blob: 565582ccb1b287c8314f322d34b9623ef7d5187e [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 Tarreauaf6d88b2021-03-27 09:42:09 +01007 2021/03/27
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
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200132
1331. Quick reminder about HTTP
134----------------------------
135
Davor Ocelice9ed2812017-12-25 17:49:28 +0100136When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200137fully analyzed and indexed, thus it becomes possible to build matching criteria
138on almost anything found in the contents.
139
140However, it is important to understand how HTTP requests and responses are
141formed, and how HAProxy decomposes them. It will then become easier to write
142correct rules and to debug existing configurations.
143
144
1451.1. The HTTP transaction model
146-------------------------------
147
148The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100149to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100150from the client to the server, a request is sent by the client through the
151connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200152will involve a new connection :
153
154 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
155
156In this mode, called the "HTTP close" mode, there are as many connection
157establishments as there are HTTP transactions. Since the connection is closed
158by the server after the response, the client does not need to know the content
159length.
160
161Due to the transactional nature of the protocol, it was possible to improve it
162to avoid closing a connection between two subsequent transactions. In this mode
163however, it is mandatory that the server indicates the content length for each
164response so that the client does not wait indefinitely. For this, a special
165header is used: "Content-length". This mode is called the "keep-alive" mode :
166
167 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
168
169Its advantages are a reduced latency between transactions, and less processing
170power required on the server side. It is generally better than the close mode,
171but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200172a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200173
Willy Tarreau95c4e142017-11-26 12:18:55 +0100174Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200175keep-alive, but the client does not wait for the first response to send the
176second request. This is useful for fetching large number of images composing a
177page :
178
179 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
180
181This can obviously have a tremendous benefit on performance because the network
182latency is eliminated between subsequent requests. Many HTTP agents do not
183correctly support pipelining since there is no way to associate a response with
184the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100185server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200186
Willy Tarreau95c4e142017-11-26 12:18:55 +0100187The next improvement is the multiplexed mode, as implemented in HTTP/2. This
188time, each transaction is assigned a single stream identifier, and all streams
189are multiplexed over an existing connection. Many requests can be sent in
190parallel by the client, and responses can arrive in any order since they also
191carry the stream identifier.
192
Willy Tarreau70dffda2014-01-30 03:07:23 +0100193By default HAProxy operates in keep-alive mode with regards to persistent
194connections: for each connection it processes each request and response, and
195leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100196start of a new request. When it receives HTTP/2 connections from a client, it
197processes all the requests in parallel and leaves the connection idling,
198waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200199
Christopher Faulet315b39c2018-09-21 16:26:19 +0200200HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100201 - keep alive : all requests and responses are processed (default)
202 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100203 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100204 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200205 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100206
Willy Tarreau95c4e142017-11-26 12:18:55 +0100207
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200208
2091.2. HTTP request
210-----------------
211
212First, let's consider this HTTP request :
213
214 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100215 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200216 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
217 2 Host: www.mydomain.com
218 3 User-agent: my small browser
219 4 Accept: image/jpeg, image/gif
220 5 Accept: image/png
221
222
2231.2.1. The Request line
224-----------------------
225
226Line 1 is the "request line". It is always composed of 3 fields :
227
228 - a METHOD : GET
229 - a URI : /serv/login.php?lang=en&profile=2
230 - a version tag : HTTP/1.1
231
232All of them are delimited by what the standard calls LWS (linear white spaces),
233which are commonly spaces, but can also be tabs or line feeds/carriage returns
234followed by spaces/tabs. The method itself cannot contain any colon (':') and
235is limited to alphabetic letters. All those various combinations make it
236desirable that HAProxy performs the splitting itself rather than leaving it to
237the user to write a complex or inaccurate regular expression.
238
239The URI itself can have several forms :
240
241 - A "relative URI" :
242
243 /serv/login.php?lang=en&profile=2
244
245 It is a complete URL without the host part. This is generally what is
246 received by servers, reverse proxies and transparent proxies.
247
248 - An "absolute URI", also called a "URL" :
249
250 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
251
252 It is composed of a "scheme" (the protocol name followed by '://'), a host
253 name or address, optionally a colon (':') followed by a port number, then
254 a relative URI beginning at the first slash ('/') after the address part.
255 This is generally what proxies receive, but a server supporting HTTP/1.1
256 must accept this form too.
257
258 - a star ('*') : this form is only accepted in association with the OPTIONS
259 method and is not relayable. It is used to inquiry a next hop's
260 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100261
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200262 - an address:port combination : 192.168.0.12:80
263 This is used with the CONNECT method, which is used to establish TCP
264 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
265 other protocols too.
266
267In a relative URI, two sub-parts are identified. The part before the question
268mark is called the "path". It is typically the relative path to static objects
269on the server. The part after the question mark is called the "query string".
270It is mostly used with GET requests sent to dynamic scripts and is very
271specific to the language, framework or application in use.
272
Willy Tarreau95c4e142017-11-26 12:18:55 +0100273HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100274assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100275
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200276
2771.2.2. The request headers
278--------------------------
279
280The headers start at the second line. They are composed of a name at the
281beginning of the line, immediately followed by a colon (':'). Traditionally,
282an LWS is added after the colon but that's not required. Then come the values.
283Multiple identical headers may be folded into one single line, delimiting the
284values with commas, provided that their order is respected. This is commonly
285encountered in the "Cookie:" field. A header may span over multiple lines if
286the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
287define a total of 3 values for the "Accept:" header.
288
Davor Ocelice9ed2812017-12-25 17:49:28 +0100289Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200290their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100291"Connection:" header). In HTTP/2, header names are always sent in lower case,
Willy Tarreau253c2512020-07-07 15:55:23 +0200292as can be seen when running in debug mode. Internally, all header names are
293normalized to lower case so that HTTP/1.x and HTTP/2 use the exact same
294representation, and they are sent as-is on the other side. This explains why an
295HTTP/1.x request typed with camel case is delivered in lower case.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200296
297The end of the headers is indicated by the first empty line. People often say
298that it's a double line feed, which is not exact, even if a double line feed
299is one valid form of empty line.
300
301Fortunately, HAProxy takes care of all these complex combinations when indexing
302headers, checking values and counting them, so there is no reason to worry
303about the way they could be written, but it is important not to accuse an
304application of being buggy if it does unusual, valid things.
305
306Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000307 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200308 in the middle of headers by LWS in order to join multi-line headers. This
309 is necessary for proper analysis and helps less capable HTTP parsers to work
310 correctly and not to be fooled by such complex constructs.
311
312
3131.3. HTTP response
314------------------
315
316An HTTP response looks very much like an HTTP request. Both are called HTTP
317messages. Let's consider this HTTP response :
318
319 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100320 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200321 1 HTTP/1.1 200 OK
322 2 Content-length: 350
323 3 Content-Type: text/html
324
Willy Tarreau816b9792009-09-15 21:25:21 +0200325As a special case, HTTP supports so called "Informational responses" as status
326codes 1xx. These messages are special in that they don't convey any part of the
327response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100328continue to post its request for instance. In the case of a status 100 response
329the requested information will be carried by the next non-100 response message
330following the informational one. This implies that multiple responses may be
331sent to a single request, and that this only works when keep-alive is enabled
332(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
333correctly forward and skip them, and only process the next non-100 response. As
334such, these messages are neither logged nor transformed, unless explicitly
335state otherwise. Status 101 messages indicate that the protocol is changing
336over the same connection and that haproxy must switch to tunnel mode, just as
337if a CONNECT had occurred. Then the Upgrade header would contain additional
338information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200339
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200340
Davor Ocelice9ed2812017-12-25 17:49:28 +01003411.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200342------------------------
343
344Line 1 is the "response line". It is always composed of 3 fields :
345
346 - a version tag : HTTP/1.1
347 - a status code : 200
348 - a reason : OK
349
350The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100351 - 1xx = informational message to be skipped (e.g. 100, 101)
352 - 2xx = OK, content is following (e.g. 200, 206)
353 - 3xx = OK, no content following (e.g. 302, 304)
354 - 4xx = error caused by the client (e.g. 401, 403, 404)
355 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200356
Lukas Tribus23953682017-04-28 13:24:30 +0000357Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100358"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200359found there, but it's a common practice to respect the well-established
360messages. It can be composed of one or multiple words, such as "OK", "Found",
361or "Authentication Required".
362
Davor Ocelice9ed2812017-12-25 17:49:28 +0100363HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200364
365 Code When / reason
366 200 access to stats page, and when replying to monitoring requests
367 301 when performing a redirection, depending on the configured code
368 302 when performing a redirection, depending on the configured code
369 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100370 307 when performing a redirection, depending on the configured code
371 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200372 400 for an invalid or too large request
373 401 when an authentication is required to perform the action (when
374 accessing the stats page)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200375 403 when a request is forbidden by a "http-request deny" rule
Florian Tham9205fea2020-01-08 13:35:30 +0100376 404 when the requested resource could not be found
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200377 408 when the request timeout strikes before the request is complete
Florian Tham272e29b2020-01-08 10:19:05 +0100378 410 when the requested resource is no longer available and will not
379 be available again
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200380 500 when haproxy encounters an unrecoverable internal error, such as a
381 memory allocation failure, which should never happen
Christopher Faulete095f312020-12-07 11:22:24 +0100382 501 when haproxy is unable to satisfy a client request because of an
383 unsupported feature
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200384 502 when the server returns an empty, invalid or incomplete response, or
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200385 when an "http-response deny" rule blocks the response.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200386 503 when no server was available to handle the request, or in response to
387 monitoring requests which match the "monitor fail" condition
388 504 when the response timeout strikes before the server responds
389
390The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3914.2).
392
393
3941.3.2. The response headers
395---------------------------
396
397Response headers work exactly like request headers, and as such, HAProxy uses
398the same parsing function for both. Please refer to paragraph 1.2.2 for more
399details.
400
401
4022. Configuring HAProxy
403----------------------
404
4052.1. Configuration file format
406------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200407
408HAProxy's configuration process involves 3 major sources of parameters :
409
410 - the arguments from the command-line, which always take precedence
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100411 - the configuration file(s), whose format is described here
Thayne McCombscdbcca92021-01-07 21:24:41 -0700412 - the running process's environment, in case some environment variables are
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100413 explicitly referenced
Willy Tarreau6a06a402007-07-15 20:15:28 +0200414
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100415The configuration file follows a fairly simple hierarchical format which obey
416a few basic rules:
Willy Tarreau0ba27502007-12-24 16:55:16 +0100417
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100418 1. a configuration file is an ordered sequence of statements
419
420 2. a statement is a single non-empty line before any unprotected "#" (hash)
421
422 3. a line is a series of tokens or "words" delimited by unprotected spaces or
423 tab characters
424
425 4. the first word or sequence of words of a line is one of the keywords or
426 keyword sequences listed in this document
427
428 5. all other words are all arguments of the first one, some being well-known
429 keywords listed in this document, others being values, references to other
430 parts of the configuration, or expressions
431
432 6. certain keywords delimit a section inside which only a subset of keywords
433 are supported
434
435 7. a section ends at the end of a file or on a special keyword starting a new
436 section
437
438This is all that is needed to know to write a simple but reliable configuration
439generator, but this is not enough to reliably parse any configuration nor to
440figure how to deal with certain corner cases.
441
442First, there are a few consequences of the rules above. Rule 6 and 7 imply that
443the keywords used to define a new section are valid everywhere and cannot have
444a different meaning in a specific section. These keywords are always a single
445word (as opposed to a sequence of words), and traditionally the section that
446follows them is designated using the same name. For example when speaking about
447the "global section", it designates the section of configuration that follows
448the "global" keyword. This usage is used a lot in error messages to help locate
449the parts that need to be addressed.
450
451A number of sections create an internal object or configuration space, which
452requires to be distinguished from other ones. In this case they will take an
453extra word which will set the name of this particular section. For some of them
454the section name is mandatory. For example "frontend foo" will create a new
455section of type "frontend" named "foo". Usually a name is specific to its
456section and two sections of different types may use the same name, but this is
457not recommended as it tends to complexify configuration management.
458
459A direct consequence of rule 7 is that when multiple files are read at once,
460each of them must start with a new section, and the end of each file will end
461a section. A file cannot contain sub-sections nor end an existing section and
462start a new one.
463
464Rule 1 mentioned that ordering matters. Indeed, some keywords create directives
465that can be repeated multiple times to create ordered sequences of rules to be
466applied in a certain order. For example "tcp-request" can be used to alternate
467"accept" and "reject" rules on varying criteria. As such, a configuration file
468processor must always preserve a section's ordering when editing a file. The
469ordering of sections usually does not matter except for the global section
470which must be placed before other sections, but it may be repeated if needed.
471In addition, some automatic identifiers may automatically be assigned to some
472of the created objects (e.g. proxies), and by reordering sections, their
473identifiers will change. These ones appear in the statistics for example. As
474such, the configuration below will assign "foo" ID number 1 and "bar" ID number
4752, which will be swapped if the two sections are reversed:
476
477 listen foo
478 bind :80
479
480 listen bar
481 bind :81
482
483Another important point is that according to rules 2 and 3 above, empty lines,
484spaces, tabs, and comments following and unprotected "#" character are not part
485of the configuration as they are just used as delimiters. This implies that the
486following configurations are strictly equivalent:
487
488 global#this is the global section
489 daemon#daemonize
490 frontend foo
491 mode http # or tcp
492
493and:
494
495 global
496 daemon
497
498 # this is the public web frontend
499 frontend foo
500 mode http
501
502The common practice is to align to the left only the keyword that initiates a
503new section, and indent (i.e. prepend a tab character or a few spaces) all
504other keywords so that it's instantly visible that they belong to the same
505section (as done in the second example above). Placing comments before a new
506section helps the reader decide if it's the desired one. Leaving a blank line
507at the end of a section also visually helps spotting the end when editing it.
508
509Tabs are very convenient for indent but they do not copy-paste well. If spaces
510are used instead, it is recommended to avoid placing too many (2 to 4) so that
511editing in field doesn't become a burden with limited editors that do not
512support automatic indent.
513
514In the early days it used to be common to see arguments split at fixed tab
515positions because most keywords would not take more than two arguments. With
516modern versions featuring complex expressions this practice does not stand
517anymore, and is not recommended.
518
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200519
William Lallemandf9873ba2015-05-05 17:37:14 +02005202.2. Quoting and escaping
521-------------------------
522
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100523In modern configurations, some arguments require the use of some characters
524that were previously considered as pure delimiters. In order to make this
525possible, HAProxy supports character escaping by prepending a backslash ('\')
526in front of the character to be escaped, weak quoting within double quotes
527('"') and strong quoting within single quotes ("'").
William Lallemandf9873ba2015-05-05 17:37:14 +0200528
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100529This is pretty similar to what is done in a number of programming languages and
530very close to what is commonly encountered in Bourne shell. The principle is
531the following: while the configuration parser cuts the lines into words, it
532also takes care of quotes and backslashes to decide whether a character is a
533delimiter or is the raw representation of this character within the current
534word. The escape character is then removed, the quotes are removed, and the
535remaining word is used as-is as a keyword or argument for example.
William Lallemandf9873ba2015-05-05 17:37:14 +0200536
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100537If a backslash is needed in a word, it must either be escaped using itself
538(i.e. double backslash) or be strongly quoted.
539
540Escaping outside quotes is achieved by preceding a special character by a
541backslash ('\'):
William Lallemandf9873ba2015-05-05 17:37:14 +0200542
543 \ to mark a space and differentiate it from a delimiter
544 \# to mark a hash and differentiate it from a comment
545 \\ to use a backslash
546 \' to use a single quote and differentiate it from strong quoting
547 \" to use a double quote and differentiate it from weak quoting
548
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100549In addition, a few non-printable characters may be emitted using their usual
550C-language representation:
551
552 \n to insert a line feed (LF, character \x0a or ASCII 10 decimal)
553 \r to insert a carriage return (CR, character \x0d or ASCII 13 decimal)
554 \t to insert a tab (character \x09 or ASCII 9 decimal)
555 \xNN to insert character having ASCII code hex NN (e.g \x0a for LF).
556
557Weak quoting is achieved by surrounding double quotes ("") around the character
558or sequence of characters to protect. Weak quoting prevents the interpretation
559of:
William Lallemandf9873ba2015-05-05 17:37:14 +0200560
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100561 space or tab as a word separator
William Lallemandf9873ba2015-05-05 17:37:14 +0200562 ' single quote as a strong quoting delimiter
563 # hash as a comment start
564
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100565Weak quoting permits the interpretation of environment variables (which are not
566evaluated outside of quotes) by preceding them with a dollar sign ('$'). If a
567dollar character is needed inside double quotes, it must be escaped using a
568backslash.
William Lallemandb2f07452015-05-12 14:27:13 +0200569
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100570Strong quoting is achieved by surrounding single quotes ('') around the
571character or sequence of characters to protect. Inside single quotes, nothing
572is interpreted, it's the efficient way to quote regular expressions.
William Lallemandf9873ba2015-05-05 17:37:14 +0200573
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100574As a result, here is the matrix indicating how special characters can be
575entered in different contexts (unprintable characters are replaced with their
576name within angle brackets). Note that some characters that may only be
577represented escaped have no possible representation inside single quotes,
578hence the '-' there:
William Lallemandf9873ba2015-05-05 17:37:14 +0200579
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100580 Character | Unquoted | Weakly quoted | Strongly quoted
581 -----------+---------------+-----------------------------+-----------------
582 <TAB> | \<TAB>, \x09 | "<TAB>", "\<TAB>", "\x09" | '<TAB>'
583 <LF> | \n, \x0a | "\n", "\x0a" | -
584 <CR> | \r, \x0d | "\r", "\x0d" | -
585 <SPC> | \<SPC>, \x20 | "<SPC>", "\<SPC>", "\x20" | '<SPC>'
586 " | \", \x22 | "\"", "\x22" | '"'
587 # | \#, \x23 | "#", "\#", "\x23" | '#'
588 $ | $, \$, \x24 | "\$", "\x24" | '$'
589 ' | \', \x27 | "'", "\'", "\x27" | -
590 \ | \\, \x5c | "\\", "\x5c" | '\'
William Lallemandf9873ba2015-05-05 17:37:14 +0200591
592 Example:
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100593 # those are all strictly equivalent:
William Lallemandf9873ba2015-05-05 17:37:14 +0200594 log-format %{+Q}o\ %t\ %s\ %{-Q}r
595 log-format "%{+Q}o %t %s %{-Q}r"
596 log-format '%{+Q}o %t %s %{-Q}r'
597 log-format "%{+Q}o %t"' %s %{-Q}r'
598 log-format "%{+Q}o %t"' %s'\ %{-Q}r
599
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100600There is one particular case where a second level of quoting or escaping may be
601necessary. Some keywords take arguments within parenthesis, sometimes delimited
602by commas. These arguments are commonly integers or predefined words, but when
603they are arbitrary strings, it may be required to perform a separate level of
604escaping to disambiguate the characters that belong to the argument from the
605characters that are used to delimit the arguments themselves. A pretty common
606case is the "regsub" converter. It takes a regular expression in argument, and
607if a closing parenthesis is needed inside, this one will require to have its
608own quotes.
609
610The keyword argument parser is exactly the same as the top-level one regarding
611quotes, except that is will not make special cases of backslashes. But what is
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500612not always obvious is that the delimiters used inside must first be escaped or
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100613quoted so that they are not resolved at the top level.
614
615Let's take this example making use of the "regsub" converter which takes 3
616arguments, one regular expression, one replacement string and one set of flags:
617
618 # replace all occurrences of "foo" with "blah" in the path:
619 http-request set-path %[path,regsub(foo,blah,g)]
620
621Here no special quoting was necessary. But if now we want to replace either
622"foo" or "bar" with "blah", we'll need the regular expression "(foo|bar)". We
623cannot write:
624
625 http-request set-path %[path,regsub((foo|bar),blah,g)]
626
627because we would like the string to cut like this:
628
629 http-request set-path %[path,regsub((foo|bar),blah,g)]
630 |---------|----|-|
631 arg1 _/ / /
632 arg2 __________/ /
633 arg3 ______________/
634
635but actually what is passed is a string between the opening and closing
636parenthesis then garbage:
637
638 http-request set-path %[path,regsub((foo|bar),blah,g)]
639 |--------|--------|
640 arg1=(foo|bar _/ /
641 trailing garbage _________/
642
643The obvious solution here seems to be that the closing parenthesis needs to be
644quoted, but alone this will not work, because as mentioned above, quotes are
645processed by the top-level parser which will resolve them before processing
646this word:
647
648 http-request set-path %[path,regsub("(foo|bar)",blah,g)]
649 ------------ -------- ----------------------------------
650 word1 word2 word3=%[path,regsub((foo|bar),blah,g)]
651
652So we didn't change anything for the argument parser at the second level which
653still sees a truncated regular expression as the only argument, and garbage at
654the end of the string. By escaping the quotes they will be passed unmodified to
655the second level:
656
657 http-request set-path %[path,regsub(\"(foo|bar)\",blah,g)]
658 ------------ -------- ------------------------------------
659 word1 word2 word3=%[path,regsub("(foo|bar)",blah,g)]
660 |---------||----|-|
661 arg1=(foo|bar) _/ / /
662 arg2=blah ___________/ /
663 arg3=g _______________/
664
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500665Another approach consists in using single quotes outside the whole string and
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100666double quotes inside (so that the double quotes are not stripped again):
667
668 http-request set-path '%[path,regsub("(foo|bar)",blah,g)]'
669 ------------ -------- ----------------------------------
670 word1 word2 word3=%[path,regsub("(foo|bar)",blah,g)]
671 |---------||----|-|
672 arg1=(foo|bar) _/ / /
673 arg2 ___________/ /
674 arg3 _______________/
675
676When using regular expressions, it can happen that the dollar ('$') character
677appears in the expression or that a backslash ('\') is used in the replacement
678string. In this case these ones will also be processed inside the double quotes
679thus single quotes are preferred (or double escaping). Example:
680
681 http-request set-path '%[path,regsub("^/(here)(/|$)","my/\1",g)]'
682 ------------ -------- -----------------------------------------
683 word1 word2 word3=%[path,regsub("^/(here)(/|$)","my/\1",g)]
684 |-------------| |-----||-|
685 arg1=(here)(/|$) _/ / /
686 arg2=my/\1 ________________/ /
687 arg3 ______________________/
688
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500689Remember that backslahes are not escape characters within single quotes and
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100690that the whole word3 above is already protected against them using the single
691quotes. Conversely, if double quotes had been used around the whole expression,
692single the dollar character and the backslashes would have been resolved at top
693level, breaking the argument contents at the second level.
694
695When in doubt, simply do not use quotes anywhere, and start to place single or
696double quotes around arguments that require a comma or a closing parenthesis,
697and think about escaping these quotes using a backslash of the string contains
698a dollar or a backslash. Again, this is pretty similar to what is used under
699a Bourne shell when double-escaping a command passed to "eval". For API writers
700the best is probably to place escaped quotes around each and every argument,
701regardless of their contents. Users will probably find that using single quotes
702around the whole expression and double quotes around each argument provides
703more readable configurations.
William Lallemandf9873ba2015-05-05 17:37:14 +0200704
705
William Lallemandb2f07452015-05-12 14:27:13 +02007062.3. Environment variables
707--------------------------
708
709HAProxy's configuration supports environment variables. Those variables are
710interpreted only within double quotes. Variables are expanded during the
711configuration parsing. Variable names must be preceded by a dollar ("$") and
712optionally enclosed with braces ("{}") similarly to what is done in Bourne
713shell. Variable names can contain alphanumerical characters or the character
Amaury Denoyellefa41cb62020-10-01 14:32:35 +0200714underscore ("_") but should not start with a digit. If the variable contains a
715list of several values separated by spaces, it can be expanded as individual
716arguments by enclosing the variable with braces and appending the suffix '[*]'
717before the closing brace.
William Lallemandb2f07452015-05-12 14:27:13 +0200718
719 Example:
720
721 bind "fd@${FD_APP1}"
722
723 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
724
725 user "$HAPROXY_USER"
726
William Lallemand4d03e432019-06-14 15:35:37 +0200727Some variables are defined by HAProxy, they can be used in the configuration
728file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200729
William Lallemand4d03e432019-06-14 15:35:37 +0200730* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
731 name of the local peer. (See "-L" in the management guide.)
732
733* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
734 separated by semicolons. Can be useful in the case you specified a
735 directory.
736
737* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
738
John Roeslerfb2fce12019-07-10 15:45:51 -0500739* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200740 processes, separated by semicolons.
741
John Roeslerfb2fce12019-07-10 15:45:51 -0500742* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200743 CLI, separated by semicolons.
744
745See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200746
Willy Tarreau4b103022021-02-12 17:59:10 +0100747
7482.4. Conditional blocks
749-----------------------
750
751It may sometimes be convenient to be able to conditionally enable or disable
752some arbitrary parts of the configuration, for example to enable/disable SSL or
753ciphers, enable or disable some pre-production listeners without modifying the
754configuration, or adjust the configuration's syntax to support two distinct
755versions of HAProxy during a migration.. HAProxy brings a set of nestable
756preprocessor-like directives which allow to integrate or ignore some blocks of
757text. These directives must be placed on their own line and they act on the
758lines that follow them. Two of them support an expression, the other ones only
759switch to an alternate block or end a current level. The 4 following directives
760are defined to form conditional blocks:
761
762 - .if <condition>
763 - .elif <condition>
764 - .else
765 - .endif
766
767The ".if" directive nests a new level, ".elif" stays at the same level, ".else"
768as well, and ".endif" closes a level. Each ".if" must be terminated by a
769matching ".endif". The ".elif" may only be placed after ".if" or ".elif", and
770there is no limit to the number of ".elif" that may be chained. There may be
771only one ".else" per ".if" and it must always be after the ".if" or the last
772".elif" of a block.
773
774Comments may be placed on the same line if needed after a '#', they will be
775ignored. The directives are tokenized like other configuration directives, and
776as such it is possible to use environment variables in conditions.
777
778The conditions are currently limited to:
779
780 - an empty string, always returns "false"
781 - the integer zero ('0'), always returns "false"
782 - a non-nul integer (e.g. '1'), always returns "true".
783
784Other patterns are not supported yet but the purpose is to bring a few
785functions to test for certain build options and supported features.
786
787Three other directives are provided to report some status:
788
789 - .notice "message" : emit this message at level NOTICE
790 - .warning "message" : emit this message at level WARNING
791 - .alert "message" : emit this message at level ALERT
792
793Messages emitted at level WARNING may cause the process to fail to start if the
794"strict-mode" is enabled. Messages emitted at level ALERT will always cause a
795fatal error. These can be used to detect some inappropriate conditions and
796provide advice to the user.
797
798Example:
799
800 .if "${A}"
801 .if "${B}"
802 .notice "A=1, B=1"
803 .elif "${C}"
804 .notice "A=1, B=0, C=1"
805 .elif "${D}"
806 .warning "A=1, B=0, C=0, D=1"
807 .else
808 .alert "A=1, B=0, C=0, D=0"
809 .endif
810 .else
811 .notice "A=0"
812 .endif
813
814
8152.5. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200816----------------
817
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100818Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100819values are generally expressed in milliseconds (unless explicitly stated
820otherwise) but may be expressed in any other unit by suffixing the unit to the
821numeric value. It is important to consider this because it will not be repeated
822for every keyword. Supported units are :
823
824 - us : microseconds. 1 microsecond = 1/1000000 second
825 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
826 - s : seconds. 1s = 1000ms
827 - m : minutes. 1m = 60s = 60000ms
828 - h : hours. 1h = 60m = 3600s = 3600000ms
829 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
830
831
Willy Tarreau4b103022021-02-12 17:59:10 +01008322.6. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200833-------------
834
835 # Simple configuration for an HTTP proxy listening on port 80 on all
836 # interfaces and forwarding requests to a single backend "servers" with a
837 # single server "server1" listening on 127.0.0.1:8000
838 global
839 daemon
840 maxconn 256
841
842 defaults
843 mode http
844 timeout connect 5000ms
845 timeout client 50000ms
846 timeout server 50000ms
847
848 frontend http-in
849 bind *:80
850 default_backend servers
851
852 backend servers
853 server server1 127.0.0.1:8000 maxconn 32
854
855
856 # The same configuration defined with a single listen block. Shorter but
857 # less expressive, especially in HTTP mode.
858 global
859 daemon
860 maxconn 256
861
862 defaults
863 mode http
864 timeout connect 5000ms
865 timeout client 50000ms
866 timeout server 50000ms
867
868 listen http-in
869 bind *:80
870 server server1 127.0.0.1:8000 maxconn 32
871
872
873Assuming haproxy is in $PATH, test these configurations in a shell with:
874
Willy Tarreauccb289d2010-12-11 20:19:38 +0100875 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200876
877
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008783. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200879--------------------
880
881Parameters in the "global" section are process-wide and often OS-specific. They
882are generally set once for all and do not need being changed once correct. Some
883of them have command-line equivalents.
884
885The following keywords are supported in the "global" section :
886
887 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200888 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200889 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200890 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200891 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200892 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200893 - description
894 - deviceatlas-json-file
895 - deviceatlas-log-level
896 - deviceatlas-separator
897 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900898 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200899 - gid
900 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100901 - hard-stop-after
Christopher Faulet98fbe952019-07-22 16:18:24 +0200902 - h1-case-adjust
903 - h1-case-adjust-file
Willy Tarreaud96f1122019-12-03 07:07:36 +0100904 - insecure-fork-wanted
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100905 - insecure-setuid-wanted
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100906 - issuers-chain-path
Dragan Dosen13cd54c2020-06-18 18:24:05 +0200907 - localpeer
Willy Tarreau6a06a402007-07-15 20:15:28 +0200908 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200909 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100910 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200911 - lua-load
Thierry Fournier59f11be2020-11-29 00:37:41 +0100912 - lua-load-per-thread
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +0100913 - lua-prepend-path
William Lallemand27edc4b2019-05-07 17:49:33 +0200914 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200915 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200916 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200917 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200918 - pidfile
Willy Tarreau119e50e2020-05-22 13:53:29 +0200919 - pp2-never-send-local
Willy Tarreau1d549722016-02-16 12:41:57 +0100920 - presetenv
921 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200922 - uid
923 - ulimit-n
924 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200925 - set-dumpable
Willy Tarreau13d2ba22021-03-26 11:38:08 +0100926 - set-var
Willy Tarreau1d549722016-02-16 12:41:57 +0100927 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200928 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200929 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200930 - ssl-default-bind-ciphersuites
Jerome Magninb203ff62020-04-03 15:28:22 +0200931 - ssl-default-bind-curves
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200932 - ssl-default-bind-options
933 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200934 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200935 - ssl-default-server-options
936 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100937 - ssl-server-verify
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +0200938 - ssl-skip-self-issued-ca
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100939 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100940 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100941 - 51degrees-data-file
942 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200943 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200944 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200945 - wurfl-data-file
946 - wurfl-information-list
947 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200948 - wurfl-cache-size
William Dauchy0fec3ab2019-10-27 20:08:11 +0100949 - strict-limits
Willy Tarreaud72758d2010-01-12 10:42:19 +0100950
Willy Tarreau6a06a402007-07-15 20:15:28 +0200951 * Performance tuning
William Dauchy0a8824f2019-10-27 20:08:09 +0100952 - busy-polling
Willy Tarreau1746eec2014-04-25 10:46:47 +0200953 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200954 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200955 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100956 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100957 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100958 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200959 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200960 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200961 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200962 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200963 - noepoll
964 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000965 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200966 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100967 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300968 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000969 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100970 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200971 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200972 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200973 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000974 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000975 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200976 - tune.buffers.limit
977 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200978 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200979 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100980 - tune.comp.maxlevel
Willy Tarreaubc52bec2020-06-18 08:58:47 +0200981 - tune.fd.edge-triggered
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200982 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200983 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200984 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100985 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200986 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200987 - tune.http.maxhdr
Willy Tarreau76cc6992020-07-01 18:49:24 +0200988 - tune.idle-pool.shared
Willy Tarreau7e312732014-02-12 16:35:14 +0100989 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100990 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100991 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100992 - tune.lua.session-timeout
993 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200994 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100995 - tune.maxaccept
996 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200997 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200998 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200999 - tune.pipesize
Willy Tarreaua8e2d972020-07-01 18:27:16 +02001000 - tune.pool-high-fd-ratio
1001 - tune.pool-low-fd-ratio
Willy Tarreaue803de22010-01-21 17:43:04 +01001002 - tune.rcvbuf.client
1003 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +01001004 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +02001005 - tune.runqueue-depth
Willy Tarreaue7723bd2020-06-24 11:11:02 +02001006 - tune.sched.low-latency
Willy Tarreaue803de22010-01-21 17:43:04 +01001007 - tune.sndbuf.client
1008 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001009 - tune.ssl.cachesize
William Lallemand7d42ef52020-07-06 11:41:30 +02001010 - tune.ssl.keylog
Willy Tarreaubfd59462013-02-21 07:46:09 +01001011 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +02001012 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +01001013 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001014 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +02001015 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001016 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001017 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +01001018 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001019 - tune.vars.reqres-max-size
1020 - tune.vars.sess-max-size
1021 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +01001022 - tune.zlib.memlevel
1023 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +01001024
Willy Tarreau6a06a402007-07-15 20:15:28 +02001025 * Debugging
Willy Tarreau6a06a402007-07-15 20:15:28 +02001026 - quiet
Willy Tarreau3eb10b82020-04-15 16:42:39 +02001027 - zero-warning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001028
1029
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010303.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +02001031------------------------------------
1032
Emeric Brunc8e8d122012-10-02 18:42:10 +02001033ca-base <dir>
1034 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +01001035 relative path is used with "ca-file", "ca-verify-file" or "crl-file"
1036 directives. Absolute locations specified in "ca-file", "ca-verify-file" and
1037 "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +02001038
Willy Tarreau6a06a402007-07-15 20:15:28 +02001039chroot <jail dir>
1040 Changes current directory to <jail dir> and performs a chroot() there before
1041 dropping privileges. This increases the security level in case an unknown
1042 vulnerability would be exploited, since it would make it very hard for the
1043 attacker to exploit the system. This only works when the process is started
1044 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +01001045 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +01001046
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001047cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
1048 On Linux 2.6 and above, it is possible to bind a process or a thread to a
1049 specific CPU set. This means that the process or the thread will never run on
1050 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
1051 sets. The first argument is a process set, eventually followed by a thread
1052 set. These sets have the format
1053
1054 all | odd | even | number[-[number]]
1055
1056 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +01001057 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001058 ignored. It is possible to specify a range with two such number delimited by
1059 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +01001060 "all", only odd numbers using "odd" or even numbers using "even", just like
1061 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +01001062 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 +01001063 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001064 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001065 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001066 specified. Each "cpu-map" directive will replace the previous ones when they
1067 overlap. A thread will be bound on the intersection of its mapping and the
1068 one of the process on which it is attached. If the intersection is null, no
1069 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +01001070
Christopher Fauletff4121f2017-11-22 16:38:49 +01001071 Ranges can be partially defined. The higher bound can be omitted. In such
1072 case, it is replaced by the corresponding maximum value, 32 or 64 depending
1073 on the machine's word size.
1074
Christopher Faulet26028f62017-11-22 15:01:51 +01001075 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001076 automatically bind a process or a thread to a CPU by incrementing
1077 process/thread and CPU sets. To be valid, both sets must have the same
1078 size. No matter the declaration order of the CPU sets, it will be bound from
1079 the lowest to the highest bound. Having a process and a thread range with the
1080 "auto:" prefix is not supported. Only one range is supported, the other one
1081 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +01001082
1083 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001084 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
1085
1086 cpu-map 1/all 0-3 # bind all threads of the first process on the
1087 # first 4 CPUs
1088
1089 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
1090 # or "cpu-map 1-32 0-31" depending on the machine's
1091 # word size.
1092
Christopher Faulet26028f62017-11-22 15:01:51 +01001093 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001094 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +01001095 cpu-map auto:1-4 0-3
1096 cpu-map auto:1-4 0-1 2-3
1097 cpu-map auto:1-4 3 2 1 0
1098
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001099 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
1100 # and so on.
1101 cpu-map auto:1/1-4 0-3
1102 cpu-map auto:1/1-4 0-1 2-3
1103 cpu-map auto:1/1-4 3 2 1 0
1104
Davor Ocelice9ed2812017-12-25 17:49:28 +01001105 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +01001106 cpu-map auto:all 0-63
1107 cpu-map auto:even 0-31
1108 cpu-map auto:odd 32-63
1109
1110 # invalid cpu-map because process and CPU sets have different sizes.
1111 cpu-map auto:1-4 0 # invalid
1112 cpu-map auto:1 0-3 # invalid
1113
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001114 # invalid cpu-map because automatic binding is used with a process range
1115 # and a thread range.
1116 cpu-map auto:all/all 0 # invalid
1117 cpu-map auto:all/1-4 0 # invalid
1118 cpu-map auto:1-4/all 0 # invalid
1119
Emeric Brunc8e8d122012-10-02 18:42:10 +02001120crt-base <dir>
1121 Assigns a default directory to fetch SSL certificates from when a relative
William Dauchy238ea3b2020-01-11 13:09:12 +01001122 path is used with "crtfile" or "crt" directives. Absolute locations specified
1123 prevail and ignore "crt-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +02001124
Willy Tarreau6a06a402007-07-15 20:15:28 +02001125daemon
1126 Makes the process fork into background. This is the recommended mode of
1127 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +01001128 disabled by the command line "-db" argument. This option is ignored in
1129 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001130
David Carlier8167f302015-06-01 13:50:06 +02001131deviceatlas-json-file <path>
1132 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001133 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +02001134
1135deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001136 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +02001137 optional and set to 0 by default if not set.
1138
1139deviceatlas-separator <char>
1140 Sets the character separator for the API properties results. This directive
1141 is optional and set to | by default if not set.
1142
Cyril Bonté0306c4a2015-10-26 22:37:38 +01001143deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001144 Sets the client cookie's name used for the detection if the DeviceAtlas
1145 Client-side component was used during the request. This directive is optional
1146 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +01001147
Simon Horman98637e52014-06-20 12:30:16 +09001148external-check
Willy Tarreaud96f1122019-12-03 07:07:36 +01001149 Allows the use of an external agent to perform health checks. This is
1150 disabled by default as a security precaution, and even when enabled, checks
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001151 may still fail unless "insecure-fork-wanted" is enabled as well. If the
1152 program launched makes use of a setuid executable (it should really not),
1153 you may also need to set "insecure-setuid-wanted" in the global section.
1154 See "option external-check", and "insecure-fork-wanted", and
1155 "insecure-setuid-wanted".
Simon Horman98637e52014-06-20 12:30:16 +09001156
Willy Tarreau6a06a402007-07-15 20:15:28 +02001157gid <number>
Thayne McCombscdbcca92021-01-07 21:24:41 -07001158 Changes the process's group ID to <number>. It is recommended that the group
Willy Tarreau6a06a402007-07-15 20:15:28 +02001159 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1160 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +01001161 Note that if haproxy is started from a user having supplementary groups, it
1162 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001163 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +01001164
Willy Tarreau11770ce2019-12-03 08:29:22 +01001165group <group name>
1166 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
1167 See also "gid" and "user".
1168
Cyril Bonté203ec5a2017-03-23 22:44:13 +01001169hard-stop-after <time>
1170 Defines the maximum time allowed to perform a clean soft-stop.
1171
1172 Arguments :
1173 <time> is the maximum time (by default in milliseconds) for which the
1174 instance will remain alive when a soft-stop is received via the
1175 SIGUSR1 signal.
1176
1177 This may be used to ensure that the instance will quit even if connections
1178 remain opened during a soft-stop (for example with long timeouts for a proxy
1179 in tcp mode). It applies both in TCP and HTTP mode.
1180
1181 Example:
1182 global
1183 hard-stop-after 30s
1184
Christopher Faulet98fbe952019-07-22 16:18:24 +02001185h1-case-adjust <from> <to>
1186 Defines the case adjustment to apply, when enabled, to the header name
1187 <from>, to change it to <to> before sending it to HTTP/1 clients or
1188 servers. <from> must be in lower case, and <from> and <to> must not differ
1189 except for their case. It may be repeated if several header names need to be
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05001190 adjusted. Duplicate entries are not allowed. If a lot of header names have to
Christopher Faulet98fbe952019-07-22 16:18:24 +02001191 be adjusted, it might be more convenient to use "h1-case-adjust-file".
1192 Please note that no transformation will be applied unless "option
1193 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
1194 specified in a proxy.
1195
1196 There is no standard case for header names because, as stated in RFC7230,
1197 they are case-insensitive. So applications must handle them in a case-
1198 insensitive manner. But some bogus applications violate the standards and
1199 erroneously rely on the cases most commonly used by browsers. This problem
1200 becomes critical with HTTP/2 because all header names must be exchanged in
1201 lower case, and HAProxy follows the same convention. All header names are
1202 sent in lower case to clients and servers, regardless of the HTTP version.
1203
1204 Applications which fail to properly process requests or responses may require
1205 to temporarily use such workarounds to adjust header names sent to them for
1206 the time it takes the application to be fixed. Please note that an
1207 application which requires such workarounds might be vulnerable to content
1208 smuggling attacks and must absolutely be fixed.
1209
1210 Example:
1211 global
1212 h1-case-adjust content-length Content-Length
1213
1214 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
1215 "option h1-case-adjust-bogus-server".
1216
1217h1-case-adjust-file <hdrs-file>
1218 Defines a file containing a list of key/value pairs used to adjust the case
1219 of some header names before sending them to HTTP/1 clients or servers. The
1220 file <hdrs-file> must contain 2 header names per line. The first one must be
1221 in lower case and both must not differ except for their case. Lines which
1222 start with '#' are ignored, just like empty lines. Leading and trailing tabs
1223 and spaces are stripped. Duplicate entries are not allowed. Please note that
1224 no transformation will be applied unless "option h1-case-adjust-bogus-client"
1225 or "option h1-case-adjust-bogus-server" is specified in a proxy.
1226
1227 If this directive is repeated, only the last one will be processed. It is an
1228 alternative to the directive "h1-case-adjust" if a lot of header names need
1229 to be adjusted. Please read the risks associated with using this.
1230
1231 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
1232 "option h1-case-adjust-bogus-server".
1233
Willy Tarreaud96f1122019-12-03 07:07:36 +01001234insecure-fork-wanted
1235 By default haproxy tries hard to prevent any thread and process creation
1236 after it starts. Doing so is particularly important when using Lua files of
1237 uncertain origin, and when experimenting with development versions which may
1238 still contain bugs whose exploitability is uncertain. And generally speaking
1239 it's good hygiene to make sure that no unexpected background activity can be
1240 triggered by traffic. But this prevents external checks from working, and may
1241 break some very specific Lua scripts which actively rely on the ability to
1242 fork. This option is there to disable this protection. Note that it is a bad
1243 idea to disable it, as a vulnerability in a library or within haproxy itself
1244 will be easier to exploit once disabled. In addition, forking from Lua or
1245 anywhere else is not reliable as the forked process may randomly embed a lock
1246 set by another thread and never manage to finish an operation. As such it is
1247 highly recommended that this option is never used and that any workload
1248 requiring such a fork be reconsidered and moved to a safer solution (such as
1249 agents instead of external checks). This option supports the "no" prefix to
1250 disable it.
1251
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001252insecure-setuid-wanted
1253 HAProxy doesn't need to call executables at run time (except when using
1254 external checks which are strongly recommended against), and is even expected
1255 to isolate itself into an empty chroot. As such, there basically is no valid
1256 reason to allow a setuid executable to be called without the user being fully
1257 aware of the risks. In a situation where haproxy would need to call external
1258 checks and/or disable chroot, exploiting a vulnerability in a library or in
1259 haproxy itself could lead to the execution of an external program. On Linux
1260 it is possible to lock the process so that any setuid bit present on such an
1261 executable is ignored. This significantly reduces the risk of privilege
1262 escalation in such a situation. This is what haproxy does by default. In case
1263 this causes a problem to an external check (for example one which would need
1264 the "ping" command), then it is possible to disable this protection by
1265 explicitly adding this directive in the global section. If enabled, it is
1266 possible to turn it back off by prefixing it with the "no" keyword.
1267
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +01001268issuers-chain-path <dir>
1269 Assigns a directory to load certificate chain for issuer completion. All
1270 files must be in PEM format. For certificates loaded with "crt" or "crt-list",
1271 if certificate chain is not included in PEM (also commonly known as
1272 intermediate certificate), haproxy will complete chain if the issuer of the
1273 certificate corresponds to the first certificate of the chain loaded with
1274 "issuers-chain-path".
1275 A "crt" file with PrivateKey+Certificate+IntermediateCA2+IntermediateCA1
1276 could be replaced with PrivateKey+Certificate. HAProxy will complete the
1277 chain if a file with IntermediateCA2+IntermediateCA1 is present in
1278 "issuers-chain-path" directory. All other certificates with the same issuer
1279 will share the chain in memory.
1280
Dragan Dosen13cd54c2020-06-18 18:24:05 +02001281localpeer <name>
1282 Sets the local instance's peer name. It will be ignored if the "-L"
1283 command line argument is specified or if used after "peers" section
1284 definitions. In such cases, a warning message will be emitted during
1285 the configuration parsing.
1286
1287 This option will also set the HAPROXY_LOCALPEER environment variable.
1288 See also "-L" in the management guide and "peers" section below.
1289
Jan Wagner3e678602020-12-17 22:22:32 +01001290log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001291 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +01001292 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +01001293 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +01001294 configured with "log global".
1295
1296 <address> can be one of:
1297
Willy Tarreau2769aa02007-12-27 18:26:09 +01001298 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +01001299 no port is specified, 514 is used by default (the standard syslog
1300 port).
1301
David du Colombier24bb5f52011-03-17 10:40:23 +01001302 - An IPv6 address followed by a colon and optionally a UDP port. If
1303 no port is specified, 514 is used by default (the standard syslog
1304 port).
1305
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001306 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +01001307 considerations for chroot (be sure the path is accessible inside
1308 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +01001309 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +01001310
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001311 - A file descriptor number in the form "fd@<number>", which may point
1312 to a pipe, terminal, or socket. In this case unbuffered logs are used
1313 and one writev() call per log is performed. This is a bit expensive
1314 but acceptable for most workloads. Messages sent this way will not be
1315 truncated but may be dropped, in which case the DroppedLogs counter
1316 will be incremented. The writev() call is atomic even on pipes for
1317 messages up to PIPE_BUF size, which POSIX recommends to be at least
1318 512 and which is 4096 bytes on most modern operating systems. Any
1319 larger message may be interleaved with messages from other processes.
1320 Exceptionally for debugging purposes the file descriptor may also be
1321 directed to a file, but doing so will significantly slow haproxy down
1322 as non-blocking calls will be ignored. Also there will be no way to
1323 purge nor rotate this file without restarting the process. Note that
1324 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +01001325 for use with a TCP syslog server. See also the "short" and "raw"
1326 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001327
1328 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
1329 "fd@2", see above.
1330
Willy Tarreauc046d162019-08-30 15:24:59 +02001331 - A ring buffer in the form "ring@<name>", which will correspond to an
1332 in-memory ring buffer accessible over the CLI using the "show events"
1333 command, which will also list existing rings and their sizes. Such
1334 buffers are lost on reload or restart but when used as a complement
1335 this can help troubleshooting by having the logs instantly available.
1336
William Lallemandb2f07452015-05-12 14:27:13 +02001337 You may want to reference some environment variables in the address
1338 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001339
Willy Tarreau18324f52014-06-27 18:10:07 +02001340 <length> is an optional maximum line length. Log lines larger than this value
1341 will be truncated before being sent. The reason is that syslog
1342 servers act differently on log line length. All servers support the
1343 default value of 1024, but some servers simply drop larger lines
1344 while others do log them. If a server supports long lines, it may
1345 make sense to set this value here in order to avoid truncating long
1346 lines. Similarly, if a server drops long lines, it is preferable to
1347 truncate them before sending them. Accepted values are 80 to 65535
1348 inclusive. The default value of 1024 is generally fine for all
1349 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +01001350 JSON-formatted logs may require larger values. You may also need to
1351 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +02001352
Dragan Dosen7ad31542015-09-28 17:16:47 +02001353 <format> is the log format used when generating syslog messages. It may be
1354 one of the following :
1355
Emeric Brun0237c4e2020-11-27 16:24:34 +01001356 local Analog to rfc3164 syslog message format except that hostname
1357 field is stripped. This is the default.
1358 Note: option "log-send-hostname" switches the default to
1359 rfc3164.
1360
1361 rfc3164 The RFC3164 syslog message format.
Dragan Dosen7ad31542015-09-28 17:16:47 +02001362 (https://tools.ietf.org/html/rfc3164)
1363
1364 rfc5424 The RFC5424 syslog message format.
1365 (https://tools.ietf.org/html/rfc5424)
1366
Emeric Brun54648852020-07-06 15:54:06 +02001367 priority A message containing only a level plus syslog facility between
1368 angle brackets such as '<63>', followed by the text. The PID,
1369 date, time, process name and system name are omitted. This is
1370 designed to be used with a local log server.
1371
Willy Tarreaue8746a02018-11-12 08:45:00 +01001372 short A message containing only a level between angle brackets such as
1373 '<3>', followed by the text. The PID, date, time, process name
1374 and system name are omitted. This is designed to be used with a
1375 local log server. This format is compatible with what the systemd
1376 logger consumes.
1377
Emeric Brun54648852020-07-06 15:54:06 +02001378 timed A message containing only a level between angle brackets such as
1379 '<3>', followed by ISO date and by the text. The PID, process
1380 name and system name are omitted. This is designed to be
1381 used with a local log server.
1382
1383 iso A message containing only the ISO date, followed by the text.
1384 The PID, process name and system name are omitted. This is
1385 designed to be used with a local log server.
1386
Willy Tarreauc1b06452018-11-12 11:57:56 +01001387 raw A message containing only the text. The level, PID, date, time,
1388 process name and system name are omitted. This is designed to be
1389 used in containers or during development, where the severity only
1390 depends on the file descriptor used (stdout/stderr).
1391
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001392 <ranges> A list of comma-separated ranges to identify the logs to sample.
1393 This is used to balance the load of the logs to send to the log
1394 server. The limits of the ranges cannot be null. They are numbered
1395 from 1. The size or period (in number of logs) of the sample must be
1396 set with <sample_size> parameter.
1397
1398 <sample_size>
1399 The size of the sample in number of logs to consider when balancing
1400 their logging loads. It is used to balance the load of the logs to
1401 send to the syslog server. This size must be greater or equal to the
1402 maximum of the high limits of the ranges.
1403 (see also <ranges> parameter).
1404
Robert Tsai81ae1952007-12-05 10:47:29 +01001405 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001406
Willy Tarreaue8746a02018-11-12 08:45:00 +01001407 kern user mail daemon auth syslog lpr news
1408 uucp cron auth2 ftp ntp audit alert cron2
1409 local0 local1 local2 local3 local4 local5 local6 local7
1410
Willy Tarreauc1b06452018-11-12 11:57:56 +01001411 Note that the facility is ignored for the "short" and "raw"
1412 formats, but still required as a positional field. It is
1413 recommended to use "daemon" in this case to make it clear that
1414 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001415
1416 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001417 all messages are sent. If a maximum level is specified, only messages with a
1418 severity at least as important as this level will be sent. An optional minimum
1419 level can be specified. If it is set, logs emitted with a more severe level
1420 than this one will be capped to this level. This is used to avoid sending
1421 "emerg" messages on all terminals on some default syslog configurations.
1422 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001423
Cyril Bontédc4d9032012-04-08 21:57:39 +02001424 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001425
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001426log-send-hostname [<string>]
1427 Sets the hostname field in the syslog header. If optional "string" parameter
1428 is set the header is set to the string contents, otherwise uses the hostname
1429 of the system. Generally used if one is not relaying logs through an
1430 intermediate syslog server or for simply customizing the hostname printed in
1431 the logs.
1432
Kevinm48936af2010-12-22 16:08:21 +00001433log-tag <string>
1434 Sets the tag field in the syslog header to this string. It defaults to the
1435 program name as launched from the command line, which usually is "haproxy".
1436 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001437 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001438
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001439lua-load <file>
Thierry Fournier59f11be2020-11-29 00:37:41 +01001440 This global directive loads and executes a Lua file in the shared context
1441 that is visible to all threads. Any variable set in such a context is visible
1442 from any thread. This is the easiest and recommended way to load Lua programs
1443 but it will not scale well if a lot of Lua calls are performed, as only one
1444 thread may be running on the global state at a time. A program loaded this
1445 way will always see 0 in the "core.thread" variable. This directive can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001446 used multiple times.
1447
Thierry Fournier59f11be2020-11-29 00:37:41 +01001448lua-load-per-thread <file>
1449 This global directive loads and executes a Lua file into each started thread.
1450 Any global variable has a thread-local visibility so that each thread could
1451 see a different value. As such it is strongly recommended not to use global
1452 variables in programs loaded this way. An independent copy is loaded and
1453 initialized for each thread, everything is done sequentially and in the
1454 thread's numeric order from 1 to nbthread. If some operations need to be
1455 performed only once, the program should check the "core.thread" variable to
1456 figure what thread is being initialized. Programs loaded this way will run
1457 concurrently on all threads and will be highly scalable. This is the
1458 recommended way to load simple functions that register sample-fetches,
1459 converters, actions or services once it is certain the program doesn't depend
1460 on global variables. For the sake of simplicity, the directive is available
1461 even if only one thread is used and even if threads are disabled (in which
1462 case it will be equivalent to lua-load). This directive can be used multiple
1463 times.
1464
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +01001465lua-prepend-path <string> [<type>]
1466 Prepends the given string followed by a semicolon to Lua's package.<type>
1467 variable.
