session has experienced a
19946 redispatch after the maximal retry count has been reached on the initial
19947 server. In this case, the server name appearing in the log is the one the
19948 connection was redispatched to, and not the first one, though both may
19949 sometimes be the same in case of hashing for instance. So as a general rule
19950 of thumb, when a '+' is present in front of the retry count, this count
19951 should not be attributed to the logged server.
19952
19953 - "srv_queue" is the total number of requests which were processed before
19954 this one in the server queue. It is zero when the request has not gone
19955 through the server queue. It makes it possible to estimate the approximate
19956 server's response time by dividing the time spent in queue by the number of
19957 requests in the queue. It is worth noting that if a session experiences a
19958 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010019959 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019960 backend queue unless a redispatch occurs.
19961
19962 - "backend_queue" is the total number of requests which were processed before
19963 this one in the backend's global queue. It is zero when the request has not
19964 gone through the global queue. It makes it possible to estimate the average
19965 queue length, which easily translates into a number of missing servers when
19966 divided by a server's "maxconn" parameter. It is worth noting that if a
19967 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010019968 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019969 through both the server queue and the backend queue unless a redispatch
19970 occurs.
19971
19972
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200199738.2.3. HTTP log format
19974----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019975
19976The HTTP format is the most complete and the best suited for HTTP proxies. It
19977is enabled by when "option httplog" is specified in the frontend. It provides
19978the same level of information as the TCP format with additional features which
19979are specific to the HTTP protocol. Just like the TCP format, the log is usually
19980emitted at the end of the session, unless "option logasap" is specified, which
19981generally only makes sense for download sites. A session which matches the
19982"monitor" rules will never logged. It is also possible not to log sessions for
19983which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019984frontend. Successful connections will not be logged if "option dontlog-normal"
19985is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019986
19987Most fields are shared with the TCP log, some being different. A few fields may
19988slightly vary depending on some configuration options. Those ones are marked
19989with a star ('*') after the field name below.
19990
19991 Example :
19992 frontend http-in
19993 mode http
19994 option httplog
19995 log global
19996 default_backend bck
19997
19998 backend static
19999 server srv1 127.0.0.1:8000
20000
20001 >>> Feb 6 12:14:14 localhost \
20002 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
20003 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 +010020004 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020005
20006 Field Format Extract from the example above
20007 1 process_name '[' pid ']:' haproxy[14389]:
20008 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020009 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020010 4 frontend_name http-in
20011 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020012 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020013 7 status_code 200
20014 8 bytes_read* 2750
20015 9 captured_request_cookie -
20016 10 captured_response_cookie -
20017 11 termination_state ----
20018 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
20019 13 srv_queue '/' backend_queue 0/0
20020 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
20021 15 '{' captured_response_headers* '}' {}
20022 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010020023
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020024Detailed fields description :
20025 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010020026 connection to haproxy. If the connection was accepted on a UNIX socket
20027 instead, the IP address would be replaced with the word "unix". Note that
20028 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010020029 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010020030 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010020031 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020032
20033 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010020034 If the connection was accepted on a UNIX socket instead, the port would be
20035 replaced with the ID of the accepting socket, which is also reported in the
20036 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020037
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020038 - "request_date" is the exact date when the first byte of the HTTP request
20039 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020040
20041 - "frontend_name" is the name of the frontend (or listener) which received
20042 and processed the connection.
20043
20044 - "backend_name" is the name of the backend (or listener) which was selected
20045 to manage the connection to the server. This will be the same as the
20046 frontend if no switching rule has been applied.
20047
20048 - "server_name" is the name of the last server to which the connection was
20049 sent, which might differ from the first one if there were connection errors
20050 and a redispatch occurred. Note that this server belongs to the backend
20051 which processed the request. If the request was aborted before reaching a
20052 server, "<NOSRV>" is indicated instead of a server name. If the request was
20053 intercepted by the stats subsystem, "<STATS>" is indicated instead.
20054
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020055 - "TR" is the total time in milliseconds spent waiting for a full HTTP
20056 request from the client (not counting body) after the first byte was
20057 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050020058 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020059 always be very small because a request generally fits in one single packet.
20060 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020020061 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
20062 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020063
20064 - "Tw" is the total time in milliseconds spent waiting in the various queues.
20065 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020020066 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020067
20068 - "Tc" is the total time in milliseconds spent waiting for the connection to
20069 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020020070 request was aborted before a connection could be established. See section
20071 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020072
20073 - "Tr" is the total time in milliseconds spent waiting for the server to send
20074 a full HTTP response, not counting data. It can be "-1" if the request was
20075 aborted before a complete response could be received. It generally matches
20076 the server's processing time for the request, though it may be altered by
20077 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020020078 "GET" requests generally indicate an overloaded server. See section 8.4
20079 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020080
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020081 - "Ta" is the time the request remained active in haproxy, which is the total
20082 time in milliseconds elapsed between the first byte of the request was
20083 received and the last byte of response was sent. It covers all possible
20084 processing except the handshake (see Th) and idle time (see Ti). There is
20085 one exception, if "option logasap" was specified, then the time counting
20086 stops at the moment the log is emitted. In this case, a '+' sign is
20087 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020020088 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020089
20090 - "status_code" is the HTTP status code returned to the client. This status
20091 is generally set by the server, but it might also be set by haproxy when
20092 the server cannot be reached or when its response is blocked by haproxy.
20093
20094 - "bytes_read" is the total number of bytes transmitted to the client when
20095 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050020096 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020097 the final one may be larger. Please note that this value is a 64-bit
20098 counter, so log analysis tools must be able to handle it without
20099 overflowing.
20100
20101 - "captured_request_cookie" is an optional "name=value" entry indicating that
20102 the client had this cookie in the request. The cookie name and its maximum
20103 length are defined by the "capture cookie" statement in the frontend
20104 configuration. The field is a single dash ('-') when the option is not
20105 set. Only one cookie may be captured, it is generally used to track session
20106 ID exchanges between a client and a server to detect session crossing
20107 between clients due to application bugs. For more details, please consult
20108 the section "Capturing HTTP headers and cookies" below.
20109
20110 - "captured_response_cookie" is an optional "name=value" entry indicating
20111 that the server has returned a cookie with its response. The cookie name
20112 and its maximum length are defined by the "capture cookie" statement in the
20113 frontend configuration. The field is a single dash ('-') when the option is
20114 not set. Only one cookie may be captured, it is generally used to track
20115 session ID exchanges between a client and a server to detect session
20116 crossing between clients due to application bugs. For more details, please
20117 consult the section "Capturing HTTP headers and cookies" below.
20118
20119 - "termination_state" is the condition the session was in when the session
20120 ended. This indicates the session state, which side caused the end of
20121 session to happen, for what reason (timeout, error, ...), just like in TCP
20122 logs, and information about persistence operations on cookies in the last
20123 two characters. The normal flags should begin with "--", indicating the
20124 session was closed by either end with no data remaining in buffers. See
20125 below "Session state at disconnection" for more details.
20126
20127 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020128 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020129 limits have been reached. For instance, if actconn is close to 512 or 1024
20130 when multiple connection errors occur, chances are high that the system
20131 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020132 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020133 system.
20134
20135 - "feconn" is the total number of concurrent connections on the frontend when
20136 the session was logged. It is useful to estimate the amount of resource
20137 required to sustain high loads, and to detect when the frontend's "maxconn"
20138 has been reached. Most often when this value increases by huge jumps, it is
20139 because there is congestion on the backend servers, but sometimes it can be
20140 caused by a denial of service attack.
20141
20142 - "beconn" is the total number of concurrent connections handled by the
20143 backend when the session was logged. It includes the total number of
20144 concurrent connections active on servers as well as the number of
20145 connections pending in queues. It is useful to estimate the amount of
20146 additional servers needed to support high loads for a given application.
20147 Most often when this value increases by huge jumps, it is because there is
20148 congestion on the backend servers, but sometimes it can be caused by a
20149 denial of service attack.
20150
20151 - "srv_conn" is the total number of concurrent connections still active on
20152 the server when the session was logged. It can never exceed the server's
20153 configured "maxconn" parameter. If this value is very often close or equal
20154 to the server's "maxconn", it means that traffic regulation is involved a
20155 lot, meaning that either the server's maxconn value is too low, or that
20156 there aren't enough servers to process the load with an optimal response
20157 time. When only one of the server's "srv_conn" is high, it usually means
20158 that this server has some trouble causing the requests to take longer to be
20159 processed than on other servers.
20160
20161 - "retries" is the number of connection retries experienced by this session
20162 when trying to connect to the server. It must normally be zero, unless a
20163 server is being stopped at the same moment the connection was attempted.
20164 Frequent retries generally indicate either a network problem between
20165 haproxy and the server, or a misconfigured system backlog on the server
20166 preventing new connections from being queued. This field may optionally be
20167 prefixed with a '+' sign, indicating that the session has experienced a
20168 redispatch after the maximal retry count has been reached on the initial
20169 server. In this case, the server name appearing in the log is the one the
20170 connection was redispatched to, and not the first one, though both may
20171 sometimes be the same in case of hashing for instance. So as a general rule
20172 of thumb, when a '+' is present in front of the retry count, this count
20173 should not be attributed to the logged server.
20174
20175 - "srv_queue" is the total number of requests which were processed before
20176 this one in the server queue. It is zero when the request has not gone
20177 through the server queue. It makes it possible to estimate the approximate
20178 server's response time by dividing the time spent in queue by the number of
20179 requests in the queue. It is worth noting that if a session experiences a
20180 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010020181 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020182 backend queue unless a redispatch occurs.
20183
20184 - "backend_queue" is the total number of requests which were processed before
20185 this one in the backend's global queue. It is zero when the request has not
20186 gone through the global queue. It makes it possible to estimate the average
20187 queue length, which easily translates into a number of missing servers when
20188 divided by a server's "maxconn" parameter. It is worth noting that if a
20189 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010020190 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020191 through both the server queue and the backend queue unless a redispatch
20192 occurs.
20193
20194 - "captured_request_headers" is a list of headers captured in the request due
20195 to the presence of the "capture request header" statement in the frontend.
20196 Multiple headers can be captured, they will be delimited by a vertical bar
20197 ('|'). When no capture is enabled, the braces do not appear, causing a
20198 shift of remaining fields. It is important to note that this field may
20199 contain spaces, and that using it requires a smarter log parser than when
20200 it's not used. Please consult the section "Capturing HTTP headers and
20201 cookies" below for more details.
20202
20203 - "captured_response_headers" is a list of headers captured in the response
20204 due to the presence of the "capture response header" statement in the
20205 frontend. Multiple headers can be captured, they will be delimited by a
20206 vertical bar ('|'). When no capture is enabled, the braces do not appear,
20207 causing a shift of remaining fields. It is important to note that this
20208 field may contain spaces, and that using it requires a smarter log parser
20209 than when it's not used. Please consult the section "Capturing HTTP headers
20210 and cookies" below for more details.
20211
20212 - "http_request" is the complete HTTP request line, including the method,
20213 request and HTTP version string. Non-printable characters are encoded (see
20214 below the section "Non-printable characters"). This is always the last
20215 field, and it is always delimited by quotes and is the only one which can
20216 contain quotes. If new fields are added to the log format, they will be
20217 added before this field. This field might be truncated if the request is
20218 huge and does not fit in the standard syslog buffer (1024 characters). This
20219 is the reason why this field must always remain the last one.
20220
20221
Cyril Bontédc4d9032012-04-08 21:57:39 +0200202228.2.4. Custom log format
20223------------------------
William Lallemand48940402012-01-30 16:47:22 +010020224
Willy Tarreau2beef582012-12-20 17:22:52 +010020225The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010020226mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010020227
Davor Ocelice9ed2812017-12-25 17:49:28 +010020228HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010020229Variables can take arguments using braces ('{}'), and multiple arguments are
20230separated by commas within the braces. Flags may be added or removed by
20231prefixing them with a '+' or '-' sign.
20232
20233Special variable "%o" may be used to propagate its flags to all other
20234variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010020235("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010020236
Willy Tarreauc8368452012-12-21 00:09:23 +010020237If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020020238as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010020239less common information such as the client's SSL certificate's DN, or to log
20240the key that would be used to store an entry into a stick table.
20241
Dragan Dosen1e3b16f2020-06-23 18:16:44 +020020242Note: spaces must be escaped. In configuration directives "log-format",
20243"log-format-sd" and "unique-id-format", spaces are considered as
20244delimiters and are merged. In order to emit a verbatim '%', it must be
20245preceded by another '%' resulting in '%%'.
William Lallemand48940402012-01-30 16:47:22 +010020246
Dragan Dosen835b9212016-02-12 13:23:03 +010020247Note: when using the RFC5424 syslog message format, the characters '"',
20248'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
20249https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
20250such cases, the use of the flag "E" should be considered.
20251
William Lallemand48940402012-01-30 16:47:22 +010020252Flags are :
20253 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020254 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010020255 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
20256 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010020257
20258 Example:
20259
20260 log-format %T\ %t\ Some\ Text
20261 log-format %{+Q}o\ %t\ %s\ %{-Q}r
20262
Dragan Dosen835b9212016-02-12 13:23:03 +010020263 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
20264
William Lallemand48940402012-01-30 16:47:22 +010020265At the moment, the default HTTP format is defined this way :
20266
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020267 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
20268 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010020269
William Lallemandbddd4fd2012-02-27 11:23:10 +010020270the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010020271
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020272 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
20273 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
20274 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010020275
William Lallemandbddd4fd2012-02-27 11:23:10 +010020276and the default TCP format is defined this way :
20277
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020278 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
20279 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010020280
William Lallemand48940402012-01-30 16:47:22 +010020281Please refer to the table below for currently defined variables :
20282
William Lallemandbddd4fd2012-02-27 11:23:10 +010020283 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020284 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020285 +---+------+-----------------------------------------------+-------------+
20286 | | %o | special variable, apply flags on all next var | |
20287 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010020288 | | %B | bytes_read (from server to client) | numeric |
20289 | H | %CC | captured_request_cookie | string |
20290 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020020291 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000020292 | H | %HM | HTTP method (ex: POST) | string |
Maciej Zdeb21acc332020-11-26 10:45:52 +000020293 | H | %HP | HTTP request URI without query string | string |
Maciej Zdebfcdfd852020-11-30 18:27:47 +000020294 | H | %HPO | HTTP path only (without host nor query string)| string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000020295 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000020296 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
20297 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010020298 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020020299 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020020300 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020301 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020302 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020020303 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080020304 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020305 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
20306 | H | %Ti | idle time before the HTTP request | numeric |
20307 | H | %Tq | Th + Ti + TR | numeric |
20308 | H | %TR | time to receive the full request from 1st byte| numeric |
20309 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020020310 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020311 | | %Tt | Tt | numeric |
Damien Claisse57c8eb92020-04-28 12:09:19 +000020312 | | %Tu | Tu | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020313 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010020314 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020315 | | %ac | actconn | numeric |
20316 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020317 | | %bc | beconn (backend concurrent connections) | numeric |
20318 | | %bi | backend_source_ip (connecting address) | IP |
20319 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020320 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010020321 | | %ci | client_ip (accepted address) | IP |
20322 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020323 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020324 | | %fc | feconn (frontend concurrent connections) | numeric |
20325 | | %fi | frontend_ip (accepting address) | IP |
20326 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020020327 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020020328 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020020329 | | %hr | captured_request_headers default style | string |
20330 | | %hrl | captured_request_headers CLF style | string list |
20331 | | %hs | captured_response_headers default style | string |
20332 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020020333 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020020334 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020335 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020336 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010020337 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020338 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020339 | | %sc | srv_conn (server concurrent connections) | numeric |
20340 | | %si | server_IP (target address) | IP |
20341 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020342 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020343 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
20344 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020345 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020346 | H | %tr | date_time of HTTP request | date |
20347 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020020348 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020349 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020350 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020351 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010020352
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020353 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010020354
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010020355
203568.2.5. Error log format
20357-----------------------
20358
20359When an incoming connection fails due to an SSL handshake or an invalid PROXY
20360protocol header, haproxy will log the event using a shorter, fixed line format.
20361By default, logs are emitted at the LOG_INFO level, unless the option
20362"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010020363will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010020364logged if the "dontlognull" option is set.
20365
20366The format looks like this :
20367
20368 >>> Dec 3 18:27:14 localhost \
20369 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
20370 Connection error during SSL handshake
20371
20372 Field Format Extract from the example above
20373 1 process_name '[' pid ']:' haproxy[6103]:
20374 2 client_ip ':' client_port 127.0.0.1:56059
20375 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
20376 4 frontend_name "/" bind_name ":" frt/f1:
20377 5 message Connection error during SSL handshake
20378
20379These fields just provide minimal information to help debugging connection
20380failures.
20381
20382
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200203838.3. Advanced logging options
20384-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020385
20386Some advanced logging options are often looked for but are not easy to find out
20387just by looking at the various options. Here is an entry point for the few
20388options which can enable better logging. Please refer to the keywords reference
20389for more information about their usage.
20390
20391
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200203928.3.1. Disabling logging of external tests
20393------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020394
20395It is quite common to have some monitoring tools perform health checks on
20396haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
20397commercial load-balancer, and sometimes it will simply be a more complete
20398monitoring system such as Nagios. When the tests are very frequent, users often
20399ask how to disable logging for those checks. There are three possibilities :
20400
20401 - if connections come from everywhere and are just TCP probes, it is often
20402 desired to simply disable logging of connections without data exchange, by
20403 setting "option dontlognull" in the frontend. It also disables logging of
20404 port scans, which may or may not be desired.
20405
Willy Tarreau9e9919d2020-10-14 15:55:23 +020020406 - it is possible to use the "http-request set-log-level silent" action using
20407 a variety of conditions (source networks, paths, user-agents, etc).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020408
20409 - if the tests are performed on a known URI, use "monitor-uri" to declare
20410 this URI as dedicated to monitoring. Any host sending this request will
20411 only get the result of a health-check, and the request will not be logged.
20412
20413
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200204148.3.2. Logging before waiting for the session to terminate
20415----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020416
20417The problem with logging at end of connection is that you have no clue about
20418what is happening during very long sessions, such as remote terminal sessions
20419or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010020420"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020421just before data transfer begins. This means that in case of TCP, it will still
20422log the connection status to the server, and in case of HTTP, it will log just
20423after processing the server headers. In this case, the number of bytes reported
20424is the number of header bytes sent to the client. In order to avoid confusion
20425with normal logs, the total time field and the number of bytes are prefixed
20426with a '+' sign which means that real numbers are certainly larger.