1468 <type> must either be "path" or "cpath". If <type> is not given it defaults
1469 to "path".
1470
1471 Lua's paths are semicolon delimited lists of patterns that specify how the
1472 `require` function attempts to find the source file of a library. Question
1473 marks (?) within a pattern will be replaced by module name. The path is
1474 evaluated left to right. This implies that paths that are prepended later
1475 will be checked earlier.
1476
1477 As an example by specifying the following path:
1478
1479 lua-prepend-path /usr/share/haproxy-lua/?/init.lua
1480 lua-prepend-path /usr/share/haproxy-lua/?.lua
1481
1482 When `require "example"` is being called Lua will first attempt to load the
1483 /usr/share/haproxy-lua/example.lua script, if that does not exist the
1484 /usr/share/haproxy-lua/example/init.lua will be attempted and the default
1485 paths if that does not exist either.
1486
1487 See https://www.lua.org/pil/8.1.html for the details within the Lua
1488 documentation.
1489
William Lallemand4cfede82017-11-24 22:02:34 +01001490master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001491 Master-worker mode. It is equivalent to the command line "-W" argument.
1492 This mode will launch a "master" which will monitor the "workers". Using
1493 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001494 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001495 or daemon mode. It is recommended to use this mode with multiprocess and
1496 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001497 By default, if a worker exits with a bad return code, in the case of a
1498 segfault for example, all workers will be killed, and the master will leave.
1499 It is convenient to combine this behavior with Restart=on-failure in a
1500 systemd unit file in order to relaunch the whole process. If you don't want
1501 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001502
William Lallemand4cfede82017-11-24 22:02:34 +01001503 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001504
William Lallemand27edc4b2019-05-07 17:49:33 +02001505mworker-max-reloads <number>
1506 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001507 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001508 number of reloads is greater than this number, the worker will receive a
1509 SIGTERM. This option helps to keep under control the number of workers.
1510 See also "show proc" in the Management Guide.
1511
Willy Tarreauf42d7942020-10-20 11:54:49 +02001512nbproc <number> (deprecated)
Willy Tarreau6a06a402007-07-15 20:15:28 +02001513 Creates <number> processes when going daemon. This requires the "daemon"
1514 mode. By default, only one process is created, which is the recommended mode
1515 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001516 process, it may be needed to fork multiple daemons. When set to a value
1517 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreauf42d7942020-10-20 11:54:49 +02001518 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. This directive is deprecated
1519 and scheduled for removal in 2.5. Please use "nbthread" instead. See also
1520 "daemon" and "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001521
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001522nbthread <number>
1523 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001524 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1525 "nbproc" historically used to be the only way to use multiple processors, it
1526 also involved a number of shortcomings related to the lack of synchronization
1527 between processes (health-checks, peers, stick-tables, stats, ...) which do
1528 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001529 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1530 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1531 when nbproc is not used, the default "nbthread" value is automatically set to
1532 the number of CPUs the process is bound to upon startup. This means that the
1533 thread count can easily be adjusted from the calling process using commands
1534 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1535 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001536
Willy Tarreau6a06a402007-07-15 20:15:28 +02001537pidfile <pidfile>
MIZUTA Takeshic32f3942020-08-26 13:46:19 +09001538 Writes PIDs of all daemons into file <pidfile> when daemon mode or writes PID
1539 of master process into file <pidfile> when master-worker mode. This option is
1540 equivalent to the "-p" command line argument. The file must be accessible to
1541 the user starting the process. See also "daemon" and "master-worker".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001542
Willy Tarreau119e50e2020-05-22 13:53:29 +02001543pp2-never-send-local
1544 A bug in the PROXY protocol v2 implementation was present in HAProxy up to
1545 version 2.1, causing it to emit a PROXY command instead of a LOCAL command
1546 for health checks. This is particularly minor but confuses some servers'
1547 logs. Sadly, the bug was discovered very late and revealed that some servers
1548 which possibly only tested their PROXY protocol implementation against
1549 HAProxy fail to properly handle the LOCAL command, and permanently remain in
1550 the "down" state when HAProxy checks them. When this happens, it is possible
1551 to enable this global option to revert to the older (bogus) behavior for the
1552 time it takes to contact the affected components' vendors and get them fixed.
1553 This option is disabled by default and acts on all servers having the
1554 "send-proxy-v2" statement.
1555
Willy Tarreau1d549722016-02-16 12:41:57 +01001556presetenv <name> <value>
1557 Sets environment variable <name> to value <value>. If the variable exists, it
1558 is NOT overwritten. The changes immediately take effect so that the next line
1559 in the configuration file sees the new value. See also "setenv", "resetenv",
1560 and "unsetenv".
1561
1562resetenv [<name> ...]
1563 Removes all environment variables except the ones specified in argument. It
1564 allows to use a clean controlled environment before setting new values with
1565 setenv or unsetenv. Please note that some internal functions may make use of
1566 some environment variables, such as time manipulation functions, but also
1567 OpenSSL or even external checks. This must be used with extreme care and only
1568 after complete validation. The changes immediately take effect so that the
1569 next line in the configuration file sees the new environment. See also
1570 "setenv", "presetenv", and "unsetenv".
1571
Christopher Fauletff4121f2017-11-22 16:38:49 +01001572stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001573 Limits the stats socket to a certain set of processes numbers. By default the
1574 stats socket is bound to all processes, causing a warning to be emitted when
1575 nbproc is greater than 1 because there is no way to select the target process
1576 when connecting. However, by using this setting, it becomes possible to pin
1577 the stats socket to a specific set of processes, typically the first one. The
1578 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001579 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001580 word size (32 or 64). Ranges can be partially defined. The higher bound can
1581 be omitted. In such case, it is replaced by the corresponding maximum
1582 value. A better option consists in using the "process" setting of the "stats
1583 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001584
Baptiste Assmann5626f482015-08-23 10:00:10 +02001585server-state-base <directory>
1586 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001587 file names which do not start with a '/'. See also "server-state-file",
1588 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001589
1590server-state-file <file>
1591 Specifies the path to the file containing state of servers. If the path starts
1592 with a slash ('/'), it is considered absolute, otherwise it is considered
1593 relative to the directory specified using "server-state-base" (if set) or to
1594 the current directory. Before reloading HAProxy, it is possible to save the
1595 servers' current state using the stats command "show servers state". The
1596 output of this command must be written in the file pointed by <file>. When
1597 starting up, before handling traffic, HAProxy will read, load and apply state
1598 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001599 configuration. See also "server-state-base" and "show servers state",
1600 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001601
Willy Tarreau13d2ba22021-03-26 11:38:08 +01001602set-var <var-name> <expr>
1603 Sets the process-wide variable '<var-name>' to the result of the evaluation
1604 of the sample expression <expr>. The variable '<var-name>' may only be a
1605 process-wide variable (using the 'proc.' prefix). It works exactly like the
1606 'set-var' action in TCP or HTTP rules except that the expression is evaluated
1607 at configuration parsing time and that the variable is instantly set. The
1608 sample fetch functions and converters permitted in the expression are only
1609 those using internal data, typically 'int(value)' or 'str(value)'. It's is
1610 possible to reference previously allocated variables as well. These variables
1611 will then be readable (and modifiable) from the regular rule sets.
1612
1613 Example:
1614 global
1615 set-var proc.current_state str(primary)
1616 set-var proc.prio int(100)
1617 set-var proc.threshold int(200),sub(proc.prio)
1618
Willy Tarreau1d549722016-02-16 12:41:57 +01001619setenv <name> <value>
1620 Sets environment variable <name> to value <value>. If the variable exists, it
1621 is overwritten. The changes immediately take effect so that the next line in
1622 the configuration file sees the new value. See also "presetenv", "resetenv",
1623 and "unsetenv".
1624
Willy Tarreau636848a2019-04-15 19:38:50 +02001625set-dumpable
1626 This option is better left disabled by default and enabled only upon a
William Dauchyec730982019-10-27 20:08:10 +01001627 developer's request. If it has been enabled, it may still be forcibly
1628 disabled by prefixing it with the "no" keyword. It has no impact on
1629 performance nor stability but will try hard to re-enable core dumps that were
1630 possibly disabled by file size limitations (ulimit -f), core size limitations
1631 (ulimit -c), or "dumpability" of a process after changing its UID/GID (such
1632 as /proc/sys/fs/suid_dumpable on Linux). Core dumps might still be limited by
1633 the current directory's permissions (check what directory the file is started
1634 from), the chroot directory's permission (it may be needed to temporarily
1635 disable the chroot directive or to move it to a dedicated writable location),
1636 or any other system-specific constraint. For example, some Linux flavours are
1637 notorious for replacing the default core file with a path to an executable
1638 not even installed on the system (check /proc/sys/kernel/core_pattern). Often,
1639 simply writing "core", "core.%p" or "/var/log/core/core.%p" addresses the
1640 issue. When trying to enable this option waiting for a rare issue to
1641 re-appear, it's often a good idea to first try to obtain such a dump by
1642 issuing, for example, "kill -11" to the haproxy process and verify that it
1643 leaves a core where expected when dying.
Willy Tarreau636848a2019-04-15 19:38:50 +02001644
Willy Tarreau610f04b2014-02-13 11:36:41 +01001645ssl-default-bind-ciphers <ciphers>
1646 This setting is only available when support for OpenSSL was built in. It sets
1647 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001648 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001649 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001650 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1651 information and recommendations see e.g.
1652 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1653 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1654 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1655 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001656
1657ssl-default-bind-ciphersuites <ciphersuites>
1658 This setting is only available when support for OpenSSL was built in and
1659 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1660 describing the list of cipher algorithms ("cipher suite") that are negotiated
1661 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1662 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001663 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1664 cipher configuration for TLSv1.2 and earlier, please check the
1665 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001666 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001667
Jerome Magninb203ff62020-04-03 15:28:22 +02001668ssl-default-bind-curves <curves>
1669 This setting is only available when support for OpenSSL was built in. It sets
1670 the default string describing the list of elliptic curves algorithms ("curve
1671 suite") that are negotiated during the SSL/TLS handshake with ECDHE. The format
1672 of the string is a colon-delimited list of curve name.
1673 Please check the "bind" keyword for more information.
1674
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001675ssl-default-bind-options [<option>]...
1676 This setting is only available when support for OpenSSL was built in. It sets
1677 default ssl-options to force on all "bind" lines. Please check the "bind"
1678 keyword to see available options.
1679
1680 Example:
1681 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001682 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001683
Willy Tarreau610f04b2014-02-13 11:36:41 +01001684ssl-default-server-ciphers <ciphers>
1685 This setting is only available when support for OpenSSL was built in. It
1686 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001687 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001688 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001689 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1690 information and recommendations see e.g.
1691 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1692 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1693 For TLSv1.3 cipher configuration, please check the
1694 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1695 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001696
1697ssl-default-server-ciphersuites <ciphersuites>
1698 This setting is only available when support for OpenSSL was built in and
1699 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1700 string describing the list of cipher algorithms that are negotiated during
1701 the TLSv1.3 handshake with the server, for all "server" lines which do not
1702 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001703 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1704 cipher configuration for TLSv1.2 and earlier, please check the
1705 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1706 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001707
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001708ssl-default-server-options [<option>]...
1709 This setting is only available when support for OpenSSL was built in. It sets
1710 default ssl-options to force on all "server" lines. Please check the "server"
1711 keyword to see available options.
1712
Remi Gacogne47783ef2015-05-29 15:53:22 +02001713ssl-dh-param-file <file>
1714 This setting is only available when support for OpenSSL was built in. It sets
1715 the default DH parameters that are used during the SSL/TLS handshake when
1716 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001717 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001718 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001719 are not specified either by using ssl-dh-param-file or by setting them
1720 directly in the certificate file, pre-generated DH parameters of the size
1721 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1722 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001723 Custom DH parameters may be generated by using the OpenSSL command
1724 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1725 parameters should not be considered secure anymore.
1726
William Lallemand8e8581e2020-10-20 17:36:46 +02001727ssl-load-extra-del-ext
1728 This setting allows to configure the way HAProxy does the lookup for the
1729 extra SSL files. By default HAProxy adds a new extension to the filename.
William Lallemand089c1382020-10-23 17:35:12 +02001730 (ex: with "foobar.crt" load "foobar.crt.key"). With this option enabled,
William Lallemand8e8581e2020-10-20 17:36:46 +02001731 HAProxy removes the extension before adding the new one (ex: with
William Lallemand089c1382020-10-23 17:35:12 +02001732 "foobar.crt" load "foobar.key").
1733
1734 Your crt file must have a ".crt" extension for this option to work.
William Lallemand8e8581e2020-10-20 17:36:46 +02001735
1736 This option is not compatible with bundle extensions (.ecdsa, .rsa. .dsa)
1737 and won't try to remove them.
1738
1739 This option is disabled by default. See also "ssl-load-extra-files".
1740
William Lallemand4c5adbf2020-02-24 14:23:22 +01001741ssl-load-extra-files <none|all|bundle|sctl|ocsp|issuer|key>*
William Lallemand3af48e72020-02-03 17:15:52 +01001742 This setting alters the way HAProxy will look for unspecified files during
Jerome Magnin587be9c2020-09-07 11:55:57 +02001743 the loading of the SSL certificates associated to "bind" lines. It does not
1744 apply to certificates used for client authentication on "server" lines.
William Lallemand3af48e72020-02-03 17:15:52 +01001745
1746 By default, HAProxy discovers automatically a lot of files not specified in
1747 the configuration, and you may want to disable this behavior if you want to
1748 optimize the startup time.
1749
1750 "none": Only load the files specified in the configuration. Don't try to load
1751 a certificate bundle if the file does not exist. In the case of a directory,
1752 it won't try to bundle the certificates if they have the same basename.
1753
1754 "all": This is the default behavior, it will try to load everything,
William Lallemand4c5adbf2020-02-24 14:23:22 +01001755 bundles, sctl, ocsp, issuer, key.
William Lallemand3af48e72020-02-03 17:15:52 +01001756
1757 "bundle": When a file specified in the configuration does not exist, HAProxy
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001758 will try to load a "cert bundle".
1759
1760 Starting from HAProxy 2.3, the bundles are not loaded in the same OpenSSL
1761 certificate store, instead it will loads each certificate in a separate
1762 store which is equivalent to declaring multiple "crt". OpenSSL 1.1.1 is
1763 required to achieve this. Which means that bundles are now used only for
1764 backward compatibility and are not mandatory anymore to do an hybrid RSA/ECC
1765 bind configuration..
1766
1767 To associate these PEM files into a "cert bundle" that is recognized by
1768 haproxy, they must be named in the following way: All PEM files that are to
1769 be bundled must have the same base name, with a suffix indicating the key
1770 type. Currently, three suffixes are supported: rsa, dsa and ecdsa. For
1771 example, if www.example.com has two PEM files, an RSA file and an ECDSA
1772 file, they must be named: "example.pem.rsa" and "example.pem.ecdsa". The
1773 first part of the filename is arbitrary; only the suffix matters. To load
1774 this bundle into haproxy, specify the base name only:
1775
1776 Example : bind :8443 ssl crt example.pem
1777
1778 Note that the suffix is not given to haproxy; this tells haproxy to look for
1779 a cert bundle.
1780
1781 HAProxy will load all PEM files in the bundle as if they were configured
1782 separately in several "crt".
1783
1784 The bundle loading does not have an impact anymore on the directory loading
1785 since files are loading separately.
1786
1787 On the CLI, bundles are seen as separate files, and the bundle extension is
1788 required to commit them.
1789
William Dauchy57dd6f12020-10-06 15:22:37 +02001790 OCSP files (.ocsp), issuer files (.issuer), Certificate Transparency (.sctl)
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001791 as well as private keys (.key) are supported with multi-cert bundling.
William Lallemand3af48e72020-02-03 17:15:52 +01001792
1793 "sctl": Try to load "<basename>.sctl" for each crt keyword.
1794
1795 "ocsp": Try to load "<basename>.ocsp" for each crt keyword.
1796
1797 "issuer": Try to load "<basename>.issuer" if the issuer of the OCSP file is
1798 not provided in the PEM file.
1799
William Lallemand4c5adbf2020-02-24 14:23:22 +01001800 "key": If the private key was not provided by the PEM file, try to load a
1801 file "<basename>.key" containing a private key.
1802
William Lallemand3af48e72020-02-03 17:15:52 +01001803 The default behavior is "all".
1804
1805 Example:
1806 ssl-load-extra-files bundle sctl
1807 ssl-load-extra-files sctl ocsp issuer
1808 ssl-load-extra-files none
1809
1810 See also: "crt", section 5.1 about bind options.
1811
Emeric Brun850efd52014-01-29 12:24:34 +01001812ssl-server-verify [none|required]
1813 The default behavior for SSL verify on servers side. If specified to 'none',
1814 servers certificates are not verified. The default is 'required' except if
1815 forced using cmdline option '-dV'.
1816
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001817ssl-skip-self-issued-ca
Daniel Corbett67a82712020-07-06 23:01:19 -04001818 Self issued CA, aka x509 root CA, is the anchor for chain validation: as a
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001819 server is useless to send it, client must have it. Standard configuration
1820 need to not include such CA in PEM file. This option allows you to keep such
1821 CA in PEM file without sending it to the client. Use case is to provide
1822 issuer for ocsp without the need for '.issuer' file and be able to share it
1823 with 'issuers-chain-path'. This concerns all certificates without intermediate
1824 certificates. It's useless for BoringSSL, .issuer is ignored because ocsp
William Lallemand9a1d8392020-08-10 17:28:23 +02001825 bits does not need it. Requires at least OpenSSL 1.0.2.
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001826
Willy Tarreauabb175f2012-09-24 12:43:26 +02001827stats socket [<address:port>|<path>] [param*]
1828 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1829 Connections to this socket will return various statistics outputs and even
1830 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001831 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001832 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001833
Willy Tarreauabb175f2012-09-24 12:43:26 +02001834 All parameters supported by "bind" lines are supported, for instance to
1835 restrict access to some users or their access rights. Please consult
1836 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001837
1838stats timeout <timeout, in milliseconds>
1839 The default timeout on the stats socket is set to 10 seconds. It is possible
1840 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001841 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001842
1843stats maxconn <connections>
1844 By default, the stats socket is limited to 10 concurrent connections. It is
1845 possible to change this value with "stats maxconn".
1846
Willy Tarreau6a06a402007-07-15 20:15:28 +02001847uid <number>
Thayne McCombscdbcca92021-01-07 21:24:41 -07001848 Changes the process's user ID to <number>. It is recommended that the user ID
Willy Tarreau6a06a402007-07-15 20:15:28 +02001849 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1850 be started with superuser privileges in order to be able to switch to another
1851 one. See also "gid" and "user".
1852
1853ulimit-n <number>
1854 Sets the maximum number of per-process file-descriptors to <number>. By
1855 default, it is automatically computed, so it is recommended not to use this
1856 option.
1857
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001858unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1859 [ group <group> ] [ gid <gid> ]
1860
1861 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1862 This is mainly used to simplify declaration of those UNIX sockets and reduce
1863 the risk of errors, since those settings are most commonly required but are
1864 also process-specific. The <prefix> setting can be used to force all socket
1865 path to be relative to that directory. This might be needed to access another
1866 component's chroot. Note that those paths are resolved before haproxy chroots
1867 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1868 all have the same meaning as their homonyms used by the "bind" statement. If
1869 both are specified, the "bind" statement has priority, meaning that the
1870 "unix-bind" settings may be seen as process-wide default settings.
1871
Willy Tarreau1d549722016-02-16 12:41:57 +01001872unsetenv [<name> ...]
1873 Removes environment variables specified in arguments. This can be useful to
1874 hide some sensitive information that are occasionally inherited from the
1875 user's environment during some operations. Variables which did not exist are
1876 silently ignored so that after the operation, it is certain that none of
1877 these variables remain. The changes immediately take effect so that the next
1878 line in the configuration file will not see these variables. See also
1879 "setenv", "presetenv", and "resetenv".
1880
Willy Tarreau6a06a402007-07-15 20:15:28 +02001881user <user name>
1882 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1883 See also "uid" and "group".
1884
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001885node <name>
1886 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1887
1888 This statement is useful in HA configurations where two or more processes or
1889 servers share the same IP address. By setting a different node-name on all
1890 nodes, it becomes easy to immediately spot what server is handling the
1891 traffic.
1892
1893description <text>
1894 Add a text that describes the instance.
1895
1896 Please note that it is required to escape certain characters (# for example)
1897 and this text is inserted into a html page so you should avoid using
1898 "<" and ">" characters.
1899
Thomas Holmesdb04f192015-05-18 13:21:39 +0100190051degrees-data-file <file path>
1901 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001902 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001903
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001904 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001905 compiled with USE_51DEGREES.
1906
Ben Shillitof25e8e52016-12-02 14:25:37 +0000190751degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001908 A list of 51Degrees property names to be load from the dataset. A full list
1909 of names is available on the 51Degrees website:
1910 https://51degrees.com/resources/property-dictionary
1911
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001912 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001913 compiled with USE_51DEGREES.
1914
Dragan Dosen93b38d92015-06-29 16:43:25 +0200191551degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001916 A char that will be appended to every property value in a response header
1917 containing 51Degrees results. If not set that will be set as ','.
1918
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001919 Please note that this option is only available when haproxy has been
1920 compiled with USE_51DEGREES.
1921
192251degrees-cache-size <number>
1923 Sets the size of the 51Degrees converter cache to <number> entries. This
1924 is an LRU cache which reminds previous device detections and their results.
1925 By default, this cache is disabled.
1926
1927 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001928 compiled with USE_51DEGREES.
1929
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001930wurfl-data-file <file path>
1931 The path of the WURFL data file to provide device detection services. The
1932 file should be accessible by HAProxy with relevant permissions.
1933
1934 Please note that this option is only available when haproxy has been compiled
1935 with USE_WURFL=1.
1936
1937wurfl-information-list [<capability>]*
1938 A space-delimited list of WURFL capabilities, virtual capabilities, property
1939 names we plan to use in injected headers. A full list of capability and
1940 virtual capability names is available on the Scientiamobile website :
1941
1942 https://www.scientiamobile.com/wurflCapability
1943
1944 Valid WURFL properties are:
1945 - wurfl_id Contains the device ID of the matched device.
1946
1947 - wurfl_root_id Contains the device root ID of the matched
1948 device.
1949
1950 - wurfl_isdevroot Tells if the matched device is a root device.
1951 Possible values are "TRUE" or "FALSE".
1952
1953 - wurfl_useragent The original useragent coming with this
1954 particular web request.
1955
1956 - wurfl_api_version Contains a string representing the currently
1957 used Libwurfl API version.
1958
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001959 - wurfl_info A string containing information on the parsed
1960 wurfl.xml and its full path.
1961
1962 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1963 WURFL has been loaded successfully.
1964
1965 - wurfl_normalized_useragent The normalized useragent.
1966
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001967 Please note that this option is only available when haproxy has been compiled
1968 with USE_WURFL=1.
1969
1970wurfl-information-list-separator <char>
1971 A char that will be used to separate values in a response header containing
1972 WURFL results. If not set that a comma (',') will be used by default.
1973
1974 Please note that this option is only available when haproxy has been compiled
1975 with USE_WURFL=1.
1976
1977wurfl-patch-file [<file path>]
1978 A list of WURFL patch file paths. Note that patches are loaded during startup
1979 thus before the chroot.
1980
1981 Please note that this option is only available when haproxy has been compiled
1982 with USE_WURFL=1.
1983
paulborilebad132c2019-04-18 11:57:04 +02001984wurfl-cache-size <size>
1985 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1986 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001987 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001988 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001989
1990 Please note that this option is only available when haproxy has been compiled
1991 with USE_WURFL=1.
1992
William Dauchy0fec3ab2019-10-27 20:08:11 +01001993strict-limits
William Dauchya5194602020-03-28 19:29:58 +01001994 Makes process fail at startup when a setrlimit fails. Haproxy tries to set the
1995 best setrlimit according to what has been calculated. If it fails, it will
1996 emit a warning. This option is here to guarantee an explicit failure of
1997 haproxy when those limits fail. It is enabled by default. It may still be
1998 forcibly disabled by prefixing it with the "no" keyword.
William Dauchy0fec3ab2019-10-27 20:08:11 +01001999
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020003.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02002001-----------------------
2002
Willy Tarreaubeb859a2018-11-22 18:07:59 +01002003busy-polling
2004 In some situations, especially when dealing with low latency on processors
2005 supporting a variable frequency or when running inside virtual machines, each
2006 time the process waits for an I/O using the poller, the processor goes back
2007 to sleep or is offered to another VM for a long time, and it causes
2008 excessively high latencies. This option provides a solution preventing the
2009 processor from sleeping by always using a null timeout on the pollers. This
2010 results in a significant latency reduction (30 to 100 microseconds observed)
2011 at the expense of a risk to overheat the processor. It may even be used with
2012 threads, in which case improperly bound threads may heavily conflict,
2013 resulting in a worse performance and high values for the CPU stolen fields
2014 in "show info" output, indicating which threads are misconfigured. It is
2015 important not to let the process run on the same processor as the network
2016 interrupts when this option is used. It is also better to avoid using it on
2017 multiple CPU threads sharing the same core. This option is disabled by
2018 default. If it has been enabled, it may still be forcibly disabled by
2019 prefixing it with the "no" keyword. It is ignored by the "select" and
2020 "poll" pollers.
2021
William Dauchy3894d972019-12-28 15:36:02 +01002022 This option is automatically disabled on old processes in the context of
2023 seamless reload; it avoids too much cpu conflicts when multiple processes
2024 stay around for some time waiting for the end of their current connections.
2025
Willy Tarreau1746eec2014-04-25 10:46:47 +02002026max-spread-checks <delay in milliseconds>
2027 By default, haproxy tries to spread the start of health checks across the
2028 smallest health check interval of all the servers in a farm. The principle is
2029 to avoid hammering services running on the same server. But when using large
2030 check intervals (10 seconds or more), the last servers in the farm take some
2031 time before starting to be tested, which can be a problem. This parameter is
2032 used to enforce an upper bound on delay between the first and the last check,
2033 even if the servers' check intervals are larger. When servers run with
2034 shorter intervals, their intervals will be respected though.
2035
Willy Tarreau6a06a402007-07-15 20:15:28 +02002036maxconn <number>
2037 Sets the maximum per-process number of concurrent connections to <number>. It
2038 is equivalent to the command-line argument "-n". Proxies will stop accepting
2039 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02002040 automatically adjusted according to this value. See also "ulimit-n". Note:
2041 the "select" poller cannot reliably use more than 1024 file descriptors on
2042 some platforms. If your platform only supports select and reports "select
2043 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01002044 below 500 in general). If this value is not set, it will automatically be
2045 calculated based on the current file descriptors limit reported by the
2046 "ulimit -n" command, possibly reduced to a lower value if a memory limit
2047 is enforced, based on the buffer size, memory allocated to compression, SSL
2048 cache size, and use or not of SSL and the associated maxsslconn (which can
2049 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02002050
Willy Tarreau81c25d02011-09-07 15:17:21 +02002051maxconnrate <number>
2052 Sets the maximum per-process number of connections per second to <number>.
2053 Proxies will stop accepting connections when this limit is reached. It can be
2054 used to limit the global capacity regardless of each frontend capacity. It is
2055 important to note that this can only be used as a service protection measure,
2056 as there will not necessarily be a fair share between frontends when the
2057 limit is reached, so it's a good idea to also limit each frontend to some
2058 value close to its expected share. Also, lowering tune.maxaccept can improve
2059 fairness.
2060
William Lallemandd85f9172012-11-09 17:05:39 +01002061maxcomprate <number>
2062 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01002063 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01002064 level will be decreased during the session. If the maximum is reached at the
2065 beginning of a session, the session will not compress at all. If the maximum
2066 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01002067 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01002068 default value.
2069
William Lallemand072a2bf2012-11-20 17:01:01 +01002070maxcompcpuusage <number>
2071 Sets the maximum CPU usage HAProxy can reach before stopping the compression
2072 for new requests or decreasing the compression level of current requests.
2073 It works like 'maxcomprate' but measures CPU usage instead of incoming data
2074 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
2075 case of multiple processes (nbproc > 1), each process manages its individual
2076 usage. A value of 100 disable the limit. The default value is 100. Setting
2077 a lower value will prevent the compression work from slowing the whole
2078 process down and from introducing high latencies.
2079
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002080maxpipes <number>
2081 Sets the maximum per-process number of pipes to <number>. Currently, pipes
2082 are only used by kernel-based tcp splicing. Since a pipe contains two file
2083 descriptors, the "ulimit-n" value will be increased accordingly. The default
2084 value is maxconn/4, which seems to be more than enough for most heavy usages.
2085 The splice code dynamically allocates and releases pipes, and can fall back
2086 to standard copy, so setting this value too low may only impact performance.
2087
Willy Tarreau93e7c002013-10-07 18:51:07 +02002088maxsessrate <number>
2089 Sets the maximum per-process number of sessions per second to <number>.
2090 Proxies will stop accepting connections when this limit is reached. It can be
2091 used to limit the global capacity regardless of each frontend capacity. It is
2092 important to note that this can only be used as a service protection measure,
2093 as there will not necessarily be a fair share between frontends when the
2094 limit is reached, so it's a good idea to also limit each frontend to some
2095 value close to its expected share. Also, lowering tune.maxaccept can improve
2096 fairness.
2097
Willy Tarreau403edff2012-09-06 11:58:37 +02002098maxsslconn <number>
2099 Sets the maximum per-process number of concurrent SSL connections to
2100 <number>. By default there is no SSL-specific limit, which means that the
2101 global maxconn setting will apply to all connections. Setting this limit
2102 avoids having openssl use too much memory and crash when malloc returns NULL
2103 (since it unfortunately does not reliably check for such conditions). Note
2104 that the limit applies both to incoming and outgoing connections, so one
2105 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01002106 If this value is not set, but a memory limit is enforced, this value will be
2107 automatically computed based on the memory limit, maxconn, the buffer size,
2108 memory allocated to compression, SSL cache size, and use of SSL in either
2109 frontends, backends or both. If neither maxconn nor maxsslconn are specified
2110 when there is a memory limit, haproxy will automatically adjust these values
2111 so that 100% of the connections can be made over SSL with no risk, and will
2112 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02002113
Willy Tarreaue43d5322013-10-07 20:01:52 +02002114maxsslrate <number>
2115 Sets the maximum per-process number of SSL sessions per second to <number>.
2116 SSL listeners will stop accepting connections when this limit is reached. It
2117 can be used to limit the global SSL CPU usage regardless of each frontend
2118 capacity. It is important to note that this can only be used as a service
2119 protection measure, as there will not necessarily be a fair share between
2120 frontends when the limit is reached, so it's a good idea to also limit each
2121 frontend to some value close to its expected share. It is also important to
2122 note that the sessions are accounted before they enter the SSL stack and not
2123 after, which also protects the stack against bad handshakes. Also, lowering
2124 tune.maxaccept can improve fairness.
2125
William Lallemand9d5f5482012-11-07 16:12:57 +01002126maxzlibmem <number>
2127 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
2128 When the maximum amount is reached, future sessions will not compress as long
2129 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01002130 The default value is 0. The value is available in bytes on the UNIX socket
2131 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
2132 "ZlibMemUsage" in bytes.
2133
Willy Tarreau6a06a402007-07-15 20:15:28 +02002134noepoll
2135 Disables the use of the "epoll" event polling system on Linux. It is
2136 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01002137 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02002138
2139nokqueue
2140 Disables the use of the "kqueue" event polling system on BSD. It is
2141 equivalent to the command-line argument "-dk". The next polling system
2142 used will generally be "poll". See also "nopoll".
2143
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00002144noevports
2145 Disables the use of the event ports event polling system on SunOS systems
2146 derived from Solaris 10 and later. It is equivalent to the command-line
2147 argument "-dv". The next polling system used will generally be "poll". See
2148 also "nopoll".
2149
Willy Tarreau6a06a402007-07-15 20:15:28 +02002150nopoll
2151 Disables the use of the "poll" event polling system. It is equivalent to the
2152 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002153 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00002154 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
2155 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02002156
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002157nosplice
2158 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002159 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002160 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002161 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002162 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
2163 be used. This option makes it easier to globally disable kernel splicing in
2164 case of doubt. See also "option splice-auto", "option splice-request" and
2165 "option splice-response".
2166
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002167nogetaddrinfo
2168 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
2169 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
2170
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00002171noreuseport
2172 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
2173 command line argument "-dR".
2174
Willy Tarreaud2d33482019-04-25 17:09:07 +02002175profiling.tasks { auto | on | off }
2176 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
2177 the profiling automatically turns on a thread when it starts to suffer from
2178 an average latency of 1000 microseconds or higher as reported in the
2179 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05002180 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02002181 loops so it does not vary quickly and tends to significantly smooth short
2182 spikes). It may also spontaneously trigger from time to time on overloaded
2183 systems, containers, or virtual machines, or when the system swaps (which
2184 must absolutely never happen on a load balancer).
2185
2186 CPU profiling per task can be very convenient to report where the time is
2187 spent and which requests have what effect on which other request. Enabling
2188 it will typically affect the overall's performance by less than 1%, thus it
2189 is recommended to leave it to the default 'auto' value so that it only
2190 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01002191 supporting the clock_gettime(2) syscall with clock identifiers
2192 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
2193 be zero. This option may be changed at run time using "set profiling" on the
2194 CLI.
2195
Willy Tarreaufe255b72007-10-14 23:09:26 +02002196spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09002197 Sometimes it is desirable to avoid sending agent and health checks to
2198 servers at exact intervals, for instance when many logical servers are
2199 located on the same physical server. With the help of this parameter, it
2200 becomes possible to add some randomness in the check interval between 0
2201 and +/- 50%. A value between 2 and 5 seems to show good results. The
2202 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02002203
Davor Ocelice9ed2812017-12-25 17:49:28 +01002204ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00002205 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01002206 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00002207 multiple times, it will simply enable multiple crypto engines. Referencing an
2208 unsupported engine will prevent haproxy from starting. Note that many engines
2209 will lead to lower HTTPS performance than pure software with recent
2210 processors. The optional command "algo" sets the default algorithms an ENGINE
2211 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01002212 of "ALL" uses the engine for all cryptographic operations. If no list of
2213 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00002214 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
2215 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
2216 openssl configuration file uses:
2217 https://www.openssl.org/docs/man1.0.2/apps/config.html
2218
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00002219ssl-mode-async
2220 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02002221 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00002222 implementation supports a maximum of 32 engines. The Openssl ASYNC API
2223 doesn't support moving read/write buffers and is not compliant with
2224 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05002225 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00002226 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00002227
Willy Tarreau33cb0652014-12-23 22:52:37 +01002228tune.buffers.limit <number>
2229 Sets a hard limit on the number of buffers which may be allocated per process.
2230 The default value is zero which means unlimited. The minimum non-zero value
2231 will always be greater than "tune.buffers.reserve" and should ideally always
2232 be about twice as large. Forcing this value can be particularly useful to
2233 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01002234 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01002235 another one to be released by another session. Since buffers are dynamically
2236 allocated and released, the waiting time is very short and not perceptible
2237 provided that limits remain reasonable. In fact sometimes reducing the limit
2238 may even increase performance by increasing the CPU cache's efficiency. Tests
2239 have shown good results on average HTTP traffic with a limit to 1/10 of the
2240 expected global maxconn setting, which also significantly reduces memory
2241 usage. The memory savings come from the fact that a number of connections
2242 will not allocate 2*tune.bufsize. It is best not to touch this value unless
2243 advised to do so by an haproxy core developer.
2244
Willy Tarreau1058ae72014-12-23 22:40:40 +01002245tune.buffers.reserve <number>
2246 Sets the number of buffers which are pre-allocated and reserved for use only
2247 during memory shortage conditions resulting in failed memory allocations. The
2248 minimum value is 2 and is also the default. There is no reason a user would
2249 want to change this value, it's mostly aimed at haproxy core developers.
2250
Willy Tarreau27a674e2009-08-17 07:23:33 +02002251tune.bufsize <number>
2252 Sets the buffer size to this size (in bytes). Lower values allow more
2253 sessions to coexist in the same amount of RAM, and higher values allow some
2254 applications with very large cookies to work. The default value is 16384 and
2255 can be changed at build time. It is strongly recommended not to change this
2256 from the default value, as very low values will break some services such as
2257 statistics, and values larger than default size will increase memory usage,
2258 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01002259 parameter should be decreased by the same factor as this one is increased. In
2260 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
2261 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04002262 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01002263 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
2264 value set using this parameter will automatically be rounded up to the next
2265 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02002266
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +01002267tune.chksize <number> (deprecated)
2268 This option is deprecated and ignored.
Willy Tarreau43961d52010-10-04 20:39:20 +02002269
William Lallemandf3747832012-11-09 12:33:10 +01002270tune.comp.maxlevel <number>
2271 Sets the maximum compression level. The compression level affects CPU
2272 usage during compression. This value affects CPU usage during compression.
2273 Each session using compression initializes the compression algorithm with
2274 this value. The default value is 1.
2275
Willy Tarreauc299e1e2019-02-27 11:35:12 +01002276tune.fail-alloc
2277 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
2278 allocation attempt fails. Must be between 0 (no failure) and 100 (no
2279 success). This is useful to debug and make sure memory failures are handled
2280 gracefully.
2281
Willy Tarreaubc52bec2020-06-18 08:58:47 +02002282tune.fd.edge-triggered { on | off } [ EXPERIMENTAL ]
2283 Enables ('on') or disables ('off') the edge-triggered polling mode for FDs
2284 that support it. This is currently only support with epoll. It may noticeably
2285 reduce the number of epoll_ctl() calls and slightly improve performance in
2286 certain scenarios. This is still experimental, it may result in frozen
2287 connections if bugs are still present, and is disabled by default.
2288
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02002289tune.h2.header-table-size <number>
2290 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
2291 cannot be larger than 65536 bytes. A larger value may help certain clients
2292 send more compact requests, depending on their capabilities. This amount of
2293 memory is consumed for each HTTP/2 connection. It is recommended not to
2294 change it.
2295
Willy Tarreaue6baec02017-07-27 11:45:11 +02002296tune.h2.initial-window-size <number>
2297 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01002298 can upload before waiting for an acknowledgment from haproxy. This setting
2299 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02002300 The default value is 65535, which roughly allows up to 5 Mbps of upload
2301 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
2302 over a 1-ms local network. It can make sense to increase this value to allow
2303 faster uploads, or to reduce it to increase fairness when dealing with many
2304 clients. It doesn't affect resource usage.
2305
Willy Tarreau5242ef82017-07-27 11:47:28 +02002306tune.h2.max-concurrent-streams <number>
2307 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
2308 number of outstanding requests on a single connection). The default value is
2309 100. A larger one may slightly improve page load time for complex sites when
2310 visited over high latency networks, but increases the amount of resources a
2311 single client may allocate. A value of zero disables the limit so a single
2312 client may create as many streams as allocatable by haproxy. It is highly
2313 recommended not to change this value.
2314
Willy Tarreaua24b35c2019-02-21 13:24:36 +01002315tune.h2.max-frame-size <number>
2316 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
2317 receive to its peers. The default value is the largest between 16384 and the
2318 buffer size (tune.bufsize). In any case, haproxy will not announce support
2319 for frame sizes larger than buffers. The main purpose of this setting is to
2320 allow to limit the maximum frame size setting when using large buffers. Too
2321 large frame sizes might have performance impact or cause some peers to
2322 misbehave. It is highly recommended not to change this value.
2323
Willy Tarreau193b8c62012-11-22 00:17:38 +01002324tune.http.cookielen <number>
2325 Sets the maximum length of captured cookies. This is the maximum value that
2326 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
2327 will automatically be truncated to this one. It is important not to set too
2328 high a value because all cookie captures still allocate this size whatever
2329 their configured value (they share a same pool). This value is per request
2330 per response, so the memory allocated is twice this value per connection.
2331 When not specified, the limit is set to 63 characters. It is recommended not
2332 to change this value.
2333
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002334tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002335 Sets the maximum length of request URI in logs. This prevents truncating long
2336 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002337 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002338 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002339 configuration directives too.
2340 The default value is 1024.
2341
Willy Tarreauac1932d2011-10-24 19:14:41 +02002342tune.http.maxhdr <number>
2343 Sets the maximum number of headers in a request. When a request comes with a
2344 number of headers greater than this value (including the first line), it is
2345 rejected with a "400 Bad Request" status code. Similarly, too large responses
2346 are blocked with "502 Bad Gateway". The default value is 101, which is enough
2347 for all usages, considering that the widely deployed Apache server uses the
2348 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02002349 a buggy application to work by the time it gets fixed. The accepted range is
2350 1..32767. Keep in mind that each new header consumes 32bits of memory for
2351 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02002352
Willy Tarreau76cc6992020-07-01 18:49:24 +02002353tune.idle-pool.shared { on | off }
2354 Enables ('on') or disables ('off') sharing of idle connection pools between
2355 threads for a same server. The default is to share them between threads in
2356 order to minimize the number of persistent connections to a server, and to
2357 optimize the connection reuse rate. But to help with debugging or when
2358 suspecting a bug in HAProxy around connection reuse, it can be convenient to
2359 forcefully disable this idle pool sharing between multiple threads, and force
Willy Tarreau0784db82021-02-19 11:45:22 +01002360 this option to "off". The default is on. It is strongly recommended against
2361 disabling this option without setting a conservative value on "pool-low-conn"
2362 for all servers relying on connection reuse to achieve a high performance
2363 level, otherwise connections might be closed very often as the thread count
2364 increases.
Willy Tarreau76cc6992020-07-01 18:49:24 +02002365
Willy Tarreau7e312732014-02-12 16:35:14 +01002366tune.idletimer <timeout>
2367 Sets the duration after which haproxy will consider that an empty buffer is
2368 probably associated with an idle stream. This is used to optimally adjust
2369 some packet sizes while forwarding large and small data alternatively. The
2370 decision to use splice() or to send large buffers in SSL is modulated by this
2371 parameter. The value is in milliseconds between 0 and 65535. A value of zero
2372 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01002373 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05002374 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01002375 tune.ssl.maxrecord below.
2376
Willy Tarreau7ac908b2019-02-27 12:02:18 +01002377tune.listener.multi-queue { on | off }
2378 Enables ('on') or disables ('off') the listener's multi-queue accept which
2379 spreads the incoming traffic to all threads a "bind" line is allowed to run
2380 on instead of taking them for itself. This provides a smoother traffic
2381 distribution and scales much better, especially in environments where threads
2382 may be unevenly loaded due to external activity (network interrupts colliding
2383 with one thread for example). This option is enabled by default, but it may
2384 be forcefully disabled for troubleshooting or for situations where it is
2385 estimated that the operating system already provides a good enough
2386 distribution and connections are extremely short-lived.
2387
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002388tune.lua.forced-yield <number>
2389 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01002390 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002391 HAProxy scheduler to process other tasks like accepting connections or
2392 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01002393 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002394 lowered. If the Lua code is quite long and its result is absolutely required
2395 to process the data, the <number> can be increased.
2396
Willy Tarreau32f61e22015-03-18 17:54:59 +01002397tune.lua.maxmem
2398 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
2399 default it is zero which means unlimited. It is important to set a limit to
2400 ensure that a bug in a script will not result in the system running out of
2401 memory.
2402
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002403tune.lua.session-timeout <timeout>
2404 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002405 preventing infinite loops or spending too much time in Lua. This timeout
2406 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002407 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002408
2409tune.lua.task-timeout <timeout>
2410 Purpose is the same as "tune.lua.session-timeout", but this timeout is
2411 dedicated to the tasks. By default, this timeout isn't set because a task may
2412 remain alive during of the lifetime of HAProxy. For example, a task used to
2413 check servers.
2414
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002415tune.lua.service-timeout <timeout>
2416 This is the execution timeout for the Lua services. This is useful for
2417 preventing infinite loops or spending too much time in Lua. This timeout
2418 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002419 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002420
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002421tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01002422 Sets the maximum number of consecutive connections a process may accept in a
2423 row before switching to other work. In single process mode, higher numbers
Willy Tarreau66161322021-02-19 15:50:27 +01002424 used to give better performance at high connection rates, though this is not
2425 the case anymore with the multi-queue. This value applies individually to
2426 each listener, so that the number of processes a listener is bound to is
2427 taken into account. This value defaults to 4 which showed best results. If a
2428 significantly higher value was inherited from an ancient config, it might be
2429 worth removing it as it will both increase performance and lower response
2430 time. In multi-process mode, it is divided by twice the number of processes
2431 the listener is bound to. Setting this value to -1 completely disables the
2432 limitation. It should normally not be needed to tweak this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002433
2434tune.maxpollevents <number>
2435 Sets the maximum amount of events that can be processed at once in a call to
2436 the polling system. The default value is adapted to the operating system. It
2437 has been noticed that reducing it below 200 tends to slightly decrease
2438 latency at the expense of network bandwidth, and increasing it above 200
2439 tends to trade latency for slightly increased bandwidth.
2440
Willy Tarreau27a674e2009-08-17 07:23:33 +02002441tune.maxrewrite <number>
2442 Sets the reserved buffer space to this size in bytes. The reserved space is
2443 used for header rewriting or appending. The first reads on sockets will never
2444 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
2445 bufsize, though that does not make much sense since there are rarely large
2446 numbers of headers to add. Setting it too high prevents processing of large
2447 requests or responses. Setting it too low prevents addition of new headers
2448 to already large requests or to POST requests. It is generally wise to set it
2449 to about 1024. It is automatically readjusted to half of bufsize if it is
2450 larger than that. This means you don't have to worry about it when changing
2451 bufsize.
2452
Willy Tarreauf3045d22015-04-29 16:24:50 +02002453tune.pattern.cache-size <number>
2454 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
2455 cache which reminds previous lookups and their results. It is used by ACLs
2456 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
2457 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
2458 strings. It applies to pattern expressions which means that it will be able
2459 to memorize the result of a lookup among all the patterns specified on a
2460 configuration line (including all those loaded from files). It automatically
2461 invalidates entries which are updated using HTTP actions or on the CLI. The
2462 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau403bfbb2019-10-23 06:59:31 +02002463 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
2464 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02002465 risk of collision in this cache, which is in the order of the size of the
2466 cache divided by 2^64. Typically, at 10000 requests per second with the
2467 default cache size of 10000 entries, there's 1% chance that a brute force
2468 attack could cause a single collision after 60 years, or 0.1% after 6 years.
2469 This is considered much lower than the risk of a memory corruption caused by
2470 aging components. If this is not acceptable, the cache can be disabled by
2471 setting this parameter to 0.
2472
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02002473tune.pipesize <number>
2474 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
2475 are the default size for the system. But sometimes when using TCP splicing,
2476 it can improve performance to increase pipe sizes, especially if it is
2477 suspected that pipes are not filled and that many calls to splice() are
2478 performed. This has an impact on the kernel's memory footprint, so this must
2479 not be changed if impacts are not understood.
2480
Olivier Houchard88698d92019-04-16 19:07:22 +02002481tune.pool-high-fd-ratio <number>
2482 This setting sets the max number of file descriptors (in percentage) used by
2483 haproxy globally against the maximum number of file descriptors haproxy can
2484 use before we start killing idle connections when we can't reuse a connection
2485 and we have to create a new one. The default is 25 (one quarter of the file
2486 descriptor will mean that roughly half of the maximum front connections can
2487 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05002488 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02002489
Willy Tarreau83ca3052020-07-01 18:30:16 +02002490tune.pool-low-fd-ratio <number>
2491 This setting sets the max number of file descriptors (in percentage) used by
2492 haproxy globally against the maximum number of file descriptors haproxy can
2493 use before we stop putting connection into the idle pool for reuse. The
2494 default is 20.
2495
Willy Tarreaue803de22010-01-21 17:43:04 +01002496tune.rcvbuf.client <number>
2497tune.rcvbuf.server <number>
2498 Forces the kernel socket receive buffer size on the client or the server side
2499 to the specified value in bytes. This value applies to all TCP/HTTP frontends
2500 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002501 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002502 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002503 order to save kernel memory by preventing it from buffering too large amounts
2504 of received data. Lower values will significantly increase CPU usage though.
2505
Willy Tarreaub22fc302015-12-14 12:04:35 +01002506tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002507 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01002508 socket buffers. One of them is that a read returns more than <recv_enough>
2509 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
2510 may be changed by this setting to better deal with workloads involving lots
2511 of short messages such as telnet or SSH sessions.
2512
Olivier Houchard1599b802018-05-24 18:59:04 +02002513tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05002514 Sets the maximum amount of task that can be processed at once when running
Willy Tarreau060a7612021-03-10 11:06:26 +01002515 tasks. The default value depends on the number of threads but sits between 35
2516 and 280, which tend to show the highest request rates and lowest latencies.
2517 Increasing it may incur latency when dealing with I/Os, making it too small
2518 can incur extra overhead. Higher thread counts benefit from lower values.
2519 When experimenting with much larger values, it may be useful to also enable
2520 tune.sched.low-latency and possibly tune.fd.edge-triggered to limit the
2521 maximum latency to the lowest possible.
Willy Tarreaue7723bd2020-06-24 11:11:02 +02002522
2523tune.sched.low-latency { on | off }
2524 Enables ('on') or disables ('off') the low-latency task scheduler. By default
2525 haproxy processes tasks from several classes one class at a time as this is
2526 the most efficient. But when running with large values of tune.runqueue-depth
2527 this can have a measurable effect on request or connection latency. When this
2528 low-latency setting is enabled, tasks of lower priority classes will always
2529 be executed before other ones if they exist. This will permit to lower the
2530 maximum latency experienced by new requests or connections in the middle of
2531 massive traffic, at the expense of a higher impact on this large traffic.
2532 For regular usage it is better to leave this off. The default value is off.
Olivier Houchard1599b802018-05-24 18:59:04 +02002533
Willy Tarreaue803de22010-01-21 17:43:04 +01002534tune.sndbuf.client <number>
2535tune.sndbuf.server <number>
2536 Forces the kernel socket send buffer size on the client or the server side to
2537 the specified value in bytes. This value applies to all TCP/HTTP frontends
2538 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002539 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002540 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002541 order to save kernel memory by preventing it from buffering too large amounts
2542 of received data. Lower values will significantly increase CPU usage though.
2543 Another use case is to prevent write timeouts with extremely slow clients due
2544 to the kernel waiting for a large part of the buffer to be read before
2545 notifying haproxy again.
2546
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002547tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01002548 Sets the size of the global SSL session cache, in a number of blocks. A block
William Dauchy9a4bbfe2021-02-12 15:58:46 +01002549 is large enough to contain an encoded session without peer certificate. An
2550 encoded session with peer certificate is stored in multiple blocks depending
2551 on the size of the peer certificate. A block uses approximately 200 bytes of
2552 memory (based on `sizeof(struct sh_ssl_sess_hdr) + SHSESS_BLOCK_MIN_SIZE`
2553 calculation used for `shctx_init` function). The default value may be forced
2554 at build time, otherwise defaults to 20000. When the cache is full, the most
2555 idle entries are purged and reassigned. Higher values reduce the occurrence
2556 of such a purge, hence the number of CPU-intensive SSL handshakes by ensuring
2557 that all users keep their session as long as possible. All entries are
2558 pre-allocated upon startup and are shared between all processes if "nbproc"
2559 is greater than 1. Setting this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002560
Emeric Brun8dc60392014-05-09 13:52:00 +02002561tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02002562 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02002563 should normally not be used since it will force many renegotiations due to
2564 clients hitting a random process. But it may be required on some operating
2565 systems where none of the SSL cache synchronization method may be used. In
2566 this case, adding a first layer of hash-based load balancing before the SSL
2567 layer might limit the impact of the lack of session sharing.
2568
William Lallemand7d42ef52020-07-06 11:41:30 +02002569tune.ssl.keylog { on | off }
2570 This option activates the logging of the TLS keys. It should be used with
2571 care as it will consume more memory per SSL session and could decrease
2572 performances. This is disabled by default.
2573
2574 These sample fetches should be used to generate the SSLKEYLOGFILE that is
2575 required to decipher traffic with wireshark.
2576
2577 https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/Key_Log_Format
2578
2579 The SSLKEYLOG is a series of lines which are formatted this way:
2580
2581 <Label> <space> <ClientRandom> <space> <Secret>
2582
2583 The ClientRandom is provided by the %[ssl_fc_client_random,hex] sample
2584 fetch, the secret and the Label could be find in the array below. You need
2585 to generate a SSLKEYLOGFILE with all the labels in this array.
2586
2587 The following sample fetches are hexadecimal strings and does not need to be
2588 converted.
2589
2590 SSLKEYLOGFILE Label | Sample fetches for the Secrets
2591 --------------------------------|-----------------------------------------
2592 CLIENT_EARLY_TRAFFIC_SECRET | %[ssl_fc_client_early_traffic_secret]
2593 CLIENT_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_client_handshake_traffic_secret]
2594 SERVER_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_server_handshake_traffic_secret]
2595 CLIENT_TRAFFIC_SECRET_0 | %[ssl_fc_client_traffic_secret_0]
2596 SERVER_TRAFFIC_SECRET_0 | %[ssl_fc_server_traffic_secret_0]
William Lallemandd742b6c2020-07-07 10:14:56 +02002597 EXPORTER_SECRET | %[ssl_fc_exporter_secret]
2598 EARLY_EXPORTER_SECRET | %[ssl_fc_early_exporter_secret]
William Lallemand7d42ef52020-07-06 11:41:30 +02002599
2600 This is only available with OpenSSL 1.1.1, and useful with TLS1.3 session.
2601
2602 If you want to generate the content of a SSLKEYLOGFILE with TLS < 1.3, you
2603 only need this line:
2604
2605 "CLIENT_RANDOM %[ssl_fc_client_random,hex] %[ssl_fc_session_key,hex]"
2606
Emeric Brun4f65bff2012-11-16 15:11:00 +01002607tune.ssl.lifetime <timeout>
2608 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002609 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01002610 does not guarantee that sessions will last that long, because if the cache is
2611 full, the longest idle sessions will be purged despite their configured
2612 lifetime. The real usefulness of this setting is to prevent sessions from
2613 being used for too long.