20427
20428
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200204298.3.3. Raising log level upon errors
20430------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020020431
20432Sometimes it is more convenient to separate normal traffic from errors logs,
20433for instance in order to ease error monitoring from log files. When the option
20434"log-separate-errors" is used, connections which experience errors, timeouts,
20435retries, redispatches or HTTP status codes 5xx will see their syslog level
20436raised from "info" to "err". This will help a syslog daemon store the log in
20437a separate file. It is very important to keep the errors in the normal traffic
20438file too, so that log ordering is not altered. You should also be careful if
20439you already have configured your syslog daemon to store all logs higher than
20440"notice" in an "admin" file, because the "err" level is higher than "notice".
20441
20442
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200204438.3.4. Disabling logging of successful connections
20444--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020020445
20446Although this may sound strange at first, some large sites have to deal with
20447multiple thousands of logs per second and are experiencing difficulties keeping
20448them intact for a long time or detecting errors within them. If the option
20449"dontlog-normal" is set on the frontend, all normal connections will not be
20450logged. In this regard, a normal connection is defined as one without any
20451error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
20452and a response with a status 5xx is not considered normal and will be logged
20453too. Of course, doing is is really discouraged as it will remove most of the
20454useful information from the logs. Do this only if you have no other
20455alternative.
20456
20457
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200204588.4. Timing events
20459------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020460
20461Timers provide a great help in troubleshooting network problems. All values are
20462reported in milliseconds (ms). These timers should be used in conjunction with
20463the session termination flags. In TCP mode with "option tcplog" set on the
20464frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020465mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
20466addition, three other measures are provided, "Th", "Ti", and "Tq".
20467
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010020468Timings events in HTTP mode:
20469
20470 first request 2nd request
20471 |<-------------------------------->|<-------------- ...
20472 t tr t tr ...
20473 ---|----|----|----|----|----|----|----|----|--
20474 : Th Ti TR Tw Tc Tr Td : Ti ...
20475 :<---- Tq ---->: :
20476 :<-------------- Tt -------------->:
Damien Claisse57c8eb92020-04-28 12:09:19 +000020477 :<-- -----Tu--------------->:
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010020478 :<--------- Ta --------->:
20479
20480Timings events in TCP mode:
20481
20482 TCP session
20483 |<----------------->|
20484 t t
20485 ---|----|----|----|----|---
20486 | Th Tw Tc Td |
20487 |<------ Tt ------->|
20488
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020489 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010020490 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020491 only happen once during the whole connection's lifetime. A large time here
20492 may indicate that the client only pre-established the connection without
20493 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010020494 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020020495 SSL handshake was very expensive to compute. Please note that this time is
20496 reported only before the first request, so it is safe to average it over
20497 all request to calculate the amortized value. The second and subsequent
20498 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020499
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020500 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
20501 counts between the end of the handshakes and the first byte of the HTTP
20502 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020020503 to count after the end of the transmission the previous response. When a
20504 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
20505 after the previous request. Some browsers pre-establish connections to a
20506 server in order to reduce the latency of a future request, and keep them
20507 pending until they need it. This delay will be reported as the idle time. A
20508 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020509
20510 - TR: total time to get the client request (HTTP mode only). It's the time
20511 elapsed between the first bytes received and the moment the proxy received
20512 the empty line marking the end of the HTTP headers. The value "-1"
20513 indicates that the end of headers has never been seen. This happens when
20514 the client closes prematurely or times out. This time is usually very short
20515 since most requests fit in a single packet. A large time may indicate a
20516 request typed by hand during a test.
20517
20518 - Tq: total time to get the client request from the accept date or since the
20519 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010020520 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 +020020521 returns -1 as well. This timer used to be very useful before the arrival of
20522 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
20523 it in favor of TR nowadays, as the idle time adds a lot of noise to the
20524 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020525
20526 - Tw: total time spent in the queues waiting for a connection slot. It
20527 accounts for backend queue as well as the server queues, and depends on the
20528 queue size, and the time needed for the server to complete previous
20529 requests. The value "-1" means that the request was killed before reaching
20530 the queue, which is generally what happens with invalid or denied requests.
20531
20532 - Tc: total time to establish the TCP connection to the server. It's the time
20533 elapsed between the moment the proxy sent the connection request, and the
20534 moment it was acknowledged by the server, or between the TCP SYN packet and
20535 the matching SYN/ACK packet in return. The value "-1" means that the
20536 connection never established.
20537
20538 - Tr: server response time (HTTP mode only). It's the time elapsed between
20539 the moment the TCP connection was established to the server and the moment
20540 the server sent its complete response headers. It purely shows its request
20541 processing time, without the network overhead due to the data transmission.
20542 It is worth noting that when the client has data to send to the server, for
20543 instance during a POST request, the time already runs, and this can distort
20544 apparent response time. For this reason, it's generally wise not to trust
20545 too much this field for POST requests initiated from clients behind an
20546 untrusted network. A value of "-1" here means that the last the response
20547 header (empty line) was never seen, most likely because the server timeout
20548 stroke before the server managed to process the request.
20549
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020550 - Ta: total active time for the HTTP request, between the moment the proxy
20551 received the first byte of the request header and the emission of the last
20552 byte of the response body. The exception is when the "logasap" option is
20553 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
20554 a '+' sign. From this field, we can deduce "Td", the data transmission time,
20555 by subtracting other timers when valid :
20556
20557 Td = Ta - (TR + Tw + Tc + Tr)
20558
20559 Timers with "-1" values have to be excluded from this equation. Note that
20560 "Ta" can never be negative.
20561
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020562 - Tt: total session duration time, between the moment the proxy accepted it
20563 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020564 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
20565 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020566 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020567
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020568 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020569
20570 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020571 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
20572 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020573
Damien Claisse57c8eb92020-04-28 12:09:19 +000020574 - Tu: total estimated time as seen from client, between the moment the proxy
20575 accepted it and the moment both ends were closed, without idle time.
20576 This is useful to roughly measure end-to-end time as a user would see it,
20577 without idle time pollution from keep-alive time between requests. This
20578 timer in only an estimation of time seen by user as it assumes network
20579 latency is the same in both directions. The exception is when the "logasap"
20580 option is specified. In this case, it only equals (Th+TR+Tw+Tc+Tr), and is
20581 prefixed with a '+' sign.
20582
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020583These timers provide precious indications on trouble causes. Since the TCP
20584protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
20585that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020586due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
20587"Tt" is close to a timeout value specified in the configuration, it often means
20588that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020589
20590Most common cases :
20591
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020592 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
20593 the client and the proxy. This is very rare on local networks but might
20594 happen when clients are on far remote networks and send large requests. It
20595 may happen that values larger than usual appear here without any network
20596 cause. Sometimes, during an attack or just after a resource starvation has
20597 ended, haproxy may accept thousands of connections in a few milliseconds.
20598 The time spent accepting these connections will inevitably slightly delay
20599 processing of other connections, and it can happen that request times in the
20600 order of a few tens of milliseconds are measured after a few thousands of
20601 new connections have been accepted at once. Using one of the keep-alive
20602 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020020603 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020604
20605 - If "Tc" is close to 3000, a packet has probably been lost between the
20606 server and the proxy during the server connection phase. This value should
20607 always be very low, such as 1 ms on local networks and less than a few tens
20608 of ms on remote networks.
20609
Willy Tarreau55165fe2009-05-10 12:02:55 +020020610 - If "Tr" is nearly always lower than 3000 except some rare values which seem
20611 to be the average majored by 3000, there are probably some packets lost
20612 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020613
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020614 - If "Ta" is large even for small byte counts, it generally is because
20615 neither the client nor the server decides to close the connection while
20616 haproxy is running in tunnel mode and both have agreed on a keep-alive
20617 connection mode. In order to solve this issue, it will be needed to specify
20618 one of the HTTP options to manipulate keep-alive or close options on either
20619 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
20620 important when connection regulation is used with the "maxconn" option on
20621 the servers, since no new connection will be sent to the server until
20622 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020623
20624Other noticeable HTTP log cases ('xx' means any value to be ignored) :
20625
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020626 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020627 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020628 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020629
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020630 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020631 or it aborted too early. Check the session termination flags
20632 then "timeout http-request" and "timeout client" settings.
20633
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020634 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020635 servers were out of order, because the request was invalid
20636 or forbidden by ACL rules. Check the session termination
20637 flags.
20638
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020639 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
20640 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020641 Check the session termination flags, then check the
20642 "timeout connect" setting. Note that the tarpit action might
20643 return similar-looking patterns, with "Tw" equal to the time
20644 the client connection was maintained open.
20645
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020646 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020647 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020648 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020649 termination flags, then check the "timeout server" setting.
20650
20651
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200206528.5. Session state at disconnection
20653-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020654
20655TCP and HTTP logs provide a session termination indicator in the
20656"termination_state" field, just before the number of active connections. It is
206572-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
20658each of which has a special meaning :
20659
20660 - On the first character, a code reporting the first event which caused the
20661 session to terminate :
20662
20663 C : the TCP session was unexpectedly aborted by the client.
20664
20665 S : the TCP session was unexpectedly aborted by the server, or the
20666 server explicitly refused it.