2614
Willy Tarreaubfd59462013-02-21 07:46:09 +01002615tune.ssl.maxrecord <number>
2616 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
2617 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
2618 data only once it has received a full record. With large records, it means
2619 that clients might have to download up to 16kB of data before starting to
2620 process them. Limiting the value can improve page load times on browsers
2621 located over high latency or low bandwidth networks. It is suggested to find
2622 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
2623 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
2624 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
2625 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002626 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01002627 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01002628
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002629tune.ssl.default-dh-param <number>
2630 Sets the maximum size of the Diffie-Hellman parameters used for generating
2631 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
2632 final size will try to match the size of the server's RSA (or DSA) key (e.g,
2633 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
Willy Tarreau3ba77d22020-05-08 09:31:18 +02002634 this maximum value. Default value if 2048. Only 1024 or higher values are
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002635 allowed. Higher values will increase the CPU load, and values greater than
2636 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02002637 used if static Diffie-Hellman parameters are supplied either directly
2638 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002639
Christopher Faulet31af49d2015-06-09 17:29:50 +02002640tune.ssl.ssl-ctx-cache-size <number>
2641 Sets the size of the cache used to store generated certificates to <number>
2642 entries. This is a LRU cache. Because generating a SSL certificate
2643 dynamically is expensive, they are cached. The default cache size is set to
2644 1000 entries.
2645
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01002646tune.ssl.capture-cipherlist-size <number>
2647 Sets the maximum size of the buffer used for capturing client-hello cipher
2648 list. If the value is 0 (default value) the capture is disabled, otherwise
2649 a buffer is allocated for each SSL/TLS connection.
2650
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002651tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002652tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002653tune.vars.reqres-max-size <size>
2654tune.vars.sess-max-size <size>
2655tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002656 These five tunes help to manage the maximum amount of memory used by the
2657 variables system. "global" limits the overall amount of memory available for
2658 all scopes. "proc" limits the memory for the process scope, "sess" limits the
2659 memory for the session scope, "txn" for the transaction scope, and "reqres"
2660 limits the memory for each request or response processing.
2661 Memory accounting is hierarchical, meaning more coarse grained limits include
2662 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
2663 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002664
Daniel Schneller0b547052016-03-21 20:46:57 +01002665 For example, when "tune.vars.sess-max-size" is limited to 100,
2666 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
2667 100 either. If we create a variable "txn.var" that contains 100 bytes,
2668 all available space is consumed.
2669 Notice that exceeding the limits at runtime will not result in an error
2670 message, but values might be cut off or corrupted. So make sure to accurately
2671 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002672
William Lallemanda509e4c2012-11-07 16:54:34 +01002673tune.zlib.memlevel <number>
2674 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002675 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002676 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002677 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002678 between 1 and 9. The default value is 8.
2679
2680tune.zlib.windowsize <number>
2681 Sets the window size (the size of the history buffer) as a parameter of the
2682 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002683 in better compression at the expense of memory usage. Can be a value between
2684 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002685
Willy Tarreauc57f0e22009-05-10 13:12:33 +020026863.3. Debugging
2687--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002688
Willy Tarreau6a06a402007-07-15 20:15:28 +02002689quiet
2690 Do not display any message during startup. It is equivalent to the command-
2691 line argument "-q".
2692
Willy Tarreau3eb10b82020-04-15 16:42:39 +02002693zero-warning
2694 When this option is set, haproxy will refuse to start if any warning was
2695 emitted while processing the configuration. It is highly recommended to set
2696 this option on configurations that are not changed often, as it helps detect
2697 subtle mistakes and keep the configuration clean and forward-compatible. Note
2698 that "haproxy -c" will also report errors in such a case. This option is
2699 equivalent to command line argument "-dW".
2700
Emeric Brunf099e792010-09-27 12:05:28 +02002701
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010027023.4. Userlists
2703--------------
2704It is possible to control access to frontend/backend/listen sections or to
2705http stats by allowing only authenticated and authorized users. To do this,
2706it is required to create at least one userlist and to define users.
2707
2708userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002709 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002710 used to store authentication & authorization data for independent customers.
2711
2712group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002713 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002714 attach users to this group by using a comma separated list of names
2715 proceeded by "users" keyword.
2716
Cyril Bontéf0c60612010-02-06 14:44:47 +01002717user <username> [password|insecure-password <password>]
2718 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002719 Adds user <username> to the current userlist. Both secure (encrypted) and
2720 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002721 evaluated using the crypt(3) function, so depending on the system's
2722 capabilities, different algorithms are supported. For example, modern Glibc
2723 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2724 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002725
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002726 Attention: Be aware that using encrypted passwords might cause significantly
2727 increased CPU usage, depending on the number of requests, and the algorithm
2728 used. For any of the hashed variants, the password for each request must
2729 be processed through the chosen algorithm, before it can be compared to the
2730 value specified in the config file. Most current algorithms are deliberately
2731 designed to be expensive to compute to achieve resistance against brute
2732 force attacks. They do not simply salt/hash the clear text password once,
2733 but thousands of times. This can quickly become a major factor in haproxy's
2734 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002735
2736 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002737 userlist L1
2738 group G1 users tiger,scott
2739 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002740
Cyril Bontéf0c60612010-02-06 14:44:47 +01002741 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2742 user scott insecure-password elgato
2743 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002744
Cyril Bontéf0c60612010-02-06 14:44:47 +01002745 userlist L2
2746 group G1
2747 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002748
Cyril Bontéf0c60612010-02-06 14:44:47 +01002749 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2750 user scott insecure-password elgato groups G1,G2
2751 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002752
2753 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002754
Emeric Brunf099e792010-09-27 12:05:28 +02002755
27563.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002757----------
Emeric Brun94900952015-06-11 18:25:54 +02002758It is possible to propagate entries of any data-types in stick-tables between
2759several haproxy instances over TCP connections in a multi-master fashion. Each
2760instance pushes its local updates and insertions to remote peers. The pushed
2761values overwrite remote ones without aggregation. Interrupted exchanges are
2762automatically detected and recovered from the last known point.
2763In addition, during a soft restart, the old process connects to the new one
2764using such a TCP connection to push all its entries before the new process
2765tries to connect to other peers. That ensures very fast replication during a
2766reload, it typically takes a fraction of a second even for large tables.
2767Note that Server IDs are used to identify servers remotely, so it is important
2768that configurations look similar or at least that the same IDs are forced on
2769each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002770
2771peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002772 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002773 which is referenced by one or more stick-tables.
2774
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002775bind [<address>]:<port_range> [, ...] [param*]
2776 Defines the binding parameters of the local peer of this "peers" section.
2777 Such lines are not supported with "peer" line in the same "peers" section.
2778
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002779disabled
2780 Disables a peers section. It disables both listening and any synchronization
2781 related to this section. This is provided to disable synchronization of stick
2782 tables without having to comment out all "peers" references.
2783
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002784default-bind [param*]
2785 Defines the binding parameters for the local peer, excepted its address.
2786
2787default-server [param*]
2788 Change default options for a server in a "peers" section.
2789
2790 Arguments:
2791 <param*> is a list of parameters for this server. The "default-server"
2792 keyword accepts an important number of options and has a complete
2793 section dedicated to it. Please refer to section 5 for more
2794 details.
2795
2796
2797 See also: "server" and section 5 about server options
2798
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002799enable
2800 This re-enables a disabled peers section which was previously disabled.
2801
Jan Wagner3e678602020-12-17 22:22:32 +01002802log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailleb6f759b2019-11-05 09:57:45 +01002803 <facility> [<level> [<minlevel>]]
2804 "peers" sections support the same "log" keyword as for the proxies to
2805 log information about the "peers" listener. See "log" option for proxies for
2806 more details.
2807
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002808peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002809 Defines a peer inside a peers section.
2810 If <peername> is set to the local peer name (by default hostname, or forced
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002811 using "-L" command line option or "localpeer" global configuration setting),
2812 haproxy will listen for incoming remote peer connection on <ip>:<port>.
2813 Otherwise, <ip>:<port> defines where to connect to in order to join the
2814 remote peer, and <peername> is used at the protocol level to identify and
2815 validate the remote peer on the server side.
Emeric Brunf099e792010-09-27 12:05:28 +02002816
2817 During a soft restart, local peer <ip>:<port> is used by the old instance to
2818 connect the new one and initiate a complete replication (teaching process).
2819
2820 It is strongly recommended to have the exact same peers declaration on all
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002821 peers and to only rely on the "-L" command line argument or the "localpeer"
2822 global configuration setting to change the local peer name. This makes it
2823 easier to maintain coherent configuration files across all peers.
Emeric Brunf099e792010-09-27 12:05:28 +02002824
William Lallemandb2f07452015-05-12 14:27:13 +02002825 You may want to reference some environment variables in the address
2826 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002827
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002828 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2829 "server" keyword explanation below).
2830
2831server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002832 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002833 with a support for all "server" parameters found in 5.2 paragraph.
2834 If the underlying peer is local, <ip>:<port> parameters must not be present.
2835 These parameters must be provided on a "bind" line (see "bind" keyword
2836 of this "peers" section).
2837 Some of these parameters are irrelevant for "peers" sections.
2838
2839
Cyril Bontédc4d9032012-04-08 21:57:39 +02002840 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002841 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002842 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002843 peer haproxy1 192.168.0.1:1024
2844 peer haproxy2 192.168.0.2:1024
2845 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002846
2847 backend mybackend
2848 mode tcp
2849 balance roundrobin
2850 stick-table type ip size 20k peers mypeers
2851 stick on src
2852
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002853 server srv1 192.168.0.30:80
2854 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002855
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002856 Example:
2857 peers mypeers
2858 bind 127.0.0.11:10001 ssl crt mycerts/pem
2859 default-server ssl verify none
2860 server hostA 127.0.0.10:10000
2861 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002862
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002863
2864table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2865 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2866
2867 Configure a stickiness table for the current section. This line is parsed
2868 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002869 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002870 mandatory first parameter to designate the stick-table. Contrary to others
2871 sections, there may be several "table" lines in "peers" sections (see also
2872 "stick-table" keyword).
2873
2874 Also be aware of the fact that "peers" sections have their own stick-table
2875 namespaces to avoid collisions between stick-table names identical in
2876 different "peers" section. This is internally handled prepending the "peers"
2877 sections names to the name of the stick-tables followed by a '/' character.
2878 If somewhere else in the configuration file you have to refer to such
2879 stick-tables declared in "peers" sections you must use the prefixed version
2880 of the stick-table name as follows:
2881
2882 peers mypeers
2883 peer A ...
2884 peer B ...
2885 table t1 ...
2886
2887 frontend fe1
2888 tcp-request content track-sc0 src table mypeers/t1
2889
2890 This is also this prefixed version of the stick-table names which must be
2891 used to refer to stick-tables through the CLI.
2892
2893 About "peers" protocol, as only "peers" belonging to the same section may
2894 communicate with each others, there is no need to do such a distinction.
2895 Several "peers" sections may declare stick-tables with the same name.
2896 This is shorter version of the stick-table name which is sent over the network.
2897 There is only a '/' character as prefix to avoid stick-table name collisions between
2898 stick-tables declared as backends and stick-table declared in "peers" sections
2899 as follows in this weird but supported configuration:
2900
2901 peers mypeers
2902 peer A ...
2903 peer B ...
2904 table t1 type string size 10m store gpc0
2905
2906 backend t1
2907 stick-table type string size 10m store gpc0 peers mypeers
2908
Daniel Corbett67a82712020-07-06 23:01:19 -04002909 Here "t1" table declared in "mypeers" section has "mypeers/t1" as global name.
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002910 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2911 level the former table is named "/t1", the latter is again named "t1".
2912
Simon Horman51a1cf62015-02-03 13:00:44 +090029133.6. Mailers
2914------------
2915It is possible to send email alerts when the state of servers changes.
2916If configured email alerts are sent to each mailer that is configured
2917in a mailers section. Email is sent to mailers using SMTP.
2918
Pieter Baauw386a1272015-08-16 15:26:24 +02002919mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002920 Creates a new mailer list with the name <mailersect>. It is an
2921 independent section which is referenced by one or more proxies.
2922
2923mailer <mailername> <ip>:<port>
2924 Defines a mailer inside a mailers section.
2925
2926 Example:
2927 mailers mymailers
2928 mailer smtp1 192.168.0.1:587
2929 mailer smtp2 192.168.0.2:587
2930
2931 backend mybackend
2932 mode tcp
2933 balance roundrobin
2934
2935 email-alert mailers mymailers
2936 email-alert from test1@horms.org
2937 email-alert to test2@horms.org
2938
2939 server srv1 192.168.0.30:80
2940 server srv2 192.168.0.31:80
2941
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002942timeout mail <time>
2943 Defines the time available for a mail/connection to be made and send to
2944 the mail-server. If not defined the default value is 10 seconds. To allow
2945 for at least two SYN-ACK packets to be send during initial TCP handshake it
2946 is advised to keep this value above 4 seconds.
2947
2948 Example:
2949 mailers mymailers
2950 timeout mail 20s
2951 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002952
William Lallemandc9515522019-06-12 16:32:11 +020029533.7. Programs
2954-------------
2955In master-worker mode, it is possible to launch external binaries with the
2956master, these processes are called programs. These programs are launched and
2957managed the same way as the workers.
2958
2959During a reload of HAProxy, those processes are dealing with the same
2960sequence as a worker:
2961
2962 - the master is re-executed
2963 - the master sends a SIGUSR1 signal to the program
2964 - if "option start-on-reload" is not disabled, the master launches a new
2965 instance of the program
2966
2967During a stop, or restart, a SIGTERM is sent to the programs.
2968
2969program <name>
2970 This is a new program section, this section will create an instance <name>
2971 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2972 the management guide).
2973
2974command <command> [arguments*]
2975 Define the command to start with optional arguments. The command is looked
2976 up in the current PATH if it does not include an absolute path. This is a
2977 mandatory option of the program section. Arguments containing spaces must
2978 be enclosed in quotes or double quotes or be prefixed by a backslash.
2979
Andrew Heberle97236962019-07-12 11:50:26 +08002980user <user name>
2981 Changes the executed command user ID to the <user name> from /etc/passwd.
2982 See also "group".
2983
2984group <group name>
2985 Changes the executed command group ID to the <group name> from /etc/group.
2986 See also "user".
2987
William Lallemandc9515522019-06-12 16:32:11 +02002988option start-on-reload
2989no option start-on-reload
2990 Start (or not) a new instance of the program upon a reload of the master.
2991 The default is to start a new instance. This option may only be used in a
2992 program section.
2993
2994
Christopher Faulet76edc0f2020-01-13 15:52:01 +010029953.8. HTTP-errors
2996----------------
2997
2998It is possible to globally declare several groups of HTTP errors, to be
2999imported afterwards in any proxy section. Same group may be referenced at
3000several places and can be fully or partially imported.
3001
3002http-errors <name>
3003 Create a new http-errors group with the name <name>. It is an independent
3004 section that may be referenced by one or more proxies using its name.
3005
3006errorfile <code> <file>
3007 Associate a file contents to an HTTP error code
3008
3009 Arguments :
3010 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02003011 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01003012 425, 429, 500, 501, 502, 503, and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003013
3014 <file> designates a file containing the full HTTP response. It is
3015 recommended to follow the common practice of appending ".http" to
3016 the filename so that people do not confuse the response with HTML
3017 error pages, and to use absolute paths, since files are read
3018 before any chroot is performed.
3019
3020 Please referrers to "errorfile" keyword in section 4 for details.
3021
3022 Example:
3023 http-errors website-1
3024 errorfile 400 /etc/haproxy/errorfiles/site1/400.http
3025 errorfile 404 /etc/haproxy/errorfiles/site1/404.http
3026 errorfile 408 /dev/null # work around Chrome pre-connect bug
3027
3028 http-errors website-2
3029 errorfile 400 /etc/haproxy/errorfiles/site2/400.http
3030 errorfile 404 /etc/haproxy/errorfiles/site2/404.http
3031 errorfile 408 /dev/null # work around Chrome pre-connect bug
3032
Emeric Brun99c453d2020-05-25 15:01:04 +020030333.9. Rings
3034----------
3035
3036It is possible to globally declare ring-buffers, to be used as target for log
3037servers or traces.
3038
3039ring <ringname>
3040 Creates a new ring-buffer with name <ringname>.
3041
3042description <text>
Daniel Corbett67a82712020-07-06 23:01:19 -04003043 The description is an optional description string of the ring. It will
Emeric Brun99c453d2020-05-25 15:01:04 +02003044 appear on CLI. By default, <name> is reused to fill this field.
3045
3046format <format>
3047 Format used to store events into the ring buffer.
3048
3049 Arguments:
3050 <format> is the log format used when generating syslog messages. It may be
3051 one of the following :
3052
3053 iso A message containing only the ISO date, followed by the text.
3054 The PID, process name and system name are omitted. This is
3055 designed to be used with a local log server.
3056
Emeric Brun0237c4e2020-11-27 16:24:34 +01003057 local Analog to rfc3164 syslog message format except that hostname
3058 field is stripped. This is the default.
3059 Note: option "log-send-hostname" switches the default to
3060 rfc3164.
3061
Emeric Brun99c453d2020-05-25 15:01:04 +02003062 raw A message containing only the text. The level, PID, date, time,
3063 process name and system name are omitted. This is designed to be
3064 used in containers or during development, where the severity
3065 only depends on the file descriptor used (stdout/stderr). This
3066 is the default.
3067
Emeric Brun0237c4e2020-11-27 16:24:34 +01003068 rfc3164 The RFC3164 syslog message format.
Emeric Brun99c453d2020-05-25 15:01:04 +02003069 (https://tools.ietf.org/html/rfc3164)
3070
3071 rfc5424 The RFC5424 syslog message format.
3072 (https://tools.ietf.org/html/rfc5424)
3073
3074 short A message containing only a level between angle brackets such as
3075 '<3>', followed by the text. The PID, date, time, process name
3076 and system name are omitted. This is designed to be used with a
3077 local log server. This format is compatible with what the systemd
3078 logger consumes.
3079
Emeric Brun54648852020-07-06 15:54:06 +02003080 priority A message containing only a level plus syslog facility between angle
3081 brackets such as '<63>', followed by the text. The PID, date, time,
3082 process name and system name are omitted. This is designed to be used
3083 with a local log server.
3084
Emeric Brun99c453d2020-05-25 15:01:04 +02003085 timed A message containing only a level between angle brackets such as
3086 '<3>', followed by ISO date and by the text. The PID, process
3087 name and system name are omitted. This is designed to be
3088 used with a local log server.
3089
3090maxlen <length>
3091 The maximum length of an event message stored into the ring,
3092 including formatted header. If an event message is longer than
3093 <length>, it will be truncated to this length.
3094
Emeric Brun494c5052020-05-28 11:13:15 +02003095server <name> <address> [param*]
3096 Used to configure a syslog tcp server to forward messages from ring buffer.
3097 This supports for all "server" parameters found in 5.2 paragraph. Some of
3098 these parameters are irrelevant for "ring" sections. Important point: there
3099 is little reason to add more than one server to a ring, because all servers
3100 will receive the exact same copy of the ring contents, and as such the ring
3101 will progress at the speed of the slowest server. If one server does not
3102 respond, it will prevent old messages from being purged and may block new
3103 messages from being inserted into the ring. The proper way to send messages
3104 to multiple servers is to use one distinct ring per log server, not to
Emeric Brun97556472020-05-30 01:42:45 +02003105 attach multiple servers to the same ring. Note that specific server directive
3106 "log-proto" is used to set the protocol used to send messages.
Emeric Brun494c5052020-05-28 11:13:15 +02003107
Emeric Brun99c453d2020-05-25 15:01:04 +02003108size <size>
3109 This is the optional size in bytes for the ring-buffer. Default value is
3110 set to BUFSIZE.
3111
Emeric Brun494c5052020-05-28 11:13:15 +02003112timeout connect <timeout>
3113 Set the maximum time to wait for a connection attempt to a server to succeed.
3114
3115 Arguments :
3116 <timeout> is the timeout value specified in milliseconds by default, but
3117 can be in any other unit if the number is suffixed by the unit,
3118 as explained at the top of this document.
3119
3120timeout server <timeout>
3121 Set the maximum time for pending data staying into output buffer.
3122
3123 Arguments :
3124 <timeout> is the timeout value specified in milliseconds by default, but
3125 can be in any other unit if the number is suffixed by the unit,
3126 as explained at the top of this document.
3127
Emeric Brun99c453d2020-05-25 15:01:04 +02003128 Example:
3129 global
3130 log ring@myring local7
3131
3132 ring myring
3133 description "My local buffer"
3134 format rfc3164
3135 maxlen 1200
3136 size 32764
Emeric Brun494c5052020-05-28 11:13:15 +02003137 timeout connect 5s
3138 timeout server 10s
Emeric Brun97556472020-05-30 01:42:45 +02003139 server mysyslogsrv 127.0.0.1:6514 log-proto octet-count
Emeric Brun99c453d2020-05-25 15:01:04 +02003140
Emeric Brun12941c82020-07-07 14:19:42 +020031413.10. Log forwarding
3142-------------------
3143
3144It is possible to declare one or multiple log forwarding section,
3145haproxy will forward all received log messages to a log servers list.
3146
3147log-forward <name>
3148 Creates a new log forwarder proxy identified as <name>.
3149
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003150backlog <conns>
3151 Give hints to the system about the approximate listen backlog desired size
3152 on connections accept.
3153
3154bind <addr> [param*]
3155 Used to configure a stream log listener to receive messages to forward.
Emeric Brunda46c1c2020-10-08 08:39:02 +02003156 This supports the "bind" parameters found in 5.1 paragraph including
3157 those about ssl but some statements such as "alpn" may be irrelevant for
3158 syslog protocol over TCP.
3159 Those listeners support both "Octet Counting" and "Non-Transparent-Framing"
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003160 modes as defined in rfc-6587.
3161
Willy Tarreau76aaa7f2020-09-16 15:07:22 +02003162dgram-bind <addr> [param*]
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003163 Used to configure a datagram log listener to receive messages to forward.
3164 Addresses must be in IPv4 or IPv6 form,followed by a port. This supports
3165 for some of the "bind" parameters found in 5.1 paragraph among which
3166 "interface", "namespace" or "transparent", the other ones being
Willy Tarreau26ff5da2020-09-16 15:22:19 +02003167 silently ignored as irrelevant for UDP/syslog case.
Emeric Brun12941c82020-07-07 14:19:42 +02003168
3169log global
Jan Wagner3e678602020-12-17 22:22:32 +01003170log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Emeric Brun12941c82020-07-07 14:19:42 +02003171 <facility> [<level> [<minlevel>]]
3172 Used to configure target log servers. See more details on proxies
3173 documentation.
3174 If no format specified, haproxy tries to keep the incoming log format.
3175 Configured facility is ignored, except if incoming message does not
3176 present a facility but one is mandatory on the outgoing format.
3177 If there is no timestamp available in the input format, but the field
3178 exists in output format, haproxy will use the local date.
3179
3180 Example:
3181 global
3182 log stderr format iso local7
3183
3184 ring myring
3185 description "My local buffer"
3186 format rfc5424
3187 maxlen 1200
3188 size 32764
3189 timeout connect 5s
3190 timeout server 10s
3191 # syslog tcp server
3192 server mysyslogsrv 127.0.0.1:514 log-proto octet-count
3193
3194 log-forward sylog-loadb
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003195 dgram-bind 127.0.0.1:1514
3196 bind 127.0.0.1:1514
Emeric Brun12941c82020-07-07 14:19:42 +02003197 # all messages on stderr
3198 log global
3199 # all messages on local tcp syslog server
3200 log ring@myring local0
3201 # load balance messages on 4 udp syslog servers
3202 log 127.0.0.1:10001 sample 1:4 local0
3203 log 127.0.0.1:10002 sample 2:4 local0
3204 log 127.0.0.1:10003 sample 3:4 local0
3205 log 127.0.0.1:10004 sample 4:4 local0
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003206
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003207maxconn <conns>
3208 Fix the maximum number of concurrent connections on a log forwarder.
3209 10 is the default.
3210
3211timeout client <timeout>
3212 Set the maximum inactivity time on the client side.
3213
Willy Tarreauc57f0e22009-05-10 13:12:33 +020032144. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02003215----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003216
Willy Tarreau6a06a402007-07-15 20:15:28 +02003217Proxy configuration can be located in a set of sections :
Willy Tarreau7c0b4d82021-02-12 14:58:08 +01003218 - defaults [<name>] [ from <defaults_name> ]
3219 - frontend <name> [ from <defaults_name> ]
3220 - backend <name> [ from <defaults_name> ]
3221 - listen <name> [ from <defaults_name> ]
Willy Tarreau6a06a402007-07-15 20:15:28 +02003222
3223A "frontend" section describes a set of listening sockets accepting client
3224connections.
3225
3226A "backend" section describes a set of servers to which the proxy will connect
3227to forward incoming connections.
3228
3229A "listen" section defines a complete proxy with its frontend and backend
3230parts combined in one section. It is generally useful for TCP-only traffic.
3231
Willy Tarreau7c0b4d82021-02-12 14:58:08 +01003232A "defaults" section resets all settings to the documented ones and presets new
3233ones for use by subsequent sections. All of "frontend", "backend" and "listen"
3234sections always take their initial settings from a defaults section, by default
3235the latest one that appears before the newly created section. It is possible to
3236explicitly designate a specific "defaults" section to load the initial settings
3237from by indicating its name on the section line after the optional keyword
3238"from". While "defaults" section do not impose a name, this use is encouraged
3239for better readability. It is also the only way to designate a specific section
3240to use instead of the default previous one. Since "defaults" section names are
3241optional, by default a very permissive check is applied on their name and these
3242are even permitted to overlap. However if a "defaults" section is referenced by
3243any other section, its name must comply with the syntax imposed on all proxy
3244names, and this name must be unique among the defaults sections. Please note
3245that regardless of what is currently permitted, it is recommended to avoid
3246duplicate section names in general and to respect the same syntax as for proxy
3247names. This rule might be enforced in a future version.
3248
3249Note that it is even possible for a defaults section to take its initial
3250settings from another one, and as such, inherit settings across multiple levels
3251of defaults sections. This can be convenient to establish certain configuration
3252profiles to carry groups of default settings (e.g. TCP vs HTTP or short vs long
3253timeouts) but can quickly become confusing to follow.
3254
Willy Tarreau0ba27502007-12-24 16:55:16 +01003255All proxy names must be formed from upper and lower case letters, digits,
3256'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
3257case-sensitive, which means that "www" and "WWW" are two different proxies.
3258
3259Historically, all proxy names could overlap, it just caused troubles in the
3260logs. Since the introduction of content switching, it is mandatory that two
3261proxies with overlapping capabilities (frontend/backend) have different names.
3262However, it is still permitted that a frontend and a backend share the same
3263name, as this configuration seems to be commonly encountered.
3264
3265Right now, two major proxy modes are supported : "tcp", also known as layer 4,
3266and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003267bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003268protocol, and can interact with it by allowing, blocking, switching, adding,
3269modifying, or removing arbitrary contents in requests or responses, based on
3270arbitrary criteria.
3271
Willy Tarreau70dffda2014-01-30 03:07:23 +01003272In HTTP mode, the processing applied to requests and responses flowing over
3273a connection depends in the combination of the frontend's HTTP options and
Julien Pivotto21ad3152019-12-10 13:11:17 +01003274the backend's. HAProxy supports 3 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01003275
3276 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
3277 requests and responses are processed, and connections remain open but idle
3278 between responses and new requests.
3279
Willy Tarreau70dffda2014-01-30 03:07:23 +01003280 - SCL: server close ("option http-server-close") : the server-facing
3281 connection is closed after the end of the response is received, but the
3282 client-facing connection remains open.
3283
Christopher Faulet315b39c2018-09-21 16:26:19 +02003284 - CLO: close ("option httpclose"): the connection is closed after the end of
3285 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003286
3287The effective mode that will be applied to a connection passing through a
3288frontend and a backend can be determined by both proxy modes according to the
3289following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02003290weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003291
Christopher Faulet315b39c2018-09-21 16:26:19 +02003292 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01003293
Christopher Faulet315b39c2018-09-21 16:26:19 +02003294 | KAL | SCL | CLO
3295 ----+-----+-----+----
3296 KAL | KAL | SCL | CLO
3297 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02003298 mode SCL | SCL | SCL | CLO
3299 ----+-----+-----+----
3300 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01003301
Christopher Faulet4d37e532021-03-26 14:44:00 +01003302It is possible to chain a TCP frontend to an HTTP backend. It is pointless if
3303only HTTP traffic is handled. But It may be used to handle several protocols
3304into the same frontend. It this case, the client's connection is first handled
3305as a raw tcp connection before being upgraded to HTTP. Before the upgrade, the
3306content processings are performend on raw data. Once upgraded, data are parsed
3307and stored using an internal representation called HTX and it is no longer
3308possible to rely on raw representation. There is no way to go back.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003309
Christopher Faulet4d37e532021-03-26 14:44:00 +01003310There are two kind of upgrades, in-place upgrades and destructive upgrades. The
3311first ones concern the TCP to HTTP/1 upgrades. In HTTP/1, the request
3312processings are serialized, thus the applicative stream can be preserved. The
3313second ones concern the TCP to HTTP/2 upgrades. Because it is a multiplexed
3314protocol, the applicative stream cannot be associated to any HTTP/2 stream and
3315is destroyed. New applicative streams are then created when HAProxy receives
3316new HTTP/2 streams at the lower level, in the H2 multiplexer. It is important
3317to understand this difference because that drastically change the way to
3318process data. When an HTTP/1 upgrade is performed, the content processings
3319already performed on raw data are neither lost nor reexecuted while for an
3320HTTP/2 upgrade, applicative streams are distinct and all frontend rules are
3321evaluated systematically on each one. And as said, the first stream, the TCP
3322one, is destroyed, but only after the frontend rules were evaluated.
3323
3324There is another importnat point to understand when HTTP processings are
3325performed from a TCP proxy. While HAProxy is able to parse HTTP/1 in-fly from
3326tcp-request content rules, it is not possible for HTTP/2. Only the HTTP/2
3327preface can be parsed. This is a huge limitation regarding the HTTP content
3328analysis in TCP. Concretely it is only possible to know if received data are
3329HTTP. For instance, it is not possible to choose a backend based on the Host
3330header value while it is trivial in HTTP/1. Hopefully, there is a solution to
3331mitigate this drawback.
3332
3333It exists two way to perform HTTP upgrades. The first one, the historical
3334method, is to select an HTTP backend. The upgrade happens when the backend is
3335set. Thus, for in-place upgrades, only the backend configuration is considered
3336in the HTTP data processing. For destructive upgrades, the applicative stream
3337is destroyed, thus its processing is stopped. With this method, possibilities
3338to choose a backend with an HTTP/2 connection are really limited, as mentioned
3339above, and a bit useless because the stream is destroyed. The second method is
3340to upgrade during the tcp-request content rules evaluation, thanks to the
3341"switch-mode http" action. In this case, the upgrade is performed in the
3342frontend context and it is possible to define HTTP directives in this
3343frontend. For in-place upgrades, it offers all the power of the HTTP analysis
3344as soon as possible. It is not that far from an HTTP frontend. For destructive
3345upgrades, it does not change anything except it is useless to choose a backend
3346on limited information. It is of course the recommended method. Thus, testing
3347the request protocol from the tcp-request content rules to perform an HTTP
3348upgrade is enough. All the remaining HTTP manipulation may be moved to the
3349frontend http-request ruleset. But keep in mind that tcp-request content rules
3350remains evaluated on each streams, that can't be changed.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003351
Willy Tarreauc57f0e22009-05-10 13:12:33 +020033524.1. Proxy keywords matrix
3353--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003354
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003355The following list of keywords is supported. Most of them may only be used in a
3356limited set of section types. Some of them are marked as "deprecated" because
3357they are inherited from an old syntax which may be confusing or functionally
3358limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01003359marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003360option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02003361and must be disabled for a specific instance. Such options may also be prefixed
3362with "default" in order to restore default settings regardless of what has been
3363specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003364
Willy Tarreau6a06a402007-07-15 20:15:28 +02003365
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003366 keyword defaults frontend listen backend
3367------------------------------------+----------+----------+---------+---------
3368acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003369backlog X X X -
3370balance X - X X
3371bind - X X -
3372bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003373capture cookie - X X -
3374capture request header - X X -
3375capture response header - X X -
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003376clitcpka-cnt X X X -
3377clitcpka-idle X X X -
3378clitcpka-intvl X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02003379compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003380cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003381declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003382default-server X - X X
3383default_backend X X X -
3384description - X X X
3385disabled X X X X
3386dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09003387email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09003388email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09003389email-alert mailers X X X X
3390email-alert myhostname X X X X
3391email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003392enabled X X X X
3393errorfile X X X X
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003394errorfiles X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003395errorloc X X X X
3396errorloc302 X X X X
3397-- keyword -------------------------- defaults - frontend - listen -- backend -
3398errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003399force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003400filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003401fullconn X - X X
3402grace X X X X
3403hash-type X - X X
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01003404http-after-response - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003405http-check comment X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003406http-check connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003407http-check disable-on-404 X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003408http-check expect X - X X
Peter Gervai8912ae62020-06-11 18:26:36 +02003409http-check send X - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02003410http-check send-state X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003411http-check set-var X - X X
3412http-check unset-var X - X X
Christopher Faulet3b967c12020-05-15 15:47:44 +02003413http-error X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003414http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003415http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02003416http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02003417http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003418id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003419ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003420load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02003421log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01003422log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02003423log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01003424log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02003425max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003426maxconn X X X -
3427mode X X X X
3428monitor fail - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003429monitor-uri X X X -
3430option abortonclose (*) X - X X
3431option accept-invalid-http-request (*) X X X -
3432option accept-invalid-http-response (*) X - X X
3433option allbackups (*) X - X X
3434option checkcache (*) X - X X
3435option clitcpka (*) X X X -
3436option contstats (*) X X X -
Christopher Faulet89aed322020-06-02 17:33:56 +02003437option disable-h2-upgrade (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003438option dontlog-normal (*) X X X -
3439option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003440-- keyword -------------------------- defaults - frontend - listen -- backend -
3441option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02003442option h1-case-adjust-bogus-client (*) X X X -
3443option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02003444option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02003445option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01003446option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02003447option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02003448option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003449option http-server-close (*) X X X X
3450option http-use-proxy-header (*) X X X -
3451option httpchk X - X X
3452option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01003453option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003454option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003455option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02003456option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003457option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003458option log-health-checks (*) X - X X
3459option log-separate-errors (*) X X X -
3460option logasap (*) X X X -
3461option mysql-check X - X X
3462option nolinger (*) X X X X
3463option originalto X X X X
3464option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02003465option pgsql-check X - X X
3466option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003467option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02003468option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003469option smtpchk X - X X
3470option socket-stats (*) X X X -
3471option splice-auto (*) X X X X
3472option splice-request (*) X X X X
3473option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01003474option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003475option srvtcpka (*) X - X X
3476option ssl-hello-chk X - X X
3477-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01003478option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003479option tcp-smart-accept (*) X X X -
3480option tcp-smart-connect (*) X - X X
3481option tcpka X X X X
3482option tcplog X X X X
3483option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003484external-check command X - X X
3485external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003486persist rdp-cookie X - X X
3487rate-limit sessions X X X -
3488redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003489-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003490retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02003491retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003492server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003493server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02003494server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003495source X - X X
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003496srvtcpka-cnt X - X X
3497srvtcpka-idle X - X X
3498srvtcpka-intvl X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02003499stats admin - X X X
3500stats auth X X X X
3501stats enable X X X X
3502stats hide-version X X X X
3503stats http-request - X X X
3504stats realm X X X X
3505stats refresh X X X X
3506stats scope X X X X
3507stats show-desc X X X X
3508stats show-legends X X X X
3509stats show-node X X X X
3510stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003511-- keyword -------------------------- defaults - frontend - listen -- backend -
3512stick match - - X X
3513stick on - - X X
3514stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02003515stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01003516stick-table - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003517tcp-check comment X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003518tcp-check connect X - X X
3519tcp-check expect X - X X
3520tcp-check send X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003521tcp-check send-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003522tcp-check send-binary X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003523tcp-check send-binary-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003524tcp-check set-var X - X X
3525tcp-check unset-var X - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02003526tcp-request connection - X X -
3527tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02003528tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02003529tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02003530tcp-response content - - X X
3531tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003532timeout check X - X X
3533timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02003534timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003535timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003536timeout http-keep-alive X X X X
3537timeout http-request X X X X
3538timeout queue X - X X
3539timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02003540timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003541timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02003542timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003543transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01003544unique-id-format X X X -
3545unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003546use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02003547use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02003548use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003549------------------------------------+----------+----------+---------+---------
3550 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02003551
Willy Tarreau0ba27502007-12-24 16:55:16 +01003552
Willy Tarreauc57f0e22009-05-10 13:12:33 +020035534.2. Alphabetically sorted keywords reference
3554---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003555
3556This section provides a description of each keyword and its usage.
3557
3558
3559acl <aclname> <criterion> [flags] [operator] <value> ...
3560 Declare or complete an access list.
3561 May be used in sections : defaults | frontend | listen | backend
3562 no | yes | yes | yes
3563 Example:
3564 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
3565 acl invalid_src src_port 0:1023
3566 acl local_dst hdr(host) -i localhost
3567
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003568 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003569
3570
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003571backlog <conns>
3572 Give hints to the system about the approximate listen backlog desired size
3573 May be used in sections : defaults | frontend | listen | backend
3574 yes | yes | yes | no
3575 Arguments :
3576 <conns> is the number of pending connections. Depending on the operating
3577 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02003578 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003579
3580 In order to protect against SYN flood attacks, one solution is to increase
3581 the system's SYN backlog size. Depending on the system, sometimes it is just
3582 tunable via a system parameter, sometimes it is not adjustable at all, and
3583 sometimes the system relies on hints given by the application at the time of
3584 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
3585 to the listen() syscall. On systems which can make use of this value, it can
3586 sometimes be useful to be able to specify a different value, hence this
3587 backlog parameter.
3588
3589 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
3590 used as a hint and the system accepts up to the smallest greater power of
3591 two, and never more than some limits (usually 32768).
3592
3593 See also : "maxconn" and the target operating system's tuning guide.
3594
3595
Willy Tarreau0ba27502007-12-24 16:55:16 +01003596balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02003597balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003598 Define the load balancing algorithm to be used in a backend.
3599 May be used in sections : defaults | frontend | listen | backend
3600 yes | no | yes | yes
3601 Arguments :
3602 <algorithm> is the algorithm used to select a server when doing load
3603 balancing. This only applies when no persistence information
3604 is available, or when a connection is redispatched to another
3605 server. <algorithm> may be one of the following :
3606
3607 roundrobin Each server is used in turns, according to their weights.
3608 This is the smoothest and fairest algorithm when the server's
3609 processing time remains equally distributed. This algorithm
3610 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02003611 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08003612 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02003613 large farms, when a server becomes up after having been down
3614 for a very short time, it may sometimes take a few hundreds
3615 requests for it to be re-integrated into the farm and start
3616 receiving traffic. This is normal, though very rare. It is
3617 indicated here in case you would have the chance to observe
3618 it, so that you don't worry.
3619
3620 static-rr Each server is used in turns, according to their weights.
3621 This algorithm is as similar to roundrobin except that it is
3622 static, which means that changing a server's weight on the
3623 fly will have no effect. On the other hand, it has no design
3624 limitation on the number of servers, and when a server goes
3625 up, it is always immediately reintroduced into the farm, once
3626 the full map is recomputed. It also uses slightly less CPU to
3627 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003628
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003629 leastconn The server with the lowest number of connections receives the
3630 connection. Round-robin is performed within groups of servers
3631 of the same load to ensure that all servers will be used. Use
3632 of this algorithm is recommended where very long sessions are
3633 expected, such as LDAP, SQL, TSE, etc... but is not very well
3634 suited for protocols using short sessions such as HTTP. This
3635 algorithm is dynamic, which means that server weights may be
Willy Tarreau8c855f62020-10-22 17:41:45 +02003636 adjusted on the fly for slow starts for instance. It will
3637 also consider the number of queued connections in addition to
3638 the established ones in order to minimize queuing.
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003639
Willy Tarreauf09c6602012-02-13 17:12:08 +01003640 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003641 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01003642 identifier to the highest (see server parameter "id"), which
3643 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02003644 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01003645 not make sense to use this algorithm without setting maxconn.
3646 The purpose of this algorithm is to always use the smallest
3647 number of servers so that extra servers can be powered off
3648 during non-intensive hours. This algorithm ignores the server
3649 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02003650 or IMAP than HTTP, though it can be useful there too. In
3651 order to use this algorithm efficiently, it is recommended
3652 that a cloud controller regularly checks server usage to turn
3653 them off when unused, and regularly checks backend queue to
3654 turn new servers on when the queue inflates. Alternatively,
3655 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01003656
Willy Tarreau0ba27502007-12-24 16:55:16 +01003657 source The source IP address is hashed and divided by the total
3658 weight of the running servers to designate which server will
3659 receive the request. This ensures that the same client IP
3660 address will always reach the same server as long as no
3661 server goes down or up. If the hash result changes due to the
3662 number of running servers changing, many clients will be
3663 directed to a different server. This algorithm is generally
3664 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003665 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01003666 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003667 static by default, which means that changing a server's
3668 weight on the fly will have no effect, but this can be
3669 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003670
Oskar Stolc8dc41842012-05-19 10:19:54 +01003671 uri This algorithm hashes either the left part of the URI (before
3672 the question mark) or the whole URI (if the "whole" parameter
3673 is present) and divides the hash value by the total weight of
3674 the running servers. The result designates which server will
3675 receive the request. This ensures that the same URI will
3676 always be directed to the same server as long as no server
3677 goes up or down. This is used with proxy caches and
3678 anti-virus proxies in order to maximize the cache hit rate.
3679 Note that this algorithm may only be used in an HTTP backend.
3680 This algorithm is static by default, which means that
3681 changing a server's weight on the fly will have no effect,
3682 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003683
Oskar Stolc8dc41842012-05-19 10:19:54 +01003684 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003685 "depth", both followed by a positive integer number. These
3686 options may be helpful when it is needed to balance servers
3687 based on the beginning of the URI only. The "len" parameter
3688 indicates that the algorithm should only consider that many
3689 characters at the beginning of the URI to compute the hash.
3690 Note that having "len" set to 1 rarely makes sense since most
3691 URIs start with a leading "/".
3692
3693 The "depth" parameter indicates the maximum directory depth
3694 to be used to compute the hash. One level is counted for each
3695 slash in the request. If both parameters are specified, the
3696 evaluation stops when either is reached.
3697
Willy Tarreau57a37412020-09-23 08:56:29 +02003698 A "path-only" parameter indicates that the hashing key starts
3699 at the first '/' of the path. This can be used to ignore the
3700 authority part of absolute URIs, and to make sure that HTTP/1
3701 and HTTP/2 URIs will provide the same hash.
3702
Willy Tarreau0ba27502007-12-24 16:55:16 +01003703 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003704 the query string of each HTTP GET request.
3705
3706 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02003707 request entity will be searched for the parameter argument,
3708 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02003709 ('?') in the URL. The message body will only start to be
3710 analyzed once either the advertised amount of data has been
3711 received or the request buffer is full. In the unlikely event
3712 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02003713 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02003714 be randomly balanced if at all. This keyword used to support
3715 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003716
3717 If the parameter is found followed by an equal sign ('=') and
3718 a value, then the value is hashed and divided by the total
3719 weight of the running servers. The result designates which
3720 server will receive the request.
3721
3722 This is used to track user identifiers in requests and ensure
3723 that a same user ID will always be sent to the same server as
3724 long as no server goes up or down. If no value is found or if
3725 the parameter is not found, then a round robin algorithm is
3726 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003727 backend. This algorithm is static by default, which means
3728 that changing a server's weight on the fly will have no
3729 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003730
Cyril Bontédc4d9032012-04-08 21:57:39 +02003731 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
3732 request. Just as with the equivalent ACL 'hdr()' function,
3733 the header name in parenthesis is not case sensitive. If the
3734 header is absent or if it does not contain any value, the
3735 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01003736
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003737 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01003738 reducing the hash algorithm to the main domain part with some
3739 specific headers such as 'Host'. For instance, in the Host
3740 value "haproxy.1wt.eu", only "1wt" will be considered.
3741
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003742 This algorithm is static by default, which means that
3743 changing a server's weight on the fly will have no effect,
3744 but this can be changed using "hash-type".
3745
Willy Tarreau21c741a2019-01-14 18:14:27 +01003746 random
3747 random(<draws>)
3748 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02003749 hashing function. This means that the servers' weights are
3750 respected, dynamic weight changes immediately take effect, as
3751 well as new server additions. Random load balancing can be
3752 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01003753 or removed as it may avoid the hammering effect that could
3754 result from roundrobin or leastconn in this situation. The
3755 hash-balance-factor directive can be used to further improve
3756 fairness of the load balancing, especially in situations
3757 where servers show highly variable response times. When an
3758 argument <draws> is present, it must be an integer value one
3759 or greater, indicating the number of draws before selecting
3760 the least loaded of these servers. It was indeed demonstrated
3761 that picking the least loaded of two servers is enough to
3762 significantly improve the fairness of the algorithm, by
3763 always avoiding to pick the most loaded server within a farm
3764 and getting rid of any bias that could be induced by the
3765 unfair distribution of the consistent list. Higher values N
3766 will take away N-1 of the highest loaded servers at the
3767 expense of performance. With very high values, the algorithm
3768 will converge towards the leastconn's result but much slower.
3769 The default value is 2, which generally shows very good
3770 distribution and performance. This algorithm is also known as
3771 the Power of Two Random Choices and is described here :
3772 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02003773
Emeric Brun736aa232009-06-30 17:56:00 +02003774 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02003775 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02003776 The RDP cookie <name> (or "mstshash" if omitted) will be
3777 looked up and hashed for each incoming TCP request. Just as
3778 with the equivalent ACL 'req_rdp_cookie()' function, the name
3779 is not case-sensitive. This mechanism is useful as a degraded
3780 persistence mode, as it makes it possible to always send the
3781 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003782 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02003783 used instead.
3784
3785 Note that for this to work, the frontend must ensure that an
3786 RDP cookie is already present in the request buffer. For this
3787 you must use 'tcp-request content accept' rule combined with
3788 a 'req_rdp_cookie_cnt' ACL.
3789
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003790 This algorithm is static by default, which means that
3791 changing a server's weight on the fly will have no effect,
3792 but this can be changed using "hash-type".
3793
Cyril Bontédc4d9032012-04-08 21:57:39 +02003794 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09003795
Willy Tarreau0ba27502007-12-24 16:55:16 +01003796 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003797 algorithms. Right now, only "url_param" and "uri" support an
3798 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003799
Willy Tarreau3cd9af22009-03-15 14:06:41 +01003800 The load balancing algorithm of a backend is set to roundrobin when no other
3801 algorithm, mode nor option have been set. The algorithm may only be set once
3802 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003803
Lukas Tribus80512b12018-10-27 20:07:40 +02003804 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05003805 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02003806 to be routed to different backend servers, breaking the invalid assumptions
3807 NTLM relies on.
3808
Willy Tarreau0ba27502007-12-24 16:55:16 +01003809 Examples :
3810 balance roundrobin
3811 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003812 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01003813 balance hdr(User-Agent)
3814 balance hdr(host)
3815 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003816
3817 Note: the following caveats and limitations on using the "check_post"
3818 extension with "url_param" must be considered :
3819
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003820 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003821 to determine if the parameters will be found in the body or entity which
3822 may contain binary data. Therefore another method may be required to
3823 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02003824 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003825
3826 - using a <max_wait> value larger than the request buffer size does not
3827 make sense and is useless. The buffer size is set at build time, and
3828 defaults to 16 kB.
3829
3830 - Content-Encoding is not supported, the parameter search will probably
3831 fail; and load balancing will fall back to Round Robin.
3832
3833 - Expect: 100-continue is not supported, load balancing will fall back to
3834 Round Robin.
3835
Lukas Tribus23953682017-04-28 13:24:30 +00003836 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003837 If the entire parameter value is not present in the first chunk, the
3838 selection of server is undefined (actually, defined by how little
3839 actually appeared in the first chunk).
3840
3841 - This feature does not support generation of a 100, 411 or 501 response.
3842
3843 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003844 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003845 white space or control characters are found, indicating the end of what
3846 might be a URL parameter list. This is probably not a concern with SGML
3847 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003848
Willy Tarreau294d0f02015-08-10 19:40:12 +02003849 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003850
3851
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003852bind [<address>]:<port_range> [, ...] [param*]
3853bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003854 Define one or several listening addresses and/or ports in a frontend.
3855 May be used in sections : defaults | frontend | listen | backend
3856 no | yes | yes | no
3857 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003858 <address> is optional and can be a host name, an IPv4 address, an IPv6
3859 address, or '*'. It designates the address the frontend will
3860 listen on. If unset, all IPv4 addresses of the system will be
3861 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01003862 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01003863 Optionally, an address family prefix may be used before the
3864 address to force the family regardless of the address format,
3865 which can be useful to specify a path to a unix socket with
3866 no slash ('/'). Currently supported prefixes are :
3867 - 'ipv4@' -> address is always IPv4
3868 - 'ipv6@' -> address is always IPv6
Emeric Brun3835c0d2020-07-07 09:46:09 +02003869 - 'udp@' -> address is resolved as IPv4 or IPv6 and
Emeric Brun12941c82020-07-07 14:19:42 +02003870 protocol UDP is used. Currently those listeners are
3871 supported only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003872 - 'udp4@' -> address is always IPv4 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003873 is used. Currently those listeners are supported
3874 only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003875 - 'udp6@' -> address is always IPv6 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003876 is used. Currently those listeners are supported
3877 only in log-forward sections.
Willy Tarreau24709282013-03-10 21:32:12 +01003878 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02003879 - 'abns@' -> address is in abstract namespace (Linux only).
3880 Note: since abstract sockets are not "rebindable", they
3881 do not cope well with multi-process mode during
3882 soft-restart, so it is better to avoid them if
3883 nbproc is greater than 1. The effect is that if the
3884 new process fails to start, only one of the old ones
3885 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01003886 - 'fd@<n>' -> use file descriptor <n> inherited from the
3887 parent. The fd must be bound and may or may not already
3888 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02003889 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
3890 connected unix socket or of a socketpair. The bind waits
3891 to receive a FD over the unix socket and uses it as if it
3892 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02003893 You may want to reference some environment variables in the
3894 address parameter, see section 2.3 about environment
3895 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003896
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003897 <port_range> is either a unique TCP port, or a port range for which the
3898 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003899 above. The port is mandatory for TCP listeners. Note that in
3900 the case of an IPv6 address, the port is always the number
3901 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003902 - a numerical port (ex: '80')
3903 - a dash-delimited ports range explicitly stating the lower
3904 and upper bounds (ex: '2000-2100') which are included in
3905 the range.
3906
3907 Particular care must be taken against port ranges, because
3908 every <address:port> couple consumes one socket (= a file
3909 descriptor), so it's easy to consume lots of descriptors
3910 with a simple range, and to run out of sockets. Also, each
3911 <address:port> couple must be used only once among all
3912 instances running on a same system. Please note that binding
3913 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003914 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003915 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003916
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003917 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01003918 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003919 receive UNIX connections on the socket located at this place.
3920 The path must begin with a slash and by default is absolute.
3921 It can be relative to the prefix defined by "unix-bind" in
3922 the global section. Note that the total length of the prefix
3923 followed by the socket path cannot exceed some system limits
3924 for UNIX sockets, which commonly are set to 107 characters.
3925
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003926 <param*> is a list of parameters common to all sockets declared on the
3927 same line. These numerous parameters depend on OS and build
3928 options and have a complete section dedicated to them. Please
3929 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003930
Willy Tarreau0ba27502007-12-24 16:55:16 +01003931 It is possible to specify a list of address:port combinations delimited by
3932 commas. The frontend will then listen on all of these addresses. There is no
3933 fixed limit to the number of addresses and ports which can be listened on in
3934 a frontend, as well as there is no limit to the number of "bind" statements
3935 in a frontend.
3936
3937 Example :
3938 listen http_proxy
3939 bind :80,:443
3940 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003941 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01003942
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003943 listen http_https_proxy
3944 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02003945 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003946
Willy Tarreau24709282013-03-10 21:32:12 +01003947 listen http_https_proxy_explicit
3948 bind ipv6@:80
3949 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
3950 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
3951
Willy Tarreaudad36a32013-03-11 01:20:04 +01003952 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02003953 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01003954
Willy Tarreau55dcaf62015-09-27 15:03:15 +02003955 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
3956 sun_path length is used for the address length. Some other programs
3957 such as socat use the string length only by default. Pass the option
3958 ",unix-tightsocklen=0" to any abstract socket definition in socat to
3959 make it compatible with HAProxy's.
3960
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003961 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003962 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003963
3964
Christopher Fauletff4121f2017-11-22 16:38:49 +01003965bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003966 Limit visibility of an instance to a certain set of processes numbers.
3967 May be used in sections : defaults | frontend | listen | backend
3968 yes | yes | yes | yes
3969 Arguments :
3970 all All process will see this instance. This is the default. It
3971 may be used to override a default value.