20667
20668 P : the session was prematurely aborted by the proxy, because of a
20669 connection limit enforcement, because a DENY filter was matched,
20670 because of a security check which detected and blocked a dangerous
20671 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010020672 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020020673
20674 L : the session was locally processed by haproxy and was not passed to
20675 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020676
20677 R : a resource on the proxy has been exhausted (memory, sockets, source
20678 ports, ...). Usually, this appears during the connection phase, and
20679 system logs should contain a copy of the precise error. If this
20680 happens, it must be considered as a very serious anomaly which
20681 should be fixed as soon as possible by any means.
20682
20683 I : an internal error was identified by the proxy during a self-check.
20684 This should NEVER happen, and you are encouraged to report any log
20685 containing this, because this would almost certainly be a bug. It
20686 would be wise to preventively restart the process after such an
20687 event too, in case it would be caused by memory corruption.
20688
Simon Horman752dc4a2011-06-21 14:34:59 +090020689 D : the session was killed by haproxy because the server was detected
20690 as down and was configured to kill all connections when going down.
20691
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070020692 U : the session was killed by haproxy on this backup server because an
20693 active server was detected as up and was configured to kill all
20694 backup connections when going up.
20695
Willy Tarreaua2a64e92011-09-07 23:01:56 +020020696 K : the session was actively killed by an admin operating on haproxy.
20697
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020698 c : the client-side timeout expired while waiting for the client to
20699 send or receive data.
20700
20701 s : the server-side timeout expired while waiting for the server to
20702 send or receive data.
20703
20704 - : normal session completion, both the client and the server closed
20705 with nothing left in the buffers.
20706
20707 - on the second character, the TCP or HTTP session state when it was closed :
20708
Willy Tarreauf7b30a92010-12-06 22:59:17 +010020709 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020710 (HTTP mode only). Nothing was sent to any server.
20711
20712 Q : the proxy was waiting in the QUEUE for a connection slot. This can
20713 only happen when servers have a 'maxconn' parameter set. It can
20714 also happen in the global queue after a redispatch consecutive to
20715 a failed attempt to connect to a dying server. If no redispatch is
20716 reported, then no connection attempt was made to any server.
20717
20718 C : the proxy was waiting for the CONNECTION to establish on the
20719 server. The server might at most have noticed a connection attempt.
20720
20721 H : the proxy was waiting for complete, valid response HEADERS from the
20722 server (HTTP only).
20723
20724 D : the session was in the DATA phase.
20725
20726 L : the proxy was still transmitting LAST data to the client while the
20727 server had already finished. This one is very rare as it can only
20728 happen when the client dies while receiving the last packets.
20729
20730 T : the request was tarpitted. It has been held open with the client
20731 during the whole "timeout tarpit" duration or until the client
20732 closed, both of which will be reported in the "Tw" timer.
20733
20734 - : normal session completion after end of data transfer.
20735
20736 - the third character tells whether the persistence cookie was provided by
20737 the client (only in HTTP mode) :
20738
20739 N : the client provided NO cookie. This is usually the case for new
20740 visitors, so counting the number of occurrences of this flag in the
20741 logs generally indicate a valid trend for the site frequentation.
20742
20743 I : the client provided an INVALID cookie matching no known server.
20744 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020020745 cookies between HTTP/HTTPS sites, persistence conditionally
20746 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020747
20748 D : the client provided a cookie designating a server which was DOWN,
20749 so either "option persist" was used and the client was sent to
20750 this server, or it was not set and the client was redispatched to
20751 another server.
20752
Willy Tarreau996a92c2010-10-13 19:30:47 +020020753 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020754 server.
20755
Willy Tarreau996a92c2010-10-13 19:30:47 +020020756 E : the client provided a valid cookie, but with a last date which was
20757 older than what is allowed by the "maxidle" cookie parameter, so
20758 the cookie is consider EXPIRED and is ignored. The request will be
20759 redispatched just as if there was no cookie.
20760
20761 O : the client provided a valid cookie, but with a first date which was
20762 older than what is allowed by the "maxlife" cookie parameter, so
20763 the cookie is consider too OLD and is ignored. The request will be
20764 redispatched just as if there was no cookie.
20765
Willy Tarreauc89ccb62012-04-05 21:18:22 +020020766 U : a cookie was present but was not used to select the server because
20767 some other server selection mechanism was used instead (typically a
20768 "use-server" rule).
20769
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020770 - : does not apply (no cookie set in configuration).
20771
20772 - the last character reports what operations were performed on the persistence
20773 cookie returned by the server (only in HTTP mode) :
20774
20775 N : NO cookie was provided by the server, and none was inserted either.
20776
20777 I : no cookie was provided by the server, and the proxy INSERTED one.
20778 Note that in "cookie insert" mode, if the server provides a cookie,
20779 it will still be overwritten and reported as "I" here.
20780
Willy Tarreau996a92c2010-10-13 19:30:47 +020020781 U : the proxy UPDATED the last date in the cookie that was presented by
20782 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020783 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020020784 date indicated in the cookie. If any other change happens, such as
20785 a redispatch, then the cookie will be marked as inserted instead.
20786
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020787 P : a cookie was PROVIDED by the server and transmitted as-is.
20788
20789 R : the cookie provided by the server was REWRITTEN by the proxy, which
20790 happens in "cookie rewrite" or "cookie prefix" modes.
20791
20792 D : the cookie provided by the server was DELETED by the proxy.
20793
20794 - : does not apply (no cookie set in configuration).
20795
Willy Tarreau996a92c2010-10-13 19:30:47 +020020796The combination of the two first flags gives a lot of information about what
20797was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020798helpful to detect server saturation, network troubles, local system resource
20799starvation, attacks, etc...
20800
20801The most common termination flags combinations are indicated below. They are
20802alphabetically sorted, with the lowercase set just after the upper case for
20803easier finding and understanding.
20804
20805 Flags Reason
20806
20807 -- Normal termination.
20808
20809 CC The client aborted before the connection could be established to the
20810 server. This can happen when haproxy tries to connect to a recently
20811 dead (or unchecked) server, and the client aborts while haproxy is
20812 waiting for the server to respond or for "timeout connect" to expire.
20813
20814 CD The client unexpectedly aborted during data transfer. This can be
20815 caused by a browser crash, by an intermediate equipment between the
20816 client and haproxy which decided to actively break the connection,
20817 by network routing issues between the client and haproxy, or by a
20818 keep-alive session between the server and the client terminated first
20819 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010020820
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020821 cD The client did not send nor acknowledge any data for as long as the
20822 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020020823 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020824
20825 CH The client aborted while waiting for the server to start responding.
20826 It might be the server taking too long to respond or the client
20827 clicking the 'Stop' button too fast.
20828
20829 cH The "timeout client" stroke while waiting for client data during a
20830 POST request. This is sometimes caused by too large TCP MSS values
20831 for PPPoE networks which cannot transport full-sized packets. It can
20832 also happen when client timeout is smaller than server timeout and
20833 the server takes too long to respond.
20834
20835 CQ The client aborted while its session was queued, waiting for a server
20836 with enough empty slots to accept it. It might be that either all the
20837 servers were saturated or that the assigned server was taking too
20838 long a time to respond.
20839
20840 CR The client aborted before sending a full HTTP request. Most likely
20841 the request was typed by hand using a telnet client, and aborted
20842 too early. The HTTP status code is likely a 400 here. Sometimes this
20843 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020020844 and the client. "option http-ignore-probes" can be used to ignore
20845 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020846
20847 cR The "timeout http-request" stroke before the client sent a full HTTP
20848 request. This is sometimes caused by too large TCP MSS values on the
20849 client side for PPPoE networks which cannot transport full-sized
20850 packets, or by clients sending requests by hand and not typing fast
20851 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020020852 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020020853 some browsers started to implement a "pre-connect" feature consisting
20854 in speculatively connecting to some recently visited web sites just
20855 in case the user would like to visit them. This results in many
20856 connections being established to web sites, which end up in 408
20857 Request Timeout if the timeout strikes first, or 400 Bad Request when
20858 the browser decides to close them first. These ones pollute the log
20859 and feed the error counters. Some versions of some browsers have even
20860 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010020861 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020020862 http-ignore-probes" in the frontend, resulting in connections with
20863 zero data transfer to be totally ignored. This will definitely hide
20864 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020865
20866 CT The client aborted while its session was tarpitted. It is important to
20867 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020020868 wrong tarpit rules have been written. If a lot of them happen, it
20869 might make sense to lower the "timeout tarpit" value to something
20870 closer to the average reported "Tw" timer, in order not to consume
20871 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020872
Willy Tarreau570f2212013-06-10 16:42:09 +020020873 LR The request was intercepted and locally handled by haproxy. Generally
20874 it means that this was a redirect or a stats request.
20875
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020876 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020877 the TCP connection (the proxy received a TCP RST or an ICMP message
20878 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010020879 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020880 or no ARP response on local network). When this happens in HTTP mode,
20881 the status code is likely a 502 or 503 here.
20882
20883 sC The "timeout connect" stroke before a connection to the server could
20884 complete. When this happens in HTTP mode, the status code is likely a
20885 503 or 504 here.