3972
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003973 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003974 option may be combined with other numbers.
3975
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003976 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003977 option may be combined with other numbers. Do not use it
3978 with less than 2 processes otherwise some instances might be
3979 missing from all processes.
3980
Christopher Fauletff4121f2017-11-22 16:38:49 +01003981 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003982 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01003983 the machine's word size. Ranges can be partially defined. The
3984 higher bound can be omitted. In such case, it is replaced by
3985 the corresponding maximum value. If a proxy is bound to
3986 process numbers greater than the configured global.nbproc, it
3987 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003988 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003989
3990 This keyword limits binding of certain instances to certain processes. This
3991 is useful in order not to have too many processes listening to the same
3992 ports. For instance, on a dual-core machine, it might make sense to set
3993 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3994 and 'even' instances.
3995
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003996 At the moment, it is not possible to reference more than 32 or 64 processes
3997 using this keyword, but this should be more than enough for most setups.
3998 Please note that 'all' really means all processes regardless of the machine's
3999 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004000
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02004001 Each "bind" line may further be limited to a subset of the proxy's processes,
4002 please consult the "process" bind keyword in section 5.1.
4003
Willy Tarreaub369a042014-09-16 13:21:03 +02004004 When a frontend has no explicit "bind-process" line, it tries to bind to all
4005 the processes referenced by its "bind" lines. That means that frontends can
4006 easily adapt to their listeners' processes.
4007
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004008 If some backends are referenced by frontends bound to other processes, the
4009 backend automatically inherits the frontend's processes.
4010
4011 Example :
4012 listen app_ip1
4013 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02004014 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004015
4016 listen app_ip2
4017 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02004018 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004019
4020 listen management
4021 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02004022 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004023
Willy Tarreau110ecc12012-11-15 17:50:01 +01004024 listen management
4025 bind 10.0.0.4:80
4026 bind-process 1-4
4027
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02004028 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01004029
4030
Willy Tarreau0ba27502007-12-24 16:55:16 +01004031capture cookie <name> len <length>
4032 Capture and log a cookie in the request and in the response.
4033 May be used in sections : defaults | frontend | listen | backend
4034 no | yes | yes | no
4035 Arguments :
4036 <name> is the beginning of the name of the cookie to capture. In order
4037 to match the exact name, simply suffix the name with an equal
4038 sign ('='). The full name will appear in the logs, which is
4039 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01004040 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01004041
4042 <length> is the maximum number of characters to report in the logs, which
4043 include the cookie name, the equal sign and the value, all in the
4044 standard "name=value" form. The string will be truncated on the
4045 right if it exceeds <length>.
4046
4047 Only the first cookie is captured. Both the "cookie" request headers and the
4048 "set-cookie" response headers are monitored. This is particularly useful to
4049 check for application bugs causing session crossing or stealing between
4050 users, because generally the user's cookies can only change on a login page.
4051
4052 When the cookie was not presented by the client, the associated log column
4053 will report "-". When a request does not cause a cookie to be assigned by the
4054 server, a "-" is reported in the response column.
4055
4056 The capture is performed in the frontend only because it is necessary that
4057 the log format does not change for a given frontend depending on the
4058 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01004059 "capture cookie" statement in a frontend. The maximum capture length is set
4060 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
4061 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004062
4063 Example:
4064 capture cookie ASPSESSION len 32
4065
4066 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004067 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004068
4069
4070capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01004071 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004072 May be used in sections : defaults | frontend | listen | backend
4073 no | yes | yes | no
4074 Arguments :
4075 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004076 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01004077 appear in the requests, with the first letter of each word in
4078 upper case. The header name will not appear in the logs, only the
4079 value is reported, but the position in the logs is respected.
4080
4081 <length> is the maximum number of characters to extract from the value and
4082 report in the logs. The string will be truncated on the right if
4083 it exceeds <length>.
4084
Willy Tarreau4460d032012-11-21 23:37:37 +01004085 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01004086 value will be added to the logs between braces ('{}'). If multiple headers
4087 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004088 in the same order they were declared in the configuration. Non-existent
4089 headers will be logged just as an empty string. Common uses for request
4090 header captures include the "Host" field in virtual hosting environments, the
4091 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004092 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004093 environments to find where the request came from.
4094
4095 Note that when capturing headers such as "User-agent", some spaces may be
4096 logged, making the log analysis more difficult. Thus be careful about what
4097 you log if you know your log parser is not smart enough to rely on the
4098 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004099
Willy Tarreau0900abb2012-11-22 00:21:46 +01004100 There is no limit to the number of captured request headers nor to their
4101 length, though it is wise to keep them low to limit memory usage per session.
4102 In order to keep log format consistent for a same frontend, header captures
4103 can only be declared in a frontend. It is not possible to specify a capture
4104 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004105
4106 Example:
4107 capture request header Host len 15
4108 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01004109 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01004110
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004111 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01004112 about logging.
4113
4114
4115capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01004116 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004117 May be used in sections : defaults | frontend | listen | backend
4118 no | yes | yes | no
4119 Arguments :
4120 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004121 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01004122 appear in the response, with the first letter of each word in
4123 upper case. The header name will not appear in the logs, only the
4124 value is reported, but the position in the logs is respected.
4125
4126 <length> is the maximum number of characters to extract from the value and
4127 report in the logs. The string will be truncated on the right if
4128 it exceeds <length>.
4129
Willy Tarreau4460d032012-11-21 23:37:37 +01004130 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01004131 result will be added to the logs between braces ('{}') after the captured
4132 request headers. If multiple headers are captured, they will be delimited by
4133 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01004134 the configuration. Non-existent headers will be logged just as an empty
4135 string. Common uses for response header captures include the "Content-length"
4136 header which indicates how many bytes are expected to be returned, the
4137 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004138
Willy Tarreau0900abb2012-11-22 00:21:46 +01004139 There is no limit to the number of captured response headers nor to their
4140 length, though it is wise to keep them low to limit memory usage per session.
4141 In order to keep log format consistent for a same frontend, header captures
4142 can only be declared in a frontend. It is not possible to specify a capture
4143 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004144
4145 Example:
4146 capture response header Content-length len 9
4147 capture response header Location len 15
4148
Willy Tarreauc57f0e22009-05-10 13:12:33 +02004149 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01004150 about logging.
4151
4152
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004153clitcpka-cnt <count>
4154 Sets the maximum number of keepalive probes TCP should send before dropping
4155 the connection on the client side.
4156 May be used in sections : defaults | frontend | listen | backend
4157 yes | yes | yes | no
4158 Arguments :
4159 <count> is the maximum number of keepalive probes.
4160
4161 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
4162 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004163 The availability of this setting depends on the operating system. It is
4164 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004165
4166 See also : "option clitcpka", "clitcpka-idle", "clitcpka-intvl".
4167
4168
4169clitcpka-idle <timeout>
4170 Sets the time the connection needs to remain idle before TCP starts sending
4171 keepalive probes, if enabled the sending of TCP keepalive packets on the
4172 client side.
4173 May be used in sections : defaults | frontend | listen | backend
4174 yes | yes | yes | no
4175 Arguments :
4176 <timeout> is the time the connection needs to remain idle before TCP starts
4177 sending keepalive probes. It is specified in seconds by default,
4178 but can be in any other unit if the number is suffixed by the
4179 unit, as explained at the top of this document.
4180
4181 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
4182 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004183 The availability of this setting depends on the operating system. It is
4184 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004185
4186 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-intvl".
4187
4188
4189clitcpka-intvl <timeout>
4190 Sets the time between individual keepalive probes on the client side.
4191 May be used in sections : defaults | frontend | listen | backend
4192 yes | yes | yes | no
4193 Arguments :
4194 <timeout> is the time between individual keepalive probes. It is specified
4195 in seconds by default, but can be in any other unit if the number
4196 is suffixed by the unit, as explained at the top of this
4197 document.
4198
4199 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
4200 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004201 The availability of this setting depends on the operating system. It is
4202 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004203
4204 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-idle".
4205
4206
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004207compression algo <algorithm> ...
4208compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02004209compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02004210 Enable HTTP compression.
4211 May be used in sections : defaults | frontend | listen | backend
4212 yes | yes | yes | yes
4213 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004214 algo is followed by the list of supported compression algorithms.
4215 type is followed by the list of MIME types that will be compressed.
4216 offload makes haproxy work as a compression offloader only (see notes).
4217
4218 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01004219 identity this is mostly for debugging, and it was useful for developing
4220 the compression feature. Identity does not apply any change on
4221 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004222
Willy Tarreauc91840a2015-03-28 17:00:39 +01004223 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01004224 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01004225
4226 deflate same as "gzip", but with deflate algorithm and zlib format.
4227 Note that this algorithm has ambiguous support on many
4228 browsers and no support at all from recent ones. It is
4229 strongly recommended not to use it for anything else than
4230 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01004231 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004232
Willy Tarreauc91840a2015-03-28 17:00:39 +01004233 raw-deflate same as "deflate" without the zlib wrapper, and used as an
4234 alternative when the browser wants "deflate". All major
4235 browsers understand it and despite violating the standards,
4236 it is known to work better than "deflate", at least on MSIE
4237 and some versions of Safari. Do not use it in conjunction
4238 with "deflate", use either one or the other since both react
4239 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01004240 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004241
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04004242 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004243 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004244 If backend servers support HTTP compression, these directives
4245 will be no-op: haproxy will see the compressed response and will not
4246 compress again. If backend servers do not support HTTP compression and
4247 there is Accept-Encoding header in request, haproxy will compress the
4248 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02004249
4250 The "offload" setting makes haproxy remove the Accept-Encoding header to
4251 prevent backend servers from compressing responses. It is strongly
4252 recommended not to do this because this means that all the compression work
4253 will be done on the single point where haproxy is located. However in some
4254 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004255 with broken HTTP compression implementation which can't be turned off.
4256 In that case haproxy can be used to prevent that gateway from emitting
4257 invalid payloads. In this case, simply removing the header in the
4258 configuration does not work because it applies before the header is parsed,
4259 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02004260 then be used for such scenarios. Note: for now, the "offload" setting is
4261 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02004262
William Lallemand05097442012-11-20 12:14:28 +01004263 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01004264 * the request does not advertise a supported compression algorithm in the
4265 "Accept-Encoding" header
4266 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01004267 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01004268 * response contain neither a "Content-Length" header nor a
4269 "Transfer-Encoding" whose last value is "chunked"
4270 * response contains a "Content-Type" header whose first value starts with
4271 "multipart"
4272 * the response contains the "no-transform" value in the "Cache-control"
4273 header
4274 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
4275 and later
4276 * The response contains a "Content-Encoding" header, indicating that the
4277 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01004278 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01004279
Tim Duesterhusb229f012019-01-29 16:38:56 +01004280 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01004281
William Lallemand82fe75c2012-10-23 10:25:10 +02004282 Examples :
4283 compression algo gzip
4284 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01004285
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004286
Willy Tarreau55165fe2009-05-10 12:02:55 +02004287cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02004288 [ postonly ] [ preserve ] [ httponly ] [ secure ]
4289 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Christopher Faulet2f533902020-01-21 11:06:48 +01004290 [ dynamic ] [ attr <value> ]*
Willy Tarreau0ba27502007-12-24 16:55:16 +01004291 Enable cookie-based persistence in a backend.
4292 May be used in sections : defaults | frontend | listen | backend
4293 yes | no | yes | yes
4294 Arguments :
4295 <name> is the name of the cookie which will be monitored, modified or
4296 inserted in order to bring persistence. This cookie is sent to
4297 the client via a "Set-Cookie" header in the response, and is
4298 brought back by the client in a "Cookie" header in all requests.
4299 Special care should be taken to choose a name which does not
4300 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01004301 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004302 HTTP/HTTPS), care should be taken to use different cookie names
4303 between all backends if persistence between them is not desired.
4304
4305 rewrite This keyword indicates that the cookie will be provided by the
4306 server and that haproxy will have to modify its value to set the
4307 server's identifier in it. This mode is handy when the management
4308 of complex combinations of "Set-cookie" and "Cache-control"
4309 headers is left to the application. The application can then
4310 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004311 cookie. Since all responses should be monitored, this mode
4312 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01004313 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004314 start with this mode for new deployments. This keyword is
4315 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004316
4317 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02004318 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004319
Willy Tarreaua79094d2010-08-31 22:54:15 +02004320 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004321 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02004322 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01004323 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004324 existing configurations running in the "rewrite" mode. The cookie
4325 will only be a session cookie and will not be stored on the
4326 client's disk. By default, unless the "indirect" option is added,
4327 the server will see the cookies emitted by the client. Due to
4328 caching effects, it is generally wise to add the "nocache" or
4329 "postonly" keywords (see below). The "insert" keyword is not
4330 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004331
4332 prefix This keyword indicates that instead of relying on a dedicated
4333 cookie for the persistence, an existing one will be completed.
4334 This may be needed in some specific environments where the client
4335 does not support more than one single cookie and the application
4336 already needs it. In this case, whenever the server sets a cookie
4337 named <name>, it will be prefixed with the server's identifier
4338 and a delimiter. The prefix will be removed from all client
4339 requests so that the server still finds the cookie it emitted.
4340 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004341 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02004342 not compatible with "rewrite" and "insert". Note: it is highly
4343 recommended not to use "indirect" with "prefix", otherwise server
4344 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004345
Willy Tarreaua79094d2010-08-31 22:54:15 +02004346 indirect When this option is specified, no cookie will be emitted to a
4347 client which already has a valid one for the server which has
4348 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004349 it will be removed, unless the "preserve" option is also set. In
4350 "insert" mode, this will additionally remove cookies from the
4351 requests transmitted to the server, making the persistence
4352 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02004353 Note: it is highly recommended not to use "indirect" with
4354 "prefix", otherwise server cookie updates would not be sent to
4355 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004356
4357 nocache This option is recommended in conjunction with the insert mode
4358 when there is a cache between the client and HAProxy, as it
4359 ensures that a cacheable response will be tagged non-cacheable if
4360 a cookie needs to be inserted. This is important because if all
4361 persistence cookies are added on a cacheable home page for
4362 instance, then all customers will then fetch the page from an
4363 outer cache and will all share the same persistence cookie,
4364 leading to one server receiving much more traffic than others.
4365 See also the "insert" and "postonly" options.
4366
4367 postonly This option ensures that cookie insertion will only be performed
4368 on responses to POST requests. It is an alternative to the
4369 "nocache" option, because POST responses are not cacheable, so
4370 this ensures that the persistence cookie will never get cached.
4371 Since most sites do not need any sort of persistence before the
4372 first POST which generally is a login request, this is a very
4373 efficient method to optimize caching without risking to find a
4374 persistence cookie in the cache.
4375 See also the "insert" and "nocache" options.
4376
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004377 preserve This option may only be used with "insert" and/or "indirect". It
4378 allows the server to emit the persistence cookie itself. In this
4379 case, if a cookie is found in the response, haproxy will leave it
4380 untouched. This is useful in order to end persistence after a
4381 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01004382 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004383 the past. By combining this mechanism with the "disable-on-404"
4384 check option, it is possible to perform a completely graceful
4385 shutdown because users will definitely leave the server after
4386 they logout.
4387
Willy Tarreau4992dd22012-05-31 21:02:17 +02004388 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
4389 when a cookie is inserted. This attribute is used so that a
4390 user agent doesn't share the cookie with non-HTTP components.
4391 Please check RFC6265 for more information on this attribute.
4392
4393 secure This option tells haproxy to add a "Secure" cookie attribute when
4394 a cookie is inserted. This attribute is used so that a user agent
4395 never emits this cookie over non-secure channels, which means
4396 that a cookie learned with this flag will be presented only over
4397 SSL/TLS connections. Please check RFC6265 for more information on
4398 this attribute.
4399
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02004400 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004401 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01004402 name. If the domain begins with a dot, the browser is allowed to
4403 use it for any host ending with that name. It is also possible to
4404 specify several domain names by invoking this option multiple
4405 times. Some browsers might have small limits on the number of
4406 domains, so be careful when doing that. For the record, sending
4407 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02004408
Willy Tarreau996a92c2010-10-13 19:30:47 +02004409 maxidle This option allows inserted cookies to be ignored after some idle
4410 time. It only works with insert-mode cookies. When a cookie is
4411 sent to the client, the date this cookie was emitted is sent too.
4412 Upon further presentations of this cookie, if the date is older
4413 than the delay indicated by the parameter (in seconds), it will
4414 be ignored. Otherwise, it will be refreshed if needed when the
4415 response is sent to the client. This is particularly useful to
4416 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004417 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02004418 this option is set and a cookie has no date, it is always
4419 accepted, but gets refreshed in the response. This maintains the
4420 ability for admins to access their sites. Cookies that have a
4421 date in the future further than 24 hours are ignored. Doing so
4422 lets admins fix timezone issues without risking kicking users off
4423 the site.
4424
4425 maxlife This option allows inserted cookies to be ignored after some life
4426 time, whether they're in use or not. It only works with insert
4427 mode cookies. When a cookie is first sent to the client, the date
4428 this cookie was emitted is sent too. Upon further presentations
4429 of this cookie, if the date is older than the delay indicated by
4430 the parameter (in seconds), it will be ignored. If the cookie in
4431 the request has no date, it is accepted and a date will be set.
4432 Cookies that have a date in the future further than 24 hours are
4433 ignored. Doing so lets admins fix timezone issues without risking
4434 kicking users off the site. Contrary to maxidle, this value is
4435 not refreshed, only the first visit date counts. Both maxidle and
4436 maxlife may be used at the time. This is particularly useful to
4437 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004438 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02004439 is stronger than the maxidle method in that it forces a
4440 redispatch after some absolute delay.
4441
Olivier Houchard4e694042017-03-14 20:01:29 +01004442 dynamic Activate dynamic cookies. When used, a session cookie is
4443 dynamically created for each server, based on the IP and port
4444 of the server, and a secret key, specified in the
4445 "dynamic-cookie-key" backend directive.
4446 The cookie will be regenerated each time the IP address change,
4447 and is only generated for IPv4/IPv6.
4448
Christopher Faulet2f533902020-01-21 11:06:48 +01004449 attr This option tells haproxy to add an extra attribute when a
4450 cookie is inserted. The attribute value can contain any
4451 characters except control ones or ";". This option may be
4452 repeated.
4453
Willy Tarreau0ba27502007-12-24 16:55:16 +01004454 There can be only one persistence cookie per HTTP backend, and it can be
4455 declared in a defaults section. The value of the cookie will be the value
4456 indicated after the "cookie" keyword in a "server" statement. If no cookie
4457 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02004458
Willy Tarreau0ba27502007-12-24 16:55:16 +01004459 Examples :
4460 cookie JSESSIONID prefix
4461 cookie SRV insert indirect nocache
4462 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02004463 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01004464
Willy Tarreau294d0f02015-08-10 19:40:12 +02004465 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004466
Willy Tarreau983e01e2010-01-11 18:42:06 +01004467
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004468declare capture [ request | response ] len <length>
4469 Declares a capture slot.
4470 May be used in sections : defaults | frontend | listen | backend
4471 no | yes | yes | no
4472 Arguments:
4473 <length> is the length allowed for the capture.
4474
4475 This declaration is only available in the frontend or listen section, but the
4476 reserved slot can be used in the backends. The "request" keyword allocates a
4477 capture slot for use in the request, and "response" allocates a capture slot
4478 for use in the response.
4479
4480 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02004481 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004482 "http-request capture" and "http-response capture".
4483
4484
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004485default-server [param*]
4486 Change default options for a server in a backend
4487 May be used in sections : defaults | frontend | listen | backend
4488 yes | no | yes | yes
4489 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01004490 <param*> is a list of parameters for this server. The "default-server"
4491 keyword accepts an important number of options and has a complete
4492 section dedicated to it. Please refer to section 5 for more
4493 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004494
Willy Tarreau983e01e2010-01-11 18:42:06 +01004495 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004496 default-server inter 1000 weight 13
4497
4498 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01004499
Willy Tarreau983e01e2010-01-11 18:42:06 +01004500
Willy Tarreau0ba27502007-12-24 16:55:16 +01004501default_backend <backend>
4502 Specify the backend to use when no "use_backend" rule has been matched.
4503 May be used in sections : defaults | frontend | listen | backend
4504 yes | yes | yes | no
4505 Arguments :
4506 <backend> is the name of the backend to use.
4507
4508 When doing content-switching between frontend and backends using the
4509 "use_backend" keyword, it is often useful to indicate which backend will be
4510 used when no rule has matched. It generally is the dynamic backend which
4511 will catch all undetermined requests.
4512
Willy Tarreau0ba27502007-12-24 16:55:16 +01004513 Example :
4514
4515 use_backend dynamic if url_dyn
4516 use_backend static if url_css url_img extension_img
4517 default_backend dynamic
4518
Willy Tarreau98d04852015-05-26 12:18:29 +02004519 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004520
Willy Tarreau0ba27502007-12-24 16:55:16 +01004521
Baptiste Assmann27f51342013-10-09 06:51:49 +02004522description <string>
4523 Describe a listen, frontend or backend.
4524 May be used in sections : defaults | frontend | listen | backend
4525 no | yes | yes | yes
4526 Arguments : string
4527
4528 Allows to add a sentence to describe the related object in the HAProxy HTML
4529 stats page. The description will be printed on the right of the object name
4530 it describes.
4531 No need to backslash spaces in the <string> arguments.
4532
4533
Willy Tarreau0ba27502007-12-24 16:55:16 +01004534disabled
4535 Disable a proxy, frontend or backend.
4536 May be used in sections : defaults | frontend | listen | backend
4537 yes | yes | yes | yes
4538 Arguments : none
4539
4540 The "disabled" keyword is used to disable an instance, mainly in order to
4541 liberate a listening port or to temporarily disable a service. The instance
4542 will still be created and its configuration will be checked, but it will be
4543 created in the "stopped" state and will appear as such in the statistics. It
4544 will not receive any traffic nor will it send any health-checks or logs. It
4545 is possible to disable many instances at once by adding the "disabled"
4546 keyword in a "defaults" section.
4547
4548 See also : "enabled"
4549
4550
Willy Tarreau5ce94572010-06-07 14:35:41 +02004551dispatch <address>:<port>
4552 Set a default server address
4553 May be used in sections : defaults | frontend | listen | backend
4554 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004555 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02004556
4557 <address> is the IPv4 address of the default server. Alternatively, a
4558 resolvable hostname is supported, but this name will be resolved
4559 during start-up.
4560
4561 <ports> is a mandatory port specification. All connections will be sent
4562 to this port, and it is not permitted to use port offsets as is
4563 possible with normal servers.
4564
Willy Tarreau787aed52011-04-15 06:45:37 +02004565 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02004566 server can take the connection. In the past it was used to forward non
4567 persistent connections to an auxiliary load balancer. Due to its simple
4568 syntax, it has also been used for simple TCP relays. It is recommended not to
4569 use it for more clarity, and to use the "server" directive instead.
4570
4571 See also : "server"
4572
Olivier Houchard4e694042017-03-14 20:01:29 +01004573
4574dynamic-cookie-key <string>
4575 Set the dynamic cookie secret key for a backend.
4576 May be used in sections : defaults | frontend | listen | backend
4577 yes | no | yes | yes
4578 Arguments : The secret key to be used.
4579
4580 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01004581 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01004582 specified on the "server" line), using a hash of the IP address of the
4583 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004584 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01004585 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02004586
Willy Tarreau0ba27502007-12-24 16:55:16 +01004587enabled
4588 Enable a proxy, frontend or backend.
4589 May be used in sections : defaults | frontend | listen | backend
4590 yes | yes | yes | yes
4591 Arguments : none
4592
4593 The "enabled" keyword is used to explicitly enable an instance, when the
4594 defaults has been set to "disabled". This is very rarely used.
4595
4596 See also : "disabled"
4597
4598
4599errorfile <code> <file>
4600 Return a file contents instead of errors generated by HAProxy
4601 May be used in sections : defaults | frontend | listen | backend
4602 yes | yes | yes | yes
4603 Arguments :
4604 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004605 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004606 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004607
4608 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004609 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01004610 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01004611 error pages, and to use absolute paths, since files are read
4612 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004613
4614 It is important to understand that this keyword is not meant to rewrite
4615 errors returned by the server, but errors detected and returned by HAProxy.
4616 This is why the list of supported errors is limited to a small set.
4617
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004618 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4619
Christopher Faulet70170672020-05-18 17:42:48 +02004620 The files are parsed when HAProxy starts and must be valid according to the
4621 HTTP specification. They should not exceed the configured buffer size
4622 (BUFSIZE), which generally is 16 kB, otherwise an internal error will be
4623 returned. It is also wise not to put any reference to local contents
4624 (e.g. images) in order to avoid loops between the client and HAProxy when all
4625 servers are down, causing an error to be returned instead of an
4626 image. Finally, The response cannot exceed (tune.bufsize - tune.maxrewrite)
4627 so that "http-after-response" rules still have room to operate (see
4628 "tune.maxrewrite").
Willy Tarreau59140a22009-02-22 12:02:30 +01004629
Willy Tarreau0ba27502007-12-24 16:55:16 +01004630 The files are read at the same time as the configuration and kept in memory.
4631 For this reason, the errors continue to be returned even when the process is
4632 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01004633 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01004634 403 status code and interrogating a blocked URL.
4635
Christopher Faulet3b967c12020-05-15 15:47:44 +02004636 See also : "http-error", "errorloc", "errorloc302", "errorloc303"
Willy Tarreau0ba27502007-12-24 16:55:16 +01004637
Willy Tarreau59140a22009-02-22 12:02:30 +01004638 Example :
4639 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01004640 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01004641 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
4642 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
4643
Willy Tarreau2769aa02007-12-27 18:26:09 +01004644
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004645errorfiles <name> [<code> ...]
4646 Import, fully or partially, the error files defined in the <name> http-errors
4647 section.
4648 May be used in sections : defaults | frontend | listen | backend
4649 yes | yes | yes | yes
4650 Arguments :
4651 <name> is the name of an existing http-errors section.
4652
4653 <code> is a HTTP status code. Several status code may be listed.
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004654 Currently, HAProxy is capable of generating codes 200, 400, 401,
Christopher Faulete095f312020-12-07 11:22:24 +01004655 403, 404, 405, 407, 408, 410, 413, 425, 429, 500, 501, 502, 503,
4656 and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004657
4658 Errors defined in the http-errors section with the name <name> are imported
4659 in the current proxy. If no status code is specified, all error files of the
4660 http-errors section are imported. Otherwise, only error files associated to
4661 the listed status code are imported. Those error files override the already
4662 defined custom errors for the proxy. And they may be overridden by following
Daniel Corbett67a82712020-07-06 23:01:19 -04004663 ones. Functionally, it is exactly the same as declaring all error files by
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004664 hand using "errorfile" directives.
4665
Christopher Faulet3b967c12020-05-15 15:47:44 +02004666 See also : "http-error", "errorfile", "errorloc", "errorloc302" ,
4667 "errorloc303" and section 3.8 about http-errors.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004668
4669 Example :
4670 errorfiles generic
4671 errorfiles site-1 403 404
4672
4673
Willy Tarreau2769aa02007-12-27 18:26:09 +01004674errorloc <code> <url>
4675errorloc302 <code> <url>
4676 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4677 May be used in sections : defaults | frontend | listen | backend
4678 yes | yes | yes | yes
4679 Arguments :
4680 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004681 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004682 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004683
4684 <url> it is the exact contents of the "Location" header. It may contain
4685 either a relative URI to an error page hosted on the same site,
4686 or an absolute URI designating an error page on another site.
4687 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004688 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004689
4690 It is important to understand that this keyword is not meant to rewrite
4691 errors returned by the server, but errors detected and returned by HAProxy.
4692 This is why the list of supported errors is limited to a small set.
4693
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004694 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4695
Willy Tarreau2769aa02007-12-27 18:26:09 +01004696 Note that both keyword return the HTTP 302 status code, which tells the
4697 client to fetch the designated URL using the same HTTP method. This can be
4698 quite problematic in case of non-GET methods such as POST, because the URL
4699 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01004700 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01004701 status code, indicating to the client that the URL must be fetched with a GET
4702 request.
4703
Christopher Faulet3b967c12020-05-15 15:47:44 +02004704 See also : "http-error", "errorfile", "errorloc303"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004705
4706
4707errorloc303 <code> <url>
4708 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4709 May be used in sections : defaults | frontend | listen | backend
4710 yes | yes | yes | yes
4711 Arguments :
4712 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004713 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004714 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004715
4716 <url> it is the exact contents of the "Location" header. It may contain
4717 either a relative URI to an error page hosted on the same site,
4718 or an absolute URI designating an error page on another site.
4719 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004720 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004721
4722 It is important to understand that this keyword is not meant to rewrite
4723 errors returned by the server, but errors detected and returned by HAProxy.
4724 This is why the list of supported errors is limited to a small set.
4725
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004726 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4727
Willy Tarreau2769aa02007-12-27 18:26:09 +01004728 Note that both keyword return the HTTP 303 status code, which tells the
4729 client to fetch the designated URL using the same HTTP GET method. This
4730 solves the usual problems associated with "errorloc" and the 302 code. It is
4731 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004732 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004733
Christopher Faulet3b967c12020-05-15 15:47:44 +02004734 See also : "http-error", "errorfile", "errorloc", "errorloc302"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004735
4736
Simon Horman51a1cf62015-02-03 13:00:44 +09004737email-alert from <emailaddr>
4738 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01004739 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09004740 May be used in sections: defaults | frontend | listen | backend
4741 yes | yes | yes | yes
4742
4743 Arguments :
4744
4745 <emailaddr> is the from email address to use when sending email alerts
4746
4747 Also requires "email-alert mailers" and "email-alert to" to be set
4748 and if so sending email alerts is enabled for the proxy.
4749
Simon Horman64e34162015-02-06 11:11:57 +09004750 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02004751 "email-alert myhostname", "email-alert to", section 3.6 about
4752 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09004753
4754
4755email-alert level <level>
4756 Declare the maximum log level of messages for which email alerts will be
4757 sent. This acts as a filter on the sending of email alerts.
4758 May be used in sections: defaults | frontend | listen | backend
4759 yes | yes | yes | yes
4760
4761 Arguments :
4762
4763 <level> One of the 8 syslog levels:
4764 emerg alert crit err warning notice info debug
4765 The above syslog levels are ordered from lowest to highest.
4766
4767 By default level is alert
4768
4769 Also requires "email-alert from", "email-alert mailers" and
4770 "email-alert to" to be set and if so sending email alerts is enabled
4771 for the proxy.
4772
Simon Horman1421e212015-04-30 13:10:35 +09004773 Alerts are sent when :
4774
4775 * An un-paused server is marked as down and <level> is alert or lower
4776 * A paused server is marked as down and <level> is notice or lower
4777 * A server is marked as up or enters the drain state and <level>
4778 is notice or lower
4779 * "option log-health-checks" is enabled, <level> is info or lower,
4780 and a health check status update occurs
4781
Simon Horman64e34162015-02-06 11:11:57 +09004782 See also : "email-alert from", "email-alert mailers",
4783 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09004784 section 3.6 about mailers.
4785
4786
4787email-alert mailers <mailersect>
4788 Declare the mailers to be used when sending email alerts
4789 May be used in sections: defaults | frontend | listen | backend
4790 yes | yes | yes | yes
4791
4792 Arguments :
4793
4794 <mailersect> is the name of the mailers section to send email alerts.
4795
4796 Also requires "email-alert from" and "email-alert to" to be set
4797 and if so sending email alerts is enabled for the proxy.
4798
Simon Horman64e34162015-02-06 11:11:57 +09004799 See also : "email-alert from", "email-alert level", "email-alert myhostname",
4800 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004801
4802
4803email-alert myhostname <hostname>
4804 Declare the to hostname address to be used when communicating with
4805 mailers.
4806 May be used in sections: defaults | frontend | listen | backend
4807 yes | yes | yes | yes
4808
4809 Arguments :
4810
Baptiste Assmann738bad92015-12-21 15:27:53 +01004811 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09004812
4813 By default the systems hostname is used.
4814
4815 Also requires "email-alert from", "email-alert mailers" and
4816 "email-alert to" to be set and if so sending email alerts is enabled
4817 for the proxy.
4818
Simon Horman64e34162015-02-06 11:11:57 +09004819 See also : "email-alert from", "email-alert level", "email-alert mailers",
4820 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004821
4822
4823email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01004824 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09004825 header of email alerts. This is the address that email alerts are sent to.
4826 May be used in sections: defaults | frontend | listen | backend
4827 yes | yes | yes | yes
4828
4829 Arguments :
4830
4831 <emailaddr> is the to email address to use when sending email alerts
4832
4833 Also requires "email-alert mailers" and "email-alert to" to be set
4834 and if so sending email alerts is enabled for the proxy.
4835
Simon Horman64e34162015-02-06 11:11:57 +09004836 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09004837 "email-alert myhostname", section 3.6 about mailers.
4838
4839
Willy Tarreau4de91492010-01-22 19:10:05 +01004840force-persist { if | unless } <condition>
4841 Declare a condition to force persistence on down servers
4842 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01004843 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01004844
4845 By default, requests are not dispatched to down servers. It is possible to
4846 force this using "option persist", but it is unconditional and redispatches
4847 to a valid server if "option redispatch" is set. That leaves with very little
4848 possibilities to force some requests to reach a server which is artificially
4849 marked down for maintenance operations.
4850
4851 The "force-persist" statement allows one to declare various ACL-based
4852 conditions which, when met, will cause a request to ignore the down status of
4853 a server and still try to connect to it. That makes it possible to start a
4854 server, still replying an error to the health checks, and run a specially
4855 configured browser to test the service. Among the handy methods, one could
4856 use a specific source IP address, or a specific cookie. The cookie also has
4857 the advantage that it can easily be added/removed on the browser from a test
4858 page. Once the service is validated, it is then possible to open the service
4859 to the world by returning a valid response to health checks.
4860
4861 The forced persistence is enabled when an "if" condition is met, or unless an
4862 "unless" condition is met. The final redispatch is always disabled when this
4863 is used.
4864
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004865 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02004866 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01004867
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004868
4869filter <name> [param*]
4870 Add the filter <name> in the filter list attached to the proxy.
4871 May be used in sections : defaults | frontend | listen | backend
4872 no | yes | yes | yes
4873 Arguments :
4874 <name> is the name of the filter. Officially supported filters are
4875 referenced in section 9.
4876
Tim Düsterhus4896c442016-11-29 02:15:19 +01004877 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004878 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01004879 filter. Please refer to the documentation of the corresponding
4880 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004881
4882 Multiple occurrences of the filter line can be used for the same proxy. The
4883 same filter can be referenced many times if needed.
4884
4885 Example:
4886 listen
4887 bind *:80
4888
4889 filter trace name BEFORE-HTTP-COMP
4890 filter compression
4891 filter trace name AFTER-HTTP-COMP
4892
4893 compression algo gzip
4894 compression offload
4895
4896 server srv1 192.168.0.1:80
4897
4898 See also : section 9.
4899
Willy Tarreau4de91492010-01-22 19:10:05 +01004900
Willy Tarreau2769aa02007-12-27 18:26:09 +01004901fullconn <conns>
4902 Specify at what backend load the servers will reach their maxconn
4903 May be used in sections : defaults | frontend | listen | backend
4904 yes | no | yes | yes
4905 Arguments :
4906 <conns> is the number of connections on the backend which will make the
4907 servers use the maximal number of connections.
4908
Willy Tarreau198a7442008-01-17 12:05:32 +01004909 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01004910 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01004911 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01004912 load. The server will then always accept at least <minconn> connections,
4913 never more than <maxconn>, and the limit will be on the ramp between both
4914 values when the backend has less than <conns> concurrent connections. This
4915 makes it possible to limit the load on the servers during normal loads, but
4916 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004917 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004918
Willy Tarreaufbb78422011-06-05 15:38:35 +02004919 Since it's hard to get this value right, haproxy automatically sets it to
4920 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01004921 backend (based on "use_backend" and "default_backend" rules). That way it's
4922 safe to leave it unset. However, "use_backend" involving dynamic names are
4923 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02004924
Willy Tarreau2769aa02007-12-27 18:26:09 +01004925 Example :
4926 # The servers will accept between 100 and 1000 concurrent connections each
4927 # and the maximum of 1000 will be reached when the backend reaches 10000
4928 # connections.
4929 backend dynamic
4930 fullconn 10000
4931 server srv1 dyn1:80 minconn 100 maxconn 1000
4932 server srv2 dyn2:80 minconn 100 maxconn 1000
4933
4934 See also : "maxconn", "server"
4935
4936
Willy Tarreauab0a5192020-10-09 19:07:01 +02004937grace <time> (deprecated)
Willy Tarreau2769aa02007-12-27 18:26:09 +01004938 Maintain a proxy operational for some time after a soft stop
4939 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01004940 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01004941 Arguments :
4942 <time> is the time (by default in milliseconds) for which the instance
4943 will remain operational with the frontend sockets still listening
4944 when a soft-stop is received via the SIGUSR1 signal.
4945
4946 This may be used to ensure that the services disappear in a certain order.
4947 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004948 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01004949 needed by the equipment to detect the failure.
4950
4951 Note that currently, there is very little benefit in using this parameter,
4952 and it may in fact complicate the soft-reconfiguration process more than
4953 simplify it.
4954
Willy Tarreau0ba27502007-12-24 16:55:16 +01004955
Andrew Rodland17be45e2016-10-25 17:04:12 -04004956hash-balance-factor <factor>
4957 Specify the balancing factor for bounded-load consistent hashing
4958 May be used in sections : defaults | frontend | listen | backend
4959 yes | no | no | yes
4960 Arguments :
4961 <factor> is the control for the maximum number of concurrent requests to
4962 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01004963 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04004964
4965 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
4966 enables an algorithm that prevents any one server from getting too many
4967 requests at once, even if some hash buckets receive many more requests than
4968 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
4969 <factor> is a percentage greater than 100. For example, if <factor> is 150,
4970 then no server will be allowed to have a load more than 1.5 times the average.
4971 If server weights are used, they will be respected.
4972
4973 If the first-choice server is disqualified, the algorithm will choose another
4974 server based on the request hash, until a server with additional capacity is
4975 found. A higher <factor> allows more imbalance between the servers, while a
4976 lower <factor> means that more servers will be checked on average, affecting
4977 performance. Reasonable values are from 125 to 200.
4978
Willy Tarreau760e81d2018-05-03 07:20:40 +02004979 This setting is also used by "balance random" which internally relies on the
4980 consistent hashing mechanism.
4981
Andrew Rodland17be45e2016-10-25 17:04:12 -04004982 See also : "balance" and "hash-type".
4983
4984
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004985hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004986 Specify a method to use for mapping hashes to servers
4987 May be used in sections : defaults | frontend | listen | backend
4988 yes | no | yes | yes
4989 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04004990 <method> is the method used to select a server from the hash computed by
4991 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004992
Bhaskar98634f02013-10-29 23:30:51 -04004993 map-based the hash table is a static array containing all alive servers.
4994 The hashes will be very smooth, will consider weights, but
4995 will be static in that weight changes while a server is up
4996 will be ignored. This means that there will be no slow start.
4997 Also, since a server is selected by its position in the array,
4998 most mappings are changed when the server count changes. This
4999 means that when a server goes up or down, or when a server is
5000 added to a farm, most connections will be redistributed to
5001 different servers. This can be inconvenient with caches for
5002 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01005003
Bhaskar98634f02013-10-29 23:30:51 -04005004 consistent the hash table is a tree filled with many occurrences of each
5005 server. The hash key is looked up in the tree and the closest
5006 server is chosen. This hash is dynamic, it supports changing
5007 weights while the servers are up, so it is compatible with the
5008 slow start feature. It has the advantage that when a server
5009 goes up or down, only its associations are moved. When a
5010 server is added to the farm, only a few part of the mappings
5011 are redistributed, making it an ideal method for caches.
5012 However, due to its principle, the distribution will never be
5013 very smooth and it may sometimes be necessary to adjust a
5014 server's weight or its ID to get a more balanced distribution.
5015 In order to get the same distribution on multiple load
5016 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005017 same IDs. Note: consistent hash uses sdbm and avalanche if no
5018 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04005019
5020 <function> is the hash function to be used :
5021
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005022 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04005023 reimplementation of ndbm) database library. It was found to do
5024 well in scrambling bits, causing better distribution of the keys
5025 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005026 function with good distribution, unless the total server weight
5027 is a multiple of 64, in which case applying the avalanche
5028 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04005029
5030 djb2 this function was first proposed by Dan Bernstein many years ago
5031 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005032 function provides a better distribution than sdbm. It generally
5033 works well with text-based inputs though it can perform extremely
5034 poorly with numeric-only input or when the total server weight is
5035 a multiple of 33, unless the avalanche modifier is also used.
5036
Willy Tarreaua0f42712013-11-14 14:30:35 +01005037 wt6 this function was designed for haproxy while testing other
5038 functions in the past. It is not as smooth as the other ones, but
5039 is much less sensible to the input data set or to the number of
5040 servers. It can make sense as an alternative to sdbm+avalanche or
5041 djb2+avalanche for consistent hashing or when hashing on numeric
5042 data such as a source IP address or a visitor identifier in a URL
5043 parameter.
5044
Willy Tarreau324f07f2015-01-20 19:44:50 +01005045 crc32 this is the most common CRC32 implementation as used in Ethernet,
5046 gzip, PNG, etc. It is slower than the other ones but may provide
5047 a better distribution or less predictable results especially when
5048 used on strings.
5049
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05005050 <modifier> indicates an optional method applied after hashing the key :
5051
5052 avalanche This directive indicates that the result from the hash
5053 function above should not be used in its raw form but that
5054 a 4-byte full avalanche hash must be applied first. The
5055 purpose of this step is to mix the resulting bits from the
5056 previous hash in order to avoid any undesired effect when
5057 the input contains some limited values or when the number of
5058 servers is a multiple of one of the hash's components (64
5059 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
5060 result less predictable, but it's also not as smooth as when
5061 using the original function. Some testing might be needed
5062 with some workloads. This hash is one of the many proposed
5063 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005064
Bhaskar98634f02013-10-29 23:30:51 -04005065 The default hash type is "map-based" and is recommended for most usages. The
5066 default function is "sdbm", the selection of a function should be based on
5067 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005068
Andrew Rodland17be45e2016-10-25 17:04:12 -04005069 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02005070
5071
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005072http-after-response <action> <options...> [ { if | unless } <condition> ]
5073 Access control for all Layer 7 responses (server, applet/service and internal
5074 ones).
5075
5076 May be used in sections: defaults | frontend | listen | backend
5077 no | yes | yes | yes
5078
5079 The http-after-response statement defines a set of rules which apply to layer
5080 7 processing. The rules are evaluated in their declaration order when they
5081 are met in a frontend, listen or backend section. Any rule may optionally be
5082 followed by an ACL-based condition, in which case it will only be evaluated
5083 if the condition is true. Since these rules apply on responses, the backend
5084 rules are applied first, followed by the frontend's rules.
5085
5086 Unlike http-response rules, these ones are applied on all responses, the
5087 server ones but also to all responses generated by HAProxy. These rules are
5088 evaluated at the end of the responses analysis, before the data forwarding.
5089
5090 The first keyword is the rule's action. The supported actions are described
5091 below.
5092
5093 There is no limit to the number of http-after-response statements per
5094 instance.
5095
Christopher Fauletd5ac6de2020-12-02 08:40:14 +01005096 Note: Errors emitted in early stage of the request parsing are handled by the
5097 multiplexer at a lower level, before any http analysis. Thus no
5098 http-after-response ruleset is evaluated on these errors.
5099
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005100 Example:
5101 http-after-response set-header Strict-Transport-Security "max-age=31536000"
5102 http-after-response set-header Cache-Control "no-store,no-cache,private"
5103 http-after-response set-header Pragma "no-cache"
5104
5105http-after-response add-header <name> <fmt> [ { if | unless } <condition> ]
5106
5107 This appends an HTTP header field whose name is specified in <name> and whose
5108 value is defined by <fmt> which follows the log-format rules (see Custom Log
5109 Format in section 8.2.4). This may be used to send a cookie to a client for
5110 example, or to pass some internal information.
5111 This rule is not final, so it is possible to add other similar rules.
5112 Note that header addition is performed immediately, so one rule might reuse
5113 the resulting header from a previous rule.
5114
5115http-after-response allow [ { if | unless } <condition> ]
5116
5117 This stops the evaluation of the rules and lets the response pass the check.
5118 No further "http-after-response" rules are evaluated.
5119
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005120http-after-response del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005121
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005122 This removes all HTTP header fields whose name is specified in <name>. <meth>
5123 is the matching method, applied on the header name. Supported matching methods
5124 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
5125 (substring match) and "reg" (regex match). If not specified, exact matching
5126 method is used.
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005127
5128http-after-response replace-header <name> <regex-match> <replace-fmt>
5129 [ { if | unless } <condition> ]
5130
5131 This works like "http-response replace-header".
5132
5133 Example:
5134 http-after-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
5135
5136 # applied to:
5137 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
5138
5139 # outputs:
5140 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
5141
5142 # assuming the backend IP is 192.168.1.20.
5143
5144http-after-response replace-value <name> <regex-match> <replace-fmt>
5145 [ { if | unless } <condition> ]
5146
5147 This works like "http-response replace-value".
5148
5149 Example:
5150 http-after-response replace-value Cache-control ^public$ private
5151
5152 # applied to:
5153 Cache-Control: max-age=3600, public
5154
5155 # outputs:
5156 Cache-Control: max-age=3600, private
5157
5158http-after-response set-header <name> <fmt> [ { if | unless } <condition> ]
5159
5160 This does the same as "add-header" except that the header name is first
5161 removed if it existed. This is useful when passing security information to
5162 the server, where the header must not be manipulated by external users.
5163
5164http-after-response set-status <status> [reason <str>]
5165 [ { if | unless } <condition> ]
5166
5167 This replaces the response status code with <status> which must be an integer
5168 between 100 and 999. Optionally, a custom reason text can be provided defined
5169 by <str>, or the default reason for the specified code will be used as a
5170 fallback.
5171
5172 Example:
5173 # return "431 Request Header Fields Too Large"
5174 http-response set-status 431
5175 # return "503 Slow Down", custom reason
5176 http-response set-status 503 reason "Slow Down"
5177
5178http-after-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5179
5180 This is used to set the contents of a variable. The variable is declared
5181 inline.
5182
5183 Arguments:
5184 <var-name> The name of the variable starts with an indication about its
5185 scope. The scopes allowed are:
5186 "proc" : the variable is shared with the whole process
5187 "sess" : the variable is shared with the whole session
5188 "txn" : the variable is shared with the transaction
5189 (request and response)
5190 "req" : the variable is shared only during request
5191 processing
5192 "res" : the variable is shared only during response
5193 processing
5194 This prefix is followed by a name. The separator is a '.'.
5195 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5196 and '_'.
5197
5198 <expr> Is a standard HAProxy expression formed by a sample-fetch
5199 followed by some converters.
5200
5201 Example:
5202 http-after-response set-var(sess.last_redir) res.hdr(location)
5203
5204http-after-response strict-mode { on | off }
5205
5206 This enables or disables the strict rewriting mode for following rules. It
5207 does not affect rules declared before it and it is only applicable on rules
5208 performing a rewrite on the responses. When the strict mode is enabled, any
5209 rewrite failure triggers an internal error. Otherwise, such errors are
5210 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005211 rewrites optional while others must be performed to continue the response
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005212 processing.
5213
5214 By default, the strict rewriting mode is enabled. Its value is also reset
5215 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04005216 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005217 rules evaluation.
5218
5219http-after-response unset-var(<var-name>) [ { if | unless } <condition> ]
5220
5221 This is used to unset a variable. See "http-after-response set-var" for
5222 details about <var-name>.
5223
5224 Example:
5225 http-after-response unset-var(sess.last_redir)
5226
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005227
5228http-check comment <string>
5229 Defines a comment for the following the http-check rule, reported in logs if
5230 it fails.
5231 May be used in sections : defaults | frontend | listen | backend
5232 yes | no | yes | yes
5233
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005234 Arguments :
5235 <string> is the comment message to add in logs if the following http-check
5236 rule fails.
5237
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005238 It only works for connect, send and expect rules. It is useful to make
5239 user-friendly error reporting.
5240
Daniel Corbett67a82712020-07-06 23:01:19 -04005241 See also : "option httpchk", "http-check connect", "http-check send" and
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005242 "http-check expect".
5243
5244
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005245http-check connect [default] [port <expr>] [addr <ip>] [send-proxy]
5246 [via-socks4] [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +02005247 [proto <name>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02005248 Opens a new connection to perform an HTTP health check
5249 May be used in sections : defaults | frontend | listen | backend
5250 yes | no | yes | yes
5251
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005252 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005253 comment <msg> defines a message to report if the rule evaluation fails.
5254
Christopher Faulete5870d82020-04-15 11:32:03 +02005255 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -04005256 checks. The server options are used only if not redefined.
Christopher Faulete5870d82020-04-15 11:32:03 +02005257
5258 port <expr> if not set, check port or server port is used.
5259 It tells HAProxy where to open the connection to.
5260 <port> must be a valid TCP port source integer, from 1 to
5261 65535 or an sample-fetch expression.
5262
5263 addr <ip> defines the IP address to do the health check.
5264
5265 send-proxy send a PROXY protocol string
5266
5267 via-socks4 enables outgoing health checks using upstream socks4 proxy.
5268
5269 ssl opens a ciphered connection
5270
5271 sni <sni> specifies the SNI to use to do health checks over SSL.
5272
5273 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
5274 list consists in a comma-delimited list of protocol names,
5275 for instance: "h2,http/1.1". If it is not set, the server ALPN
5276 is used.
5277
Christopher Fauletedc6ed92020-04-23 16:27:59 +02005278 proto <name> forces the multiplexer's protocol to use for this connection.
5279 It must be an HTTP mux protocol and it must be usable on the
5280 backend side. The list of available protocols is reported in
5281 haproxy -vv.
5282
Christopher Faulete5870d82020-04-15 11:32:03 +02005283 linger cleanly close the connection instead of using a single RST.
5284
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005285 Just like tcp-check health checks, it is possible to configure the connection
5286 to use to perform HTTP health check. This directive should also be used to
5287 describe a scenario involving several request/response exchanges, possibly on
5288 different ports or with different servers.
5289
5290 When there are no TCP port configured on the server line neither server port
5291 directive, then the first step of the http-check sequence must be to specify
5292 the port with a "http-check connect".
5293
5294 In an http-check ruleset a 'connect' is required, it is also mandatory to start
5295 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
5296 do.
5297
5298 When a connect must start the ruleset, if may still be preceded by set-var,
5299 unset-var or comment rules.
5300
5301 Examples :
Christopher Faulete5870d82020-04-15 11:32:03 +02005302 # check HTTP and HTTPs services on a server.
5303 # first open port 80 thanks to server line port directive, then
5304 # tcp-check opens port 443, ciphered and run a request on it:
5305 option httpchk
5306
5307 http-check connect
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02005308 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005309 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02005310 http-check connect port 443 ssl sni haproxy.1wt.eu
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02005311 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005312 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02005313
5314 server www 10.0.0.1 check port 80
5315
5316 See also : "option httpchk", "http-check send", "http-check expect"
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005317
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005318
Willy Tarreau0ba27502007-12-24 16:55:16 +01005319http-check disable-on-404
5320 Enable a maintenance mode upon HTTP/404 response to health-checks
5321 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01005322 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01005323 Arguments : none
5324
5325 When this option is set, a server which returns an HTTP code 404 will be
5326 excluded from further load-balancing, but will still receive persistent
5327 connections. This provides a very convenient method for Web administrators
5328 to perform a graceful shutdown of their servers. It is also important to note
5329 that a server which is detected as failed while it was in this mode will not
5330 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
5331 will immediately be reinserted into the farm. The status on the stats page
5332 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01005333 option only works in conjunction with the "httpchk" option. If this option
5334 is used with "http-check expect", then it has precedence over it so that 404
Christopher Fauletfa8b89a2020-11-20 18:54:13 +01005335 responses will still be considered as soft-stop. Note also that a stopped
5336 server will stay stopped even if it replies 404s. This option is only
5337 evaluated for running servers.
Willy Tarreaubd741542010-03-16 18:46:54 +01005338
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005339 See also : "option httpchk" and "http-check expect".
Willy Tarreaubd741542010-03-16 18:46:54 +01005340
5341
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005342http-check expect [min-recv <int>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02005343 [ok-status <st>] [error-status <st>] [tout-status <st>]
5344 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
5345 [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005346 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01005347 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02005348 yes | no | yes | yes
Christopher Faulete5870d82020-04-15 11:32:03 +02005349
Willy Tarreaubd741542010-03-16 18:46:54 +01005350 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005351 comment <msg> defines a message to report if the rule evaluation fails.
5352
Christopher Faulete5870d82020-04-15 11:32:03 +02005353 min-recv is optional and can define the minimum amount of data required to
5354 evaluate the current expect rule. If the number of received bytes
5355 is under this limit, the check will wait for more data. This
5356 option can be used to resolve some ambiguous matching rules or to
5357 avoid executing costly regex matches on content known to be still
5358 incomplete. If an exact string is used, the minimum between the
5359 string length and this parameter is used. This parameter is
5360 ignored if it is set to -1. If the expect rule does not match,
5361 the check will wait for more data. If set to 0, the evaluation
5362 result is always conclusive.