20886
20887 SD The connection to the server died with an error during the data
20888 transfer. This usually means that haproxy has received an RST from
20889 the server or an ICMP message from an intermediate equipment while
20890 exchanging data with the server. This can be caused by a server crash
20891 or by a network issue on an intermediate equipment.
20892
20893 sD The server did not send nor acknowledge any data for as long as the
20894 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010020895 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020896 load-balancers, ...), as well as keep-alive sessions maintained
20897 between the client and the server expiring first on haproxy.
20898
20899 SH The server aborted before sending its full HTTP response headers, or
20900 it crashed while processing the request. Since a server aborting at
20901 this moment is very rare, it would be wise to inspect its logs to
20902 control whether it crashed and why. The logged request may indicate a
20903 small set of faulty requests, demonstrating bugs in the application.
20904 Sometimes this might also be caused by an IDS killing the connection
20905 between haproxy and the server.
20906
20907 sH The "timeout server" stroke before the server could return its
20908 response headers. This is the most common anomaly, indicating too
20909 long transactions, probably caused by server or database saturation.
20910 The immediate workaround consists in increasing the "timeout server"
20911 setting, but it is important to keep in mind that the user experience
20912 will suffer from these long response times. The only long term
20913 solution is to fix the application.
20914
20915 sQ The session spent too much time in queue and has been expired. See
20916 the "timeout queue" and "timeout connect" settings to find out how to
20917 fix this if it happens too often. If it often happens massively in
20918 short periods, it may indicate general problems on the affected
20919 servers due to I/O or database congestion, or saturation caused by
20920 external attacks.
20921
20922 PC The proxy refused to establish a connection to the server because the
Thayne McCombscdbcca92021-01-07 21:24:41 -070020923 process's socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020020924 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020925 so that it does not happen anymore. This status is very rare and
20926 might happen when the global "ulimit-n" parameter is forced by hand.
20927
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020928 PD The proxy blocked an incorrectly formatted chunked encoded message in
20929 a request or a response, after the server has emitted its headers. In
20930 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010020931 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020020932 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020933
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020934 PH The proxy blocked the server's response, because it was invalid,
20935 incomplete, dangerous (cache control), or matched a security filter.
20936 In any case, an HTTP 502 error is sent to the client. One possible
20937 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020938 containing unauthorized characters. It is also possible but quite
20939 rare, that the proxy blocked a chunked-encoding request from the
20940 client due to an invalid syntax, before the server responded. In this
20941 case, an HTTP 400 error is sent to the client and reported in the
20942 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020943
20944 PR The proxy blocked the client's HTTP request, either because of an
20945 invalid HTTP syntax, in which case it returned an HTTP 400 error to
20946 the client, or because a deny filter matched, in which case it
20947 returned an HTTP 403 error.
20948
20949 PT The proxy blocked the client's request and has tarpitted its
20950 connection before returning it a 500 server error. Nothing was sent
20951 to the server. The connection was maintained open for as long as
20952 reported by the "Tw" timer field.
20953
20954 RC A local resource has been exhausted (memory, sockets, source ports)
20955 preventing the connection to the server from establishing. The error
20956 logs will tell precisely what was missing. This is very rare and can
20957 only be solved by proper system tuning.
20958
Willy Tarreau996a92c2010-10-13 19:30:47 +020020959The combination of the two last flags gives a lot of information about how
20960persistence was handled by the client, the server and by haproxy. This is very
20961important to troubleshoot disconnections, when users complain they have to
20962re-authenticate. The commonly encountered flags are :
20963
20964 -- Persistence cookie is not enabled.
20965
20966 NN No cookie was provided by the client, none was inserted in the
20967 response. For instance, this can be in insert mode with "postonly"
20968 set on a GET request.
20969
20970 II A cookie designating an invalid server was provided by the client,
20971 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020972 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020020973 value can be presented by a client when no other server knows it.
20974
20975 NI No cookie was provided by the client, one was inserted in the
20976 response. This typically happens for first requests from every user
20977 in "insert" mode, which makes it an easy way to count real users.
20978
20979 VN A cookie was provided by the client, none was inserted in the
20980 response. This happens for most responses for which the client has
20981 already got a cookie.
20982
20983 VU A cookie was provided by the client, with a last visit date which is
20984 not completely up-to-date, so an updated cookie was provided in
20985 response. This can also happen if there was no date at all, or if
20986 there was a date but the "maxidle" parameter was not set, so that the
20987 cookie can be switched to unlimited time.
20988
20989 EI A cookie was provided by the client, with a last visit date which is
20990 too old for the "maxidle" parameter, so the cookie was ignored and a
20991 new cookie was inserted in the response.
20992
20993 OI A cookie was provided by the client, with a first visit date which is
20994 too old for the "maxlife" parameter, so the cookie was ignored and a
20995 new cookie was inserted in the response.
20996
20997 DI The server designated by the cookie was down, a new server was
20998 selected and a new cookie was emitted in the response.
20999
21000 VI The server designated by the cookie was not marked dead but could not
21001 be reached. A redispatch happened and selected another one, which was
21002 then advertised in the response.
21003
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021004
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200210058.6. Non-printable characters
21006-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021007
21008In order not to cause trouble to log analysis tools or terminals during log
21009consulting, non-printable characters are not sent as-is into log files, but are
21010converted to the two-digits hexadecimal representation of their ASCII code,
21011prefixed by the character '#'. The only characters that can be logged without
21012being escaped are comprised between 32 and 126 (inclusive). Obviously, the
21013escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
21014is the same for the character '"' which becomes "#22", as well as '{', '|' and
21015'}' when logging headers.
21016
21017Note that the space character (' ') is not encoded in headers, which can cause
21018issues for tools relying on space count to locate fields. A typical header
21019containing spaces is "User-Agent".
21020
21021Last, it has been observed that some syslog daemons such as syslog-ng escape
21022the quote ('"') with a backslash ('\'). The reverse operation can safely be
21023performed since no quote may appear anywhere else in the logs.
21024
21025
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200210268.7. Capturing HTTP cookies
21027---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021028
21029Cookie capture simplifies the tracking a complete user session. This can be
21030achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021031section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021032cookie will simultaneously be checked in the request ("Cookie:" header) and in
21033the response ("Set-Cookie:" header). The respective values will be reported in
21034the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021035locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021036not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
21037user switches to a new session for example, because the server will reassign it
21038a new cookie. It is also possible to detect if a server unexpectedly sets a
21039wrong cookie to a client, leading to session crossing.
21040
21041 Examples :
21042 # capture the first cookie whose name starts with "ASPSESSION"
21043 capture cookie ASPSESSION len 32
21044
21045 # capture the first cookie whose name is exactly "vgnvisitor"
21046 capture cookie vgnvisitor= len 32
21047
21048
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200210498.8. Capturing HTTP headers
21050---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021051
21052Header captures are useful to track unique request identifiers set by an upper
21053proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
21054the response, one can search for information about the response length, how the
21055server asked the cache to behave, or an object location during a redirection.
21056
21057Header captures are performed using the "capture request header" and "capture
21058response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021059section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021060
21061It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010021062time. Non-existent headers are logged as empty strings, and if one header
21063appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021064are grouped within braces '{' and '}' in the same order as they were declared,
21065and delimited with a vertical bar '|' without any space. Response headers
21066follow the same representation, but are displayed after a space following the
21067request headers block. These blocks are displayed just before the HTTP request
21068in the logs.
21069
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020021070As a special case, it is possible to specify an HTTP header capture in a TCP
21071frontend. The purpose is to enable logging of headers which will be parsed in
21072an HTTP backend if the request is then switched to this HTTP backend.
21073
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021074 Example :
21075 # This instance chains to the outgoing proxy
21076 listen proxy-out
21077 mode http
21078 option httplog
21079 option logasap
21080 log global
21081 server cache1 192.168.1.1:3128
21082
21083 # log the name of the virtual server
21084 capture request header Host len 20
21085
21086 # log the amount of data uploaded during a POST
21087 capture request header Content-Length len 10
21088
21089 # log the beginning of the referrer
21090 capture request header Referer len 20
21091
21092 # server name (useful for outgoing proxies only)
21093 capture response header Server len 20
21094
21095 # logging the content-length is useful with "option logasap"
21096 capture response header Content-Length len 10
21097
Davor Ocelice9ed2812017-12-25 17:49:28 +010021098 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021099 capture response header Cache-Control len 8
21100
21101 # the Via header will report the next proxy's name
21102 capture response header Via len 20
21103
21104 # log the URL location during a redirection
21105 capture response header Location len 20
21106
21107 >>> Aug 9 20:26:09 localhost \
21108 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
21109 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
21110 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
21111 "GET http://fr.adserver.yahoo.com/"
21112
21113 >>> Aug 9 20:30:46 localhost \
21114 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
21115 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
21116 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010021117 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021118
21119 >>> Aug 9 20:30:46 localhost \
21120 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
21121 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
21122 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
21123 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010021124 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021125
21126
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200211278.9. Examples of logs
21128---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021129
21130These are real-world examples of logs accompanied with an explanation. Some of
21131them have been made up by hand. The syslog part has been removed for better
21132reading. Their sole purpose is to explain how to decipher them.