5363
5364 ok-status <st> is optional and can be used to set the check status if
5365 the expect rule is successfully evaluated and if it is
5366 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005367 "L6OK" and "L4OK" are supported :
5368 - L7OK : check passed on layer 7
Christopher Faulet83662b52020-11-20 17:47:47 +01005369 - L7OKC : check conditionally passed on layer 7, set
5370 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005371 - L6OK : check passed on layer 6
5372 - L4OK : check passed on layer 4
5373 By default "L7OK" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02005374
5375 error-status <st> is optional and can be used to set the check status if
5376 an error occurred during the expect rule evaluation.
Christopher Faulet83662b52020-11-20 17:47:47 +01005377 "L7OKC", "L7RSP", "L7STS", "L6RSP" and "L4CON" are
5378 supported :
5379 - L7OKC : check conditionally passed on layer 7, set
5380 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005381 - L7RSP : layer 7 invalid response - protocol error
5382 - L7STS : layer 7 response error, for example HTTP 5xx
5383 - L6RSP : layer 6 invalid response - protocol error
5384 - L4CON : layer 1-4 connection problem
5385 By default "L7RSP" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02005386
5387 tout-status <st> is optional and can be used to set the check status if
5388 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005389 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
5390 - L7TOUT : layer 7 (HTTP/SMTP) timeout
5391 - L6TOUT : layer 6 (SSL) timeout
5392 - L4TOUT : layer 1-4 timeout
Christopher Faulete5870d82020-04-15 11:32:03 +02005393 By default "L7TOUT" is used.
5394
5395 on-success <fmt> is optional and can be used to customize the
5396 informational message reported in logs if the expect
5397 rule is successfully evaluated and if it is the last rule
5398 in the tcp-check ruleset. <fmt> is a log-format string.
5399
5400 on-error <fmt> is optional and can be used to customize the
5401 informational message reported in logs if an error
5402 occurred during the expect rule evaluation. <fmt> is a
5403 log-format string.
5404
Willy Tarreaubd741542010-03-16 18:46:54 +01005405 <match> is a keyword indicating how to look for a specific pattern in the
Christopher Fauletb5594262020-05-05 20:23:13 +02005406 response. The keyword may be one of "status", "rstatus", "hdr",
5407 "fhdr", "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01005408 exclamation mark ("!") to negate the match. Spaces are allowed
5409 between the exclamation mark and the keyword. See below for more
5410 details on the supported keywords.
5411
Christopher Faulet39708192020-05-05 10:47:36 +02005412 <pattern> is the pattern to look for. It may be a string, a regular
5413 expression or a more complex pattern with several arguments. If
5414 the string pattern contains spaces, they must be escaped with the
5415 usual backslash ('\').
Willy Tarreaubd741542010-03-16 18:46:54 +01005416
5417 By default, "option httpchk" considers that response statuses 2xx and 3xx
5418 are valid, and that others are invalid. When "http-check expect" is used,
5419 it defines what is considered valid or invalid. Only one "http-check"
5420 statement is supported in a backend. If a server fails to respond or times
5421 out, the check obviously fails. The available matches are :
5422
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005423 status <codes> : test the status codes found parsing <codes> string. it
5424 must be a comma-separated list of status codes or range
5425 codes. A health check response will be considered as
5426 valid if the response's status code matches any status
5427 code or is inside any range of the list. If the "status"
5428 keyword is prefixed with "!", then the response will be
5429 considered invalid if the status code matches.
Willy Tarreaubd741542010-03-16 18:46:54 +01005430
5431 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005432 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005433 response's status code matches the expression. If the
5434 "rstatus" keyword is prefixed with "!", then the response
5435 will be considered invalid if the status code matches.
5436 This is mostly used to check for multiple codes.
5437
Christopher Fauletb5594262020-05-05 20:23:13 +02005438 hdr { name | name-lf } [ -m <meth> ] <name>
5439 [ { value | value-lf } [ -m <meth> ] <value> :
Christopher Faulet39708192020-05-05 10:47:36 +02005440 test the specified header pattern on the HTTP response
5441 headers. The name pattern is mandatory but the value
5442 pattern is optional. If not specified, only the header
5443 presence is verified. <meth> is the matching method,
5444 applied on the header name or the header value. Supported
5445 matching methods are "str" (exact match), "beg" (prefix
5446 match), "end" (suffix match), "sub" (substring match) or
5447 "reg" (regex match). If not specified, exact matching
Christopher Fauletb5594262020-05-05 20:23:13 +02005448 method is used. If the "name-lf" parameter is used,
5449 <name> is evaluated as a log-format string. If "value-lf"
5450 parameter is used, <value> is evaluated as a log-format
5451 string. These parameters cannot be used with the regex
5452 matching method. Finally, the header value is considered
5453 as comma-separated list. Note that matchings are case
5454 insensitive on the header names.
5455
5456 fhdr { name | name-lf } [ -m <meth> ] <name>
5457 [ { value | value-lf } [ -m <meth> ] <value> :
5458 test the specified full header pattern on the HTTP
5459 response headers. It does exactly the same than "hdr"
5460 keyword, except the full header value is tested, commas
5461 are not considered as delimiters.
Christopher Faulet39708192020-05-05 10:47:36 +02005462
Willy Tarreaubd741542010-03-16 18:46:54 +01005463 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005464 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005465 response's body contains this exact string. If the
5466 "string" keyword is prefixed with "!", then the response
5467 will be considered invalid if the body contains this
5468 string. This can be used to look for a mandatory word at
5469 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005470 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01005471 trace).
5472
5473 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005474 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005475 response's body matches this expression. If the "rstring"
5476 keyword is prefixed with "!", then the response will be
5477 considered invalid if the body matches the expression.
5478 This can be used to look for a mandatory word at the end
5479 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01005480 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01005481
Christopher Fauletaaab0832020-05-05 15:54:22 +02005482 string-lf <fmt> : test a log-format string match in the HTTP response body.
5483 A health check response will be considered valid if the
5484 response's body contains the string resulting of the
5485 evaluation of <fmt>, which follows the log-format rules.
5486 If prefixed with "!", then the response will be
5487 considered invalid if the body contains the string.
5488
Willy Tarreaubd741542010-03-16 18:46:54 +01005489 It is important to note that the responses will be limited to a certain size
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +01005490 defined by the global "tune.bufsize" option, which defaults to 16384 bytes.
Willy Tarreaubd741542010-03-16 18:46:54 +01005491 Thus, too large responses may not contain the mandatory pattern when using
5492 "string" or "rstring". If a large response is absolutely required, it is
5493 possible to change the default max size by setting the global variable.
5494 However, it is worth keeping in mind that parsing very large responses can
5495 waste some CPU cycles, especially when regular expressions are used, and that
5496 it is always better to focus the checks on smaller resources.
5497
Christopher Faulete5870d82020-04-15 11:32:03 +02005498 In an http-check ruleset, the last expect rule may be implicit. If no expect
5499 rule is specified after the last "http-check send", an implicit expect rule
5500 is defined to match on 2xx or 3xx status codes. It means this rule is also
5501 defined if there is no "http-check" rule at all, when only "option httpchk"
5502 is set.
Cyril Bonté32602d22015-01-30 00:07:07 +01005503
Willy Tarreaubd741542010-03-16 18:46:54 +01005504 Last, if "http-check expect" is combined with "http-check disable-on-404",
5505 then this last one has precedence when the server responds with 404.
5506
5507 Examples :
5508 # only accept status 200 as valid
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005509 http-check expect status 200,201,300-310
Willy Tarreaubd741542010-03-16 18:46:54 +01005510
Christopher Faulet39708192020-05-05 10:47:36 +02005511 # be sure a sessid coookie is set
5512 http-check expect header name "set-cookie" value -m beg "sessid="
5513
Willy Tarreaubd741542010-03-16 18:46:54 +01005514 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005515 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01005516
5517 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005518 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01005519
5520 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005521 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01005522
Christopher Faulete5870d82020-04-15 11:32:03 +02005523 See also : "option httpchk", "http-check connect", "http-check disable-on-404"
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005524 and "http-check send".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005525
5526
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005527http-check send [meth <method>] [{ uri <uri> | uri-lf <fmt> }>] [ver <version>]
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005528 [hdr <name> <fmt>]* [{ body <string> | body-lf <fmt> }]
5529 [comment <msg>]
Christopher Faulet8acb1282020-04-09 08:44:06 +02005530 Add a possible list of headers and/or a body to the request sent during HTTP
5531 health checks.
5532 May be used in sections : defaults | frontend | listen | backend
5533 yes | no | yes | yes
5534 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005535 comment <msg> defines a message to report if the rule evaluation fails.
5536
Christopher Faulete5870d82020-04-15 11:32:03 +02005537 meth <method> is the optional HTTP method used with the requests. When not
5538 set, the "OPTIONS" method is used, as it generally requires
5539 low server processing and is easy to filter out from the
5540 logs. Any method may be used, though it is not recommended
5541 to invent non-standard ones.
5542
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005543 uri <uri> is optional and set the URI referenced in the HTTP requests
5544 to the string <uri>. It defaults to "/" which is accessible
5545 by default on almost any server, but may be changed to any
5546 other URI. Query strings are permitted.
5547
5548 uri-lf <fmt> is optional and set the URI referenced in the HTTP requests
5549 using the log-format string <fmt>. It defaults to "/" which
5550 is accessible by default on almost any server, but may be
5551 changed to any other URI. Query strings are permitted.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005552
Christopher Faulet907701b2020-04-28 09:37:00 +02005553 ver <version> is the optional HTTP version string. It defaults to
Christopher Faulete5870d82020-04-15 11:32:03 +02005554 "HTTP/1.0" but some servers might behave incorrectly in HTTP
Daniel Corbett67a82712020-07-06 23:01:19 -04005555 1.0, so turning it to HTTP/1.1 may sometimes help. Note that
Christopher Faulete5870d82020-04-15 11:32:03 +02005556 the Host field is mandatory in HTTP/1.1, use "hdr" argument
5557 to add it.
5558
5559 hdr <name> <fmt> adds the HTTP header field whose name is specified in
5560 <name> and whose value is defined by <fmt>, which follows
5561 to the log-format rules.
5562
5563 body <string> add the body defined by <string> to the request sent during
5564 HTTP health checks. If defined, the "Content-Length" header
5565 is thus automatically added to the request.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005566
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005567 body-lf <fmt> add the body defined by the log-format string <fmt> to the
5568 request sent during HTTP health checks. If defined, the
5569 "Content-Length" header is thus automatically added to the
5570 request.
5571
Christopher Faulet8acb1282020-04-09 08:44:06 +02005572 In addition to the request line defined by the "option httpchk" directive,
5573 this one is the valid way to add some headers and optionally a body to the
5574 request sent during HTTP health checks. If a body is defined, the associate
Christopher Faulet9df910c2020-04-29 14:20:47 +02005575 "Content-Length" header is automatically added. Thus, this header or
5576 "Transfer-encoding" header should not be present in the request provided by
5577 "http-check send". If so, it will be ignored. The old trick consisting to add
5578 headers after the version string on the "option httpchk" line is now
Amaury Denoyelle6d975f02020-12-22 14:08:52 +01005579 deprecated.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005580
Christopher Faulete5870d82020-04-15 11:32:03 +02005581 Also "http-check send" doesn't support HTTP keep-alive. Keep in mind that it
Amaury Denoyelle6d975f02020-12-22 14:08:52 +01005582 will automatically append a "Connection: close" header, unless a Connection
5583 header has already already been configured via a hdr entry.
Christopher Faulet9df910c2020-04-29 14:20:47 +02005584
5585 Note that the Host header and the request authority, when both defined, are
5586 automatically synchronized. It means when the HTTP request is sent, when a
5587 Host is inserted in the request, the request authority is accordingly
5588 updated. Thus, don't be surprised if the Host header value overwrites the
5589 configured request authority.
5590
5591 Note also for now, no Host header is automatically added in HTTP/1.1 or above
5592 requests. You should add it explicitly.
Christopher Faulete5870d82020-04-15 11:32:03 +02005593
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005594 See also : "option httpchk", "http-check send-state" and "http-check expect".
Christopher Faulet8acb1282020-04-09 08:44:06 +02005595
5596
Willy Tarreauef781042010-01-27 11:53:01 +01005597http-check send-state
5598 Enable emission of a state header with HTTP health checks
5599 May be used in sections : defaults | frontend | listen | backend
5600 yes | no | yes | yes
5601 Arguments : none
5602
5603 When this option is set, haproxy will systematically send a special header
5604 "X-Haproxy-Server-State" with a list of parameters indicating to each server
5605 how they are seen by haproxy. This can be used for instance when a server is
5606 manipulated without access to haproxy and the operator needs to know whether
5607 haproxy still sees it up or not, or if the server is the last one in a farm.
5608
5609 The header is composed of fields delimited by semi-colons, the first of which
5610 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
5611 checks on the total number before transition, just as appears in the stats
5612 interface. Next headers are in the form "<variable>=<value>", indicating in
5613 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08005614 - a variable "address", containing the address of the backend server.
5615 This corresponds to the <address> field in the server declaration. For
5616 unix domain sockets, it will read "unix".
5617
5618 - a variable "port", containing the port of the backend server. This
5619 corresponds to the <port> field in the server declaration. For unix
5620 domain sockets, it will read "unix".
5621
Willy Tarreauef781042010-01-27 11:53:01 +01005622 - a variable "name", containing the name of the backend followed by a slash
5623 ("/") then the name of the server. This can be used when a server is
5624 checked in multiple backends.
5625
5626 - a variable "node" containing the name of the haproxy node, as set in the
5627 global "node" variable, otherwise the system's hostname if unspecified.
5628
5629 - a variable "weight" indicating the weight of the server, a slash ("/")
5630 and the total weight of the farm (just counting usable servers). This
5631 helps to know if other servers are available to handle the load when this
5632 one fails.
5633
5634 - a variable "scur" indicating the current number of concurrent connections
5635 on the server, followed by a slash ("/") then the total number of
5636 connections on all servers of the same backend.
5637
5638 - a variable "qcur" indicating the current number of requests in the
5639 server's queue.
5640
5641 Example of a header received by the application server :
5642 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
5643 scur=13/22; qcur=0
5644
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005645 See also : "option httpchk", "http-check disable-on-404" and
5646 "http-check send".
Willy Tarreauef781042010-01-27 11:53:01 +01005647
Christopher Faulete5870d82020-04-15 11:32:03 +02005648
5649http-check set-var(<var-name>) <expr>
Christopher Faulete5870d82020-04-15 11:32:03 +02005650 This operation sets the content of a variable. The variable is declared inline.
Christopher Faulete5870d82020-04-15 11:32:03 +02005651 May be used in sections: defaults | frontend | listen | backend
5652 yes | no | yes | yes
5653
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005654 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005655 <var-name> The name of the variable starts with an indication about its
5656 scope. The scopes allowed for http-check are:
5657 "proc" : the variable is shared with the whole process.
5658 "sess" : the variable is shared with the tcp-check session.
5659 "check": the variable is declared for the lifetime of the tcp-check.
5660 This prefix is followed by a name. The separator is a '.'.
5661 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5662 and '-'.
5663
5664 <expr> Is a sample-fetch expression potentially followed by converters.
5665
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005666 Examples :
5667 http-check set-var(check.port) int(1234)
Christopher Faulete5870d82020-04-15 11:32:03 +02005668
5669
5670http-check unset-var(<var-name>)
Christopher Faulete5870d82020-04-15 11:32:03 +02005671 Free a reference to a variable within its scope.
Christopher Faulete5870d82020-04-15 11:32:03 +02005672 May be used in sections: defaults | frontend | listen | backend
5673 yes | no | yes | yes
5674
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005675 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005676 <var-name> The name of the variable starts with an indication about its
5677 scope. The scopes allowed for http-check are:
5678 "proc" : the variable is shared with the whole process.
5679 "sess" : the variable is shared with the tcp-check session.
5680 "check": the variable is declared for the lifetime of the tcp-check.
5681 This prefix is followed by a name. The separator is a '.'.
5682 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5683 and '-'.
5684
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005685 Examples :
5686 http-check unset-var(check.port)
Christopher Faulete5870d82020-04-15 11:32:03 +02005687
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005688
Christopher Faulet3b967c12020-05-15 15:47:44 +02005689http-error status <code> [content-type <type>]
5690 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5691 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5692 [ hdr <name> <fmt> ]*
5693 Defines a custom error message to use instead of errors generated by HAProxy.
5694 May be used in sections : defaults | frontend | listen | backend
5695 yes | yes | yes | yes
5696 Arguments :
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005697 status <code> is the HTTP status code. It must be specified.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005698 Currently, HAProxy is capable of generating codes
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005699 200, 400, 401, 403, 404, 405, 407, 408, 410, 413, 425,
Christopher Faulete095f312020-12-07 11:22:24 +01005700 429, 500, 501, 502, 503, and 504.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005701
5702 content-type <type> is the response content type, for instance
5703 "text/plain". This parameter is ignored and should be
5704 omitted when an errorfile is configured or when the
5705 payload is empty. Otherwise, it must be defined.
5706
5707 default-errorfiles Reset the previously defined error message for current
5708 proxy for the status <code>. If used on a backend, the
5709 frontend error message is used, if defined. If used on
5710 a frontend, the default error message is used.
5711
5712 errorfile <file> designates a file containing the full HTTP response.
5713 It is recommended to follow the common practice of
5714 appending ".http" to the filename so that people do
5715 not confuse the response with HTML error pages, and to
5716 use absolute paths, since files are read before any
5717 chroot is performed.
5718
5719 errorfiles <name> designates the http-errors section to use to import
5720 the error message with the status code <code>. If no
5721 such message is found, the proxy's error messages are
5722 considered.
5723
5724 file <file> specifies the file to use as response payload. If the
5725 file is not empty, its content-type must be set as
5726 argument to "content-type", otherwise, any
5727 "content-type" argument is ignored. <file> is
5728 considered as a raw string.
5729
5730 string <str> specifies the raw string to use as response payload.
5731 The content-type must always be set as argument to
5732 "content-type".
5733
5734 lf-file <file> specifies the file to use as response payload. If the
5735 file is not empty, its content-type must be set as
5736 argument to "content-type", otherwise, any
5737 "content-type" argument is ignored. <file> is
5738 evaluated as a log-format string.
5739
5740 lf-string <str> specifies the log-format string to use as response
5741 payload. The content-type must always be set as
5742 argument to "content-type".
5743
5744 hdr <name> <fmt> adds to the response the HTTP header field whose name
5745 is specified in <name> and whose value is defined by
5746 <fmt>, which follows to the log-format rules.
5747 This parameter is ignored if an errorfile is used.
5748
5749 This directive may be used instead of "errorfile", to define a custom error
5750 message. As "errorfile" directive, it is used for errors detected and
5751 returned by HAProxy. If an errorfile is defined, it is parsed when HAProxy
5752 starts and must be valid according to the HTTP standards. The generated
5753 response must not exceed the configured buffer size (BUFFSIZE), otherwise an
5754 internal error will be returned. Finally, if you consider to use some
5755 http-after-response rules to rewrite these errors, the reserved buffer space
5756 should be available (see "tune.maxrewrite").
5757
5758 The files are read at the same time as the configuration and kept in memory.
5759 For this reason, the errors continue to be returned even when the process is
5760 chrooted, and no file change is considered while the process is running.
5761
Christopher Fauletd5ac6de2020-12-02 08:40:14 +01005762 Note: 400/408/500 errors emitted in early stage of the request parsing are
5763 handled by the multiplexer at a lower level. No custom formatting is
5764 supported at this level. Thus only static error messages, defined with
5765 "errorfile" directive, are supported. However, this limitation only
5766 exists during the request headers parsing or between two transactions.
5767
Christopher Faulet3b967c12020-05-15 15:47:44 +02005768 See also : "errorfile", "errorfiles", "errorloc", "errorloc302",
5769 "errorloc303" and section 3.8 about http-errors.
5770
5771
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005772http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005773 Access control for Layer 7 requests
5774
5775 May be used in sections: defaults | frontend | listen | backend
5776 no | yes | yes | yes
5777
Willy Tarreau20b0de52012-12-24 15:45:22 +01005778 The http-request statement defines a set of rules which apply to layer 7
5779 processing. The rules are evaluated in their declaration order when they are
5780 met in a frontend, listen or backend section. Any rule may optionally be
5781 followed by an ACL-based condition, in which case it will only be evaluated
5782 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005783
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005784 The first keyword is the rule's action. The supported actions are described
5785 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005786
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005787 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005788
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005789 Example:
5790 acl nagios src 192.168.129.3
5791 acl local_net src 192.168.0.0/16
5792 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01005793
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005794 http-request allow if nagios
5795 http-request allow if local_net auth_ok
5796 http-request auth realm Gimme if local_net auth_ok
5797 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01005798
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005799 Example:
5800 acl key req.hdr(X-Add-Acl-Key) -m found
5801 acl add path /addacl
5802 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01005803
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005804 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01005805
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005806 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
5807 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02005808
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005809 Example:
5810 acl value req.hdr(X-Value) -m found
5811 acl setmap path /setmap
5812 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005813
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005814 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06005815
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005816 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
5817 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005818
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005819 See also : "stats http-request", section 3.4 about userlists and section 7
5820 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06005821
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005822http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005823
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005824 This is used to add a new entry into an ACL. The ACL must be loaded from a
5825 file (even a dummy empty file). The file name of the ACL to be updated is
5826 passed between parentheses. It takes one argument: <key fmt>, which follows
5827 log-format rules, to collect content of the new entry. It performs a lookup
5828 in the ACL before insertion, to avoid duplicated (or more) values. This
5829 lookup is done by a linear search and can be expensive with large lists!
5830 It is the equivalent of the "add acl" command from the stats socket, but can
5831 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06005832
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005833http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005834
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005835 This appends an HTTP header field whose name is specified in <name> and
5836 whose value is defined by <fmt> which follows the log-format rules (see
5837 Custom Log Format in section 8.2.4). This is particularly useful to pass
5838 connection-specific information to the server (e.g. the client's SSL
5839 certificate), or to combine several headers into one. This rule is not
5840 final, so it is possible to add other similar rules. Note that header
5841 addition is performed immediately, so one rule might reuse the resulting
5842 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06005843
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005844http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005845
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005846 This stops the evaluation of the rules and lets the request pass the check.
5847 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06005848
Sasha Pachev218f0642014-06-16 12:05:59 -06005849
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005850http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005851
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005852 This stops the evaluation of the rules and immediately responds with an
5853 HTTP 401 or 407 error code to invite the user to present a valid user name
5854 and password. No further "http-request" rules are evaluated. An optional
5855 "realm" parameter is supported, it sets the authentication realm that is
5856 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06005857
Christopher Faulet612f2ea2020-05-27 09:57:28 +02005858 The corresponding proxy's error message is used. It may be customized using
5859 an "errorfile" or an "http-error" directive. For 401 responses, all
5860 occurrences of the WWW-Authenticate header are removed and replaced by a new
5861 one with a basic authentication challenge for realm "<realm>". For 407
5862 responses, the same is done on the Proxy-Authenticate header. If the error
5863 message must not be altered, consider to use "http-request return" rule
5864 instead.
5865
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005866 Example:
5867 acl auth_ok http_auth_group(L1) G1
5868 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06005869
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02005870http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005871
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005872 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005873
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005874http-request capture <sample> [ len <length> | id <id> ]
5875 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005876
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005877 This captures sample expression <sample> from the request buffer, and
5878 converts it to a string of at most <len> characters. The resulting string is
5879 stored into the next request "capture" slot, so it will possibly appear next
5880 to some captured HTTP headers. It will then automatically appear in the logs,
5881 and it will be possible to extract it using sample fetch rules to feed it
5882 into headers or anything. The length should be limited given that this size
5883 will be allocated for each capture during the whole session life.
5884 Please check section 7.3 (Fetching samples) and "capture request header" for
5885 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005886
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005887 If the keyword "id" is used instead of "len", the action tries to store the
5888 captured string in a previously declared capture slot. This is useful to run
5889 captures in backends. The slot id can be declared by a previous directive
Baptiste Assmann19a69b32020-01-16 14:34:22 +01005890 "http-request capture" or with the "declare capture" keyword.
5891
5892 When using this action in a backend, double check that the relevant
5893 frontend(s) have the required capture slots otherwise, this rule will be
5894 ignored at run time. This can't be detected at configuration parsing time
5895 due to HAProxy's ability to dynamically resolve backend name at runtime.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005896
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005897http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005898
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005899 This is used to delete an entry from an ACL. The ACL must be loaded from a
5900 file (even a dummy empty file). The file name of the ACL to be updated is
5901 passed between parentheses. It takes one argument: <key fmt>, which follows
5902 log-format rules, to collect content of the entry to delete.
5903 It is the equivalent of the "del acl" command from the stats socket, but can
5904 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005905
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005906http-request del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02005907
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005908 This removes all HTTP header fields whose name is specified in <name>. <meth>
5909 is the matching method, applied on the header name. Supported matching methods
5910 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
5911 (substring match) and "reg" (regex match). If not specified, exact matching
5912 method is used.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02005913
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005914http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02005915
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005916 This is used to delete an entry from a MAP. The MAP must be loaded from a
5917 file (even a dummy empty file). The file name of the MAP to be updated is
5918 passed between parentheses. It takes one argument: <key fmt>, which follows
5919 log-format rules, to collect content of the entry to delete.
5920 It takes one argument: "file name" It is the equivalent of the "del map"
5921 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02005922
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005923http-request deny [deny_status <status>] [ { if | unless } <condition> ]
5924http-request deny [ { status | deny_status } <code>] [content-type <type>]
5925 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5926 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5927 [ hdr <name> <fmt> ]*
5928 [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04005929
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005930 This stops the evaluation of the rules and immediately rejects the request.
5931 By default an HTTP 403 error is returned. But the response may be customized
5932 using same syntax than "http-request return" rules. Thus, see "http-request
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005933 return" for details. For compatibility purpose, when no argument is defined,
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005934 or only "deny_status", the argument "default-errorfiles" is implied. It means
5935 "http-request deny [deny_status <status>]" is an alias of
5936 "http-request deny [status <status>] default-errorfiles".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005937 No further "http-request" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005938 See also "http-request return".
Patrick Hemmer268a7072018-05-11 12:52:31 -04005939
Olivier Houchard602bf7d2019-05-10 13:59:15 +02005940http-request disable-l7-retry [ { if | unless } <condition> ]
5941 This disables any attempt to retry the request if it fails for any other
5942 reason than a connection failure. This can be useful for example to make
5943 sure POST requests aren't retried on failure.
5944
Baptiste Assmann333939c2019-01-21 08:34:50 +01005945http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
5946
5947 This action performs a DNS resolution of the output of <expr> and stores
5948 the result in the variable <var>. It uses the DNS resolvers section
5949 pointed by <resolvers>.
5950 It is possible to choose a resolution preference using the optional
5951 arguments 'ipv4' or 'ipv6'.
5952 When performing the DNS resolution, the client side connection is on
5953 pause waiting till the end of the resolution.
5954 If an IP address can be found, it is stored into <var>. If any kind of
5955 error occurs, then <var> is not set.
5956 One can use this action to discover a server IP address at run time and
5957 based on information found in the request (IE a Host header).
5958 If this action is used to find the server's IP address (using the
5959 "set-dst" action), then the server IP address in the backend must be set
5960 to 0.0.0.0.
5961
5962 Example:
5963 resolvers mydns
5964 nameserver local 127.0.0.53:53
5965 nameserver google 8.8.8.8:53
5966 timeout retry 1s
5967 hold valid 10s
5968 hold nx 3s
5969 hold other 3s
5970 hold obsolete 0s
5971 accepted_payload_size 8192
5972
5973 frontend fe
5974 bind 10.42.0.1:80
5975 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
5976 http-request capture var(txn.myip) len 40
5977
5978 # return 503 when the variable is not set,
5979 # which mean DNS resolution error
5980 use_backend b_503 unless { var(txn.myip) -m found }
5981
5982 default_backend be
5983
5984 backend b_503
5985 # dummy backend used to return 503.
5986 # one can use the errorfile directive to send a nice
5987 # 503 error page to end users
5988
5989 backend be
5990 # rule to prevent HAProxy from reconnecting to services
5991 # on the local network (forged DNS name used to scan the network)
5992 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
5993 http-request set-dst var(txn.myip)
5994 server clear 0.0.0.0:0
5995
5996 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
5997 be used to scan the network or worst won't loop over itself...
5998
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01005999http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
6000
6001 This is used to build an HTTP 103 Early Hints response prior to any other one.
6002 This appends an HTTP header field to this response whose name is specified in
6003 <name> and whose value is defined by <fmt> which follows the log-format rules
6004 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01006005 to the client some Link headers to preload resources required to render the
6006 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01006007
6008 See RFC 8297 for more information.
6009
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006010http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006011
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006012 This performs an HTTP redirection based on a redirect rule. This is exactly
6013 the same as the "redirect" statement except that it inserts a redirect rule
6014 which can be processed in the middle of other "http-request" rules and that
6015 these rules use the "log-format" strings. See the "redirect" keyword for the
6016 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006017
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006018http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006019
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006020 This stops the evaluation of the rules and immediately closes the connection
6021 without sending any response. It acts similarly to the
6022 "tcp-request content reject" rules. It can be useful to force an immediate
6023 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006024
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006025http-request replace-header <name> <match-regex> <replace-fmt>
6026 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02006027
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006028 This matches the value of all occurrences of header field <name> against
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006029 <match-regex>. Matching is performed case-sensitively. Matching values are
6030 completely replaced by <replace-fmt>. Format characters are allowed in
6031 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
6032 Standard back-references using the backslash ('\') followed by a number are
6033 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02006034
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006035 This action acts on whole header lines, regardless of the number of values
6036 they may contain. Thus it is well-suited to process headers naturally
6037 containing commas in their value, such as If-Modified-Since. Headers that
6038 contain a comma-separated list of values, such as Accept, should be processed
6039 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01006040
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006041 Example:
6042 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
6043
6044 # applied to:
6045 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
6046
6047 # outputs:
6048 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
6049
6050 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02006051
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006052 http-request replace-header User-Agent curl foo
6053
6054 # applied to:
6055 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02006056
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006057 # outputs:
6058 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02006059
Willy Tarreau262c3f12019-12-17 06:52:51 +01006060http-request replace-path <match-regex> <replace-fmt>
6061 [ { if | unless } <condition> ]
6062
6063 This works like "replace-header" except that it works on the request's path
6064 component instead of a header. The path component starts at the first '/'
Christopher Faulet82c83322020-09-02 14:16:59 +02006065 after an optional scheme+authority and ends before the question mark. Thus,
6066 the replacement does not modify the scheme, the authority and the
6067 query-string.
Willy Tarreau262c3f12019-12-17 06:52:51 +01006068
6069 It is worth noting that regular expressions may be more expensive to evaluate
6070 than certain ACLs, so rare replacements may benefit from a condition to avoid
6071 performing the evaluation at all if it does not match.
6072
6073 Example:
6074 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
6075 http-request replace-path (.*) /foo\1
6076
Willy Tarreau262c3f12019-12-17 06:52:51 +01006077 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
6078 http-request replace-path /foo/(.*) /\1
6079 # or more efficient if only some requests match :
6080 http-request replace-path /foo/(.*) /\1 if { url_beg /foo/ }
6081
Christopher Faulet312294f2020-09-02 17:17:44 +02006082http-request replace-pathq <match-regex> <replace-fmt>
6083 [ { if | unless } <condition> ]
6084
6085 This does the same as "http-request replace-path" except that the path
6086 contains the query-string if any is present. Thus, the path and the
6087 query-string are replaced.
6088
6089 Example:
6090 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
6091 http-request replace-pathq ([^?]*)(\?(.*))? \1/foo\2
6092
Willy Tarreau33810222019-06-12 17:44:02 +02006093http-request replace-uri <match-regex> <replace-fmt>
6094 [ { if | unless } <condition> ]
6095
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006096 This works like "replace-header" except that it works on the request's URI part
6097 instead of a header. The URI part may contain an optional scheme, authority or
6098 query string. These are considered to be part of the value that is matched
6099 against.
6100
6101 It is worth noting that regular expressions may be more expensive to evaluate
6102 than certain ACLs, so rare replacements may benefit from a condition to avoid
6103 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02006104
Willy Tarreau62b59132019-12-17 06:51:20 +01006105 IMPORTANT NOTE: historically in HTTP/1.x, the vast majority of requests sent
6106 by browsers use the "origin form", which differs from the "absolute form" in
6107 that they do not contain a scheme nor authority in the URI portion. Mostly
6108 only requests sent to proxies, those forged by hand and some emitted by
6109 certain applications use the absolute form. As such, "replace-uri" usually
6110 works fine most of the time in HTTP/1.x with rules starting with a "/". But
6111 with HTTP/2, clients are encouraged to send absolute URIs only, which look
6112 like the ones HTTP/1 clients use to talk to proxies. Such partial replace-uri
6113 rules may then fail in HTTP/2 when they work in HTTP/1. Either the rules need
Willy Tarreau262c3f12019-12-17 06:52:51 +01006114 to be adapted to optionally match a scheme and authority, or replace-path
6115 should be used.
Willy Tarreau33810222019-06-12 17:44:02 +02006116
Willy Tarreau62b59132019-12-17 06:51:20 +01006117 Example:
6118 # rewrite all "http" absolute requests to "https":
6119 http-request replace-uri ^http://(.*) https://\1
Willy Tarreau33810222019-06-12 17:44:02 +02006120
Willy Tarreau62b59132019-12-17 06:51:20 +01006121 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
6122 http-request replace-uri ([^/:]*://[^/]*)?(.*) \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02006123
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006124http-request replace-value <name> <match-regex> <replace-fmt>
6125 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02006126
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006127 This works like "replace-header" except that it matches the regex against
6128 every comma-delimited value of the header field <name> instead of the
6129 entire header. This is suited for all headers which are allowed to carry
6130 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02006131
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006132 Example:
6133 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02006134
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006135 # applied to:
6136 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02006137
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006138 # outputs:
6139 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01006140
Christopher Faulet24231ab2020-01-24 17:44:23 +01006141http-request return [status <code>] [content-type <type>]
6142 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6143 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006144 [ hdr <name> <fmt> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01006145 [ { if | unless } <condition> ]
6146
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006147 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01006148 default status code used for the response is 200. It can be optionally
6149 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04006150 specified as an argument to "content-type". Finally the response itself may
Sébastien Grossab877122020-10-08 10:06:03 +02006151 be defined. It can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006152 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01006153 are followed to create the response :
6154
6155 * If neither the errorfile nor the payload to use is defined, a dummy
6156 response is returned. Only the "status" argument is considered. It can be
6157 any code in the range [200, 599]. The "content-type" argument, if any, is
6158 ignored.
6159
6160 * If "default-errorfiles" argument is set, the proxy's errorfiles are
6161 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04006162 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Christopher Faulete095f312020-12-07 11:22:24 +01006163 425, 429, 500, 501, 502, 503, and 504). The "content-type" argument, if
6164 any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006165
6166 * If a specific errorfile is defined, with an "errorfile" argument, the
6167 corresponding file, containing a full HTTP response, is returned. Only the
6168 "status" argument is considered. It must be one of the status code handled
Christopher Faulete095f312020-12-07 11:22:24 +01006169 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 501,
6170 502, 503, and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006171
6172 * If an http-errors section is defined, with an "errorfiles" argument, the
6173 corresponding file in the specified http-errors section, containing a full
6174 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04006175 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Christopher Faulete095f312020-12-07 11:22:24 +01006176 408, 410, 413, 425, 429, 500, 501, 502, 503, and 504). The "content-type"
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006177 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006178
6179 * If a "file" or a "lf-file" argument is specified, the file's content is
6180 used as the response payload. If the file is not empty, its content-type
6181 must be set as argument to "content-type". Otherwise, any "content-type"
6182 argument is ignored. With a "lf-file" argument, the file's content is
6183 evaluated as a log-format string. With a "file" argument, it is considered
6184 as a raw content.
6185
6186 * If a "string" or "lf-string" argument is specified, the defined string is
6187 used as the response payload. The content-type must always be set as
6188 argument to "content-type". With a "lf-string" argument, the string is
6189 evaluated as a log-format string. With a "string" argument, it is
6190 considered as a raw string.
6191
Sébastien Grossab877122020-10-08 10:06:03 +02006192 When the response is not based on an errorfile, it is possible to append HTTP
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006193 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
6194 arguments are ignored. For each one, the header name is specified in <name>
6195 and its value is defined by <fmt> which follows the log-format rules.
6196
Christopher Faulet24231ab2020-01-24 17:44:23 +01006197 Note that the generated response must be smaller than a buffer. And to avoid
6198 any warning, when an errorfile or a raw file is loaded, the buffer space
Sébastien Grossab877122020-10-08 10:06:03 +02006199 reserved for the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006200
6201 No further "http-request" rules are evaluated.
6202
6203 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006204 http-request return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006205 if { path /ping }
6206
6207 http-request return content-type image/x-icon file /var/www/favicon.ico \
6208 if { path /favicon.ico }
6209
6210 http-request return status 403 content-type text/plain \
6211 lf-string "Access denied. IP %[src] is blacklisted." \
6212 if { src -f /etc/haproxy/blacklist.lst }
6213
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006214http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6215http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006216
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006217 This actions increments the GPC0 or GPC1 counter according with the sticky
6218 counter designated by <sc-id>. If an error occurs, this action silently fails
6219 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006220
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006221http-request sc-set-gpt0(<sc-id>) { <int> | <expr> }
6222 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006223
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006224 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6225 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
6226 boolean. If an error occurs, this action silently fails and the actions
6227 evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006228
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006229http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006230
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006231 This is used to set the destination IP address to the value of specified
6232 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
6233 but provides the correct IP in a HTTP header; or you want to mask the IP for
6234 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
6235 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01006236
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006237 Arguments:
6238 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6239 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01006240
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006241 Example:
6242 http-request set-dst hdr(x-dst)
6243 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01006244
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006245 When possible, set-dst preserves the original destination port as long as the
6246 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006247
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006248http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006249
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006250 This is used to set the destination port address to the value of specified
6251 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
6252 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006253
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006254 Arguments:
6255 <expr> Is a standard HAProxy expression formed by a sample-fetch
6256 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006257
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006258 Example:
6259 http-request set-dst-port hdr(x-port)
6260 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006261
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006262 When possible, set-dst-port preserves the original destination address as
6263 long as the address family supports a port, otherwise it forces the
6264 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02006265
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006266http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02006267
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006268 This does the same as "http-request add-header" except that the header name
6269 is first removed if it existed. This is useful when passing security
6270 information to the server, where the header must not be manipulated by
6271 external users. Note that the new value is computed before the removal so it
6272 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02006273
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006274 Example:
6275 http-request set-header X-Haproxy-Current-Date %T
6276 http-request set-header X-SSL %[ssl_fc]
6277 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
6278 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
6279 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
6280 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
6281 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
6282 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
6283 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02006284
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006285http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02006286
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006287 This is used to change the log level of the current request when a certain
6288 condition is met. Valid levels are the 8 syslog levels (see the "log"
6289 keyword) plus the special level "silent" which disables logging for this
6290 request. This rule is not final so the last matching rule wins. This rule
6291 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006292
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006293http-request set-map(<file-name>) <key fmt> <value fmt>
6294 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006295
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006296 This is used to add a new entry into a MAP. The MAP must be loaded from a
6297 file (even a dummy empty file). The file name of the MAP to be updated is
6298 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
6299 log-format rules, used to collect MAP key, and <value fmt>, which follows
6300 log-format rules, used to collect content for the new entry.
6301 It performs a lookup in the MAP before insertion, to avoid duplicated (or
6302 more) values. This lookup is done by a linear search and can be expensive
6303 with large lists! It is the equivalent of the "set map" command from the
6304 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006305
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006306http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006307
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006308 This is used to set the Netfilter MARK on all packets sent to the client to
6309 the value passed in <mark> on platforms which support it. This value is an
6310 unsigned 32 bit value which can be matched by netfilter and by the routing
6311 table. It can be expressed both in decimal or hexadecimal format (prefixed by
6312 "0x"). This can be useful to force certain packets to take a different route
6313 (for example a cheaper network path for bulk downloads). This works on Linux
6314 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02006315
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006316http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006317
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006318 This rewrites the request method with the result of the evaluation of format
6319 string <fmt>. There should be very few valid reasons for having to do so as
6320 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006321
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006322http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006323
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006324 This sets the "nice" factor of the current request being processed. It only
6325 has effect against the other requests being processed at the same time.
6326 The default value is 0, unless altered by the "nice" setting on the "bind"
6327 line. The accepted range is -1024..1024. The higher the value, the nicest
6328 the request will be. Lower values will make the request more important than
6329 other ones. This can be useful to improve the speed of some requests, or
6330 lower the priority of non-important requests. Using this setting without
6331 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006332
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006333http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02006334
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006335 This rewrites the request path with the result of the evaluation of format
6336 string <fmt>. The query string, if any, is left intact. If a scheme and
6337 authority is found before the path, they are left intact as well. If the
6338 request doesn't have a path ("*"), this one is replaced with the format.
6339 This can be used to prepend a directory component in front of a path for
6340 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02006341
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006342 Example :
6343 # prepend the host name before the path
6344 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006345
Christopher Faulet312294f2020-09-02 17:17:44 +02006346http-request set-pathq <fmt> [ { if | unless } <condition> ]
6347
6348 This does the same as "http-request set-path" except that the query-string is
6349 also rewritten. It may be used to remove the query-string, including the
6350 question mark (it is not possible using "http-request set-query").
6351
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006352http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02006353
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006354 This is used to set the queue priority class of the current request.
6355 The value must be a sample expression which converts to an integer in the
6356 range -2047..2047. Results outside this range will be truncated.
6357 The priority class determines the order in which queued requests are
6358 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006359
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006360http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006361
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006362 This is used to set the queue priority timestamp offset of the current
6363 request. The value must be a sample expression which converts to an integer
6364 in the range -524287..524287. Results outside this range will be truncated.
6365 When a request is queued, it is ordered first by the priority class, then by
6366 the current timestamp adjusted by the given offset in milliseconds. Lower
6367 values have higher priority.
6368 Note that the resulting timestamp is is only tracked with enough precision
6369 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
6370 adjusted timestamp exceeds this value, it will be misidentified as highest
6371 priority. Thus it is important to set "timeout queue" to a value, where when
6372 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006373
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006374http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01006375
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006376 This rewrites the request's query string which appears after the first
6377 question mark ("?") with the result of the evaluation of format string <fmt>.
6378 The part prior to the question mark is left intact. If the request doesn't
6379 contain a question mark and the new value is not empty, then one is added at
6380 the end of the URI, followed by the new value. If a question mark was
6381 present, it will never be removed even if the value is empty. This can be
6382 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08006383
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006384 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006385
6386 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006387 # replace "%3D" with "=" in the query string
6388 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006389
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006390http-request set-src <expr> [ { if | unless } <condition> ]
6391 This is used to set the source IP address to the value of specified
6392 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
6393 provides the correct IP in a HTTP header; or you want to mask source IP for
Olivier Doucet56e31202020-04-21 09:32:56 +02006394 privacy. All subsequent calls to "src" fetch will return this value
6395 (see example).
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006396
6397 Arguments :
6398 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6399 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006400
Olivier Doucet56e31202020-04-21 09:32:56 +02006401 See also "option forwardfor".
6402
Cyril Bonté78caf842010-03-10 22:41:43 +01006403 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006404 http-request set-src hdr(x-forwarded-for)
6405 http-request set-src src,ipmask(24)
6406
Olivier Doucet56e31202020-04-21 09:32:56 +02006407 # After the masking this will track connections
6408 # based on the IP address with the last byte zeroed out.
6409 http-request track-sc0 src
6410
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006411 When possible, set-src preserves the original source port as long as the
6412 address family allows it, otherwise the source port is set to 0.
6413
6414http-request set-src-port <expr> [ { if | unless } <condition> ]
6415
6416 This is used to set the source port address to the value of specified
6417 expression.
6418
6419 Arguments:
6420 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6421 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006422
Willy Tarreau20b0de52012-12-24 15:45:22 +01006423 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006424 http-request set-src-port hdr(x-port)
6425 http-request set-src-port int(4000)
6426
6427 When possible, set-src-port preserves the original source address as long as
6428 the address family supports a port, otherwise it forces the source address to
6429 IPv4 "0.0.0.0" before rewriting the port.
6430
Amaury Denoyelle8d228232020-12-10 13:43:54 +01006431http-request set-timeout server|tunnel { <timeout> | <expr> }
6432 [ { if | unless } <condition> ]
6433
6434 This action overrides the specified "server" or "tunnel" timeout for the
6435 current stream only. The timeout can be specified in millisecond or with any
6436 other unit if the number is suffixed by the unit as explained at the top of
6437 this document. It is also possible to write an expression which must returns
6438 a number interpreted as a timeout in millisecond.
6439
6440 Note that the server/tunnel timeouts are only relevant on the backend side
6441 and thus this rule is only available for the proxies with backend
6442 capabilities. Also the timeout value must be non-null to obtain the expected
6443 results.
6444
6445 Example:
6446 http-request set-timeout server 5s
6447 http-request set-timeout hdr(host),map_int(host.lst)
6448
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006449http-request set-tos <tos> [ { if | unless } <condition> ]
6450
6451 This is used to set the TOS or DSCP field value of packets sent to the client
6452 to the value passed in <tos> on platforms which support this. This value
6453 represents the whole 8 bits of the IP TOS field, and can be expressed both in
6454 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
6455 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
6456 be used to adjust some routing behavior on border routers based on some
6457 information from the request.
6458
6459 See RFC 2474, 2597, 3260 and 4594 for more information.
6460
6461http-request set-uri <fmt> [ { if | unless } <condition> ]
6462
6463 This rewrites the request URI with the result of the evaluation of format
6464 string <fmt>. The scheme, authority, path and query string are all replaced
6465 at once. This can be used to rewrite hosts in front of proxies, or to
6466 perform complex modifications to the URI such as moving parts between the
6467 path and the query string.
6468 See also "http-request set-path" and "http-request set-query".
6469
6470http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
6471
6472 This is used to set the contents of a variable. The variable is declared
6473 inline.
6474
6475 Arguments:
6476 <var-name> The name of the variable starts with an indication about its
6477 scope. The scopes allowed are:
6478 "proc" : the variable is shared with the whole process
6479 "sess" : the variable is shared with the whole session
6480 "txn" : the variable is shared with the transaction
6481 (request and response)
6482 "req" : the variable is shared only during request
6483 processing
6484 "res" : the variable is shared only during response
6485 processing
6486 This prefix is followed by a name. The separator is a '.'.
6487 The name may only contain characters 'a-z', 'A-Z', '0-9'
6488 and '_'.
6489
6490 <expr> Is a standard HAProxy expression formed by a sample-fetch
6491 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01006492
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006493 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006494 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006495
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006496http-request send-spoe-group <engine-name> <group-name>
6497 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006498
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006499 This action is used to trigger sending of a group of SPOE messages. To do so,
6500 the SPOE engine used to send messages must be defined, as well as the SPOE
6501 group to send. Of course, the SPOE engine must refer to an existing SPOE
6502 filter. If not engine name is provided on the SPOE filter line, the SPOE
6503 agent name must be used.
6504
6505 Arguments:
6506 <engine-name> The SPOE engine name.
6507
6508 <group-name> The SPOE group name as specified in the engine
6509 configuration.
6510
6511http-request silent-drop [ { if | unless } <condition> ]
6512
6513 This stops the evaluation of the rules and makes the client-facing connection
6514 suddenly disappear using a system-dependent way that tries to prevent the
6515 client from being notified. The effect it then that the client still sees an
6516 established connection while there's none on HAProxy. The purpose is to
6517 achieve a comparable effect to "tarpit" except that it doesn't use any local
6518 resource at all on the machine running HAProxy. It can resist much higher
6519 loads than "tarpit", and slow down stronger attackers. It is important to
6520 understand the impact of using this mechanism. All stateful equipment placed
6521 between the client and HAProxy (firewalls, proxies, load balancers) will also
6522 keep the established connection for a long time and may suffer from this
6523 action.
6524 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6525 option is used to block the emission of a TCP reset. On other systems, the
6526 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6527 router, though it's still delivered to local networks. Do not use it unless
6528 you fully understand how it works.
6529
Christopher Faulet46f95542019-12-20 10:07:22 +01006530http-request strict-mode { on | off }
6531
6532 This enables or disables the strict rewriting mode for following rules. It
6533 does not affect rules declared before it and it is only applicable on rules
6534 performing a rewrite on the requests. When the strict mode is enabled, any
6535 rewrite failure triggers an internal error. Otherwise, such errors are
6536 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006537 rewrites optional while others must be performed to continue the request
Christopher Faulet46f95542019-12-20 10:07:22 +01006538 processing.
6539
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006540 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006541 when a ruleset evaluation ends. So, for instance, if you change the mode on
6542 the frontend, the default mode is restored when HAProxy starts the backend
6543 rules evaluation.
6544
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006545http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
6546http-request tarpit [ { status | deny_status } <code>] [content-type <type>]
6547 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6548 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6549 [ hdr <name> <fmt> ]*
6550 [ { if | unless } <condition> ]
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006551
6552 This stops the evaluation of the rules and immediately blocks the request
6553 without responding for a delay specified by "timeout tarpit" or
6554 "timeout connect" if the former is not set. After that delay, if the client
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006555 is still connected, a response is returned so that the client does not
6556 suspect it has been tarpitted. Logs will report the flags "PT". The goal of
6557 the tarpit rule is to slow down robots during an attack when they're limited
6558 on the number of concurrent requests. It can be very efficient against very
6559 dumb robots, and will significantly reduce the load on firewalls compared to
6560 a "deny" rule. But when facing "correctly" developed robots, it can make
6561 things worse by forcing haproxy and the front firewall to support insane
6562 number of concurrent connections. By default an HTTP error 500 is returned.
6563 But the response may be customized using same syntax than
6564 "http-request return" rules. Thus, see "http-request return" for details.
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006565 For compatibility purpose, when no argument is defined, or only "deny_status",
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006566 the argument "default-errorfiles" is implied. It means
6567 "http-request tarpit [deny_status <status>]" is an alias of
6568 "http-request tarpit [status <status>] default-errorfiles".
6569 No further "http-request" rules are evaluated.
6570 See also "http-request return" and "http-request silent-drop".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006571
6572http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6573http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6574http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
6575
6576 This enables tracking of sticky counters from current request. These rules do
6577 not stop evaluation and do not change default action. The number of counters
6578 that may be simultaneously tracked by the same connection is set in
6579 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
Matteo Contrini1857b8c2020-10-16 17:35:54 +02006580 so the track-sc number is between 0 and (MAX_SESS_STKCTR-1). The first
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006581 "track-sc0" rule executed enables tracking of the counters of the specified
6582 table as the first set. The first "track-sc1" rule executed enables tracking
6583 of the counters of the specified table as the second set. The first
6584 "track-sc2" rule executed enables tracking of the counters of the specified
6585 table as the third set. It is a recommended practice to use the first set of
6586 counters for the per-frontend counters and the second set for the per-backend
6587 ones. But this is just a guideline, all may be used everywhere.
6588
6589 Arguments :
6590 <key> is mandatory, and is a sample expression rule as described in
6591 section 7.3. It describes what elements of the incoming request or
6592 connection will be analyzed, extracted, combined, and used to
6593 select which table entry to update the counters.
6594
6595 <table> is an optional table to be used instead of the default one, which
6596 is the stick-table declared in the current proxy. All the counters
6597 for the matches and updates for the key will then be performed in
6598 that table until the session ends.
6599
6600 Once a "track-sc*" rule is executed, the key is looked up in the table and if
6601 it is not found, an entry is allocated for it. Then a pointer to that entry
6602 is kept during all the session's life, and this entry's counters are updated
6603 as often as possible, every time the session's counters are updated, and also
6604 systematically when the session ends. Counters are only updated for events
6605 that happen after the tracking has been started. As an exception, connection
6606 counters and request counters are systematically updated so that they reflect
6607 useful information.
6608
6609 If the entry tracks concurrent connection counters, one connection is counted
6610 for as long as the entry is tracked, and the entry will not expire during
6611 that time. Tracking counters also provides a performance advantage over just
6612 checking the keys, because only one table lookup is performed for all ACL
6613 checks that make use of it.
6614
6615http-request unset-var(<var-name>) [ { if | unless } <condition> ]
6616
6617 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006618
6619 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006620 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006621
Christopher Faulet579d83b2019-11-22 15:34:17 +01006622http-request use-service <service-name> [ { if | unless } <condition> ]
6623
6624 This directive executes the configured HTTP service to reply to the request
6625 and stops the evaluation of the rules. An HTTP service may choose to reply by
6626 sending any valid HTTP response or it may immediately close the connection
6627 without sending any response. Outside natives services, for instance the
6628 Prometheus exporter, it is possible to write your own services in Lua. No
6629 further "http-request" rules are evaluated.
6630
6631 Arguments :
6632 <service-name> is mandatory. It is the service to call
6633
6634 Example:
6635 http-request use-service prometheus-exporter if { path /metrics }
6636
Christopher Faulet021a8e42021-03-29 10:46:38 +02006637http-request wait-for-body time <time> [ at-least <bytes> ]
6638 [ { if | unless } <condition> ]
6639
6640 This will delay the processing of the request waiting for the payload for at
6641 most <time> milliseconds. if "at-least" argument is specified, HAProxy stops
6642 to wait the payload when the first <bytes> bytes are received. 0 means no
6643 limit, it is the default value. Regardless the "at-least" argument value,
6644 HAProxy stops to wait if the whole payload is received or if the request
6645 buffer is full. This action may be used as a replacement to "option
6646 http-buffer-request".
6647
6648 Arguments :
6649
6650 <time> is mandatory. It is the maximum time to wait for the body. It
6651 follows the HAProxy time format and is expressed in milliseconds.