21133
21134 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
21135 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
21136 "HEAD / HTTP/1.0"
21137
21138 => long request (6.5s) entered by hand through 'telnet'. The server replied
21139 in 147 ms, and the session ended normally ('----')
21140
21141 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
21142 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
21143 0/9 "HEAD / HTTP/1.0"
21144
21145 => Idem, but the request was queued in the global queue behind 9 other
21146 requests, and waited there for 1230 ms.
21147
21148 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
21149 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
21150 "GET /image.iso HTTP/1.0"
21151
21152 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010021153 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021154 14 ms, 243 bytes of headers were sent to the client, and total time from
21155 accept to first data byte is 30 ms.
21156
21157 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
21158 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
21159 "GET /cgi-bin/bug.cgi? HTTP/1.0"
21160
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020021161 => the proxy blocked a server response either because of an "http-response
21162 deny" rule, or because the response was improperly formatted and not
21163 HTTP-compliant, or because it blocked sensitive information which risked
21164 being cached. In this case, the response is replaced with a "502 bad
21165 gateway". The flags ("PH--") tell us that it was haproxy who decided to
21166 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021167
21168 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010021169 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 +010021170
21171 => the client never completed its request and aborted itself ("C---") after
21172 8.5s, while the proxy was waiting for the request headers ("-R--").
21173 Nothing was sent to any server.
21174
21175 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
21176 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
21177
21178 => The client never completed its request, which was aborted by the
21179 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010021180 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021181 send a 408 return code to the client.
21182
21183 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
21184 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
21185
21186 => This log was produced with "option tcplog". The client timed out after
21187 5 seconds ("c----").
21188
21189 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
21190 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010021191 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021192
21193 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021194 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021195 (config says 'retries 3'), and no redispatch (otherwise we would have
21196 seen "/+3"). Status code 503 was returned to the client. There were 115
21197 connections on this server, 202 connections on this proxy, and 205 on
21198 the global process. It is possible that the server refused the
21199 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010021200
Willy Tarreau52b2d222011-09-07 23:48:48 +020021201
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200212029. Supported filters
21203--------------------
21204
21205Here are listed officially supported filters with the list of parameters they
21206accept. Depending on compile options, some of these filters might be
21207unavailable. The list of available filters is reported in haproxy -vv.
21208
21209See also : "filter"
21210
212119.1. Trace
21212----------
21213
Christopher Fauletc41d8bd2020-11-17 10:43:26 +010021214filter trace [name <name>] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021215
21216 Arguments:
21217 <name> is an arbitrary name that will be reported in
21218 messages. If no name is provided, "TRACE" is used.
21219
Christopher Faulet96a577a2020-11-17 10:45:05 +010021220 <quiet> inhibits trace messages.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021221
Davor Ocelice9ed2812017-12-25 17:49:28 +010021222 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021223 default, this filter forwards all previously parsed
21224 data. With this parameter, it only forwards a random
21225 amount of the parsed data.
21226
Davor Ocelice9ed2812017-12-25 17:49:28 +010021227 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010021228
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021229This filter can be used as a base to develop new filters. It defines all
21230callbacks and print a message on the standard error stream (stderr) with useful
21231information for all of them. It may be useful to debug the activity of other
21232filters or, quite simply, HAProxy's activity.
21233
21234Using <random-parsing> and/or <random-forwarding> parameters is a good way to
21235tests the behavior of a filter that parses data exchanged between a client and
21236a server by adding some latencies in the processing.
21237
21238
212399.2. HTTP compression
21240---------------------
21241
21242filter compression
21243
21244The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
21245keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020021246when no other filter is used, it is enough. When used with the cache or the
21247fcgi-app enabled, it is also enough. In this case, the compression is always
21248done after the response is stored in the cache. But it is mandatory to
21249explicitly use a filter line to enable the HTTP compression when at least one
21250filter other than the cache or the fcgi-app is used for the same
21251listener/frontend/backend. This is important to know the filters evaluation
21252order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021253
Christopher Fauletb30b3102019-09-12 23:03:09 +020021254See also : "compression", section 9.4 about the cache filter and section 9.5
21255 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021256
21257
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200212589.3. Stream Processing Offload Engine (SPOE)
21259--------------------------------------------
21260
21261filter spoe [engine <name>] config <file>
21262
21263 Arguments :
21264
21265 <name> is the engine name that will be used to find the right scope in
21266 the configuration file. If not provided, all the file will be
21267 parsed.
21268
21269 <file> is the path of the engine configuration file. This file can
21270 contain configuration of several engines. In this case, each
21271 part must be placed in its own scope.
21272
21273The Stream Processing Offload Engine (SPOE) is a filter communicating with
21274external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010021275streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020021276exchanged with them are configured in dedicated files, for the main part. It
21277also requires dedicated backends, defined in HAProxy configuration.
21278
21279SPOE communicates with external components using an in-house binary protocol,
21280the Stream Processing Offload Protocol (SPOP).
21281
Tim Düsterhus4896c442016-11-29 02:15:19 +010021282For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020021283"doc/SPOE.txt".
21284
Christopher Faulet99a17a22018-12-11 09:18:27 +0100212859.4. Cache
21286----------
21287
21288filter cache <name>
21289
21290 Arguments :
21291
21292 <name> is name of the cache section this filter will use.
21293
21294The cache uses a filter to store cacheable responses. The HTTP rules
21295"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050021296cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020021297other filters than fcgi-app or compression are used, it is enough. In such
21298case, the compression filter is always evaluated after the cache filter. But it
21299is mandatory to explicitly use a filter line to use a cache when at least one
21300filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010021301listener/frontend/backend. This is important to know the filters evaluation
21302order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010021303
Christopher Fauletb30b3102019-09-12 23:03:09 +020021304See also : section 9.2 about the compression filter, section 9.5 about the
21305 fcgi-app filter and section 6 about cache.
21306
21307
213089.5. Fcgi-app
21309-------------
21310
Daniel Corbett67a82712020-07-06 23:01:19 -040021311filter fcgi-app <name>
Christopher Fauletb30b3102019-09-12 23:03:09 +020021312
21313 Arguments :
21314
21315 <name> is name of the fcgi-app section this filter will use.
21316
21317The FastCGI application uses a filter to evaluate all custom parameters on the
21318request path, and to process the headers on the response path. the <name> must
21319reference an existing fcgi-app section. The directive "use-fcgi-app" should be
21320used to define the application to use. By default the corresponding filter is
21321implicitly defined. And when no other filters than cache or compression are
21322used, it is enough. But it is mandatory to explicitly use a filter line to a
21323fcgi-app when at least one filter other than the compression or the cache is
21324used for the same backend. This is important to know the filters evaluation
21325order.
21326
21327See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
21328 about the cache filter and section 10 about FastCGI application.
21329
21330
Miroslav Zagoracdc32cd92020-12-13 18:32:57 +0100213319.6. OpenTracing
21332----------------
21333
21334The OpenTracing filter adds native support for using distributed tracing in
21335HAProxy. This is enabled by sending an OpenTracing compliant request to one
21336of the supported tracers such as Datadog, Jaeger, Lightstep and Zipkin tracers.
21337Please note: tracers are not listed by any preference, but alphabetically.
21338
21339This feature is only enabled when haproxy was built with USE_OT=1.
21340
21341The OpenTracing filter activation is done explicitly by specifying it in the
21342HAProxy configuration. If this is not done, the OpenTracing filter in no way
21343participates in the work of HAProxy.
21344
21345filter opentracing [id <id>] config <file>
21346
21347 Arguments :
21348
21349 <id> is the OpenTracing filter id that will be used to find the
21350 right scope in the configuration file. If no filter id is
21351 specified, 'ot-filter' is used as default. If scope is not
21352 specified in the configuration file, it applies to all defined
21353 OpenTracing filters.
21354
21355 <file> is the path of the OpenTracing configuration file. The same
21356 file can contain configurations for multiple OpenTracing
21357 filters simultaneously. In that case we do not need to define
21358 scope so the same configuration applies to all filters or each
21359 filter must have its own scope defined.
21360
21361More detailed documentation related to the operation, configuration and use
Willy Tarreaua63d1a02021-04-02 17:16:46 +020021362of the filter can be found in the addons/ot directory.
Miroslav Zagoracdc32cd92020-12-13 18:32:57 +010021363
21364
Christopher Fauletb30b3102019-09-12 23:03:09 +02002136510. FastCGI applications
21366-------------------------
21367
21368HAProxy is able to send HTTP requests to Responder FastCGI applications. This
21369feature was added in HAProxy 2.1. To do so, servers must be configured to use
21370the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
21371FastCGI application must be configured and used by the backend managing these
21372servers (using the keyword "use-fcgi-app" into the proxy section). Several
21373FastCGI applications may be defined, but only one can be used at a time by a
21374backend.
21375
21376HAProxy implements all features of the FastCGI specification for Responder
21377application. Especially it is able to multiplex several requests on a simple
21378connection.
21379
2138010.1. Setup
21381-----------
21382
2138310.1.1. Fcgi-app section
21384--------------------------
21385
21386fcgi-app <name>
21387 Declare a FastCGI application named <name>. To be valid, at least the
21388 document root must be defined.
21389
21390acl <aclname> <criterion> [flags] [operator] <value> ...
21391 Declare or complete an access list.
21392
21393 See "acl" keyword in section 4.2 and section 7 about ACL usage for
21394 details. ACLs defined for a FastCGI application are private. They cannot be
21395 used by any other application or by any proxy. In the same way, ACLs defined
21396 in any other section are not usable by a FastCGI application. However,
21397 Pre-defined ACLs are available.
21398
21399docroot <path>
21400 Define the document root on the remote host. <path> will be used to build
21401 the default value of FastCGI parameters SCRIPT_FILENAME and
21402 PATH_TRANSLATED. It is a mandatory setting.
21403
21404index <script-name>
21405 Define the script name that will be appended after an URI that ends with a
21406 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
21407 is an optional setting.
21408
21409 Example :
21410 index index.php
21411
21412log-stderr global
21413log-stderr <address> [len <length>] [format <format>]
Jan Wagner3e678602020-12-17 22:22:32 +010021414 [sample <ranges>:<sample_size>] <facility> [<level> [<minlevel>]]
Christopher Fauletb30b3102019-09-12 23:03:09 +020021415 Enable logging of STDERR messages reported by the FastCGI application.
21416
21417 See "log" keyword in section 4.2 for details. It is an optional setting. By
21418 default STDERR messages are ignored.
21419
21420pass-header <name> [ { if | unless } <condition> ]
21421 Specify the name of a request header which will be passed to the FastCGI
21422 application. It may optionally be followed by an ACL-based condition, in
21423 which case it will only be evaluated if the condition is true.
21424
21425 Most request headers are already available to the FastCGI application,
21426 prefixed with "HTTP_". Thus, this directive is only required to pass headers
21427 that are purposefully omitted. Currently, the headers "Authorization",
21428 "Proxy-Authorization" and hop-by-hop headers are omitted.
21429
21430 Note that the headers "Content-type" and "Content-length" are never passed to
21431 the FastCGI application because they are already converted into parameters.
21432
21433path-info <regex>
Christopher Faulet28cb3662020-02-14 14:47:37 +010021434 Define a regular expression to extract the script-name and the path-info from
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010021435 the URL-decoded path. Thus, <regex> may have two captures: the first one to
21436 capture the script name and the second one to capture the path-info. The
21437 first one is mandatory, the second one is optional. This way, it is possible
21438 to extract the script-name from the path ignoring the path-info. It is an
21439 optional setting. If it is not defined, no matching is performed on the
21440 path. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not
21441 filled.
Christopher Faulet28cb3662020-02-14 14:47:37 +010021442
21443 For security reason, when this regular expression is defined, the newline and
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050021444 the null characters are forbidden from the path, once URL-decoded. The reason
Christopher Faulet28cb3662020-02-14 14:47:37 +010021445 to such limitation is because otherwise the matching always fails (due to a
21446 limitation one the way regular expression are executed in HAProxy). So if one
21447 of these two characters is found in the URL-decoded path, an error is
21448 returned to the client. The principle of least astonishment is applied here.
Christopher Fauletb30b3102019-09-12 23:03:09 +020021449
21450 Example :
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010021451 path-info ^(/.+\.php)(/.*)?$ # both script-name and path-info may be set
21452 path-info ^(/.+\.php) # the path-info is ignored
Christopher Fauletb30b3102019-09-12 23:03:09 +020021453
21454option get-values
21455no option get-values
21456 Enable or disable the retrieve of variables about connection management.
21457
Daniel Corbett67a82712020-07-06 23:01:19 -040021458 HAProxy is able to send the record FCGI_GET_VALUES on connection
Christopher Fauletb30b3102019-09-12 23:03:09 +020021459 establishment to retrieve the value for following variables:
21460
21461 * FCGI_MAX_REQS The maximum number of concurrent requests this
21462 application will accept.
21463
William Lallemand93e548e2019-09-30 13:54:02 +020021464 * FCGI_MPXS_CONNS "0" if this application does not multiplex connections,
21465 "1" otherwise.
Christopher Fauletb30b3102019-09-12 23:03:09 +020021466
21467 Some FastCGI applications does not support this feature. Some others close
Ilya Shipitsin11057a32020-06-21 21:18:27 +050021468 the connection immediately after sending their response. So, by default, this
Christopher Fauletb30b3102019-09-12 23:03:09 +020021469 option is disabled.
21470
21471 Note that the maximum number of concurrent requests accepted by a FastCGI
21472 application is a connection variable. It only limits the number of streams
21473 per connection. If the global load must be limited on the application, the
21474 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
21475 an application does not support connection multiplexing, the maximum number
21476 of concurrent requests is automatically set to 1.
21477
21478option keep-conn
21479no option keep-conn
21480 Instruct the FastCGI application to keep the connection open or not after
21481 sending a response.
21482
21483 If disabled, the FastCGI application closes the connection after responding
21484 to this request. By default, this option is enabled.
21485
21486option max-reqs <reqs>
21487 Define the maximum number of concurrent requests this application will
21488 accept.
21489
21490 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
21491 during connection establishment. Furthermore, if the application does not
21492 support connection multiplexing, this option will be ignored. By default set
21493 to 1.
21494
21495option mpxs-conns
21496no option mpxs-conns
21497 Enable or disable the support of connection multiplexing.
21498
21499 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
21500 during connection establishment. It is disabled by default.
21501
21502set-param <name> <fmt> [ { if | unless } <condition> ]
21503 Set a FastCGI parameter that should be passed to this application. Its
21504 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
21505 "Custom Log format"). It may optionally be followed by an ACL-based
21506 condition, in which case it will only be evaluated if the condition is true.
21507
21508 With this directive, it is possible to overwrite the value of default FastCGI
21509 parameters. If the value is evaluated to an empty string, the rule is
21510 ignored. These directives are evaluated in their declaration order.
21511
21512 Example :
21513 # PHP only, required if PHP was built with --enable-force-cgi-redirect
21514 set-param REDIRECT_STATUS 200
21515
21516 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
21517
21518
2151910.1.2. Proxy section
21520---------------------
21521
21522use-fcgi-app <name>
21523 Define the FastCGI application to use for the backend.
21524
21525 Arguments :
21526 <name> is the name of the FastCGI application to use.
21527
21528 This keyword is only available for HTTP proxies with the backend capability
21529 and with at least one FastCGI server. However, FastCGI servers can be mixed
21530 with HTTP servers. But except there is a good reason to do so, it is not
21531 recommended (see section 10.3 about the limitations for details). Only one
21532 application may be defined at a time per backend.
21533
21534 Note that, once a FastCGI application is referenced for a backend, depending
21535 on the configuration some processing may be done even if the request is not
21536 sent to a FastCGI server. Rules to set parameters or pass headers to an
21537 application are evaluated.
21538
21539
2154010.1.3. Example
21541---------------
21542
21543 frontend front-http
21544 mode http
21545 bind *:80
21546 bind *:
21547
21548 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
21549 default_backend back-static
21550
21551 backend back-static
21552 mode http
21553 server www A.B.C.D:80
21554
21555 backend back-dynamic
21556 mode http
21557 use-fcgi-app php-fpm
21558 server php-fpm A.B.C.D:9000 proto fcgi
21559
21560 fcgi-app php-fpm
21561 log-stderr global
21562 option keep-conn
21563
21564 docroot /var/www/my-app
21565 index index.php
21566 path-info ^(/.+\.php)(/.*)?$
21567
21568
2156910.2. Default parameters
21570------------------------
21571
21572A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
21573the CGI/1.1 specification (RFC3875), several variables must be passed to the
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050021574script. So HAProxy set them and some others commonly used by FastCGI
Christopher Fauletb30b3102019-09-12 23:03:09 +020021575applications. All these variables may be overwritten, with caution though.