6652
6653 <bytes> is optional. It is the minimum payload size to receive to stop to
6654 wait. It fallows the HAProxy size format and is expressed in
6655 bytes.
6656
6657 Example:
6658 http-request wait-for-body time 1s at-least 1k if METH_POST
6659
6660 See also : "option http-buffer-request"
6661
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006662http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006663
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006664 This will delay the processing of the request until the SSL handshake
6665 happened. This is mostly useful to delay processing early data until we're
6666 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006667
Willy Tarreauef781042010-01-27 11:53:01 +01006668
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006669http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006670 Access control for Layer 7 responses
6671
6672 May be used in sections: defaults | frontend | listen | backend
6673 no | yes | yes | yes
6674
6675 The http-response statement defines a set of rules which apply to layer 7
6676 processing. The rules are evaluated in their declaration order when they are
6677 met in a frontend, listen or backend section. Any rule may optionally be
6678 followed by an ACL-based condition, in which case it will only be evaluated
6679 if the condition is true. Since these rules apply on responses, the backend
6680 rules are applied first, followed by the frontend's rules.
6681
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006682 The first keyword is the rule's action. The supported actions are described
6683 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006684
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006685 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006686
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006687 Example:
6688 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02006689
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006690 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06006691
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006692 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
6693 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06006694
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006695 Example:
6696 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06006697
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006698 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06006699
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006700 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
6701 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06006702
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006703 See also : "http-request", section 3.4 about userlists and section 7 about
6704 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06006705
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006706http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006707
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006708 This is used to add a new entry into an ACL. The ACL must be loaded from a
6709 file (even a dummy empty file). The file name of the ACL to be updated is
6710 passed between parentheses. It takes one argument: <key fmt>, which follows
6711 log-format rules, to collect content of the new entry. It performs a lookup
6712 in the ACL before insertion, to avoid duplicated (or more) values.
6713 This lookup is done by a linear search and can be expensive with large lists!
6714 It is the equivalent of the "add acl" command from the stats socket, but can
6715 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06006716
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006717http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006718
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006719 This appends an HTTP header field whose name is specified in <name> and whose
6720 value is defined by <fmt> which follows the log-format rules (see Custom Log
6721 Format in section 8.2.4). This may be used to send a cookie to a client for
6722 example, or to pass some internal information.
6723 This rule is not final, so it is possible to add other similar rules.
6724 Note that header addition is performed immediately, so one rule might reuse
6725 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06006726
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006727http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006728
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006729 This stops the evaluation of the rules and lets the response pass the check.
6730 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06006731
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02006732http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006733
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006734 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06006735
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006736http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006737
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006738 This captures sample expression <sample> from the response buffer, and
6739 converts it to a string. The resulting string is stored into the next request
6740 "capture" slot, so it will possibly appear next to some captured HTTP
6741 headers. It will then automatically appear in the logs, and it will be
6742 possible to extract it using sample fetch rules to feed it into headers or
6743 anything. Please check section 7.3 (Fetching samples) and
6744 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006745
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006746 The keyword "id" is the id of the capture slot which is used for storing the
6747 string. The capture slot must be defined in an associated frontend.
6748 This is useful to run captures in backends. The slot id can be declared by a
6749 previous directive "http-response capture" or with the "declare capture"
6750 keyword.
Baptiste Assmann19a69b32020-01-16 14:34:22 +01006751
6752 When using this action in a backend, double check that the relevant
6753 frontend(s) have the required capture slots otherwise, this rule will be
6754 ignored at run time. This can't be detected at configuration parsing time
6755 due to HAProxy's ability to dynamically resolve backend name at runtime.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006756
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006757http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006758
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006759 This is used to delete an entry from an ACL. The ACL must be loaded from a
6760 file (even a dummy empty file). The file name of the ACL to be updated is
6761 passed between parentheses. It takes one argument: <key fmt>, which follows
6762 log-format rules, to collect content of the entry to delete.
6763 It is the equivalent of the "del acl" command from the stats socket, but can
6764 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02006765
Maciej Zdebebdd4c52020-11-20 13:58:48 +00006766http-response del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02006767
Maciej Zdebebdd4c52020-11-20 13:58:48 +00006768 This removes all HTTP header fields whose name is specified in <name>. <meth>
6769 is the matching method, applied on the header name. Supported matching methods
6770 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
6771 (substring match) and "reg" (regex match). If not specified, exact matching
6772 method is used.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02006773
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006774http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02006775
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006776 This is used to delete an entry from a MAP. The MAP must be loaded from a
6777 file (even a dummy empty file). The file name of the MAP to be updated is
6778 passed between parentheses. It takes one argument: <key fmt>, which follows
6779 log-format rules, to collect content of the entry to delete.
6780 It takes one argument: "file name" It is the equivalent of the "del map"
6781 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006782
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006783http-response deny [deny_status <status>] [ { if | unless } <condition> ]
6784http-response deny [ { status | deny_status } <code>] [content-type <type>]
6785 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6786 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6787 [ hdr <name> <fmt> ]*
6788 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006789
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006790 This stops the evaluation of the rules and immediately rejects the response.
6791 By default an HTTP 502 error is returned. But the response may be customized
6792 using same syntax than "http-response return" rules. Thus, see
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006793 "http-response return" for details. For compatibility purpose, when no
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006794 argument is defined, or only "deny_status", the argument "default-errorfiles"
6795 is implied. It means "http-response deny [deny_status <status>]" is an alias
6796 of "http-response deny [status <status>] default-errorfiles".
Christopher Faulet040c8cd2020-01-13 16:43:45 +01006797 No further "http-response" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006798 See also "http-response return".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006799
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006800http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006801
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006802 This performs an HTTP redirection based on a redirect rule.
6803 This supports a format string similarly to "http-request redirect" rules,
6804 with the exception that only the "location" type of redirect is possible on
6805 the response. See the "redirect" keyword for the rule's syntax. When a
6806 redirect rule is applied during a response, connections to the server are
6807 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006808
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006809http-response replace-header <name> <regex-match> <replace-fmt>
6810 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006811
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006812 This works like "http-request replace-header" except that it works on the
6813 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01006814
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006815 Example:
6816 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02006817
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006818 # applied to:
6819 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006820
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006821 # outputs:
6822 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 +02006823
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006824 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006825
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006826http-response replace-value <name> <regex-match> <replace-fmt>
6827 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006828
Tim Duesterhus6bd909b2020-01-17 15:53:18 +01006829 This works like "http-request replace-value" except that it works on the
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006830 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006831
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006832 Example:
6833 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006834
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006835 # applied to:
6836 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01006837
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006838 # outputs:
6839 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006840
Christopher Faulet24231ab2020-01-24 17:44:23 +01006841http-response return [status <code>] [content-type <type>]
6842 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6843 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006844 [ hdr <name> <value> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01006845 [ { if | unless } <condition> ]
6846
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006847 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01006848 default status code used for the response is 200. It can be optionally
6849 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04006850 specified as an argument to "content-type". Finally the response itself may
Christopher Faulet24231ab2020-01-24 17:44:23 +01006851 be defined. If can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006852 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01006853 are followed to create the response :
6854
6855 * If neither the errorfile nor the payload to use is defined, a dummy
6856 response is returned. Only the "status" argument is considered. It can be
6857 any code in the range [200, 599]. The "content-type" argument, if any, is
6858 ignored.
6859
6860 * If "default-errorfiles" argument is set, the proxy's errorfiles are
6861 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04006862 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Christopher Faulete095f312020-12-07 11:22:24 +01006863 425, 429, 500, 501, 502, 503, and 504). The "content-type" argument, if
6864 any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006865
6866 * If a specific errorfile is defined, with an "errorfile" argument, the
6867 corresponding file, containing a full HTTP response, is returned. Only the
6868 "status" argument is considered. It must be one of the status code handled
Christopher Faulete095f312020-12-07 11:22:24 +01006869 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 501,
6870 502, 503, and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006871
6872 * If an http-errors section is defined, with an "errorfiles" argument, the
6873 corresponding file in the specified http-errors section, containing a full
6874 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04006875 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Christopher Faulete095f312020-12-07 11:22:24 +01006876 408, 410, 413, 425, 429, 500, 501, 502, 503, and 504). The "content-type"
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006877 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006878
6879 * If a "file" or a "lf-file" argument is specified, the file's content is
6880 used as the response payload. If the file is not empty, its content-type
6881 must be set as argument to "content-type". Otherwise, any "content-type"
6882 argument is ignored. With a "lf-file" argument, the file's content is
6883 evaluated as a log-format string. With a "file" argument, it is considered
6884 as a raw content.
6885
6886 * If a "string" or "lf-string" argument is specified, the defined string is
6887 used as the response payload. The content-type must always be set as
6888 argument to "content-type". With a "lf-string" argument, the string is
6889 evaluated as a log-format string. With a "string" argument, it is
6890 considered as a raw string.
6891
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006892 When the response is not based an errorfile, it is possible to appends HTTP
6893 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
6894 arguments are ignored. For each one, the header name is specified in <name>
6895 and its value is defined by <fmt> which follows the log-format rules.
6896
Christopher Faulet24231ab2020-01-24 17:44:23 +01006897 Note that the generated response must be smaller than a buffer. And to avoid
6898 any warning, when an errorfile or a raw file is loaded, the buffer space
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006899 reserved to the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006900
6901 No further "http-response" rules are evaluated.
6902
6903 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006904 http-response return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006905 if { status eq 404 }
6906
6907 http-response return content-type text/plain \
6908 string "This is the end !" \
6909 if { status eq 500 }
6910
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006911http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6912http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08006913
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006914 This action increments the GPC0 or GPC1 counter according with the sticky
6915 counter designated by <sc-id>. If an error occurs, this action silently fails
6916 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02006917
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006918http-response sc-set-gpt0(<sc-id>) { <int> | <expr> }
6919 [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02006920
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006921 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6922 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
6923 boolean. If an error occurs, this action silently fails and the actions
6924 evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01006925
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006926http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02006927
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006928 This action is used to trigger sending of a group of SPOE messages. To do so,
6929 the SPOE engine used to send messages must be defined, as well as the SPOE
6930 group to send. Of course, the SPOE engine must refer to an existing SPOE
6931 filter. If not engine name is provided on the SPOE filter line, the SPOE
6932 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006933
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006934 Arguments:
6935 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006936
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006937 <group-name> The SPOE group name as specified in the engine
6938 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006939
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006940http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006941
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006942 This does the same as "add-header" except that the header name is first
6943 removed if it existed. This is useful when passing security information to
6944 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006945
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006946http-response set-log-level <level> [ { if | unless } <condition> ]
6947
6948 This is used to change the log level of the current request when a certain
6949 condition is met. Valid levels are the 8 syslog levels (see the "log"
6950 keyword) plus the special level "silent" which disables logging for this
6951 request. This rule is not final so the last matching rule wins. This rule can
6952 be useful to disable health checks coming from another equipment.
6953
6954http-response set-map(<file-name>) <key fmt> <value fmt>
6955
6956 This is used to add a new entry into a MAP. The MAP must be loaded from a
6957 file (even a dummy empty file). The file name of the MAP to be updated is
6958 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
6959 log-format rules, used to collect MAP key, and <value fmt>, which follows
6960 log-format rules, used to collect content for the new entry. It performs a
6961 lookup in the MAP before insertion, to avoid duplicated (or more) values.
6962 This lookup is done by a linear search and can be expensive with large lists!
6963 It is the equivalent of the "set map" command from the stats socket, but can
6964 be triggered by an HTTP response.
6965
6966http-response set-mark <mark> [ { if | unless } <condition> ]
6967
6968 This is used to set the Netfilter MARK on all packets sent to the client to
6969 the value passed in <mark> on platforms which support it. This value is an
6970 unsigned 32 bit value which can be matched by netfilter and by the routing
6971 table. It can be expressed both in decimal or hexadecimal format (prefixed
6972 by "0x"). This can be useful to force certain packets to take a different
6973 route (for example a cheaper network path for bulk downloads). This works on
6974 Linux kernels 2.6.32 and above and requires admin privileges.
6975
6976http-response set-nice <nice> [ { if | unless } <condition> ]
6977
6978 This sets the "nice" factor of the current request being processed.
6979 It only has effect against the other requests being processed at the same
6980 time. The default value is 0, unless altered by the "nice" setting on the
6981 "bind" line. The accepted range is -1024..1024. The higher the value, the
6982 nicest the request will be. Lower values will make the request more important
6983 than other ones. This can be useful to improve the speed of some requests, or
6984 lower the priority of non-important requests. Using this setting without
6985 prior experimentation can cause some major slowdown.
6986
6987http-response set-status <status> [reason <str>]
6988 [ { if | unless } <condition> ]
6989
6990 This replaces the response status code with <status> which must be an integer
6991 between 100 and 999. Optionally, a custom reason text can be provided defined
6992 by <str>, or the default reason for the specified code will be used as a
6993 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08006994
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006995 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006996 # return "431 Request Header Fields Too Large"
6997 http-response set-status 431
6998 # return "503 Slow Down", custom reason
6999 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007000
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007001http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007002
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007003 This is used to set the TOS or DSCP field value of packets sent to the client
7004 to the value passed in <tos> on platforms which support this.
7005 This value represents the whole 8 bits of the IP TOS field, and can be
7006 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
7007 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
7008 always 0. This can be used to adjust some routing behavior on border routers
7009 based on some information from the request.
7010
7011 See RFC 2474, 2597, 3260 and 4594 for more information.
7012
7013http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
7014
7015 This is used to set the contents of a variable. The variable is declared
7016 inline.
7017
7018 Arguments:
7019 <var-name> The name of the variable starts with an indication about its
7020 scope. The scopes allowed are:
7021 "proc" : the variable is shared with the whole process
7022 "sess" : the variable is shared with the whole session
7023 "txn" : the variable is shared with the transaction
7024 (request and response)
7025 "req" : the variable is shared only during request
7026 processing
7027 "res" : the variable is shared only during response
7028 processing
7029 This prefix is followed by a name. The separator is a '.'.
7030 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
7031 and '_'.
7032
7033 <expr> Is a standard HAProxy expression formed by a sample-fetch
7034 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007035
7036 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007037 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007038
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007039http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007040
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007041 This stops the evaluation of the rules and makes the client-facing connection
7042 suddenly disappear using a system-dependent way that tries to prevent the
7043 client from being notified. The effect it then that the client still sees an
7044 established connection while there's none on HAProxy. The purpose is to
7045 achieve a comparable effect to "tarpit" except that it doesn't use any local
7046 resource at all on the machine running HAProxy. It can resist much higher
7047 loads than "tarpit", and slow down stronger attackers. It is important to
7048 understand the impact of using this mechanism. All stateful equipment placed
7049 between the client and HAProxy (firewalls, proxies, load balancers) will also
7050 keep the established connection for a long time and may suffer from this
7051 action.
7052 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
7053 option is used to block the emission of a TCP reset. On other systems, the
7054 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
7055 router, though it's still delivered to local networks. Do not use it unless
7056 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02007057
Christopher Faulet46f95542019-12-20 10:07:22 +01007058http-response strict-mode { on | off }
7059
7060 This enables or disables the strict rewriting mode for following rules. It
7061 does not affect rules declared before it and it is only applicable on rules
7062 performing a rewrite on the responses. When the strict mode is enabled, any
7063 rewrite failure triggers an internal error. Otherwise, such errors are
7064 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05007065 rewrites optional while others must be performed to continue the response
Christopher Faulet46f95542019-12-20 10:07:22 +01007066 processing.
7067
Christopher Faulet1aea50e2020-01-17 16:03:53 +01007068 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01007069 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04007070 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet46f95542019-12-20 10:07:22 +01007071 rules evaluation.
7072
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007073http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
7074http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
7075http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02007076
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02007077 This enables tracking of sticky counters from current response. Please refer
7078 to "http-request track-sc" for a complete description. The only difference
7079 from "http-request track-sc" is the <key> sample expression can only make use
7080 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
7081 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
7082 supported, haproxy will fail and warn while parsing the config.
7083
7084http-response unset-var(<var-name>) [ { if | unless } <condition> ]
7085
7086 This is used to unset a variable. See "http-response set-var" for details
7087 about <var-name>.
7088
7089 Example:
7090 http-response unset-var(sess.last_redir)
7091
Christopher Faulet021a8e42021-03-29 10:46:38 +02007092http-response wait-for-body time <time> [ at-least <bytes> ]
7093 [ { if | unless } <condition> ]
7094
7095 This will delay the processing of the response waiting for the payload for at
7096 most <time> milliseconds. if "at-least" argument is specified, HAProxy stops
7097 to wait the payload when the first <bytes> bytes are received. 0 means no
7098 limit, it is the default value. Regardless the "at-least" argument value,
7099 HAProxy stops to wait if the whole payload is received or if the response
7100 buffer is full.
7101
7102 Arguments :
7103
7104 <time> is mandatory. It is the maximum time to wait for the body. It
7105 follows the HAProxy time format and is expressed in milliseconds.
7106
7107 <bytes> is optional. It is the minimum payload size to receive to stop to
7108 wait. It fallows the HAProxy size format and is expressed in
7109 bytes.
7110
7111 Example:
7112 http-response wait-for-body time 1s at-least 10k
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02007113
Willy Tarreau30631952015-08-06 15:05:24 +02007114http-reuse { never | safe | aggressive | always }
7115 Declare how idle HTTP connections may be shared between requests
7116
7117 May be used in sections: defaults | frontend | listen | backend
7118 yes | no | yes | yes
7119
7120 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01007121 which is considered safe for reuse is moved back to the server's idle
7122 connections pool so that any other request can make use of it. This is the
7123 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02007124
7125 The argument indicates the desired connection reuse strategy :
7126
Olivier Houchard86006a52018-12-14 19:37:49 +01007127 - "never" : idle connections are never shared between sessions. This mode
7128 may be enforced to cancel a different strategy inherited from
7129 a defaults section or for troubleshooting. For example, if an
7130 old bogus application considers that multiple requests over
7131 the same connection come from the same client and it is not
7132 possible to fix the application, it may be desirable to
7133 disable connection sharing in a single backend. An example of
7134 such an application could be an old haproxy using cookie
7135 insertion in tunnel mode and not checking any request past the
7136 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02007137
Olivier Houchard86006a52018-12-14 19:37:49 +01007138 - "safe" : this is the default and the recommended strategy. The first
7139 request of a session is always sent over its own connection,
7140 and only subsequent requests may be dispatched over other
7141 existing connections. This ensures that in case the server
7142 closes the connection when the request is being sent, the
7143 browser can decide to silently retry it. Since it is exactly
7144 equivalent to regular keep-alive, there should be no side
Amaury Denoyelle27179652020-10-14 18:17:12 +02007145 effects. There is also a special handling for the connections
7146 using protocols subject to Head-of-line blocking (backend with
7147 h2 or fcgi). In this case, when at least one stream is
7148 processed, the used connection is reserved to handle streams
7149 of the same session. When no more streams are processed, the
7150 connection is released and can be reused.
Willy Tarreau30631952015-08-06 15:05:24 +02007151
7152 - "aggressive" : this mode may be useful in webservices environments where
7153 all servers are not necessarily known and where it would be
7154 appreciable to deliver most first requests over existing
7155 connections. In this case, first requests are only delivered
7156 over existing connections that have been reused at least once,
7157 proving that the server correctly supports connection reuse.
7158 It should only be used when it's sure that the client can
7159 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02007160 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02007161 downsides of rare connection failures.
7162
7163 - "always" : this mode is only recommended when the path to the server is
7164 known for never breaking existing connections quickly after
7165 releasing them. It allows the first request of a session to be
7166 sent to an existing connection. This can provide a significant
7167 performance increase over the "safe" strategy when the backend
7168 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01007169 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02007170 sharing. It is recommended that the "http-keep-alive" timeout
7171 remains low in this mode so that no dead connections remain
7172 usable. In most cases, this will lead to the same performance
7173 gains as "aggressive" but with more risks. It should only be
7174 used when it improves the situation over "aggressive".
7175
7176 When http connection sharing is enabled, a great care is taken to respect the
Amaury Denoyelled773a4e2021-01-29 15:18:49 +01007177 connection properties and compatibility. Indeed, some properties are specific
7178 and it is not possibly to reuse it blindly. Those are the SSL SNI, source
7179 and destination address and proxy protocol block. A connection is reused only
7180 if it shares the same set of properties with the request.
Willy Tarreau30631952015-08-06 15:05:24 +02007181
Amaury Denoyelled773a4e2021-01-29 15:18:49 +01007182 Also note that connections with certain bogus authentication schemes (relying
7183 on the connection) like NTLM are marked private and never shared.
Willy Tarreau30631952015-08-06 15:05:24 +02007184
Lukas Tribuse8adfeb2019-11-06 11:50:25 +01007185 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02007186
7187 Note: connection reuse improves the accuracy of the "server maxconn" setting,
7188 because almost no new connection will be established while idle connections
7189 remain available. This is particularly true with the "always" strategy.
7190
7191 See also : "option http-keep-alive", "server maxconn"
7192
7193
Mark Lamourinec2247f02012-01-04 13:02:01 -05007194http-send-name-header [<header>]
7195 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05007196 May be used in sections: defaults | frontend | listen | backend
7197 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05007198 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05007199 <header> The header string to use to send the server name
7200
Willy Tarreau81bef7e2019-10-07 14:58:02 +02007201 The "http-send-name-header" statement causes the header field named <header>
7202 to be set to the name of the target server at the moment the request is about
7203 to be sent on the wire. Any existing occurrences of this header are removed.
7204 Upon retries and redispatches, the header field is updated to always reflect
7205 the server being attempted to connect to. Given that this header is modified
7206 very late in the connection setup, it may have unexpected effects on already
7207 modified headers. For example using it with transport-level header such as
7208 connection, content-length, transfer-encoding and so on will likely result in
7209 invalid requests being sent to the server. Additionally it has been reported
7210 that this directive is currently being used as a way to overwrite the Host
7211 header field in outgoing requests; while this trick has been known to work
7212 as a side effect of the feature for some time, it is not officially supported
7213 and might possibly not work anymore in a future version depending on the
7214 technical difficulties this feature induces. A long-term solution instead
7215 consists in fixing the application which required this trick so that it binds
7216 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05007217
7218 See also : "server"
7219
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01007220id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02007221 Set a persistent ID to a proxy.
7222 May be used in sections : defaults | frontend | listen | backend
7223 no | yes | yes | yes
7224 Arguments : none
7225
7226 Set a persistent ID for the proxy. This ID must be unique and positive.
7227 An unused ID will automatically be assigned if unset. The first assigned
7228 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01007229
7230
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007231ignore-persist { if | unless } <condition>
7232 Declare a condition to ignore persistence
7233 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01007234 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007235
7236 By default, when cookie persistence is enabled, every requests containing
7237 the cookie are unconditionally persistent (assuming the target server is up
7238 and running).
7239
7240 The "ignore-persist" statement allows one to declare various ACL-based
7241 conditions which, when met, will cause a request to ignore persistence.
7242 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007243 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007244 persistence for a specific User-Agent (for example, some web crawler bots).
7245
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007246 The persistence is ignored when an "if" condition is met, or unless an
7247 "unless" condition is met.
7248
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03007249 Example:
7250 acl url_static path_beg /static /images /img /css
7251 acl url_static path_end .gif .png .jpg .css .js
7252 ignore-persist if url_static
7253
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007254 See also : "force-persist", "cookie", and section 7 about ACL usage.
7255
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007256load-server-state-from-file { global | local | none }
7257 Allow seamless reload of HAProxy
7258 May be used in sections: defaults | frontend | listen | backend
7259 yes | no | yes | yes
7260
7261 This directive points HAProxy to a file where server state from previous
7262 running process has been saved. That way, when starting up, before handling
7263 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01007264 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007265 to tell haproxy which file to use. For now, only 2 arguments to either prevent
7266 loading state or load states from a file containing all backends and servers.
7267 The state file can be generated by running the command "show servers state"
7268 over the stats socket and redirect output.
7269
Davor Ocelice9ed2812017-12-25 17:49:28 +01007270 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007271 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02007272 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007273
7274 Arguments:
7275 global load the content of the file pointed by the global directive
7276 named "server-state-file".
7277
7278 local load the content of the file pointed by the directive
7279 "server-state-file-name" if set. If not set, then the backend
7280 name is used as a file name.
7281
7282 none don't load any stat for this backend
7283
7284 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01007285 - server's IP address is preserved across reloads by default, but the
7286 order can be changed thanks to the server's "init-addr" setting. This
7287 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007288 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01007289 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007290
7291 - server's weight is applied from previous running process unless it has
7292 has changed between previous and new configuration files.
7293
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007294 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007295
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007296 global
7297 stats socket /tmp/socket
7298 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007299
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007300 defaults
7301 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007302
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007303 backend bk
7304 server s1 127.0.0.1:22 check weight 11
7305 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007306
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007307
7308 Then one can run :
7309
7310 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
7311
7312 Content of the file /tmp/server_state would be like this:
7313
7314 1
7315 # <field names skipped for the doc example>
7316 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
7317 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
7318
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007319 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007320
7321 global
7322 stats socket /tmp/socket
7323 server-state-base /etc/haproxy/states
7324
7325 defaults
7326 load-server-state-from-file local
7327
7328 backend bk
7329 server s1 127.0.0.1:22 check weight 11
7330 server s2 127.0.0.1:22 check weight 12
7331
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007332
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007333 Then one can run :
7334
7335 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
7336
7337 Content of the file /etc/haproxy/states/bk would be like this:
7338
7339 1
7340 # <field names skipped for the doc example>
7341 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
7342 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
7343
7344 See also: "server-state-file", "server-state-file-name", and
7345 "show servers state"
7346
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007347
Willy Tarreau2769aa02007-12-27 18:26:09 +01007348log global
Jan Wagner3e678602020-12-17 22:22:32 +01007349log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02007350 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02007351no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01007352 Enable per-instance logging of events and traffic.
7353 May be used in sections : defaults | frontend | listen | backend
7354 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02007355
7356 Prefix :
7357 no should be used when the logger list must be flushed. For example,
7358 if you don't want to inherit from the default logger list. This
7359 prefix does not allow arguments.
7360
Willy Tarreau2769aa02007-12-27 18:26:09 +01007361 Arguments :
7362 global should be used when the instance's logging parameters are the
7363 same as the global ones. This is the most common usage. "global"
7364 replaces <address>, <facility> and <level> with those of the log
7365 entries found in the "global" section. Only one "log global"
7366 statement may be used per instance, and this form takes no other
7367 parameter.
7368
7369 <address> indicates where to send the logs. It takes the same format as
7370 for the "global" section's logs, and can be one of :
7371
7372 - An IPv4 address optionally followed by a colon (':') and a UDP
7373 port. If no port is specified, 514 is used by default (the
7374 standard syslog port).
7375
David du Colombier24bb5f52011-03-17 10:40:23 +01007376 - An IPv6 address followed by a colon (':') and optionally a UDP
7377 port. If no port is specified, 514 is used by default (the
7378 standard syslog port).
7379
Willy Tarreau2769aa02007-12-27 18:26:09 +01007380 - A filesystem path to a UNIX domain socket, keeping in mind
7381 considerations for chroot (be sure the path is accessible
7382 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007383 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007384
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007385 - A file descriptor number in the form "fd@<number>", which may
7386 point to a pipe, terminal, or socket. In this case unbuffered
7387 logs are used and one writev() call per log is performed. This
7388 is a bit expensive but acceptable for most workloads. Messages
7389 sent this way will not be truncated but may be dropped, in
7390 which case the DroppedLogs counter will be incremented. The
7391 writev() call is atomic even on pipes for messages up to
7392 PIPE_BUF size, which POSIX recommends to be at least 512 and
7393 which is 4096 bytes on most modern operating systems. Any
7394 larger message may be interleaved with messages from other
7395 processes. Exceptionally for debugging purposes the file
7396 descriptor may also be directed to a file, but doing so will
7397 significantly slow haproxy down as non-blocking calls will be
7398 ignored. Also there will be no way to purge nor rotate this
7399 file without restarting the process. Note that the configured
7400 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01007401 with a TCP syslog server. See also the "short" and "raw"
7402 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007403
7404 - "stdout" / "stderr", which are respectively aliases for "fd@1"
7405 and "fd@2", see above.
7406
Willy Tarreauc046d162019-08-30 15:24:59 +02007407 - A ring buffer in the form "ring@<name>", which will correspond
7408 to an in-memory ring buffer accessible over the CLI using the
7409 "show events" command, which will also list existing rings and
7410 their sizes. Such buffers are lost on reload or restart but
7411 when used as a complement this can help troubleshooting by
7412 having the logs instantly available.
7413
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007414 You may want to reference some environment variables in the
7415 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01007416
Willy Tarreau18324f52014-06-27 18:10:07 +02007417 <length> is an optional maximum line length. Log lines larger than this
7418 value will be truncated before being sent. The reason is that
7419 syslog servers act differently on log line length. All servers
7420 support the default value of 1024, but some servers simply drop
7421 larger lines while others do log them. If a server supports long
7422 lines, it may make sense to set this value here in order to avoid
7423 truncating long lines. Similarly, if a server drops long lines,
7424 it is preferable to truncate them before sending them. Accepted
7425 values are 80 to 65535 inclusive. The default value of 1024 is
7426 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01007427 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02007428
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02007429 <ranges> A list of comma-separated ranges to identify the logs to sample.
7430 This is used to balance the load of the logs to send to the log
7431 server. The limits of the ranges cannot be null. They are numbered
7432 from 1. The size or period (in number of logs) of the sample must
7433 be set with <sample_size> parameter.
7434
7435 <sample_size>
7436 The size of the sample in number of logs to consider when balancing
7437 their logging loads. It is used to balance the load of the logs to
7438 send to the syslog server. This size must be greater or equal to the
7439 maximum of the high limits of the ranges.
7440 (see also <ranges> parameter).
7441
Willy Tarreauadb345d2018-11-12 07:56:13 +01007442 <format> is the log format used when generating syslog messages. It may be
7443 one of the following :
7444
Emeric Brun0237c4e2020-11-27 16:24:34 +01007445 local Analog to rfc3164 syslog message format except that hostname
7446 field is stripped. This is the default.
7447 Note: option "log-send-hostname" switches the default to
7448 rfc3164.
7449
7450 rfc3164 The RFC3164 syslog message format.
Willy Tarreauadb345d2018-11-12 07:56:13 +01007451 (https://tools.ietf.org/html/rfc3164)
7452
7453 rfc5424 The RFC5424 syslog message format.
7454 (https://tools.ietf.org/html/rfc5424)
7455
Emeric Brun54648852020-07-06 15:54:06 +02007456 priority A message containing only a level plus syslog facility between
7457 angle brackets such as '<63>', followed by the text. The PID,
7458 date, time, process name and system name are omitted. This is
7459 designed to be used with a local log server.
7460
Willy Tarreaue8746a02018-11-12 08:45:00 +01007461 short A message containing only a level between angle brackets such as
7462 '<3>', followed by the text. The PID, date, time, process name
7463 and system name are omitted. This is designed to be used with a
7464 local log server. This format is compatible with what the
7465 systemd logger consumes.
7466
Emeric Brun54648852020-07-06 15:54:06 +02007467 timed A message containing only a level between angle brackets such as
7468 '<3>', followed by ISO date and by the text. The PID, process
7469 name and system name are omitted. This is designed to be
7470 used with a local log server.
7471
7472 iso A message containing only the ISO date, followed by the text.
7473 The PID, process name and system name are omitted. This is
7474 designed to be used with a local log server.
7475
Willy Tarreauc1b06452018-11-12 11:57:56 +01007476 raw A message containing only the text. The level, PID, date, time,
7477 process name and system name are omitted. This is designed to
7478 be used in containers or during development, where the severity
7479 only depends on the file descriptor used (stdout/stderr).
7480
Willy Tarreau2769aa02007-12-27 18:26:09 +01007481 <facility> must be one of the 24 standard syslog facilities :
7482
Willy Tarreaue8746a02018-11-12 08:45:00 +01007483 kern user mail daemon auth syslog lpr news
7484 uucp cron auth2 ftp ntp audit alert cron2
7485 local0 local1 local2 local3 local4 local5 local6 local7
7486
Willy Tarreauc1b06452018-11-12 11:57:56 +01007487 Note that the facility is ignored for the "short" and "raw"
7488 formats, but still required as a positional field. It is
7489 recommended to use "daemon" in this case to make it clear that
7490 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007491
7492 <level> is optional and can be specified to filter outgoing messages. By
7493 default, all messages are sent. If a level is specified, only
7494 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007495 will be sent. An optional minimum level can be specified. If it
7496 is set, logs emitted with a more severe level than this one will
7497 be capped to this level. This is used to avoid sending "emerg"
7498 messages on all terminals on some default syslog configurations.
7499 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007500
7501 emerg alert crit err warning notice info debug
7502
William Lallemand0f99e342011-10-12 17:50:54 +02007503 It is important to keep in mind that it is the frontend which decides what to
7504 log from a connection, and that in case of content switching, the log entries
7505 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01007506
7507 However, backend log declaration define how and where servers status changes
7508 will be logged. Level "notice" will be used to indicate a server going up,
7509 "warning" will be used for termination signals and definitive service
7510 termination, and "alert" will be used for when a server goes down.
7511
7512 Note : According to RFC3164, messages are truncated to 1024 bytes before
7513 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007514
7515 Example :
7516 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01007517 log stdout format short daemon # send log to systemd
7518 log stdout format raw daemon # send everything to stdout
7519 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007520 log 127.0.0.1:514 local0 notice # only send important events
7521 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02007522 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01007523
Willy Tarreau2769aa02007-12-27 18:26:09 +01007524
William Lallemand48940402012-01-30 16:47:22 +01007525log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007526 Specifies the log format string to use for traffic logs
7527 May be used in sections: defaults | frontend | listen | backend
7528 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01007529
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007530 This directive specifies the log format string that will be used for all logs
7531 resulting from traffic passing through the frontend using this line. If the
7532 directive is used in a defaults section, all subsequent frontends will use
7533 the same log format. Please see section 8.2.4 which covers the log format
7534 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01007535
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007536 "log-format" directive overrides previous "option tcplog", "log-format" and
7537 "option httplog" directives.
7538
Dragan Dosen7ad31542015-09-28 17:16:47 +02007539log-format-sd <string>
7540 Specifies the RFC5424 structured-data log format string
7541 May be used in sections: defaults | frontend | listen | backend
7542 yes | yes | yes | no
7543
7544 This directive specifies the RFC5424 structured-data log format string that
7545 will be used for all logs resulting from traffic passing through the frontend
7546 using this line. If the directive is used in a defaults section, all
7547 subsequent frontends will use the same log format. Please see section 8.2.4
7548 which covers the log format string in depth.
7549
7550 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
7551 about the RFC5424 structured-data part.
7552
7553 Note : This log format string will be used only for loggers that have set
7554 log format to "rfc5424".
7555
7556 Example :
7557 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
7558
7559
Willy Tarreau094af4e2015-01-07 15:03:42 +01007560log-tag <string>
7561 Specifies the log tag to use for all outgoing logs
7562 May be used in sections: defaults | frontend | listen | backend
7563 yes | yes | yes | yes
7564
7565 Sets the tag field in the syslog header to this string. It defaults to the
7566 log-tag set in the global section, otherwise the program name as launched
7567 from the command line, which usually is "haproxy". Sometimes it can be useful
7568 to differentiate between multiple processes running on the same host, or to
7569 differentiate customer instances running in the same process. In the backend,
7570 logs about servers up/down will use this tag. As a hint, it can be convenient
7571 to set a log-tag related to a hosted customer in a defaults section then put
7572 all the frontends and backends for that customer, then start another customer
7573 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007574
Willy Tarreauc35362a2014-04-25 13:58:37 +02007575max-keep-alive-queue <value>
7576 Set the maximum server queue size for maintaining keep-alive connections
7577 May be used in sections: defaults | frontend | listen | backend
7578 yes | no | yes | yes
7579
7580 HTTP keep-alive tries to reuse the same server connection whenever possible,
7581 but sometimes it can be counter-productive, for example if a server has a lot
7582 of connections while other ones are idle. This is especially true for static
7583 servers.
7584
7585 The purpose of this setting is to set a threshold on the number of queued
7586 connections at which haproxy stops trying to reuse the same server and prefers
7587 to find another one. The default value, -1, means there is no limit. A value
7588 of zero means that keep-alive requests will never be queued. For very close
7589 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007590 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02007591 use a value of 10 or less). For remote servers suffering from a high latency,
7592 higher values might be needed to cover for the latency and/or the cost of
7593 picking a different server.
7594
7595 Note that this has no impact on responses which are maintained to the same
7596 server consecutively to a 401 response. They will still go to the same server
7597 even if they have to be queued.
7598
7599 See also : "option http-server-close", "option prefer-last-server", server
7600 "maxconn" and cookie persistence.
7601
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01007602max-session-srv-conns <nb>
7603 Set the maximum number of outgoing connections we can keep idling for a given
7604 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
7605 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02007606
Willy Tarreau2769aa02007-12-27 18:26:09 +01007607maxconn <conns>
7608 Fix the maximum number of concurrent connections on a frontend
7609 May be used in sections : defaults | frontend | listen | backend
7610 yes | yes | yes | no
7611 Arguments :
7612 <conns> is the maximum number of concurrent connections the frontend will
7613 accept to serve. Excess connections will be queued by the system
7614 in the socket's listen queue and will be served once a connection
7615 closes.
7616
7617 If the system supports it, it can be useful on big sites to raise this limit
7618 very high so that haproxy manages connection queues, instead of leaving the
7619 clients with unanswered connection attempts. This value should not exceed the
7620 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01007621 of tune.bufsize (16kB by default) each, as well as some other data resulting
7622 in about 33 kB of RAM being consumed per established connection. That means
7623 that a medium system equipped with 1GB of RAM can withstand around
7624 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007625
7626 Also, when <conns> is set to large values, it is possible that the servers
7627 are not sized to accept such loads, and for this reason it is generally wise
7628 to assign them some reasonable connection limits.
7629
Willy Tarreauc8d5b952019-02-27 17:25:52 +01007630 When this value is set to zero, which is the default, the global "maxconn"
7631 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02007632
Willy Tarreau2769aa02007-12-27 18:26:09 +01007633 See also : "server", global section's "maxconn", "fullconn"
7634
7635
Willy Tarreau77e0dae2020-10-14 15:44:27 +02007636mode { tcp|http }
Willy Tarreau2769aa02007-12-27 18:26:09 +01007637 Set the running mode or protocol of the instance
7638 May be used in sections : defaults | frontend | listen | backend
7639 yes | yes | yes | yes
7640 Arguments :
7641 tcp The instance will work in pure TCP mode. A full-duplex connection
7642 will be established between clients and servers, and no layer 7
7643 examination will be performed. This is the default mode. It
7644 should be used for SSL, SSH, SMTP, ...
7645
7646 http The instance will work in HTTP mode. The client request will be
7647 analyzed in depth before connecting to any server. Any request
7648 which is not RFC-compliant will be rejected. Layer 7 filtering,
7649 processing and switching will be possible. This is the mode which
7650 brings HAProxy most of its value.
7651
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007652 When doing content switching, it is mandatory that the frontend and the
7653 backend are in the same mode (generally HTTP), otherwise the configuration
7654 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007655
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007656 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007657 defaults http_instances
7658 mode http
7659
Willy Tarreau0ba27502007-12-24 16:55:16 +01007660
Cyril Bontéf0c60612010-02-06 14:44:47 +01007661monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01007662 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007663 May be used in sections : defaults | frontend | listen | backend
7664 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01007665 Arguments :
7666 if <cond> the monitor request will fail if the condition is satisfied,
7667 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007668 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01007669 are met, for instance a low number of servers both in a
7670 backend and its backup.
7671
7672 unless <cond> the monitor request will succeed only if the condition is
7673 satisfied, and will fail otherwise. Such a condition may be
7674 based on a test on the presence of a minimum number of active
7675 servers in a list of backends.
7676
7677 This statement adds a condition which can force the response to a monitor
7678 request to report a failure. By default, when an external component queries
7679 the URI dedicated to monitoring, a 200 response is returned. When one of the
7680 conditions above is met, haproxy will return 503 instead of 200. This is
7681 very useful to report a site failure to an external component which may base
7682 routing advertisements between multiple sites on the availability reported by
7683 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02007684 criterion. Note that "monitor fail" only works in HTTP mode. Both status
7685 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007686
7687 Example:
7688 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01007689 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01007690 acl site_dead nbsrv(dynamic) lt 2
7691 acl site_dead nbsrv(static) lt 2
7692 monitor-uri /site_alive
7693 monitor fail if site_dead
7694
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007695 See also : "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007696
7697
Willy Tarreau2769aa02007-12-27 18:26:09 +01007698monitor-uri <uri>
7699 Intercept a URI used by external components' monitor requests
7700 May be used in sections : defaults | frontend | listen | backend
7701 yes | yes | yes | no
7702 Arguments :
7703 <uri> is the exact URI which we want to intercept to return HAProxy's
7704 health status instead of forwarding the request.
7705
7706 When an HTTP request referencing <uri> will be received on a frontend,
7707 HAProxy will not forward it nor log it, but instead will return either
7708 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
7709 conditions defined with "monitor fail". This is normally enough for any
7710 front-end HTTP probe to detect that the service is UP and running without
7711 forwarding the request to a backend server. Note that the HTTP method, the
7712 version and all headers are ignored, but the request must at least be valid
7713 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
7714
Willy Tarreau721d8e02017-12-01 18:25:08 +01007715 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007716 and even before any "http-request". The only rulesets applied before are the
7717 tcp-request ones. They cannot be logged either, and it is the intended
7718 purpose. They are only used to report HAProxy's health to an upper component,
7719 nothing more. However, it is possible to add any number of conditions using
7720 "monitor fail" and ACLs so that the result can be adjusted to whatever check
7721 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007722
Christopher Faulet6072beb2020-02-18 15:34:58 +01007723 Note: if <uri> starts by a slash ('/'), the matching is performed against the
7724 request's path instead of the request's uri. It is a workaround to let
7725 the HTTP/2 requests match the monitor-uri. Indeed, in HTTP/2, clients
7726 are encouraged to send absolute URIs only.
7727
Willy Tarreau2769aa02007-12-27 18:26:09 +01007728 Example :
7729 # Use /haproxy_test to report haproxy's status
7730 frontend www
7731 mode http
7732 monitor-uri /haproxy_test
7733
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007734 See also : "monitor fail"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007735
Willy Tarreau0ba27502007-12-24 16:55:16 +01007736
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007737option abortonclose
7738no option abortonclose
7739 Enable or disable early dropping of aborted requests pending in queues.
7740 May be used in sections : defaults | frontend | listen | backend
7741 yes | no | yes | yes
7742 Arguments : none
7743
7744 In presence of very high loads, the servers will take some time to respond.
7745 The per-instance connection queue will inflate, and the response time will
7746 increase respective to the size of the queue times the average per-session
7747 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01007748 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007749 the queue, and slowing down other users, and the servers as well, because the
7750 request will eventually be served, then aborted at the first error
7751 encountered while delivering the response.
7752
7753 As there is no way to distinguish between a full STOP and a simple output
7754 close on the client side, HTTP agents should be conservative and consider
7755 that the client might only have closed its output channel while waiting for
7756 the response. However, this introduces risks of congestion when lots of users
7757 do the same, and is completely useless nowadays because probably no client at
7758 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01007759 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007760 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01007761 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007762 of being the single component to break rare but valid traffic is extremely
7763 low, which adds to the temptation to be able to abort a session early while
7764 still not served and not pollute the servers.
7765
Davor Ocelice9ed2812017-12-25 17:49:28 +01007766 In HAProxy, the user can choose the desired behavior using the option
7767 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007768 compliant and aborted requests will be served. But when the option is
7769 specified, a session with an incoming channel closed will be aborted while
7770 it is still possible, either pending in the queue for a connection slot, or
7771 during the connection establishment if the server has not yet acknowledged
7772 the connection request. This considerably reduces the queue size and the load
7773 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01007774 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007775
7776 If this option has been enabled in a "defaults" section, it can be disabled
7777 in a specific instance by prepending the "no" keyword before it.
7778
7779 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
7780
7781
Willy Tarreau4076a152009-04-02 15:18:36 +02007782option accept-invalid-http-request
7783no option accept-invalid-http-request
7784 Enable or disable relaxing of HTTP request parsing
7785 May be used in sections : defaults | frontend | listen | backend
7786 yes | yes | yes | no
7787 Arguments : none
7788
Willy Tarreau91852eb2015-05-01 13:26:00 +02007789 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007790 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007791 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007792 forbidden characters are essentially used to build attacks exploiting server
7793 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7794 server will emit invalid header names for whatever reason (configuration,
7795 implementation) and the issue will not be immediately fixed. In such a case,
7796 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01007797 even if that does not make sense, by specifying this option. Similarly, the
7798 list of characters allowed to appear in a URI is well defined by RFC3986, and
7799 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
7800 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01007801 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02007802 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02007803 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
7804 to pass through (no version specified) and multiple digits for both the major
7805 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007806
7807 This option should never be enabled by default as it hides application bugs
7808 and open security breaches. It should only be deployed after a problem has
7809 been confirmed.
7810
7811 When this option is enabled, erroneous header names will still be accepted in
7812 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01007813 analysis using the "show errors" request on the UNIX stats socket. Similarly,
7814 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02007815 also helps confirming that the issue has been solved.
7816
7817 If this option has been enabled in a "defaults" section, it can be disabled
7818 in a specific instance by prepending the "no" keyword before it.
7819
7820 See also : "option accept-invalid-http-response" and "show errors" on the
7821 stats socket.
7822
7823
7824option accept-invalid-http-response
7825no option accept-invalid-http-response
7826 Enable or disable relaxing of HTTP response parsing
7827 May be used in sections : defaults | frontend | listen | backend
7828 yes | no | yes | yes
7829 Arguments : none
7830
Willy Tarreau91852eb2015-05-01 13:26:00 +02007831 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007832 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007833 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007834 forbidden characters are essentially used to build attacks exploiting server
7835 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7836 server will emit invalid header names for whatever reason (configuration,
7837 implementation) and the issue will not be immediately fixed. In such a case,
7838 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02007839 even if that does not make sense, by specifying this option. This option also
7840 relaxes the test on the HTTP version format, it allows multiple digits for
7841 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007842
7843 This option should never be enabled by default as it hides application bugs
7844 and open security breaches. It should only be deployed after a problem has
7845 been confirmed.
7846
7847 When this option is enabled, erroneous header names will still be accepted in
7848 responses, but the complete response will be captured in order to permit
7849 later analysis using the "show errors" request on the UNIX stats socket.
7850 Doing this also helps confirming that the issue has been solved.
7851
7852 If this option has been enabled in a "defaults" section, it can be disabled
7853 in a specific instance by prepending the "no" keyword before it.
7854
7855 See also : "option accept-invalid-http-request" and "show errors" on the
7856 stats socket.
7857
7858
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007859option allbackups
7860no option allbackups
7861 Use either all backup servers at a time or only the first one
7862 May be used in sections : defaults | frontend | listen | backend
7863 yes | no | yes | yes
7864 Arguments : none
7865
7866 By default, the first operational backup server gets all traffic when normal
7867 servers are all down. Sometimes, it may be preferred to use multiple backups
7868 at once, because one will not be enough. When "option allbackups" is enabled,
7869 the load balancing will be performed among all backup servers when all normal
7870 ones are unavailable. The same load balancing algorithm will be used and the
7871 servers' weights will be respected. Thus, there will not be any priority
7872 order between the backup servers anymore.
7873
7874 This option is mostly used with static server farms dedicated to return a
7875 "sorry" page when an application is completely offline.
7876
7877 If this option has been enabled in a "defaults" section, it can be disabled
7878 in a specific instance by prepending the "no" keyword before it.
7879
7880
7881option checkcache
7882no option checkcache
Godbach7056a352013-12-11 20:01:07 +08007883 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007884 May be used in sections : defaults | frontend | listen | backend
7885 yes | no | yes | yes
7886 Arguments : none
7887
7888 Some high-level frameworks set application cookies everywhere and do not
7889 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007890 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007891 high risk of session crossing or stealing between users traversing the same
7892 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007893 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007894
7895 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007896 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01007897 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007898 response to check if there's a risk of caching a cookie on a client-side
7899 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01007900 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007901 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01007902 - all those with a return code other than 200, 203, 204, 206, 300, 301,
7903 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01007904 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01007905 - all those that result from a request using a method other than GET, HEAD,
7906 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01007907 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007908 - those with a 'Pragma: no-cache' header
7909 - those with a 'Cache-control: private' header
7910 - those with a 'Cache-control: no-store' header
7911 - those with a 'Cache-control: max-age=0' header
7912 - those with a 'Cache-control: s-maxage=0' header
7913 - those with a 'Cache-control: no-cache' header
7914 - those with a 'Cache-control: no-cache="set-cookie"' header
7915 - those with a 'Cache-control: no-cache="set-cookie,' header
7916 (allowing other fields after set-cookie)
7917
7918 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007919 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
7920 gateway". The session state shows "PH--" meaning that the proxy blocked the
7921 response during headers processing. Additionally, an alert will be sent in
7922 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007923
7924 Due to the high impact on the application, the application should be tested
7925 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007926 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01007927 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007928
7929 If this option has been enabled in a "defaults" section, it can be disabled
7930 in a specific instance by prepending the "no" keyword before it.
7931
7932
7933option clitcpka
7934no option clitcpka
7935 Enable or disable the sending of TCP keepalive packets on the client side
7936 May be used in sections : defaults | frontend | listen | backend
7937 yes | yes | yes | no
7938 Arguments : none
7939
7940 When there is a firewall or any session-aware component between a client and
7941 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007942 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007943 components decides to expire a session which has remained idle for too long.
7944
7945 Enabling socket-level TCP keep-alives makes the system regularly send packets
7946 to the other end of the connection, leaving it active. The delay between
7947 keep-alive probes is controlled by the system only and depends both on the
7948 operating system and its tuning parameters.
7949
7950 It is important to understand that keep-alive packets are neither emitted nor
7951 received at the application level. It is only the network stacks which sees
7952 them. For this reason, even if one side of the proxy already uses keep-alives
7953 to maintain its connection alive, those keep-alive packets will not be
7954 forwarded to the other side of the proxy.
7955
7956 Please note that this has nothing to do with HTTP keep-alive.
7957
7958 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
7959 client side of a connection, which should help when session expirations are
7960 noticed between HAProxy and a client.
7961
7962 If this option has been enabled in a "defaults" section, it can be disabled
7963 in a specific instance by prepending the "no" keyword before it.
7964
7965 See also : "option srvtcpka", "option tcpka"
7966
7967
Willy Tarreau0ba27502007-12-24 16:55:16 +01007968option contstats
7969 Enable continuous traffic statistics updates
7970 May be used in sections : defaults | frontend | listen | backend
7971 yes | yes | yes | no
7972 Arguments : none
7973
7974 By default, counters used for statistics calculation are incremented
7975 only when a session finishes. It works quite well when serving small
7976 objects, but with big ones (for example large images or archives) or
7977 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01007978 a hedgehog. With this option enabled counters get incremented frequently
7979 along the session, typically every 5 seconds, which is often enough to
7980 produce clean graphs. Recounting touches a hotpath directly so it is not
7981 not enabled by default, as it can cause a lot of wakeups for very large
7982 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007983
Christopher Faulet89aed322020-06-02 17:33:56 +02007984option disable-h2-upgrade
7985no option disable-h2-upgrade
7986 Enable or disable the implicit HTTP/2 upgrade from an HTTP/1.x client
7987 connection.
7988 May be used in sections : defaults | frontend | listen | backend
7989 yes | yes | yes | no
7990 Arguments : none
7991
7992 By default, HAProxy is able to implicitly upgrade an HTTP/1.x client
7993 connection to an HTTP/2 connection if the first request it receives from a
7994 given HTTP connection matches the HTTP/2 connection preface (i.e. the string
7995 "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 +01007996 HTTP/1.x and HTTP/2 clients on a non-SSL connections. This option must be
7997 used to disable the implicit upgrade. Note this implicit upgrade is only
7998 supported for HTTP proxies, thus this option too. Note also it is possible to
7999 force the HTTP/2 on clear connections by specifying "proto h2" on the bind
8000 line. Finally, this option is applied on all bind lines. To disable implicit
8001 HTTP/2 upgrades for a specific bind line, it is possible to use "proto h1".
Christopher Faulet89aed322020-06-02 17:33:56 +02008002
8003 If this option has been enabled in a "defaults" section, it can be disabled
8004 in a specific instance by prepending the "no" keyword before it.
Willy Tarreau0ba27502007-12-24 16:55:16 +01008005
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008006option dontlog-normal
8007no option dontlog-normal
8008 Enable or disable logging of normal, successful connections
8009 May be used in sections : defaults | frontend | listen | backend
8010 yes | yes | yes | no
8011 Arguments : none
8012
8013 There are large sites dealing with several thousand connections per second
8014 and for which logging is a major pain. Some of them are even forced to turn
8015 logs off and cannot debug production issues. Setting this option ensures that
8016 normal connections, those which experience no error, no timeout, no retry nor
8017 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
8018 mode, the response status code is checked and return codes 5xx will still be
8019 logged.
8020
8021 It is strongly discouraged to use this option as most of the time, the key to
8022 complex issues is in the normal logs which will not be logged here. If you
8023 need to separate logs, see the "log-separate-errors" option instead.
8024
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008025 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008026 logging.