21576
21577 +-------------------+-----------------------------------------------------+
21578 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
21579 | | to authenticate the user. Concretely, only the |
21580 | | BASIC authentication mechanism is supported. |
21581 | | |
21582 +-------------------+-----------------------------------------------------+
21583 | CONTENT_LENGTH | Contains the size of the message-body attached to |
21584 | | the request. It means only requests with a known |
21585 | | size are considered as valid and sent to the |
21586 | | application. |
21587 | | |
21588 +-------------------+-----------------------------------------------------+
21589 | CONTENT_TYPE | Contains the type of the message-body attached to |
21590 | | the request. It may not be set. |
21591 | | |
21592 +-------------------+-----------------------------------------------------+
21593 | DOCUMENT_ROOT | Contains the document root on the remote host under |
21594 | | which the script should be executed, as defined in |
21595 | | the application's configuration. |
21596 | | |
21597 +-------------------+-----------------------------------------------------+
21598 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
21599 | | to communicate with the FastCGI application. |
21600 | | Concretely, it is set to "CGI/1.1". |
21601 | | |
21602 +-------------------+-----------------------------------------------------+
21603 | PATH_INFO | Contains the portion of the URI path hierarchy |
21604 | | following the part that identifies the script |
21605 | | itself. To be set, the directive "path-info" must |
21606 | | be defined. |
21607 | | |
21608 +-------------------+-----------------------------------------------------+
21609 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
21610 | | It is the concatenation of DOCUMENT_ROOT and |
21611 | | PATH_INFO. If PATH_INFO is not set, this parameters |
21612 | | is not set too. |
21613 | | |
21614 +-------------------+-----------------------------------------------------+
21615 | QUERY_STRING | Contains the request's query string. It may not be |
21616 | | set. |
21617 | | |
21618 +-------------------+-----------------------------------------------------+
21619 | REMOTE_ADDR | Contains the network address of the client sending |
21620 | | the request. |
21621 | | |
21622 +-------------------+-----------------------------------------------------+
21623 | REMOTE_USER | Contains the user identification string supplied by |
21624 | | client as part of user authentication. |
21625 | | |
21626 +-------------------+-----------------------------------------------------+
21627 | REQUEST_METHOD | Contains the method which should be used by the |
21628 | | script to process the request. |
21629 | | |
21630 +-------------------+-----------------------------------------------------+
21631 | REQUEST_URI | Contains the request's URI. |
21632 | | |
21633 +-------------------+-----------------------------------------------------+
21634 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
21635 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
21636 | | |
21637 +-------------------+-----------------------------------------------------+
21638 | SCRIPT_NAME | Contains the name of the script. If the directive |
21639 | | "path-info" is defined, it is the first part of the |
21640 | | URI path hierarchy, ending with the script name. |
21641 | | Otherwise, it is the entire URI path. |
21642 | | |
21643 +-------------------+-----------------------------------------------------+
21644 | SERVER_NAME | Contains the name of the server host to which the |
21645 | | client request is directed. It is the value of the |
21646 | | header "Host", if defined. Otherwise, the |
21647 | | destination address of the connection on the client |
21648 | | side. |
21649 | | |
21650 +-------------------+-----------------------------------------------------+
21651 | SERVER_PORT | Contains the destination TCP port of the connection |
21652 | | on the client side, which is the port the client |
21653 | | connected to. |
21654 | | |
21655 +-------------------+-----------------------------------------------------+
21656 | SERVER_PROTOCOL | Contains the request's protocol. |
21657 | | |
21658 +-------------------+-----------------------------------------------------+
21659 | HTTPS | Set to a non-empty value ("on") if the script was |
21660 | | queried through the HTTPS protocol. |
21661 | | |
21662 +-------------------+-----------------------------------------------------+
21663
21664
2166510.3. Limitations
21666------------------
21667
21668The current implementation have some limitations. The first one is about the
21669way some request headers are hidden to the FastCGI applications. This happens
21670during the headers analysis, on the backend side, before the connection
21671establishment. At this stage, HAProxy know the backend is using a FastCGI
21672application but it don't know if the request will be routed to a FastCGI server
21673or not. But to hide request headers, it simply removes them from the HTX
21674message. So, if the request is finally routed to an HTTP server, it never see
21675these headers. For this reason, it is not recommended to mix FastCGI servers
21676and HTTP servers under the same backend.
21677
21678Similarly, the rules "set-param" and "pass-header" are evaluated during the
21679request headers analysis. So the evaluation is always performed, even if the
21680requests is finally forwarded to an HTTP server.
21681
21682About the rules "set-param", when a rule is applied, a pseudo header is added
21683into the HTX message. So, the same way than for HTTP header rewrites, it may
21684fail if the buffer is full. The rules "set-param" will compete with
21685"http-request" ones.
21686
21687Finally, all FastCGI params and HTTP headers are sent into a unique record
21688FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
21689processing error is returned. It means the record FCGI_PARAM, once encoded,
21690must not exceeds the size of a buffer. However, there is no reserve to respect
21691here.
William Lallemand86d0df02017-11-24 21:36:45 +010021692
Emeric Brunce325c42021-04-02 17:05:09 +020021693
2169411. Address formats
21695-------------------
21696
21697Several statements as "bind, "server", "nameserver" and "log" requires an
21698address.
21699
21700This address can be a host name, an IPv4 address, an IPv6 address, or '*'.
21701The '*' is equal to the special address "0.0.0.0" and can be used, in the case
21702of "bind" or "dgram-bind" to listen on all IPv4 of the system.The IPv6
21703equivalent is '::'.
21704
21705Depending of the statement, a port or port range follows the IP address. This
21706is mandatory on 'bind' statement, optional on 'server'.
21707
21708This address can also begin with a slash '/'. It is considered as the "unix"
21709family, and '/' and following characters must be present the path.
21710
21711Default socket type or transport method "datagram" or "stream" depends on the
21712configuration statement showing the address. Indeed, 'bind' and 'server' will
21713use a "stream" socket type by default whereas 'log', 'nameserver' or
21714'dgram-bind' will use a "datagram".
21715
21716Optionally, a prefix could be used to force the address family and/or the
21717socket type and the transport method.
21718
21719
2172011.1 Address family prefixes
21721----------------------------
21722
21723'abns@<name>' following <name> is an abstract namespace (Linux only).
21724
21725'fd@<n>' following address is a file descriptor <n> inherited from the
21726 parent. The fd must be bound and may or may not already be
21727 listening.
21728
21729'ip@<address>[:port1[-port2]]' following <address> is considered as an IPv4 or
21730 IPv6 address depending on the syntax. Depending
21731 on the statement using this address, a port or
21732 a port range may or must be specified.
21733
21734'ipv4@<address>[:port1[-port2]]' following <address> is always considered as
21735 an IPv4 address. Depending on the statement
21736 using this address, a port or a port range
21737 may or must be specified.
21738
21739'ipv6@<address>[:port1[-port2]]' following <address> is always considered as
21740 an IPv6 address. Depending on the statement
21741 using this address, a port or a port range
21742 may or must be specified.
21743
21744'sockpair@<n>' following address is the file descriptor of a connected unix
21745 socket or of a socketpair. During a connection, the initiator
21746 creates a pair of connected sockets, and passes one of them
21747 over the FD to the other end. The listener waits to receive
21748 the FD from the unix socket and uses it as if it were the FD
21749 of an accept(). Should be used carefully.
21750
21751'unix@<path>' following string is considered as a UNIX socket <path>. this
21752 prefix is useful to declare an UNIX socket path which don't
21753 start by slash '/'.
21754
21755
2175611.2 Socket type prefixes
21757-------------------------
21758
21759Previous "Address family prefixes" can also be prefixed to force the socket
21760type and the transport method. The default depends of the statement using
21761this address but in some cases the user may force it to a different one.
21762This is the case for "log" statement where the default is syslog over UDP
21763but we could force to use syslog over TCP.
21764
21765Those prefixes were designed for internal purpose and users should
21766instead use aliases of the next section "11.5.3 Protocol prefixes".
21767
21768If users need one those prefixes to perform what they expect because
21769they can not configure the same using the protocol prefixes, they should
21770report this to the maintainers.
21771
21772'stream+<family>@<address>' forces socket type and transport method
21773 to "stream"
21774
21775'dgram+<family>@<address>' forces socket type and transport method
21776 to "datagram".
21777
21778
2177911.3 Protocol prefixes
21780----------------------
21781
21782'tcp@<address>[:port1[-port2]]' following <address> is considered as an IPv4
21783 or IPv6 address depending of the syntax but
21784 socket type and transport method is forced to
21785 "stream". Depending on the statement using
21786 this address, a port or a port range can or
21787 must be specified. It is considered as an alias
21788 of 'stream+ip@'.
21789
21790'tcp4@<address>[:port1[-port2]]' following <address> is always considered as
21791 an IPv4 address but socket type and transport
21792 method is forced to "stream". Depending on the
21793 statement using this address, a port or port
21794 range can or must be specified.
21795 It is considered as an alias of 'stream+ipv4@'.
21796
21797'tcp6@<address>[:port1[-port2]]' following <address> is always considered as
21798 an IPv6 address but socket type and transport
21799 method is forced to "stream". Depending on the
21800 statement using this address, a port or port
21801 range can or must be specified.
21802 It is considered as an alias of 'stream+ipv4@'.
21803
21804'udp@<address>[:port1[-port2]]' following <address> is considered as an IPv4
21805 or IPv6 address depending of the syntax but
21806 socket type and transport method is forced to
21807 "datagram". Depending on the statement using
21808 this address, a port or a port range can or
21809 must be specified. It is considered as an alias
21810 of 'dgram+ip@'.
21811
21812'udp4@<address>[:port1[-port2]]' following <address> is always considered as
21813 an IPv4 address but socket type and transport
21814 method is forced to "datagram". Depending on
21815 the statement using this address, a port or
21816 port range can or must be specified.
21817 It is considered as an alias of 'stream+ipv4@'.
21818
21819'udp6@<address>[:port1[-port2]]' following <address> is always considered as
21820 an IPv6 address but socket type and transport
21821 method is forced to "datagram". Depending on
21822 the statement using this address, a port or
21823 port range can or must be specified.
21824 It is considered as an alias of 'stream+ipv4@'.
21825
21826'uxdg@<path>' following string is considered as a unix socket <path> but
21827 transport method is forced to "datagram". It is considered as
21828 an alias of 'dgram+unix@'.
21829
21830'uxst@<path>' following string is considered as a unix socket <path> but
21831 transport method is forced to "stream". It is considered as
21832 an alias of 'stream+unix@'.
21833
21834In future versions, other prefixes could be used to specify protocols like
21835QUIC which proposes stream transport based on socket of type "datagram".
21836
Willy Tarreau0ba27502007-12-24 16:55:16 +010021837/*
21838 * Local variables:
21839 * fill-column: 79
21840 * End:
21841 */