8027
8028
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008029option dontlognull
8030no option dontlognull
8031 Enable or disable logging of null connections
8032 May be used in sections : defaults | frontend | listen | backend
8033 yes | yes | yes | no
8034 Arguments : none
8035
8036 In certain environments, there are components which will regularly connect to
8037 various systems to ensure that they are still alive. It can be the case from
8038 another load balancer as well as from monitoring systems. By default, even a
8039 simple port probe or scan will produce a log. If those connections pollute
8040 the logs too much, it is possible to enable option "dontlognull" to indicate
8041 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02008042 which typically corresponds to those probes. Note that errors will still be
8043 returned to the client and accounted for in the stats. If this is not what is
8044 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008045
8046 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01008047 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008048 would not be logged.
8049
8050 If this option has been enabled in a "defaults" section, it can be disabled
8051 in a specific instance by prepending the "no" keyword before it.
8052
Willy Tarreau9e9919d2020-10-14 15:55:23 +02008053 See also : "log", "http-ignore-probes", "monitor-uri", and
Willy Tarreau0f228a02015-05-01 15:37:53 +02008054 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008055
Willy Tarreaubf1f8162007-12-28 17:42:56 +01008056
Willy Tarreau87cf5142011-08-19 22:57:24 +02008057option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01008058 Enable insertion of the X-Forwarded-For header to requests sent to servers
8059 May be used in sections : defaults | frontend | listen | backend
8060 yes | yes | yes | yes
8061 Arguments :
8062 <network> is an optional argument used to disable this option for sources
8063 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02008064 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01008065 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008066
8067 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
8068 their client address. This is sometimes annoying when the client's IP address
8069 is expected in server logs. To solve this problem, the well-known HTTP header
8070 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
8071 This header contains a value representing the client's IP address. Since this
8072 header is always appended at the end of the existing header list, the server
8073 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02008074 the server's manual to find how to enable use of this standard header. Note
8075 that only the last occurrence of the header must be used, since it is really
8076 possible that the client has already brought one.
8077
Willy Tarreaud72758d2010-01-12 10:42:19 +01008078 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02008079 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01008080 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01008081 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008082 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02008083 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01008084
8085 Sometimes, a same HAProxy instance may be shared between a direct client
8086 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
8087 used to decrypt HTTPS traffic). It is possible to disable the addition of the
8088 header for a known source address or network by adding the "except" keyword
8089 followed by the network address. In this case, any source IP matching the
8090 network will not cause an addition of this header. Most common uses are with
Christopher Faulet5d1def62021-02-26 09:19:15 +01008091 private networks or 127.0.0.1. IPv4 and IPv6 are both supported.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008092
Willy Tarreau87cf5142011-08-19 22:57:24 +02008093 Alternatively, the keyword "if-none" states that the header will only be
8094 added if it is not present. This should only be used in perfectly trusted
8095 environment, as this might cause a security issue if headers reaching haproxy
8096 are under the control of the end-user.
8097
Willy Tarreauc27debf2008-01-06 08:57:02 +01008098 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02008099 least one of them uses it, the header will be added. Note that the backend's
8100 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02008101 both are defined. In the case of the "if-none" argument, if at least one of
8102 the frontend or the backend does not specify it, it wants the addition to be
8103 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008104
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02008105 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01008106 # Public HTTP address also used by stunnel on the same machine
8107 frontend www
8108 mode http
8109 option forwardfor except 127.0.0.1 # stunnel already adds the header
8110
Ross Westaf72a1d2008-08-03 10:51:45 +02008111 # Those servers want the IP Address in X-Client
8112 backend www
8113 mode http
8114 option forwardfor header X-Client
8115
Willy Tarreau87cf5142011-08-19 22:57:24 +02008116 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02008117 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01008118
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008119
Christopher Faulet98fbe952019-07-22 16:18:24 +02008120option h1-case-adjust-bogus-client
8121no option h1-case-adjust-bogus-client
8122 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
8123 May be used in sections : defaults | frontend | listen | backend
8124 yes | yes | yes | no
8125 Arguments : none
8126
8127 There is no standard case for header names because, as stated in RFC7230,
8128 they are case-insensitive. So applications must handle them in a case-
8129 insensitive manner. But some bogus applications violate the standards and
8130 erroneously rely on the cases most commonly used by browsers. This problem
8131 becomes critical with HTTP/2 because all header names must be exchanged in
8132 lower case, and HAProxy follows the same convention. All header names are
8133 sent in lower case to clients and servers, regardless of the HTTP version.
8134
8135 When HAProxy receives an HTTP/1 response, its header names are converted to
8136 lower case and manipulated and sent this way to the clients. If a client is
8137 known to violate the HTTP standards and to fail to process a response coming
8138 from HAProxy, it is possible to transform the lower case header names to a
8139 different format when the response is formatted and sent to the client, by
8140 enabling this option and specifying the list of headers to be reformatted
8141 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
8142 must only be a temporary workaround for the time it takes the client to be
8143 fixed, because clients which require such workarounds might be vulnerable to
8144 content smuggling attacks and must absolutely be fixed.
8145
8146 Please note that this option will not affect standards-compliant clients.
8147
8148 If this option has been enabled in a "defaults" section, it can be disabled
8149 in a specific instance by prepending the "no" keyword before it.
8150
8151 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
8152 "h1-case-adjust-file".
8153
8154
8155option h1-case-adjust-bogus-server
8156no option h1-case-adjust-bogus-server
8157 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
8158 May be used in sections : defaults | frontend | listen | backend
8159 yes | no | yes | yes
8160 Arguments : none
8161
8162 There is no standard case for header names because, as stated in RFC7230,
8163 they are case-insensitive. So applications must handle them in a case-
8164 insensitive manner. But some bogus applications violate the standards and
8165 erroneously rely on the cases most commonly used by browsers. This problem
8166 becomes critical with HTTP/2 because all header names must be exchanged in
8167 lower case, and HAProxy follows the same convention. All header names are
8168 sent in lower case to clients and servers, regardless of the HTTP version.
8169
8170 When HAProxy receives an HTTP/1 request, its header names are converted to
8171 lower case and manipulated and sent this way to the servers. If a server is
8172 known to violate the HTTP standards and to fail to process a request coming
8173 from HAProxy, it is possible to transform the lower case header names to a
8174 different format when the request is formatted and sent to the server, by
8175 enabling this option and specifying the list of headers to be reformatted
8176 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
8177 must only be a temporary workaround for the time it takes the server to be
8178 fixed, because servers which require such workarounds might be vulnerable to
8179 content smuggling attacks and must absolutely be fixed.
8180
8181 Please note that this option will not affect standards-compliant servers.
8182
8183 If this option has been enabled in a "defaults" section, it can be disabled
8184 in a specific instance by prepending the "no" keyword before it.
8185
8186 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
8187 "h1-case-adjust-file".
8188
8189
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008190option http-buffer-request
8191no option http-buffer-request
8192 Enable or disable waiting for whole HTTP request body before proceeding
8193 May be used in sections : defaults | frontend | listen | backend
8194 yes | yes | yes | yes
8195 Arguments : none
8196
8197 It is sometimes desirable to wait for the body of an HTTP request before
8198 taking a decision. This is what is being done by "balance url_param" for
8199 example. The first use case is to buffer requests from slow clients before
8200 connecting to the server. Another use case consists in taking the routing
8201 decision based on the request body's contents. This option placed in a
8202 frontend or backend forces the HTTP processing to wait until either the whole
Christopher Faulet6db8a2e2019-11-19 16:27:25 +01008203 body is received or the request buffer is full. It can have undesired side
8204 effects with some applications abusing HTTP by expecting unbuffered
8205 transmissions between the frontend and the backend, so this should definitely
8206 not be used by default.
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008207
Christopher Faulet021a8e42021-03-29 10:46:38 +02008208 See also : "option http-no-delay", "timeout http-request",
8209 "http-request wait-for-body"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008210
8211
Willy Tarreau0f228a02015-05-01 15:37:53 +02008212option http-ignore-probes
8213no option http-ignore-probes
8214 Enable or disable logging of null connections and request timeouts
8215 May be used in sections : defaults | frontend | listen | backend
8216 yes | yes | yes | no
8217 Arguments : none
8218
8219 Recently some browsers started to implement a "pre-connect" feature
8220 consisting in speculatively connecting to some recently visited web sites
8221 just in case the user would like to visit them. This results in many
8222 connections being established to web sites, which end up in 408 Request
8223 Timeout if the timeout strikes first, or 400 Bad Request when the browser
8224 decides to close them first. These ones pollute the log and feed the error
8225 counters. There was already "option dontlognull" but it's insufficient in
8226 this case. Instead, this option does the following things :
8227 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01008228 was received over a connection before it was closed;
8229 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02008230 - prevent any error counter from being incremented
8231
8232 That way the empty connection is silently ignored. Note that it is better
8233 not to use this unless it is clear that it is needed, because it will hide
8234 real problems. The most common reason for not receiving a request and seeing
8235 a 408 is due to an MTU inconsistency between the client and an intermediary
8236 element such as a VPN, which blocks too large packets. These issues are
8237 generally seen with POST requests as well as GET with large cookies. The logs
8238 are often the only way to detect them.
8239
8240 If this option has been enabled in a "defaults" section, it can be disabled
8241 in a specific instance by prepending the "no" keyword before it.
8242
8243 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
8244
8245
Willy Tarreau16bfb022010-01-16 19:48:41 +01008246option http-keep-alive
8247no option http-keep-alive
8248 Enable or disable HTTP keep-alive from client to server
8249 May be used in sections : defaults | frontend | listen | backend
8250 yes | yes | yes | yes
8251 Arguments : none
8252
Willy Tarreau70dffda2014-01-30 03:07:23 +01008253 By default HAProxy operates in keep-alive mode with regards to persistent
8254 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02008255 leaves the connection idle on both sides between the end of a response and
8256 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008257 as "option http-server-close" or "option httpclose". This option allows to
8258 set back the keep-alive mode, which can be useful when another mode was used
8259 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01008260
8261 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
8262 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01008263 network) and the fastest session reuse on the server side at the expense
8264 of maintaining idle connections to the servers. In general, it is possible
8265 with this option to achieve approximately twice the request rate that the
8266 "http-server-close" option achieves on small objects. There are mainly two
8267 situations where this option may be useful :
8268
8269 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008270 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01008271
8272 - when the cost of establishing the connection to the server is significant
8273 compared to the cost of retrieving the associated object from the server.
8274
8275 This last case can happen when the server is a fast static server of cache.
8276 In this case, the server will need to be properly tuned to support high enough
8277 connection counts because connections will last until the client sends another
8278 request.
8279
8280 If the client request has to go to another backend or another server due to
8281 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01008282 immediately be closed and a new one re-opened. Option "prefer-last-server" is
8283 available to try optimize server selection so that if the server currently
8284 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01008285
Willy Tarreau16bfb022010-01-16 19:48:41 +01008286 At the moment, logs will not indicate whether requests came from the same
8287 session or not. The accept date reported in the logs corresponds to the end
8288 of the previous request, and the request time corresponds to the time spent
8289 waiting for a new request. The keep-alive request time is still bound to the
8290 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
8291 not set.
8292
Christopher Faulet159e6672019-07-16 15:09:52 +02008293 This option disables and replaces any previous "option httpclose" or "option
8294 http-server-close". When backend and frontend options differ, all of these 4
8295 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01008296
Christopher Faulet315b39c2018-09-21 16:26:19 +02008297 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01008298 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01008299 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01008300
8301
Willy Tarreau96e31212011-05-30 18:10:30 +02008302option http-no-delay
8303no option http-no-delay
8304 Instruct the system to favor low interactive delays over performance in HTTP
8305 May be used in sections : defaults | frontend | listen | backend
8306 yes | yes | yes | yes
8307 Arguments : none
8308
8309 In HTTP, each payload is unidirectional and has no notion of interactivity.
8310 Any agent is expected to queue data somewhat for a reasonably low delay.
8311 There are some very rare server-to-server applications that abuse the HTTP
8312 protocol and expect the payload phase to be highly interactive, with many
8313 interleaved data chunks in both directions within a single request. This is
8314 absolutely not supported by the HTTP specification and will not work across
8315 most proxies or servers. When such applications attempt to do this through
8316 haproxy, it works but they will experience high delays due to the network
8317 optimizations which favor performance by instructing the system to wait for
8318 enough data to be available in order to only send full packets. Typical
8319 delays are around 200 ms per round trip. Note that this only happens with
8320 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
8321 affected.
8322
8323 When "option http-no-delay" is present in either the frontend or the backend
8324 used by a connection, all such optimizations will be disabled in order to
8325 make the exchanges as fast as possible. Of course this offers no guarantee on
8326 the functionality, as it may break at any other place. But if it works via
8327 HAProxy, it will work as fast as possible. This option should never be used
8328 by default, and should never be used at all unless such a buggy application
8329 is discovered. The impact of using this option is an increase of bandwidth
8330 usage and CPU usage, which may significantly lower performance in high
8331 latency environments.
8332
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008333 See also : "option http-buffer-request"
8334
Willy Tarreau96e31212011-05-30 18:10:30 +02008335
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008336option http-pretend-keepalive
8337no option http-pretend-keepalive
8338 Define whether haproxy will announce keepalive to the server or not
8339 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02008340 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008341 Arguments : none
8342
Christopher Faulet315b39c2018-09-21 16:26:19 +02008343 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008344 adds a "Connection: close" header to the request forwarded to the server.
8345 Unfortunately, when some servers see this header, they automatically refrain
8346 from using the chunked encoding for responses of unknown length, while this
8347 is totally unrelated. The immediate effect is that this prevents haproxy from
8348 maintaining the client connection alive. A second effect is that a client or
8349 a cache could receive an incomplete response without being aware of it, and
8350 consider the response complete.
8351
8352 By setting "option http-pretend-keepalive", haproxy will make the server
8353 believe it will keep the connection alive. The server will then not fall back
8354 to the abnormal undesired above. When haproxy gets the whole response, it
8355 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02008356 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008357 connection is correctly closed on the server side.
8358
8359 It is recommended not to enable this option by default, because most servers
8360 will more efficiently close the connection themselves after the last packet,
8361 and release its buffers slightly earlier. Also, the added packet on the
8362 network could slightly reduce the overall peak performance. However it is
8363 worth noting that when this option is enabled, haproxy will have slightly
8364 less work to do. So if haproxy is the bottleneck on the whole architecture,
8365 enabling this option might save a few CPU cycles.
8366
Christopher Faulet98db9762018-09-21 10:25:19 +02008367 This option may be set in backend and listen sections. Using it in a frontend
8368 section will be ignored and a warning will be reported during startup. It is
8369 a backend related option, so there is no real reason to set it on a
8370 frontend. This option may be combined with "option httpclose", which will
8371 cause keepalive to be announced to the server and close to be announced to
8372 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008373
8374 If this option has been enabled in a "defaults" section, it can be disabled
8375 in a specific instance by prepending the "no" keyword before it.
8376
Christopher Faulet315b39c2018-09-21 16:26:19 +02008377 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01008378 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008379
Willy Tarreauc27debf2008-01-06 08:57:02 +01008380
Willy Tarreaub608feb2010-01-02 22:47:18 +01008381option http-server-close
8382no option http-server-close
8383 Enable or disable HTTP connection closing on the server side
8384 May be used in sections : defaults | frontend | listen | backend
8385 yes | yes | yes | yes
8386 Arguments : none
8387
Willy Tarreau70dffda2014-01-30 03:07:23 +01008388 By default HAProxy operates in keep-alive mode with regards to persistent
8389 connections: for each connection it processes each request and response, and
8390 leaves the connection idle on both sides between the end of a response and
8391 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008392 as "option http-server-close" or "option httpclose". Setting "option
8393 http-server-close" enables HTTP connection-close mode on the server side
8394 while keeping the ability to support HTTP keep-alive and pipelining on the
8395 client side. This provides the lowest latency on the client side (slow
8396 network) and the fastest session reuse on the server side to save server
8397 resources, similarly to "option httpclose". It also permits non-keepalive
8398 capable servers to be served in keep-alive mode to the clients if they
8399 conform to the requirements of RFC7230. Please note that some servers do not
8400 always conform to those requirements when they see "Connection: close" in the
8401 request. The effect will be that keep-alive will never be used. A workaround
8402 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01008403
8404 At the moment, logs will not indicate whether requests came from the same
8405 session or not. The accept date reported in the logs corresponds to the end
8406 of the previous request, and the request time corresponds to the time spent
8407 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01008408 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
8409 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01008410
8411 This option may be set both in a frontend and in a backend. It is enabled if
8412 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008413 It disables and replaces any previous "option httpclose" or "option
8414 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8415 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01008416
8417 If this option has been enabled in a "defaults" section, it can be disabled
8418 in a specific instance by prepending the "no" keyword before it.
8419
Christopher Faulet315b39c2018-09-21 16:26:19 +02008420 See also : "option httpclose", "option http-pretend-keepalive",
8421 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01008422
Willy Tarreau88d349d2010-01-25 12:15:43 +01008423option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01008424no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01008425 Make use of non-standard Proxy-Connection header instead of Connection
8426 May be used in sections : defaults | frontend | listen | backend
8427 yes | yes | yes | no
8428 Arguments : none
8429
Lukas Tribus23953682017-04-28 13:24:30 +00008430 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01008431 Connection header to indicate their wish of persistent or non-persistent
8432 connections, both browsers and proxies ignore this header for proxied
8433 connections and make use of the undocumented, non-standard Proxy-Connection
8434 header instead. The issue begins when trying to put a load balancer between
8435 browsers and such proxies, because there will be a difference between what
8436 haproxy understands and what the client and the proxy agree on.
8437
8438 By setting this option in a frontend, haproxy can automatically switch to use
8439 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008440 defined here as one where the URI begins with neither a '/' nor a '*'. This
8441 is incompatible with the HTTP tunnel mode. Note that this option can only be
8442 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01008443
Willy Tarreau844a7e72010-01-31 21:46:18 +01008444 Also, when this option is set, a request which requires authentication will
8445 automatically switch to use proxy authentication headers if it is itself a
8446 proxied request. That makes it possible to check or enforce authentication in
8447 front of an existing proxy.
8448
Willy Tarreau88d349d2010-01-25 12:15:43 +01008449 This option should normally never be used, except in front of a proxy.
8450
Christopher Faulet315b39c2018-09-21 16:26:19 +02008451 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01008452
Willy Tarreaud63335a2010-02-26 12:56:52 +01008453option httpchk
8454option httpchk <uri>
8455option httpchk <method> <uri>
8456option httpchk <method> <uri> <version>
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008457 Enables HTTP protocol to check on the servers health
Willy Tarreaud63335a2010-02-26 12:56:52 +01008458 May be used in sections : defaults | frontend | listen | backend
8459 yes | no | yes | yes
8460 Arguments :
8461 <method> is the optional HTTP method used with the requests. When not set,
8462 the "OPTIONS" method is used, as it generally requires low server
8463 processing and is easy to filter out from the logs. Any method
8464 may be used, though it is not recommended to invent non-standard
8465 ones.
8466
8467 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
8468 which is accessible by default on almost any server, but may be
8469 changed to any other URI. Query strings are permitted.
8470
8471 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
8472 but some servers might behave incorrectly in HTTP 1.0, so turning
8473 it to HTTP/1.1 may sometimes help. Note that the Host field is
Christopher Faulet8acb1282020-04-09 08:44:06 +02008474 mandatory in HTTP/1.1, use "http-check send" directive to add it.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008475
8476 By default, server health checks only consist in trying to establish a TCP
8477 connection. When "option httpchk" is specified, a complete HTTP request is
8478 sent once the TCP connection is established, and responses 2xx and 3xx are
8479 considered valid, while all other ones indicate a server failure, including
8480 the lack of any response.
8481
Christopher Faulete5870d82020-04-15 11:32:03 +02008482 Combined with "http-check" directives, it is possible to customize the
8483 request sent during the HTTP health checks or the matching rules on the
8484 response. It is also possible to configure a send/expect sequence, just like
8485 with the directive "tcp-check" for TCP health checks.
8486
8487 The server configuration is used by default to open connections to perform
8488 HTTP health checks. By it is also possible to overwrite server parameters
8489 using "http-check connect" rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008490
Christopher Faulete5870d82020-04-15 11:32:03 +02008491 "httpchk" option does not necessarily require an HTTP backend, it also works
8492 with plain TCP backends. This is particularly useful to check simple scripts
Christopher Faulet14cd3162020-04-16 14:50:06 +02008493 bound to some dedicated ports using the inetd daemon. However, it will always
Daniel Corbett67a82712020-07-06 23:01:19 -04008494 internally relies on an HTX multiplexer. Thus, it means the request
Christopher Faulet14cd3162020-04-16 14:50:06 +02008495 formatting and the response parsing will be strict.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008496
Christopher Faulet8acb1282020-04-09 08:44:06 +02008497 Note : For a while, there was no way to add headers or body in the request
8498 used for HTTP health checks. So a workaround was to hide it at the end
8499 of the version string with a "\r\n" after the version. It is now
8500 deprecated. The directive "http-check send" must be used instead.
8501
Willy Tarreaud63335a2010-02-26 12:56:52 +01008502 Examples :
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008503 # Relay HTTPS traffic to Apache instance and check service availability
8504 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
8505 backend https_relay
8506 mode tcp
8507 option httpchk OPTIONS * HTTP/1.1
8508 http-check send hdr Host www
8509 server apache1 192.168.1.1:443 check port 80
Willy Tarreaud63335a2010-02-26 12:56:52 +01008510
Simon Hormanafc47ee2013-11-25 10:46:35 +09008511 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
8512 "option pgsql-check", "http-check" and the "check", "port" and
8513 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008514
8515
Willy Tarreauc27debf2008-01-06 08:57:02 +01008516option httpclose
8517no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02008518 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01008519 May be used in sections : defaults | frontend | listen | backend
8520 yes | yes | yes | yes
8521 Arguments : none
8522
Willy Tarreau70dffda2014-01-30 03:07:23 +01008523 By default HAProxy operates in keep-alive mode with regards to persistent
8524 connections: for each connection it processes each request and response, and
8525 leaves the connection idle on both sides between the end of a response and
8526 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008527 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01008528
Christopher Faulet315b39c2018-09-21 16:26:19 +02008529 If "option httpclose" is set, HAProxy will close connections with the server
8530 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05008531 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02008532 and will add one if missing. Any "Connection" header different from "close"
8533 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008534
Christopher Faulet315b39c2018-09-21 16:26:19 +02008535 This option may also be combined with "option http-pretend-keepalive", which
8536 will disable sending of the "Connection: close" header, but will still cause
8537 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008538
8539 This option may be set both in a frontend and in a backend. It is enabled if
8540 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008541 It disables and replaces any previous "option http-server-close" or "option
8542 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8543 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008544
8545 If this option has been enabled in a "defaults" section, it can be disabled
8546 in a specific instance by prepending the "no" keyword before it.
8547
Christopher Faulet315b39c2018-09-21 16:26:19 +02008548 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01008549
8550
Emeric Brun3a058f32009-06-30 18:26:00 +02008551option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01008552 Enable logging of HTTP request, session state and timers
8553 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01008554 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02008555 Arguments :
8556 clf if the "clf" argument is added, then the output format will be
8557 the CLF format instead of HAProxy's default HTTP format. You can
8558 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01008559 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02008560 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008561
8562 By default, the log output format is very poor, as it only contains the
8563 source and destination addresses, and the instance name. By specifying
8564 "option httplog", each log line turns into a much richer format including,
8565 but not limited to, the HTTP request, the connection timers, the session
8566 status, the connections numbers, the captured headers and cookies, the
8567 frontend, backend and server name, and of course the source address and
8568 ports.
8569
PiBa-NLbd556bf2014-12-11 21:31:54 +01008570 Specifying only "option httplog" will automatically clear the 'clf' mode
8571 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02008572
Guillaume de Lafond29f45602017-03-31 19:52:15 +02008573 "option httplog" overrides any previous "log-format" directive.
8574
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008575 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008576
Willy Tarreau55165fe2009-05-10 12:02:55 +02008577
8578option http_proxy
8579no option http_proxy
8580 Enable or disable plain HTTP proxy mode
8581 May be used in sections : defaults | frontend | listen | backend
8582 yes | yes | yes | yes
8583 Arguments : none
8584
8585 It sometimes happens that people need a pure HTTP proxy which understands
8586 basic proxy requests without caching nor any fancy feature. In this case,
8587 it may be worth setting up an HAProxy instance with the "option http_proxy"
8588 set. In this mode, no server is declared, and the connection is forwarded to
8589 the IP address and port found in the URL after the "http://" scheme.
8590
8591 No host address resolution is performed, so this only works when pure IP
8592 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008593 it will probably be used only by experts who know they need exactly it. This
8594 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02008595
8596 If this option has been enabled in a "defaults" section, it can be disabled
8597 in a specific instance by prepending the "no" keyword before it.
8598
8599 Example :
8600 # this backend understands HTTP proxy requests and forwards them directly.
8601 backend direct_forward
8602 option httpclose
8603 option http_proxy
8604
8605 See also : "option httpclose"
8606
Willy Tarreau211ad242009-10-03 21:45:07 +02008607
Jamie Gloudon801a0a32012-08-25 00:18:33 -04008608option independent-streams
8609no option independent-streams
8610 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008611 May be used in sections : defaults | frontend | listen | backend
8612 yes | yes | yes | yes
8613 Arguments : none
8614
8615 By default, when data is sent over a socket, both the write timeout and the
8616 read timeout for that socket are refreshed, because we consider that there is
8617 activity on that socket, and we have no other means of guessing if we should
8618 receive data or not.
8619
Davor Ocelice9ed2812017-12-25 17:49:28 +01008620 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008621 exists a situation where it is desirable to disable it, and only refresh the
8622 read timeout if there are incoming data. This happens on sessions with large
8623 timeouts and low amounts of exchanged data such as telnet session. If the
8624 server suddenly disappears, the output data accumulates in the system's
8625 socket buffers, both timeouts are correctly refreshed, and there is no way
8626 to know the server does not receive them, so we don't timeout. However, when
8627 the underlying protocol always echoes sent data, it would be enough by itself
8628 to detect the issue using the read timeout. Note that this problem does not
8629 happen with more verbose protocols because data won't accumulate long in the
8630 socket buffers.
8631
8632 When this option is set on the frontend, it will disable read timeout updates
8633 on data sent to the client. There probably is little use of this case. When
8634 the option is set on the backend, it will disable read timeout updates on
8635 data sent to the server. Doing so will typically break large HTTP posts from
8636 slow lines, so use it with caution.
8637
Willy Tarreauce887fd2012-05-12 12:50:00 +02008638 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008639
8640
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02008641option ldap-check
8642 Use LDAPv3 health checks for server testing
8643 May be used in sections : defaults | frontend | listen | backend
8644 yes | no | yes | yes
8645 Arguments : none
8646
8647 It is possible to test that the server correctly talks LDAPv3 instead of just
8648 testing that it accepts the TCP connection. When this option is set, an
8649 LDAPv3 anonymous simple bind message is sent to the server, and the response
8650 is analyzed to find an LDAPv3 bind response message.
8651
8652 The server is considered valid only when the LDAP response contains success
8653 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
8654
8655 Logging of bind requests is server dependent see your documentation how to
8656 configure it.
8657
8658 Example :
8659 option ldap-check
8660
8661 See also : "option httpchk"
8662
8663
Simon Horman98637e52014-06-20 12:30:16 +09008664option external-check
8665 Use external processes for server health checks
8666 May be used in sections : defaults | frontend | listen | backend
8667 yes | no | yes | yes
8668
8669 It is possible to test the health of a server using an external command.
8670 This is achieved by running the executable set using "external-check
8671 command".
8672
8673 Requires the "external-check" global to be set.
8674
8675 See also : "external-check", "external-check command", "external-check path"
8676
8677
Willy Tarreau211ad242009-10-03 21:45:07 +02008678option log-health-checks
8679no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02008680 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02008681 May be used in sections : defaults | frontend | listen | backend
8682 yes | no | yes | yes
8683 Arguments : none
8684
Willy Tarreaubef1b322014-05-13 21:01:39 +02008685 By default, failed health check are logged if server is UP and successful
8686 health checks are logged if server is DOWN, so the amount of additional
8687 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02008688
Willy Tarreaubef1b322014-05-13 21:01:39 +02008689 When this option is enabled, any change of the health check status or to
8690 the server's health will be logged, so that it becomes possible to know
8691 that a server was failing occasional checks before crashing, or exactly when
8692 it failed to respond a valid HTTP status, then when the port started to
8693 reject connections, then when the server stopped responding at all.
8694
Davor Ocelice9ed2812017-12-25 17:49:28 +01008695 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02008696 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02008697
Willy Tarreaubef1b322014-05-13 21:01:39 +02008698 See also: "option httpchk", "option ldap-check", "option mysql-check",
8699 "option pgsql-check", "option redis-check", "option smtpchk",
8700 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02008701
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008702
8703option log-separate-errors
8704no option log-separate-errors
8705 Change log level for non-completely successful connections
8706 May be used in sections : defaults | frontend | listen | backend
8707 yes | yes | yes | no
8708 Arguments : none
8709
8710 Sometimes looking for errors in logs is not easy. This option makes haproxy
8711 raise the level of logs containing potentially interesting information such
8712 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
8713 level changes from "info" to "err". This makes it possible to log them
8714 separately to a different file with most syslog daemons. Be careful not to
8715 remove them from the original file, otherwise you would lose ordering which
8716 provides very important information.
8717
8718 Using this option, large sites dealing with several thousand connections per
8719 second may log normal traffic to a rotating buffer and only archive smaller
8720 error logs.
8721
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008722 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008723 logging.
8724
Willy Tarreauc27debf2008-01-06 08:57:02 +01008725
8726option logasap
8727no option logasap
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008728 Enable or disable early logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008729 May be used in sections : defaults | frontend | listen | backend
8730 yes | yes | yes | no
8731 Arguments : none
8732
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008733 By default, logs are emitted when all the log format variables and sample
8734 fetches used in the definition of the log-format string return a value, or
8735 when the session is terminated. This allows the built in log-format strings
8736 to account for the transfer time, or the number of bytes in log messages.
8737
8738 When handling long lived connections such as large file transfers or RDP,
8739 it may take a while for the request or connection to appear in the logs.
8740 Using "option logasap", the log message is created as soon as the server
8741 connection is established in mode tcp, or as soon as the server sends the
8742 complete headers in mode http. Missing information in the logs will be the
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008743 total number of bytes which will only indicate the amount of data transferred
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008744 before the message was created and the total time which will not take the
8745 remainder of the connection life or transfer time into account. For the case
8746 of HTTP, it is good practice to capture the Content-Length response header
8747 so that the logs at least indicate how many bytes are expected to be
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008748 transferred.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008749
Willy Tarreaucc6c8912009-02-22 10:53:55 +01008750 Examples :
8751 listen http_proxy 0.0.0.0:80
8752 mode http
8753 option httplog
8754 option logasap
8755 log 192.168.2.200 local3
8756
8757 >>> Feb 6 12:14:14 localhost \
8758 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
8759 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
8760 "GET /image.iso HTTP/1.0"
8761
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008762 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01008763 logging.
8764
8765
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008766option mysql-check [ user <username> [ { post-41 | pre-41 } ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008767 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008768 May be used in sections : defaults | frontend | listen | backend
8769 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008770 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008771 <username> This is the username which will be used when connecting to MySQL
8772 server.
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008773 post-41 Send post v4.1 client compatible checks (the default)
8774 pre-41 Send pre v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008775
8776 If you specify a username, the check consists of sending two MySQL packet,
8777 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01008778 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008779 Error packet. It is a basic but useful test which does not produce error nor
8780 aborted connect on the server. However, it requires adding an authorization
8781 in the MySQL table, like this :
8782
8783 USE mysql;
8784 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
8785 FLUSH PRIVILEGES;
8786
8787 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008788 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008789 Error packet, we don't send anything in this mode. It was reported that it
8790 can generate lockout if check is too frequent and/or if there is not enough
8791 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
8792 value as if a connection is established successfully within fewer than MySQL
8793 "max_connect_errors" attempts after a previous connection was interrupted,
8794 the error count for the host is cleared to zero. If HAProxy's server get
8795 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
8796
8797 Remember that this does not check database presence nor database consistency.
8798 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008799
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02008800 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008801
8802 Most often, an incoming MySQL server needs to see the client's IP address for
8803 various purposes, including IP privilege matching and connection logging.
8804 When possible, it is often wise to masquerade the client's IP address when
8805 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02008806 which requires the transparent proxy feature to be compiled in, and the MySQL
8807 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008808
8809 See also: "option httpchk"
8810
8811
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008812option nolinger
8813no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008814 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008815 May be used in sections: defaults | frontend | listen | backend
8816 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008817 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008818
Davor Ocelice9ed2812017-12-25 17:49:28 +01008819 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008820 physically disconnected), the session timeouts triggers and the session is
8821 closed. But it will remain in FIN_WAIT1 state for some time in the system,
8822 using some resources and possibly limiting the ability to establish newer
8823 connections.
8824
8825 When this happens, it is possible to activate "option nolinger" which forces
8826 the system to immediately remove any socket's pending data on close. Thus,
Willy Tarreau4a321032020-10-16 04:55:19 +02008827 a TCP RST is emitted, any pending data are truncated, and the session is
8828 instantly purged from the system's tables. The generally visible effect for
8829 a client is that responses are truncated if the close happens with a last
8830 block of data (e.g. on a redirect or error response). On the server side,
8831 it may help release the source ports immediately when forwarding a client
8832 aborts in tunnels. In both cases, TCP resets are emitted and given that
8833 the session is instantly destroyed, there will be no retransmit. On a lossy
8834 network this can increase problems, especially when there is a firewall on
8835 the lossy side, because the firewall might see and process the reset (hence
8836 purge its session) and block any further traffic for this session,, including
8837 retransmits from the other side. So if the other side doesn't receive it,
8838 it will never receive any RST again, and the firewall might log many blocked
8839 packets.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008840
Willy Tarreau4a321032020-10-16 04:55:19 +02008841 For all these reasons, it is strongly recommended NOT to use this option,
8842 unless absolutely needed as a last resort. In most situations, using the
8843 "client-fin" or "server-fin" timeouts achieves similar results with a more
8844 reliable behavior. On Linux it's also possible to use the "tcp-ut" bind or
8845 server setting.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008846
8847 This option may be used both on frontends and backends, depending on the side
8848 where it is required. Use it on the frontend for clients, and on the backend
Willy Tarreau4a321032020-10-16 04:55:19 +02008849 for servers. While this option is technically supported in "defaults"
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +05008850 sections, it must really not be used there as it risks to accidentally
Willy Tarreau4a321032020-10-16 04:55:19 +02008851 propagate to sections that must no use it and to cause problems there.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008852
8853 If this option has been enabled in a "defaults" section, it can be disabled
8854 in a specific instance by prepending the "no" keyword before it.
8855
Willy Tarreau4a321032020-10-16 04:55:19 +02008856 See also: "timeout client-fin", "timeout server-fin", "tcp-ut" bind or server
8857 keywords.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008858
Willy Tarreau55165fe2009-05-10 12:02:55 +02008859option originalto [ except <network> ] [ header <name> ]
8860 Enable insertion of the X-Original-To header to requests sent to servers
8861 May be used in sections : defaults | frontend | listen | backend
8862 yes | yes | yes | yes
8863 Arguments :
8864 <network> is an optional argument used to disable this option for sources
8865 matching <network>
8866 <name> an optional argument to specify a different "X-Original-To"
8867 header name.
8868
8869 Since HAProxy can work in transparent mode, every request from a client can
8870 be redirected to the proxy and HAProxy itself can proxy every request to a
8871 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
8872 be lost. This is annoying when you want access rules based on destination ip
8873 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
8874 added by HAProxy to all requests sent to the server. This header contains a
8875 value representing the original destination IP address. Since this must be
8876 configured to always use the last occurrence of this header only. Note that
8877 only the last occurrence of the header must be used, since it is really
8878 possible that the client has already brought one.
8879
8880 The keyword "header" may be used to supply a different header name to replace
8881 the default "X-Original-To". This can be useful where you might already
8882 have a "X-Original-To" header from a different application, and you need
8883 preserve it. Also if your backend server doesn't use the "X-Original-To"
8884 header and requires different one.
8885
8886 Sometimes, a same HAProxy instance may be shared between a direct client
8887 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
8888 used to decrypt HTTPS traffic). It is possible to disable the addition of the
Amaury Denoyellef8b42922021-03-04 18:41:14 +01008889 header for a known destination address or network by adding the "except"
8890 keyword followed by the network address. In this case, any destination IP
8891 matching the network will not cause an addition of this header. Most common
8892 uses are with private networks or 127.0.0.1. IPv4 and IPv6 are both
8893 supported.
Willy Tarreau55165fe2009-05-10 12:02:55 +02008894
8895 This option may be specified either in the frontend or in the backend. If at
8896 least one of them uses it, the header will be added. Note that the backend's
8897 setting of the header subargument takes precedence over the frontend's if
8898 both are defined.
8899
Willy Tarreau55165fe2009-05-10 12:02:55 +02008900 Examples :
8901 # Original Destination address
8902 frontend www
8903 mode http
8904 option originalto except 127.0.0.1
8905
8906 # Those servers want the IP Address in X-Client-Dst
8907 backend www
8908 mode http
8909 option originalto header X-Client-Dst
8910
Christopher Faulet315b39c2018-09-21 16:26:19 +02008911 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02008912
8913
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008914option persist
8915no option persist
8916 Enable or disable forced persistence on down servers
8917 May be used in sections: defaults | frontend | listen | backend
8918 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008919 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008920
8921 When an HTTP request reaches a backend with a cookie which references a dead
8922 server, by default it is redispatched to another server. It is possible to
8923 force the request to be sent to the dead server first using "option persist"
8924 if absolutely needed. A common use case is when servers are under extreme
8925 load and spend their time flapping. In this case, the users would still be
8926 directed to the server they opened the session on, in the hope they would be
8927 correctly served. It is recommended to use "option redispatch" in conjunction
8928 with this option so that in the event it would not be possible to connect to
8929 the server at all (server definitely dead), the client would finally be
8930 redirected to another valid server.
8931
8932 If this option has been enabled in a "defaults" section, it can be disabled
8933 in a specific instance by prepending the "no" keyword before it.
8934
Willy Tarreau4de91492010-01-22 19:10:05 +01008935 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008936
8937
Willy Tarreau0c122822013-12-15 18:49:01 +01008938option pgsql-check [ user <username> ]
8939 Use PostgreSQL health checks for server testing
8940 May be used in sections : defaults | frontend | listen | backend
8941 yes | no | yes | yes
8942 Arguments :
8943 <username> This is the username which will be used when connecting to
8944 PostgreSQL server.
8945
8946 The check sends a PostgreSQL StartupMessage and waits for either
8947 Authentication request or ErrorResponse message. It is a basic but useful
8948 test which does not produce error nor aborted connect on the server.
8949 This check is identical with the "mysql-check".
8950
8951 See also: "option httpchk"
8952
8953
Willy Tarreau9420b122013-12-15 18:58:25 +01008954option prefer-last-server
8955no option prefer-last-server
8956 Allow multiple load balanced requests to remain on the same server
8957 May be used in sections: defaults | frontend | listen | backend
8958 yes | no | yes | yes
8959 Arguments : none
8960
8961 When the load balancing algorithm in use is not deterministic, and a previous
8962 request was sent to a server to which haproxy still holds a connection, it is
8963 sometimes desirable that subsequent requests on a same session go to the same
8964 server as much as possible. Note that this is different from persistence, as
8965 we only indicate a preference which haproxy tries to apply without any form
8966 of warranty. The real use is for keep-alive connections sent to servers. When
8967 this option is used, haproxy will try to reuse the same connection that is
8968 attached to the server instead of rebalancing to another server, causing a
8969 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01008970 not make much sense to use this in combination with hashing algorithms. Note,
8971 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02008972 to a proxy which sends a 407 (authentication required), when the load
8973 balancing algorithm is not deterministic. This is mandatory for use with the
8974 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01008975 troubleshooting some faulty applications. Option prefer-last-server might be
8976 desirable in these environments as well, to avoid redistributing the traffic
8977 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01008978
8979 If this option has been enabled in a "defaults" section, it can be disabled
8980 in a specific instance by prepending the "no" keyword before it.
8981
8982 See also: "option http-keep-alive"
8983
8984
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008985option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07008986option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008987no option redispatch
8988 Enable or disable session redistribution in case of connection failure
8989 May be used in sections: defaults | frontend | listen | backend
8990 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07008991 Arguments :
8992 <interval> The optional integer value that controls how often redispatches
8993 occur when retrying connections. Positive value P indicates a
8994 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01008995 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07008996 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008997 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07008998 positive value of 1 would indicate a redispatch on every retry,
8999 and a positive value of 3 would indicate a redispatch on every
9000 third retry. You can disable redispatches with a value of 0.
9001
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009002
9003 In HTTP mode, if a server designated by a cookie is down, clients may
9004 definitely stick to it because they cannot flush the cookie, so they will not
9005 be able to access the service anymore.
9006
Willy Tarreau59884a62019-01-02 14:48:31 +01009007 Specifying "option redispatch" will allow the proxy to break cookie or
9008 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009009
Olivier Carrère6e6f59b2020-04-15 11:30:18 +02009010 Active servers are selected from a subset of the list of available
9011 servers. Active servers that are not down or in maintenance (i.e., whose
9012 health is not checked or that have been checked as "up"), are selected in the
9013 following order:
9014
9015 1. Any active, non-backup server, if any, or,
9016
9017 2. If the "allbackups" option is not set, the first backup server in the
9018 list, or
9019
9020 3. If the "allbackups" option is set, any backup server.
9021
9022 When a retry occurs, HAProxy tries to select another server than the last
9023 one. The new server is selected from the current list of servers.
9024
9025 Sometimes, if the list is updated between retries (e.g., if numerous retries
9026 occur and last longer than the time needed to check that a server is down,
9027 remove it from the list and fall back on the list of backup servers),
9028 connections may be redirected to a backup server, though.
9029
Joseph Lynch726ab712015-05-11 23:25:34 -07009030 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009031 connection failures. Of course, it requires having "retries" set to a nonzero
9032 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01009033
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009034 If this option has been enabled in a "defaults" section, it can be disabled
9035 in a specific instance by prepending the "no" keyword before it.
9036
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02009037 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009038
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009039
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02009040option redis-check
9041 Use redis health checks for server testing
9042 May be used in sections : defaults | frontend | listen | backend
9043 yes | no | yes | yes
9044 Arguments : none
9045
9046 It is possible to test that the server correctly talks REDIS protocol instead
9047 of just testing that it accepts the TCP connection. When this option is set,
9048 a PING redis command is sent to the server, and the response is analyzed to
9049 find the "+PONG" response message.
9050
9051 Example :
9052 option redis-check
9053
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009054 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02009055
9056
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009057option smtpchk
9058option smtpchk <hello> <domain>
9059 Use SMTP health checks for server testing
9060 May be used in sections : defaults | frontend | listen | backend
9061 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01009062 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009063 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02009064 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009065 values will be turned into the default command ("HELO").
9066
9067 <domain> is the domain name to present to the server. It may only be
9068 specified (and is mandatory) if the hello command has been
9069 specified. By default, "localhost" is used.
9070
9071 When "option smtpchk" is set, the health checks will consist in TCP
9072 connections followed by an SMTP command. By default, this command is
9073 "HELO localhost". The server's return code is analyzed and only return codes
9074 starting with a "2" will be considered as valid. All other responses,
9075 including a lack of response will constitute an error and will indicate a
9076 dead server.
9077
9078 This test is meant to be used with SMTP servers or relays. Depending on the
9079 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01009080 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009081 25 is often easier than adjusting the configuration.
9082
9083 Most often, an incoming SMTP server needs to see the client's IP address for
9084 various purposes, including spam filtering, anti-spoofing and logging. When
9085 possible, it is often wise to masquerade the client's IP address when
9086 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02009087 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009088
9089 Example :
9090 option smtpchk HELO mydomain.org
9091
9092 See also : "option httpchk", "source"
9093
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01009094
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02009095option socket-stats
9096no option socket-stats
9097
9098 Enable or disable collecting & providing separate statistics for each socket.
9099 May be used in sections : defaults | frontend | listen | backend
9100 yes | yes | yes | no
9101
9102 Arguments : none
9103
9104
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009105option splice-auto
9106no option splice-auto
9107 Enable or disable automatic kernel acceleration on sockets in both directions
9108 May be used in sections : defaults | frontend | listen | backend
9109 yes | yes | yes | yes
9110 Arguments : none
9111
9112 When this option is enabled either on a frontend or on a backend, haproxy
9113 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01009114 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009115 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009116 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009117 are not much aggressive in order to limit excessive use of splicing. This
9118 option requires splicing to be enabled at compile time, and may be globally
9119 disabled with the global option "nosplice". Since splice uses pipes, using it
9120 requires that there are enough spare pipes.
9121
9122 Important note: kernel-based TCP splicing is a Linux-specific feature which
9123 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
9124 transfer data between sockets without copying these data to user-space, thus
9125 providing noticeable performance gains and CPU cycles savings. Since many
9126 early implementations are buggy, corrupt data and/or are inefficient, this
9127 feature is not enabled by default, and it should be used with extreme care.
9128 While it is not possible to detect the correctness of an implementation,
9129 2.6.29 is the first version offering a properly working implementation. In
9130 case of doubt, splicing may be globally disabled using the global "nosplice"
9131 keyword.
9132
9133 Example :
9134 option splice-auto
9135
9136 If this option has been enabled in a "defaults" section, it can be disabled
9137 in a specific instance by prepending the "no" keyword before it.
9138
9139 See also : "option splice-request", "option splice-response", and global
9140 options "nosplice" and "maxpipes"
9141
9142
9143option splice-request
9144no option splice-request
9145 Enable or disable automatic kernel acceleration on sockets for requests
9146 May be used in sections : defaults | frontend | listen | backend
9147 yes | yes | yes | yes
9148 Arguments : none
9149
9150 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009151 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009152 the client to the server. It might still use the recv/send scheme if there
9153 are no spare pipes left. This option requires splicing to be enabled at
9154 compile time, and may be globally disabled with the global option "nosplice".
9155 Since splice uses pipes, using it requires that there are enough spare pipes.
9156
9157 Important note: see "option splice-auto" for usage limitations.
9158
9159 Example :
9160 option splice-request
9161
9162 If this option has been enabled in a "defaults" section, it can be disabled
9163 in a specific instance by prepending the "no" keyword before it.
9164
9165 See also : "option splice-auto", "option splice-response", and global options
9166 "nosplice" and "maxpipes"
9167
9168
9169option splice-response
9170no option splice-response
9171 Enable or disable automatic kernel acceleration on sockets for responses
9172 May be used in sections : defaults | frontend | listen | backend
9173 yes | yes | yes | yes
9174 Arguments : none
9175
9176 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009177 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01009178 the server to the client. It might still use the recv/send scheme if there
9179 are no spare pipes left. This option requires splicing to be enabled at
9180 compile time, and may be globally disabled with the global option "nosplice".
9181 Since splice uses pipes, using it requires that there are enough spare pipes.
9182
9183 Important note: see "option splice-auto" for usage limitations.
9184
9185 Example :
9186 option splice-response
9187
9188 If this option has been enabled in a "defaults" section, it can be disabled
9189 in a specific instance by prepending the "no" keyword before it.
9190
9191 See also : "option splice-auto", "option splice-request", and global options
9192 "nosplice" and "maxpipes"
9193
9194
Christopher Fauletba7bc162016-11-07 21:07:38 +01009195option spop-check
9196 Use SPOP health checks for server testing
9197 May be used in sections : defaults | frontend | listen | backend
9198 no | no | no | yes
9199 Arguments : none
9200
9201 It is possible to test that the server correctly talks SPOP protocol instead
9202 of just testing that it accepts the TCP connection. When this option is set,
9203 a HELLO handshake is performed between HAProxy and the server, and the
9204 response is analyzed to check no error is reported.
9205
9206 Example :
9207 option spop-check
9208
9209 See also : "option httpchk"
9210
9211
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009212option srvtcpka
9213no option srvtcpka
9214 Enable or disable the sending of TCP keepalive packets on the server side
9215 May be used in sections : defaults | frontend | listen | backend
9216 yes | no | yes | yes
9217 Arguments : none
9218
9219 When there is a firewall or any session-aware component between a client and
9220 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01009221 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009222 components decides to expire a session which has remained idle for too long.
9223
9224 Enabling socket-level TCP keep-alives makes the system regularly send packets
9225 to the other end of the connection, leaving it active. The delay between
9226 keep-alive probes is controlled by the system only and depends both on the
9227 operating system and its tuning parameters.
9228
9229 It is important to understand that keep-alive packets are neither emitted nor
9230 received at the application level. It is only the network stacks which sees
9231 them. For this reason, even if one side of the proxy already uses keep-alives
9232 to maintain its connection alive, those keep-alive packets will not be
9233 forwarded to the other side of the proxy.
9234
9235 Please note that this has nothing to do with HTTP keep-alive.
9236
9237 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
9238 server side of a connection, which should help when session expirations are
9239 noticed between HAProxy and a server.
9240
9241 If this option has been enabled in a "defaults" section, it can be disabled
9242 in a specific instance by prepending the "no" keyword before it.
9243
9244 See also : "option clitcpka", "option tcpka"
9245
9246
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009247option ssl-hello-chk
9248 Use SSLv3 client hello health checks for server testing
9249 May be used in sections : defaults | frontend | listen | backend
9250 yes | no | yes | yes
9251 Arguments : none
9252
9253 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
9254 possible to test that the server correctly talks SSL instead of just testing
9255 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
9256 SSLv3 client hello messages are sent once the connection is established to
9257 the server, and the response is analyzed to find an SSL server hello message.
9258 The server is considered valid only when the response contains this server
9259 hello message.
9260
9261 All servers tested till there correctly reply to SSLv3 client hello messages,
9262 and most servers tested do not even log the requests containing only hello
9263 messages, which is appreciable.
9264
Willy Tarreau763a95b2012-10-04 23:15:39 +02009265 Note that this check works even when SSL support was not built into haproxy
9266 because it forges the SSL message. When SSL support is available, it is best
9267 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009268
Willy Tarreau763a95b2012-10-04 23:15:39 +02009269 See also: "option httpchk", "check-ssl"
9270
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009271
Willy Tarreaued179852013-12-16 01:07:00 +01009272option tcp-check
9273 Perform health checks using tcp-check send/expect sequences
9274 May be used in sections: defaults | frontend | listen | backend
9275 yes | no | yes | yes
9276
9277 This health check method is intended to be combined with "tcp-check" command
9278 lists in order to support send/expect types of health check sequences.
9279
9280 TCP checks currently support 4 modes of operations :
9281 - no "tcp-check" directive : the health check only consists in a connection
9282 attempt, which remains the default mode.
9283
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009284 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01009285 used to send a string along with a connection opening. With some
9286 protocols, it helps sending a "QUIT" message for example that prevents
9287 the server from logging a connection error for each health check. The
9288 check result will still be based on the ability to open the connection
9289 only.
9290
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009291 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01009292 The connection is opened and haproxy waits for the server to present some
9293 contents which must validate some rules. The check result will be based
9294 on the matching between the contents and the rules. This is suited for
9295 POP, IMAP, SMTP, FTP, SSH, TELNET.
9296
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009297 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009298 used to test a hello-type protocol. HAProxy sends a message, the server
9299 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009300 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01009301 suited for protocols which require a binding or a request/response model.
9302 LDAP, MySQL, Redis and SSL are example of such protocols, though they
9303 already all have their dedicated checks with a deeper understanding of
9304 the respective protocols.
9305 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01009306 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01009307
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009308 A fifth mode can be used to insert comments in different steps of the script.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009309
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009310 For each tcp-check rule you create, you can add a "comment" directive,
9311 followed by a string. This string will be reported in the log and stderr in
9312 debug mode. It is useful to make user-friendly error reporting. The
9313 "comment" is of course optional.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009314
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009315 During the execution of a health check, a variable scope is made available to
9316 store data samples, using the "tcp-check set-var" operation. Freeing those
9317 variable is possible using "tcp-check unset-var".
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +01009318
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009319
Willy Tarreaued179852013-12-16 01:07:00 +01009320 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01009321 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01009322 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009323 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01009324
Davor Ocelice9ed2812017-12-25 17:49:28 +01009325 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01009326 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009327 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01009328
9329 # look for the redis master server after ensuring it speaks well
9330 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009331 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01009332 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009333 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01009334 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009335 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009336 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01009337 tcp-check send info\ replication\r\n
9338 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009339 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01009340 tcp-check send QUIT\r\n
9341 tcp-check expect string +OK
9342
Davor Ocelice9ed2812017-12-25 17:49:28 +01009343 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01009344 (send many headers before analyzing)
9345 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009346 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01009347 tcp-check send HEAD\ /\ HTTP/1.1\r\n
9348 tcp-check send Host:\ www.mydomain.com\r\n
9349 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
9350 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009351 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01009352
9353
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009354 See also : "tcp-check connect", "tcp-check expect" and "tcp-check send".
Willy Tarreaued179852013-12-16 01:07:00 +01009355
9356
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009357option tcp-smart-accept
9358no option tcp-smart-accept
9359 Enable or disable the saving of one ACK packet during the accept sequence
9360 May be used in sections : defaults | frontend | listen | backend
9361 yes | yes | yes | no
9362 Arguments : none
9363
9364 When an HTTP connection request comes in, the system acknowledges it on
9365 behalf of HAProxy, then the client immediately sends its request, and the
9366 system acknowledges it too while it is notifying HAProxy about the new
9367 connection. HAProxy then reads the request and responds. This means that we
9368 have one TCP ACK sent by the system for nothing, because the request could
9369 very well be acknowledged by HAProxy when it sends its response.
9370
9371 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
9372 sending this useless ACK on platforms which support it (currently at least
9373 Linux). It must not cause any problem, because the system will send it anyway
9374 after 40 ms if the response takes more time than expected to come.
9375
9376 During complex network debugging sessions, it may be desirable to disable
9377 this optimization because delayed ACKs can make troubleshooting more complex
9378 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01009379 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009380
9381 It is also possible to force it for non-HTTP proxies by simply specifying
9382 "option tcp-smart-accept". For instance, it can make sense with some services
9383 such as SMTP where the server speaks first.
9384
9385 It is recommended to avoid forcing this option in a defaults section. In case
9386 of doubt, consider setting it back to automatic values by prepending the
9387 "default" keyword before it, or disabling it using the "no" keyword.
9388
Willy Tarreaud88edf22009-06-14 15:48:17 +02009389 See also : "option tcp-smart-connect"
9390
9391
9392option tcp-smart-connect
9393no option tcp-smart-connect
9394 Enable or disable the saving of one ACK packet during the connect sequence
9395 May be used in sections : defaults | frontend | listen | backend
9396 yes | no | yes | yes
9397 Arguments : none
9398
9399 On certain systems (at least Linux), HAProxy can ask the kernel not to
9400 immediately send an empty ACK upon a connection request, but to directly
9401 send the buffer request instead. This saves one packet on the network and
9402 thus boosts performance. It can also be useful for some servers, because they
9403 immediately get the request along with the incoming connection.
9404
9405 This feature is enabled when "option tcp-smart-connect" is set in a backend.
9406 It is not enabled by default because it makes network troubleshooting more
9407 complex.
9408
9409 It only makes sense to enable it with protocols where the client speaks first
9410 such as HTTP. In other situations, if there is no data to send in place of
9411 the ACK, a normal ACK is sent.
9412
9413 If this option has been enabled in a "defaults" section, it can be disabled
9414 in a specific instance by prepending the "no" keyword before it.
9415
9416 See also : "option tcp-smart-accept"
9417
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009418
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009419option tcpka
9420 Enable or disable the sending of TCP keepalive packets on both sides
9421 May be used in sections : defaults | frontend | listen | backend
9422 yes | yes | yes | yes
9423 Arguments : none
9424
9425 When there is a firewall or any session-aware component between a client and
9426 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01009427 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009428 components decides to expire a session which has remained idle for too long.
9429
9430 Enabling socket-level TCP keep-alives makes the system regularly send packets
9431 to the other end of the connection, leaving it active. The delay between
9432 keep-alive probes is controlled by the system only and depends both on the
9433 operating system and its tuning parameters.
9434
9435 It is important to understand that keep-alive packets are neither emitted nor
9436 received at the application level. It is only the network stacks which sees
9437 them. For this reason, even if one side of the proxy already uses keep-alives
9438 to maintain its connection alive, those keep-alive packets will not be
9439 forwarded to the other side of the proxy.
9440
9441 Please note that this has nothing to do with HTTP keep-alive.
9442
9443 Using option "tcpka" enables the emission of TCP keep-alive probes on both
9444 the client and server sides of a connection. Note that this is meaningful
9445 only in "defaults" or "listen" sections. If this option is used in a
9446 frontend, only the client side will get keep-alives, and if this option is
9447 used in a backend, only the server side will get keep-alives. For this
9448 reason, it is strongly recommended to explicitly use "option clitcpka" and
9449 "option srvtcpka" when the configuration is split between frontends and
9450 backends.
9451
9452 See also : "option clitcpka", "option srvtcpka"
9453
Willy Tarreau844e3c52008-01-11 16:28:18 +01009454
9455option tcplog
9456 Enable advanced logging of TCP connections with session state and timers
9457 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01009458 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01009459 Arguments : none
9460
9461 By default, the log output format is very poor, as it only contains the
9462 source and destination addresses, and the instance name. By specifying
9463 "option tcplog", each log line turns into a much richer format including, but
9464 not limited to, the connection timers, the session status, the connections
9465 numbers, the frontend, backend and server name, and of course the source
9466 address and ports. This option is useful for pure TCP proxies in order to
9467 find which of the client or server disconnects or times out. For normal HTTP
9468 proxies, it's better to use "option httplog" which is even more complete.
9469
Guillaume de Lafond29f45602017-03-31 19:52:15 +02009470 "option tcplog" overrides any previous "log-format" directive.
9471
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009472 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009473
9474
Willy Tarreau844e3c52008-01-11 16:28:18 +01009475option transparent
9476no option transparent
9477 Enable client-side transparent proxying
9478 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01009479 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01009480 Arguments : none
9481
9482 This option was introduced in order to provide layer 7 persistence to layer 3
9483 load balancers. The idea is to use the OS's ability to redirect an incoming
9484 connection for a remote address to a local process (here HAProxy), and let
9485 this process know what address was initially requested. When this option is
9486 used, sessions without cookies will be forwarded to the original destination
9487 IP address of the incoming request (which should match that of another
9488 equipment), while requests with cookies will still be forwarded to the
9489 appropriate server.
9490
9491 Note that contrary to a common belief, this option does NOT make HAProxy
9492 present the client's IP to the server when establishing the connection.
9493
Willy Tarreaua1146052011-03-01 09:51:54 +01009494 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009495 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009496
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009497
Simon Horman98637e52014-06-20 12:30:16 +09009498external-check command <command>
9499 Executable to run when performing an external-check
9500 May be used in sections : defaults | frontend | listen | backend
9501 yes | no | yes | yes
9502
9503 Arguments :
9504 <command> is the external command to run
9505
Simon Horman98637e52014-06-20 12:30:16 +09009506 The arguments passed to the to the command are:
9507
Cyril Bonté777be862014-12-02 21:21:35 +01009508 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09009509
Cyril Bonté777be862014-12-02 21:21:35 +01009510 The <proxy_address> and <proxy_port> are derived from the first listener
9511 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
9512 listener the proxy_address will be the path of the socket and the
9513 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
9514 possible to determine a listener, and both <proxy_address> and <proxy_port>
9515 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09009516
Cyril Bonté72cda2a2014-12-27 22:28:39 +01009517 Some values are also provided through environment variables.
9518
9519 Environment variables :
9520 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
9521 applicable, for example in a "backend" section).
9522
9523 HAPROXY_PROXY_ID The backend id.
9524
9525 HAPROXY_PROXY_NAME The backend name.
9526
9527 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
9528 applicable, for example in a "backend" section or
9529 for a UNIX socket).
9530
9531 HAPROXY_SERVER_ADDR The server address.
9532
9533 HAPROXY_SERVER_CURCONN The current number of connections on the server.
9534
9535 HAPROXY_SERVER_ID The server id.
9536
9537 HAPROXY_SERVER_MAXCONN The server max connections.
9538
9539 HAPROXY_SERVER_NAME The server name.
9540
9541 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
9542 socket).
9543
9544 PATH The PATH environment variable used when executing
9545 the command may be set using "external-check path".
9546
William Lallemand4d03e432019-06-14 15:35:37 +02009547 See also "2.3. Environment variables" for other variables.
9548
Simon Horman98637e52014-06-20 12:30:16 +09009549 If the command executed and exits with a zero status then the check is
9550 considered to have passed, otherwise the check is considered to have
9551 failed.
9552
9553 Example :
9554 external-check command /bin/true
9555
9556 See also : "external-check", "option external-check", "external-check path"
9557
9558
9559external-check path <path>
9560 The value of the PATH environment variable used when running an external-check
9561 May be used in sections : defaults | frontend | listen | backend
9562 yes | no | yes | yes
9563
9564 Arguments :
9565 <path> is the path used when executing external command to run
9566
9567 The default path is "".
9568
9569 Example :
9570 external-check path "/usr/bin:/bin"
9571
9572 See also : "external-check", "option external-check",
9573 "external-check command"
9574
9575
Emeric Brun647caf12009-06-30 17:57:00 +02009576persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02009577persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02009578 Enable RDP cookie-based persistence
9579 May be used in sections : defaults | frontend | listen | backend
9580 yes | no | yes | yes
9581 Arguments :
9582 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02009583 default cookie name "msts" will be used. There currently is no
9584 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02009585
9586 This statement enables persistence based on an RDP cookie. The RDP cookie
9587 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01009588 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02009589 and if an RDP cookie is found, it is decoded. If it matches a known server
9590 which is still UP (or if "option persist" is set), then the connection is
9591 forwarded to this server.
9592
9593 Note that this only makes sense in a TCP backend, but for this to work, the
9594 frontend must have waited long enough to ensure that an RDP cookie is present
9595 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009596 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02009597 a single "listen" section.
9598
Willy Tarreau61e28f22010-05-16 22:31:05 +02009599 Also, it is important to understand that the terminal server will emit this
9600 RDP cookie only if it is configured for "token redirection mode", which means
9601 that the "IP address redirection" option is disabled.
9602
Emeric Brun647caf12009-06-30 17:57:00 +02009603 Example :
9604 listen tse-farm
9605 bind :3389
9606 # wait up to 5s for an RDP cookie in the request
9607 tcp-request inspect-delay 5s
9608 tcp-request content accept if RDP_COOKIE
9609 # apply RDP cookie persistence
9610 persist rdp-cookie
9611 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02009612 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02009613 balance rdp-cookie
9614 server srv1 1.1.1.1:3389
9615 server srv2 1.1.1.2:3389
9616
Simon Hormanab814e02011-06-24 14:50:20 +09009617 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
9618 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02009619
9620
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009621rate-limit sessions <rate>
9622 Set a limit on the number of new sessions accepted per second on a frontend
9623 May be used in sections : defaults | frontend | listen | backend
9624 yes | yes | yes | no
9625 Arguments :
9626 <rate> The <rate> parameter is an integer designating the maximum number
9627 of new sessions per second to accept on the frontend.
9628
9629 When the frontend reaches the specified number of new sessions per second, it
9630 stops accepting new connections until the rate drops below the limit again.
9631 During this time, the pending sessions will be kept in the socket's backlog
9632 (in system buffers) and haproxy will not even be aware that sessions are
9633 pending. When applying very low limit on a highly loaded service, it may make
9634 sense to increase the socket's backlog using the "backlog" keyword.
9635
9636 This feature is particularly efficient at blocking connection-based attacks
9637 or service abuse on fragile servers. Since the session rate is measured every
9638 millisecond, it is extremely accurate. Also, the limit applies immediately,
9639 no delay is needed at all to detect the threshold.
9640
9641 Example : limit the connection rate on SMTP to 10 per second max
9642 listen smtp
9643 mode tcp
9644 bind :25
9645 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02009646 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009647
Willy Tarreaua17c2d92011-07-25 08:16:20 +02009648 Note : when the maximum rate is reached, the frontend's status is not changed
9649 but its sockets appear as "WAITING" in the statistics if the
9650 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009651
9652 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
9653
9654
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009655redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
9656redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
9657redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009658 Return an HTTP redirection if/unless a condition is matched
9659 May be used in sections : defaults | frontend | listen | backend
9660 no | yes | yes | yes
9661
9662 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01009663 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009664
Willy Tarreau0140f252008-11-19 21:07:09 +01009665 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009666 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009667 the HTTP "Location" header. When used in an "http-request" rule,
9668 <loc> value follows the log-format rules and can include some
9669 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009670
9671 <pfx> With "redirect prefix", the "Location" header is built from the
9672 concatenation of <pfx> and the complete URI path, including the
9673 query string, unless the "drop-query" option is specified (see
9674 below). As a special case, if <pfx> equals exactly "/", then
9675 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009676 redirect to the same URL (for instance, to insert a cookie). When
9677 used in an "http-request" rule, <pfx> value follows the log-format
9678 rules and can include some dynamic values (see Custom Log Format
9679 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009680
9681 <sch> With "redirect scheme", then the "Location" header is built by
9682 concatenating <sch> with "://" then the first occurrence of the
9683 "Host" header, and then the URI path, including the query string
9684 unless the "drop-query" option is specified (see below). If no
9685 path is found or if the path is "*", then "/" is used instead. If
9686 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009687 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009688 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009689 HTTPS. When used in an "http-request" rule, <sch> value follows
9690 the log-format rules and can include some dynamic values (see
9691 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01009692
9693 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009694 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
9695 with 302 used by default if no code is specified. 301 means
9696 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02009697 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009698 cache the redirection. 303 is equivalent to 302 except that the
9699 browser will fetch the location with a GET method. 307 is just
9700 like 302 but makes it clear that the same method must be reused.
9701 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01009702
9703 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009704 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01009705
9706 - "drop-query"
9707 When this keyword is used in a prefix-based redirection, then the
9708 location will be set without any possible query-string, which is useful
9709 for directing users to a non-secure page for instance. It has no effect
9710 with a location-type redirect.
9711
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009712 - "append-slash"
9713 This keyword may be used in conjunction with "drop-query" to redirect
9714 users who use a URL not ending with a '/' to the same one with the '/'.
9715 It can be useful to ensure that search engines will only see one URL.
9716 For this, a return code 301 is preferred.
9717
Willy Tarreau0140f252008-11-19 21:07:09 +01009718 - "set-cookie NAME[=value]"
9719 A "Set-Cookie" header will be added with NAME (and optionally "=value")
9720 to the response. This is sometimes used to indicate that a user has
9721 been seen, for instance to protect against some types of DoS. No other
9722 cookie option is added, so the cookie will be a session cookie. Note
9723 that for a browser, a sole cookie name without an equal sign is
9724 different from a cookie with an equal sign.
9725
9726 - "clear-cookie NAME[=]"
9727 A "Set-Cookie" header will be added with NAME (and optionally "="), but
9728 with the "Max-Age" attribute set to zero. This will tell the browser to
9729 delete this cookie. It is useful for instance on logout pages. It is
9730 important to note that clearing the cookie "NAME" will not remove a
9731 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
9732 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009733
9734 Example: move the login URL only to HTTPS.
9735 acl clear dst_port 80
9736 acl secure dst_port 8080
9737 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01009738 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01009739 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01009740 acl cookie_set hdr_sub(cookie) SEEN=1
9741
9742 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01009743 redirect prefix https://mysite.com if login_page !secure
9744 redirect prefix http://mysite.com drop-query if login_page !uid_given
9745 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01009746 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009747
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009748 Example: send redirects for request for articles without a '/'.
9749 acl missing_slash path_reg ^/article/[^/]*$
9750 redirect code 301 prefix / drop-query append-slash if missing_slash
9751
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009752 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01009753 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009754
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009755 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02009756 http-request redirect code 301 location \
9757 http://www.%[hdr(host)]%[capture.req.uri] \
9758 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009759
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009760 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009761
Willy Tarreau303c0352008-01-17 19:01:39 +01009762
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009763retries <value>
9764 Set the number of retries to perform on a server after a connection failure
9765 May be used in sections: defaults | frontend | listen | backend
9766 yes | no | yes | yes
9767 Arguments :
9768 <value> is the number of times a connection attempt should be retried on
9769 a server when a connection either is refused or times out. The
9770 default value is 3.
9771
9772 It is important to understand that this value applies to the number of
9773 connection attempts, not full requests. When a connection has effectively
9774 been established to a server, there will be no more retry.
9775
9776 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07009777 a turn-around timer of min("timeout connect", one second) is applied before
9778 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009779
9780 When "option redispatch" is set, the last retry may be performed on another
9781 server even if a cookie references a different server.
9782
9783 See also : "option redispatch"
9784
9785
Olivier Houcharda254a372019-04-05 15:30:12 +02009786retry-on [list of keywords]
Jerome Magnin5ce3c142020-05-13 20:09:57 +02009787 Specify when to attempt to automatically retry a failed request.
9788 This setting is only valid when "mode" is set to http and is silently ignored
9789 otherwise.
Olivier Houcharda254a372019-04-05 15:30:12 +02009790 May be used in sections: defaults | frontend | listen | backend
9791 yes | no | yes | yes
9792 Arguments :
9793 <keywords> is a list of keywords or HTTP status codes, each representing a
9794 type of failure event on which an attempt to retry the request
9795 is desired. Please read the notes at the bottom before changing
9796 this setting. The following keywords are supported :
9797
9798 none never retry
9799
9800 conn-failure retry when the connection or the SSL handshake failed
9801 and the request could not be sent. This is the default.
9802
9803 empty-response retry when the server connection was closed after part
9804 of the request was sent, and nothing was received from
9805 the server. This type of failure may be caused by the
9806 request timeout on the server side, poor network
9807 condition, or a server crash or restart while
9808 processing the request.
9809
Olivier Houcharde3249a92019-05-03 23:01:47 +02009810 junk-response retry when the server returned something not looking
9811 like a complete HTTP response. This includes partial
9812 responses headers as well as non-HTTP contents. It
9813 usually is a bad idea to retry on such events, which
9814 may be caused a configuration issue (wrong server port)
9815 or by the request being harmful to the server (buffer
9816 overflow attack for example).
9817
Olivier Houcharda254a372019-04-05 15:30:12 +02009818 response-timeout the server timeout stroke while waiting for the server
9819 to respond to the request. This may be caused by poor
9820 network condition, the reuse of an idle connection
9821 which has expired on the path, or by the request being
9822 extremely expensive to process. It generally is a bad
9823 idea to retry on such events on servers dealing with
9824 heavy database processing (full scans, etc) as it may
9825 amplify denial of service attacks.
9826
Olivier Houchard865d8392019-05-03 22:46:27 +02009827 0rtt-rejected retry requests which were sent over early data and were
9828 rejected by the server. These requests are generally
9829 considered to be safe to retry.
9830
Julien Pivotto2de240a2020-11-12 11:14:05 +01009831 <status> any HTTP status code among "401" (Unauthorized), "403"
9832 (Forbidden), "404" (Not Found), "408" (Request Timeout),
9833 "425" (Too Early), "500" (Server Error), "501" (Not
9834 Implemented), "502" (Bad Gateway), "503" (Service
9835 Unavailable), "504" (Gateway Timeout).
Olivier Houcharda254a372019-04-05 15:30:12 +02009836
Olivier Houchardddf0e032019-05-10 18:05:40 +02009837 all-retryable-errors
9838 retry request for any error that are considered
9839 retryable. This currently activates "conn-failure",
9840 "empty-response", "junk-response", "response-timeout",
9841 "0rtt-rejected", "500", "502", "503", and "504".
9842
Olivier Houcharda254a372019-04-05 15:30:12 +02009843 Using this directive replaces any previous settings with the new ones; it is
9844 not cumulative.
9845
9846 Please note that using anything other than "none" and "conn-failure" requires
9847 to allocate a buffer and copy the whole request into it, so it has memory and
9848 performance impacts. Requests not fitting in a single buffer will never be
9849 retried (see the global tune.bufsize setting).
9850
9851 You have to make sure the application has a replay protection mechanism built
9852 in such as a unique transaction IDs passed in requests, or that replaying the
9853 same request has no consequence, or it is very dangerous to use any retry-on
9854 value beside "conn-failure" and "none". Static file servers and caches are
9855 generally considered safe against any type of retry. Using a status code can
9856 be useful to quickly leave a server showing an abnormal behavior (out of
9857 memory, file system issues, etc), but in this case it may be a good idea to
9858 immediately redispatch the connection to another server (please see "option
9859 redispatch" for this). Last, it is important to understand that most causes
9860 of failures are the requests themselves and that retrying a request causing a
9861 server to misbehave will often make the situation even worse for this server,
9862 or for the whole service in case of redispatch.
9863
9864 Unless you know exactly how the application deals with replayed requests, you
9865 should not use this directive.
9866
9867 The default is "conn-failure".
9868
9869 See also: "retries", "option redispatch", "tune.bufsize"
9870
David du Colombier486df472011-03-17 10:40:26 +01009871server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009872 Declare a server in a backend
9873 May be used in sections : defaults | frontend | listen | backend
9874 no | no | yes | yes
9875 Arguments :
9876 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009877 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05009878 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009879
David du Colombier486df472011-03-17 10:40:26 +01009880 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
9881 resolvable hostname is supported, but this name will be resolved
9882 during start-up. Address "0.0.0.0" or "*" has a special meaning.
9883 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02009884 address as the one from the client connection. This is useful in
9885 transparent proxy architectures where the client's connection is
9886 intercepted and haproxy must forward to the original destination
9887 address. This is more or less what the "transparent" keyword does
9888 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01009889 to report statistics. Optionally, an address family prefix may be
9890 used before the address to force the family regardless of the
9891 address format, which can be useful to specify a path to a unix
9892 socket with no slash ('/'). Currently supported prefixes are :
9893 - 'ipv4@' -> address is always IPv4
9894 - 'ipv6@' -> address is always IPv6
9895 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02009896 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02009897 - 'sockpair@' -> address is the FD of a connected unix
9898 socket or of a socketpair. During a connection, the
9899 backend creates a pair of connected sockets, and passes
9900 one of them over the FD. The bind part will use the
9901 received socket as the client FD. Should be used
9902 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02009903 You may want to reference some environment variables in the
9904 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01009905 variables. The "init-addr" setting can be used to modify the way
9906 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009907
Willy Tarreaub6205fd2012-09-24 12:27:33 +02009908 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009909 be sent to this port. If unset, the same port the client
9910 connected to will be used. The port may also be prefixed by a "+"
9911 or a "-". In this case, the server's port will be determined by
9912 adding this value to the client's port.
9913
9914 <param*> is a list of parameters for this server. The "server" keywords
9915 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009916 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009917
9918 Examples :
9919 server first 10.1.1.1:1080 cookie first check inter 1000
9920 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01009921 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02009922 server backup "${SRV_BACKUP}:1080" backup
9923 server www1_dc1 "${LAN_DC1}.101:80"
9924 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009925
Willy Tarreau55dcaf62015-09-27 15:03:15 +02009926 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
9927 sun_path length is used for the address length. Some other programs
9928 such as socat use the string length only by default. Pass the option
9929 ",unix-tightsocklen=0" to any abstract socket definition in socat to
9930 make it compatible with HAProxy's.
9931
Mark Lamourinec2247f02012-01-04 13:02:01 -05009932 See also: "default-server", "http-send-name-header" and section 5 about
9933 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009934
Christopher Faulet583b6de2021-02-12 09:27:10 +01009935server-state-file-name [ { use-backend-name | <file> } ]
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009936 Set the server state file to read, load and apply to servers available in
Christopher Faulet583b6de2021-02-12 09:27:10 +01009937 this backend.
9938 May be used in sections: defaults | frontend | listen | backend
9939 no | no | yes | yes
9940
9941 It only applies when the directive "load-server-state-from-file" is set to
9942 "local". When <file> is not provided, if "use-backend-name" is used or if
9943 this directive is not set, then backend name is used. If <file> starts with a
9944 slash '/', then it is considered as an absolute path. Otherwise, <file> is
9945 concatenated to the global directive "server-state-base".
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009946
9947 Example: the minimal configuration below would make HAProxy look for the
9948 state server file '/etc/haproxy/states/bk':
9949
9950 global
9951 server-state-file-base /etc/haproxy/states
9952
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +01009953 backend bk
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009954 load-server-state-from-file
9955
Christopher Faulet583b6de2021-02-12 09:27:10 +01009956 See also: "server-state-base", "load-server-state-from-file", and
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009957 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009958
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02009959server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
9960 Set a template to initialize servers with shared parameters.
9961 The names of these servers are built from <prefix> and <num | range> parameters.
9962 May be used in sections : defaults | frontend | listen | backend
9963 no | no | yes | yes
9964
9965 Arguments:
9966 <prefix> A prefix for the server names to be built.
9967
9968 <num | range>
9969 If <num> is provided, this template initializes <num> servers
9970 with 1 up to <num> as server name suffixes. A range of numbers
9971 <num_low>-<num_high> may also be used to use <num_low> up to
9972 <num_high> as server name suffixes.
9973
9974 <fqdn> A FQDN for all the servers this template initializes.
9975
9976 <port> Same meaning as "server" <port> argument (see "server" keyword).
9977
9978 <params*>
9979 Remaining server parameters among all those supported by "server"
9980 keyword.
9981
9982 Examples:
9983 # Initializes 3 servers with srv1, srv2 and srv3 as names,
9984 # google.com as FQDN, and health-check enabled.
9985 server-template srv 1-3 google.com:80 check
9986
9987 # or
9988 server-template srv 3 google.com:80 check
9989
9990 # would be equivalent to:
9991 server srv1 google.com:80 check
9992 server srv2 google.com:80 check
9993 server srv3 google.com:80 check
9994
9995
9996
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009997source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02009998source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01009999source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010000 Set the source address for outgoing connections
10001 May be used in sections : defaults | frontend | listen | backend
10002 yes | no | yes | yes
10003 Arguments :
10004 <addr> is the IPv4 address HAProxy will bind to before connecting to a
10005 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +010010006
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010007 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +010010008 the most appropriate address to reach its destination. Optionally
10009 an address family prefix may be used before the address to force
10010 the family regardless of the address format, which can be useful
10011 to specify a path to a unix socket with no slash ('/'). Currently
10012 supported prefixes are :
10013 - 'ipv4@' -> address is always IPv4
10014 - 'ipv6@' -> address is always IPv6
10015 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +020010016 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +020010017 You may want to reference some environment variables in the
10018 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010019
10020 <port> is an optional port. It is normally not needed but may be useful
10021 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020010022 the system will select a free port. Note that port ranges are not
10023 supported in the backend. If you want to force port ranges, you
10024 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010025
10026 <addr2> is the IP address to present to the server when connections are
10027 forwarded in full transparent proxy mode. This is currently only
10028 supported on some patched Linux kernels. When this address is
10029 specified, clients connecting to the server will be presented
10030 with this address, while health checks will still use the address
10031 <addr>.
10032
10033 <port2> is the optional port to present to the server when connections
10034 are forwarded in full transparent proxy mode (see <addr2> above).
10035 The default value of zero means the system will select a free
10036 port.
10037
Willy Tarreaubce70882009-09-07 11:51:47 +020010038 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
10039 This is the name of a comma-separated header list which can
10040 contain multiple IP addresses. By default, the last occurrence is
10041 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +010010042 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +020010043 by previous proxy, typically Stunnel. In order to use another
10044 occurrence from the last one, please see the <occ> parameter
10045 below. When the header (or occurrence) is not found, no binding
10046 is performed so that the proxy's default IP address is used. Also
10047 keep in mind that the header name is case insensitive, as for any
10048 HTTP header.
10049
10050 <occ> is the occurrence number of a value to be used in a multi-value
10051 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010052 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +020010053 address. Positive values indicate a position from the first
10054 occurrence, 1 being the first one. Negative values indicate
10055 positions relative to the last one, -1 being the last one. This
10056 is helpful for situations where an X-Forwarded-For header is set
10057 at the entry point of an infrastructure and must be used several
10058 proxy layers away. When this value is not specified, -1 is
10059 assumed. Passing a zero here disables the feature.
10060
Willy Tarreaud53f96b2009-02-04 18:46:54 +010010061 <name> is an optional interface name to which to bind to for outgoing
10062 traffic. On systems supporting this features (currently, only
10063 Linux), this allows one to bind all traffic to the server to
10064 this interface even if it is not the one the system would select
10065 based on routing tables. This should be used with extreme care.
10066 Note that using this option requires root privileges.
10067
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010068 The "source" keyword is useful in complex environments where a specific
10069 address only is allowed to connect to the servers. It may be needed when a
10070 private address must be used through a public gateway for instance, and it is
10071 known that the system cannot determine the adequate source address by itself.
10072
10073 An extension which is available on certain patched Linux kernels may be used
10074 through the "usesrc" optional keyword. It makes it possible to connect to the
10075 servers with an IP address which does not belong to the system itself. This
10076 is called "full transparent proxy mode". For this to work, the destination
10077 servers have to route their traffic back to this address through the machine
10078 running HAProxy, and IP forwarding must generally be enabled on this machine.
10079
10080 In this "full transparent proxy" mode, it is possible to force a specific IP
10081 address to be presented to the servers. This is not much used in fact. A more
10082 common use is to tell HAProxy to present the client's IP address. For this,
10083 there are two methods :
10084
10085 - present the client's IP and port addresses. This is the most transparent
10086 mode, but it can cause problems when IP connection tracking is enabled on
10087 the machine, because a same connection may be seen twice with different
10088 states. However, this solution presents the huge advantage of not
10089 limiting the system to the 64k outgoing address+port couples, because all
10090 of the client ranges may be used.
10091
10092 - present only the client's IP address and select a spare port. This
10093 solution is still quite elegant but slightly less transparent (downstream
10094 firewalls logs will not match upstream's). It also presents the downside
10095 of limiting the number of concurrent connections to the usual 64k ports.
10096 However, since the upstream and downstream ports are different, local IP
10097 connection tracking on the machine will not be upset by the reuse of the
10098 same session.
10099
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010100 This option sets the default source for all servers in the backend. It may
10101 also be specified in a "defaults" section. Finer source address specification
10102 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010103 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010104
Baptiste Assmann91bd3372015-07-17 21:59:42 +020010105 In order to work, "usesrc" requires root privileges.
10106
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010107 Examples :
10108 backend private
10109 # Connect to the servers using our 192.168.1.200 source address
10110 source 192.168.1.200
10111
10112 backend transparent_ssl1
10113 # Connect to the SSL farm from the client's source address
10114 source 192.168.1.200 usesrc clientip
10115
10116 backend transparent_ssl2
10117 # Connect to the SSL farm from the client's source address and port
10118 # not recommended if IP conntrack is present on the local machine.
10119 source 192.168.1.200 usesrc client
10120
10121 backend transparent_ssl3
10122 # Connect to the SSL farm from the client's source address. It
10123 # is more conntrack-friendly.
10124 source 192.168.1.200 usesrc clientip
10125
10126 backend transparent_smtp
10127 # Connect to the SMTP farm from the client's source address/port
10128 # with Tproxy version 4.
10129 source 0.0.0.0 usesrc clientip
10130
Willy Tarreaubce70882009-09-07 11:51:47 +020010131 backend transparent_http
10132 # Connect to the servers using the client's IP as seen by previous
10133 # proxy.
10134 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
10135
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010136 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010137 the Linux kernel on www.balabit.com, the "bind" keyword.
10138
Willy Tarreau844e3c52008-01-11 16:28:18 +010010139
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010140srvtcpka-cnt <count>
10141 Sets the maximum number of keepalive probes TCP should send before dropping
10142 the connection on the server side.
10143 May be used in sections : defaults | frontend | listen | backend
10144 yes | no | yes | yes
10145 Arguments :
10146 <count> is the maximum number of keepalive probes.
10147
10148 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
10149 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +020010150 The availability of this setting depends on the operating system. It is
10151 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010152
10153 See also : "option srvtcpka", "srvtcpka-idle", "srvtcpka-intvl".
10154
10155
10156srvtcpka-idle <timeout>
10157 Sets the time the connection needs to remain idle before TCP starts sending
10158 keepalive probes, if enabled the sending of TCP keepalive packets on the
10159 server side.
10160 May be used in sections : defaults | frontend | listen | backend
10161 yes | no | yes | yes
10162 Arguments :
10163 <timeout> is the time the connection needs to remain idle before TCP starts
10164 sending keepalive probes. It is specified in seconds by default,
10165 but can be in any other unit if the number is suffixed by the
10166 unit, as explained at the top of this document.
10167
10168 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
10169 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +020010170 The availability of this setting depends on the operating system. It is
10171 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010172
10173 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-intvl".
10174
10175
10176srvtcpka-intvl <timeout>
10177 Sets the time between individual keepalive probes on the server side.
10178 May be used in sections : defaults | frontend | listen | backend
10179 yes | no | yes | yes
10180 Arguments :
10181 <timeout> is the time between individual keepalive probes. It is specified
10182 in seconds by default, but can be in any other unit if the number
10183 is suffixed by the unit, as explained at the top of this
10184 document.
10185
10186 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
10187 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +020010188 The availability of this setting depends on the operating system. It is
10189 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +090010190
10191 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-idle".
10192
10193
Cyril Bonté66c327d2010-10-12 00:14:37 +020010194stats admin { if | unless } <cond>
10195 Enable statistics admin level if/unless a condition is matched
10196 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010197 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +020010198
10199 This statement enables the statistics admin level if/unless a condition is
10200 matched.
10201
10202 The admin level allows to enable/disable servers from the web interface. By
10203 default, statistics page is read-only for security reasons.
10204
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010205 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10206 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010207 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010208
Cyril Bonté23b39d92011-02-10 22:54:44 +010010209 Currently, the POST request is limited to the buffer size minus the reserved
10210 buffer space, which means that if the list of servers is too long, the
10211 request won't be processed. It is recommended to alter few servers at a
10212 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +020010213
10214 Example :
10215 # statistics admin level only for localhost
10216 backend stats_localhost
10217 stats enable
10218 stats admin if LOCALHOST
10219
10220 Example :
10221 # statistics admin level always enabled because of the authentication
10222 backend stats_auth
10223 stats enable
10224 stats auth admin:AdMiN123
10225 stats admin if TRUE
10226
10227 Example :
10228 # statistics admin level depends on the authenticated user
10229 userlist stats-auth
10230 group admin users admin
10231 user admin insecure-password AdMiN123
10232 group readonly users haproxy
10233 user haproxy insecure-password haproxy
10234
10235 backend stats_auth
10236 stats enable
10237 acl AUTH http_auth(stats-auth)
10238 acl AUTH_ADMIN http_auth_group(stats-auth) admin
10239 stats http-request auth unless AUTH
10240 stats admin if AUTH_ADMIN
10241
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010242 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
10243 "bind-process", section 3.4 about userlists and section 7 about
10244 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +020010245
10246
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010247stats auth <user>:<passwd>
10248 Enable statistics with authentication and grant access to an account
10249 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010250 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010251 Arguments :
10252 <user> is a user name to grant access to
10253
10254 <passwd> is the cleartext password associated to this user
10255
10256 This statement enables statistics with default settings, and restricts access
10257 to declared users only. It may be repeated as many times as necessary to
10258 allow as many users as desired. When a user tries to access the statistics
10259 without a valid account, a "401 Forbidden" response will be returned so that
10260 the browser asks the user to provide a valid user and password. The real
10261 which will be returned to the browser is configurable using "stats realm".
10262
10263 Since the authentication method is HTTP Basic Authentication, the passwords
10264 circulate in cleartext on the network. Thus, it was decided that the
10265 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020010266 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010267
10268 It is also possible to reduce the scope of the proxies which appear in the
10269 report using "stats scope".
10270
10271 Though this statement alone is enough to enable statistics reporting, it is
10272 recommended to set all other settings in order to avoid relying on default
10273 unobvious parameters.
10274
10275 Example :
10276 # public access (limited to this backend only)
10277 backend public_www
10278 server srv1 192.168.0.1:80
10279 stats enable
10280 stats hide-version
10281 stats scope .
10282 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010283 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010284 stats auth admin1:AdMiN123
10285 stats auth admin2:AdMiN321
10286
10287 # internal monitoring access (unlimited)
10288 backend private_monitoring
10289 stats enable
10290 stats uri /admin?stats
10291 stats refresh 5s
10292
10293 See also : "stats enable", "stats realm", "stats scope", "stats uri"
10294
10295
10296stats enable
10297 Enable statistics reporting with default settings
10298 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010299 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010300 Arguments : none
10301
10302 This statement enables statistics reporting with default settings defined
10303 at build time. Unless stated otherwise, these settings are used :
10304 - stats uri : /haproxy?stats
10305 - stats realm : "HAProxy Statistics"
10306 - stats auth : no authentication
10307 - stats scope : no restriction
10308
10309 Though this statement alone is enough to enable statistics reporting, it is
10310 recommended to set all other settings in order to avoid relying on default
10311 unobvious parameters.
10312
10313 Example :
10314 # public access (limited to this backend only)
10315 backend public_www
10316 server srv1 192.168.0.1:80
10317 stats enable
10318 stats hide-version
10319 stats scope .
10320 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010321 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010322 stats auth admin1:AdMiN123
10323 stats auth admin2:AdMiN321
10324
10325 # internal monitoring access (unlimited)
10326 backend private_monitoring
10327 stats enable
10328 stats uri /admin?stats
10329 stats refresh 5s
10330
10331 See also : "stats auth", "stats realm", "stats uri"
10332
10333
Willy Tarreaud63335a2010-02-26 12:56:52 +010010334stats hide-version
10335 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010336 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010337 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010338 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010339
Willy Tarreaud63335a2010-02-26 12:56:52 +010010340 By default, the stats page reports some useful status information along with
10341 the statistics. Among them is HAProxy's version. However, it is generally
10342 considered dangerous to report precise version to anyone, as it can help them
10343 target known weaknesses with specific attacks. The "stats hide-version"
10344 statement removes the version from the statistics report. This is recommended
10345 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010346
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +020010347 Though this statement alone is enough to enable statistics reporting, it is
10348 recommended to set all other settings in order to avoid relying on default
10349 unobvious parameters.
10350
Willy Tarreaud63335a2010-02-26 12:56:52 +010010351 Example :
10352 # public access (limited to this backend only)
10353 backend public_www
10354 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +020010355 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +010010356 stats hide-version
10357 stats scope .
10358 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010359 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +010010360 stats auth admin1:AdMiN123
10361 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010362
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010363 # internal monitoring access (unlimited)
10364 backend private_monitoring
10365 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +010010366 stats uri /admin?stats
10367 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +010010368
Willy Tarreaud63335a2010-02-26 12:56:52 +010010369 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010370
Willy Tarreau983e01e2010-01-11 18:42:06 +010010371
Cyril Bonté2be1b3f2010-09-30 23:46:30 +020010372stats http-request { allow | deny | auth [realm <realm>] }
10373 [ { if | unless } <condition> ]
10374 Access control for statistics
10375
10376 May be used in sections: defaults | frontend | listen | backend
10377 no | no | yes | yes
10378
10379 As "http-request", these set of options allow to fine control access to
10380 statistics. Each option may be followed by if/unless and acl.
10381 First option with matched condition (or option without condition) is final.
10382 For "deny" a 403 error will be returned, for "allow" normal processing is
10383 performed, for "auth" a 401/407 error code is returned so the client
10384 should be asked to enter a username and password.
10385
10386 There is no fixed limit to the number of http-request statements per
10387 instance.
10388
10389 See also : "http-request", section 3.4 about userlists and section 7
10390 about ACL usage.
10391
10392
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010393stats realm <realm>
10394 Enable statistics and set authentication realm
10395 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010396 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010397 Arguments :
10398 <realm> is the name of the HTTP Basic Authentication realm reported to
10399 the browser. The browser uses it to display it in the pop-up
10400 inviting the user to enter a valid username and password.
10401
10402 The realm is read as a single word, so any spaces in it should be escaped
10403 using a backslash ('\').
10404
10405 This statement is useful only in conjunction with "stats auth" since it is
10406 only related to authentication.
10407
10408 Though this statement alone is enough to enable statistics reporting, it is
10409 recommended to set all other settings in order to avoid relying on default
10410 unobvious parameters.
10411
10412 Example :
10413 # public access (limited to this backend only)
10414 backend public_www
10415 server srv1 192.168.0.1:80
10416 stats enable
10417 stats hide-version
10418 stats scope .
10419 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010420 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010421 stats auth admin1:AdMiN123
10422 stats auth admin2:AdMiN321
10423
10424 # internal monitoring access (unlimited)
10425 backend private_monitoring
10426 stats enable
10427 stats uri /admin?stats
10428 stats refresh 5s
10429
10430 See also : "stats auth", "stats enable", "stats uri"
10431
10432
10433stats refresh <delay>
10434 Enable statistics with automatic refresh
10435 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010436 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010437 Arguments :
10438 <delay> is the suggested refresh delay, specified in seconds, which will
10439 be returned to the browser consulting the report page. While the
10440 browser is free to apply any delay, it will generally respect it
10441 and refresh the page this every seconds. The refresh interval may
10442 be specified in any other non-default time unit, by suffixing the
10443 unit after the value, as explained at the top of this document.
10444
10445 This statement is useful on monitoring displays with a permanent page
10446 reporting the load balancer's activity. When set, the HTML report page will
10447 include a link "refresh"/"stop refresh" so that the user can select whether
Jackie Tapia749f74c2020-07-22 18:59:40 -050010448 they want automatic refresh of the page or not.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010449
10450 Though this statement alone is enough to enable statistics reporting, it is
10451 recommended to set all other settings in order to avoid relying on default
10452 unobvious parameters.
10453
10454 Example :
10455 # public access (limited to this backend only)
10456 backend public_www
10457 server srv1 192.168.0.1:80
10458 stats enable
10459 stats hide-version
10460 stats scope .
10461 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010462 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010463 stats auth admin1:AdMiN123
10464 stats auth admin2:AdMiN321
10465
10466 # internal monitoring access (unlimited)
10467 backend private_monitoring
10468 stats enable
10469 stats uri /admin?stats
10470 stats refresh 5s
10471
10472 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10473
10474
10475stats scope { <name> | "." }
10476 Enable statistics and limit access scope
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 <name> is the name of a listen, frontend or backend section to be
10481 reported. The special name "." (a single dot) designates the
10482 section in which the statement appears.
10483
10484 When this statement is specified, only the sections enumerated with this
10485 statement will appear in the report. All other ones will be hidden. This
10486 statement may appear as many times as needed if multiple sections need to be
10487 reported. Please note that the name checking is performed as simple string
10488 comparisons, and that it is never checked that a give section name really
10489 exists.
10490
10491 Though this statement alone is enough to enable statistics reporting, it is
10492 recommended to set all other settings in order to avoid relying on default
10493 unobvious parameters.
10494
10495 Example :
10496 # public access (limited to this backend only)
10497 backend public_www
10498 server srv1 192.168.0.1:80
10499 stats enable
10500 stats hide-version
10501 stats scope .
10502 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010503 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010504 stats auth admin1:AdMiN123
10505 stats auth admin2:AdMiN321
10506
10507 # internal monitoring access (unlimited)
10508 backend private_monitoring
10509 stats enable
10510 stats uri /admin?stats
10511 stats refresh 5s
10512
10513 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10514
Willy Tarreaud63335a2010-02-26 12:56:52 +010010515
Willy Tarreauc9705a12010-07-27 20:05:50 +020010516stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +010010517 Enable reporting of a description on the statistics page.
10518 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010519 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010520
Willy Tarreauc9705a12010-07-27 20:05:50 +020010521 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +010010522 description from global section is automatically used instead.
10523
10524 This statement is useful for users that offer shared services to their
10525 customers, where node or description should be different for each customer.
10526
10527 Though this statement alone is enough to enable statistics reporting, it is
10528 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010529 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010530
10531 Example :
10532 # internal monitoring access (unlimited)
10533 backend private_monitoring
10534 stats enable
10535 stats show-desc Master node for Europe, Asia, Africa
10536 stats uri /admin?stats
10537 stats refresh 5s
10538
10539 See also: "show-node", "stats enable", "stats uri" and "description" in
10540 global section.
10541
10542
10543stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +020010544 Enable reporting additional information on the statistics page
10545 May be used in sections : defaults | frontend | listen | backend
10546 yes | yes | yes | yes
10547 Arguments : none
10548
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010549 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +010010550 - cap: capabilities (proxy)
10551 - mode: one of tcp, http or health (proxy)
10552 - id: SNMP ID (proxy, socket, server)
10553 - IP (socket, server)
10554 - cookie (backend, server)
10555
10556 Though this statement alone is enough to enable statistics reporting, it is
10557 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010558 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010559
10560 See also: "stats enable", "stats uri".
10561
10562
Amaury Denoyelle0b70a8a2020-10-05 11:49:45 +020010563stats show-modules
10564 Enable display of extra statistics module on the statistics page
10565 May be used in sections : defaults | frontend | listen | backend
10566 yes | yes | yes | yes
10567 Arguments : none
10568
10569 New columns are added at the end of the line containing the extra statistics
10570 values as a tooltip.
10571
10572 Though this statement alone is enough to enable statistics reporting, it is
10573 recommended to set all other settings in order to avoid relying on default
10574 unobvious parameters. Default behavior is not to show this information.
10575
10576 See also: "stats enable", "stats uri".
10577
10578
Willy Tarreaud63335a2010-02-26 12:56:52 +010010579stats show-node [ <name> ]
10580 Enable reporting of a host name on the statistics page.
10581 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010582 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010583 Arguments:
10584 <name> is an optional name to be reported. If unspecified, the
10585 node name from global section is automatically used instead.
10586
10587 This statement is useful for users that offer shared services to their
10588 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +010010589 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010590
10591 Though this statement alone is enough to enable statistics reporting, it is
10592 recommended to set all other settings in order to avoid relying on default
10593 unobvious parameters.
10594
10595 Example:
10596 # internal monitoring access (unlimited)
10597 backend private_monitoring
10598 stats enable
10599 stats show-node Europe-1
10600 stats uri /admin?stats
10601 stats refresh 5s
10602
10603 See also: "show-desc", "stats enable", "stats uri", and "node" in global
10604 section.
10605
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010606
10607stats uri <prefix>
10608 Enable statistics and define the URI prefix to access them
10609 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010610 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010611 Arguments :
10612 <prefix> is the prefix of any URI which will be redirected to stats. This
10613 prefix may contain a question mark ('?') to indicate part of a
10614 query string.
10615
10616 The statistics URI is intercepted on the relayed traffic, so it appears as a
10617 page within the normal application. It is strongly advised to ensure that the
10618 selected URI will never appear in the application, otherwise it will never be
10619 possible to reach it in the application.
10620
10621 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010622 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010623 It is generally a good idea to include a question mark in the URI so that
10624 intermediate proxies refrain from caching the results. Also, since any string
10625 beginning with the prefix will be accepted as a stats request, the question
10626 mark helps ensuring that no valid URI will begin with the same words.
10627
10628 It is sometimes very convenient to use "/" as the URI prefix, and put that
10629 statement in a "listen" instance of its own. That makes it easy to dedicate
10630 an address or a port to statistics only.
10631
10632 Though this statement alone is enough to enable statistics reporting, it is
10633 recommended to set all other settings in order to avoid relying on default
10634 unobvious parameters.
10635
10636 Example :
10637 # public access (limited to this backend only)
10638 backend public_www
10639 server srv1 192.168.0.1:80
10640 stats enable
10641 stats hide-version
10642 stats scope .
10643 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010644 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010645 stats auth admin1:AdMiN123
10646 stats auth admin2:AdMiN321
10647
10648 # internal monitoring access (unlimited)
10649 backend private_monitoring
10650 stats enable
10651 stats uri /admin?stats
10652 stats refresh 5s
10653
10654 See also : "stats auth", "stats enable", "stats realm"
10655
10656
Willy Tarreaud63335a2010-02-26 12:56:52 +010010657stick match <pattern> [table <table>] [{if | unless} <cond>]
10658 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010659 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +010010660 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010661
10662 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010663 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010664 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010665 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010666 stickiness table. This rule is mandatory.
10667
10668 <table> is an optional stickiness table name. If unspecified, the same
10669 backend's table is used. A stickiness table is declared using
10670 the "stick-table" statement.
10671
10672 <cond> is an optional matching condition. It makes it possible to match
10673 on a certain criterion only when other conditions are met (or
10674 not met). For instance, it could be used to match on a source IP
10675 address except when a request passes through a known proxy, in
10676 which case we'd match on a header containing that IP address.
10677
10678 Some protocols or applications require complex stickiness rules and cannot
10679 always simply rely on cookies nor hashing. The "stick match" statement
10680 describes a rule to extract the stickiness criterion from an incoming request
10681 or connection. See section 7 for a complete list of possible patterns and
10682 transformation rules.
10683
10684 The table has to be declared using the "stick-table" statement. It must be of
10685 a type compatible with the pattern. By default it is the one which is present
10686 in the same backend. It is possible to share a table with other backends by
10687 referencing it using the "table" keyword. If another table is referenced,
10688 the server's ID inside the backends are used. By default, all server IDs
10689 start at 1 in each backend, so the server ordering is enough. But in case of
10690 doubt, it is highly recommended to force server IDs using their "id" setting.
10691
10692 It is possible to restrict the conditions where a "stick match" statement
10693 will apply, using "if" or "unless" followed by a condition. See section 7 for
10694 ACL based conditions.
10695
10696 There is no limit on the number of "stick match" statements. The first that
10697 applies and matches will cause the request to be directed to the same server
10698 as was used for the request which created the entry. That way, multiple
10699 matches can be used as fallbacks.
10700
10701 The stick rules are checked after the persistence cookies, so they will not
10702 affect stickiness if a cookie has already been used to select a server. That
10703 way, it becomes very easy to insert cookies and match on IP addresses in
10704 order to maintain stickiness between HTTP and HTTPS.
10705
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010706 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10707 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010708 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010709
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010710 Example :
10711 # forward SMTP users to the same server they just used for POP in the
10712 # last 30 minutes
10713 backend pop
10714 mode tcp
10715 balance roundrobin
10716 stick store-request src
10717 stick-table type ip size 200k expire 30m
10718 server s1 192.168.1.1:110
10719 server s2 192.168.1.1:110
10720
10721 backend smtp
10722 mode tcp
10723 balance roundrobin
10724 stick match src table pop
10725 server s1 192.168.1.1:25
10726 server s2 192.168.1.1:25
10727
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010728 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010729 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010730
10731
10732stick on <pattern> [table <table>] [{if | unless} <condition>]
10733 Define a request pattern to associate a user to a server
10734 May be used in sections : defaults | frontend | listen | backend
10735 no | no | yes | yes
10736
10737 Note : This form is exactly equivalent to "stick match" followed by
10738 "stick store-request", all with the same arguments. Please refer
10739 to both keywords for details. It is only provided as a convenience
10740 for writing more maintainable configurations.
10741
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010742 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10743 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010744 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010745
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010746 Examples :
10747 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +010010748 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010749
10750 # ...is strictly equivalent to this one :
10751 stick match src table pop if !localhost
10752 stick store-request src table pop if !localhost
10753
10754
10755 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
10756 # well as HTTP without cookie. Share the same table between both accesses.
10757 backend http
10758 mode http
10759 balance roundrobin
10760 stick on src table https
10761 cookie SRV insert indirect nocache
10762 server s1 192.168.1.1:80 cookie s1
10763 server s2 192.168.1.1:80 cookie s2
10764
10765 backend https
10766 mode tcp
10767 balance roundrobin
10768 stick-table type ip size 200k expire 30m
10769 stick on src
10770 server s1 192.168.1.1:443
10771 server s2 192.168.1.1:443
10772
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010773 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010774
10775
10776stick store-request <pattern> [table <table>] [{if | unless} <condition>]
10777 Define a request pattern used to create an entry in a stickiness table
10778 May be used in sections : defaults | frontend | listen | backend
10779 no | no | yes | yes
10780
10781 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010782 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010783 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010784 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010785 server is selected.
10786
10787 <table> is an optional stickiness table name. If unspecified, the same
10788 backend's table is used. A stickiness table is declared using
10789 the "stick-table" statement.
10790
10791 <cond> is an optional storage condition. It makes it possible to store
10792 certain criteria only when some conditions are met (or not met).
10793 For instance, it could be used to store the source IP address
10794 except when the request passes through a known proxy, in which
10795 case we'd store a converted form of a header containing that IP
10796 address.
10797
10798 Some protocols or applications require complex stickiness rules and cannot
10799 always simply rely on cookies nor hashing. The "stick store-request" statement
10800 describes a rule to decide what to extract from the request and when to do
10801 it, in order to store it into a stickiness table for further requests to
10802 match it using the "stick match" statement. Obviously the extracted part must
10803 make sense and have a chance to be matched in a further request. Storing a
10804 client's IP address for instance often makes sense. Storing an ID found in a
10805 URL parameter also makes sense. Storing a source port will almost never make
10806 any sense because it will be randomly matched. See section 7 for a complete
10807 list of possible patterns and transformation rules.
10808
10809 The table has to be declared using the "stick-table" statement. It must be of
10810 a type compatible with the pattern. By default it is the one which is present
10811 in the same backend. It is possible to share a table with other backends by
10812 referencing it using the "table" keyword. If another table is referenced,
10813 the server's ID inside the backends are used. By default, all server IDs
10814 start at 1 in each backend, so the server ordering is enough. But in case of
10815 doubt, it is highly recommended to force server IDs using their "id" setting.
10816
10817 It is possible to restrict the conditions where a "stick store-request"
10818 statement will apply, using "if" or "unless" followed by a condition. This
10819 condition will be evaluated while parsing the request, so any criteria can be
10820 used. See section 7 for ACL based conditions.
10821
10822 There is no limit on the number of "stick store-request" statements, but
10823 there is a limit of 8 simultaneous stores per request or response. This
10824 makes it possible to store up to 8 criteria, all extracted from either the
10825 request or the response, regardless of the number of rules. Only the 8 first
10826 ones which match will be kept. Using this, it is possible to feed multiple
10827 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010010828 another protocol or access method. Using multiple store-request rules with
10829 the same table is possible and may be used to find the best criterion to rely
10830 on, by arranging the rules by decreasing preference order. Only the first
10831 extracted criterion for a given table will be stored. All subsequent store-
10832 request rules referencing the same table will be skipped and their ACLs will
10833 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010834
10835 The "store-request" rules are evaluated once the server connection has been
10836 established, so that the table will contain the real server that processed
10837 the request.
10838
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010839 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10840 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010841 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010842
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010843 Example :
10844 # forward SMTP users to the same server they just used for POP in the
10845 # last 30 minutes
10846 backend pop
10847 mode tcp
10848 balance roundrobin
10849 stick store-request src
10850 stick-table type ip size 200k expire 30m
10851 server s1 192.168.1.1:110
10852 server s2 192.168.1.1:110
10853
10854 backend smtp
10855 mode tcp
10856 balance roundrobin
10857 stick match src table pop
10858 server s1 192.168.1.1:25
10859 server s2 192.168.1.1:25
10860
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010861 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010862 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010863
10864
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010865stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Thayne McCombs92149f92020-11-20 01:28:26 -070010866 size <size> [expire <expire>] [nopurge] [peers <peersect>] [srvkey <srvkey>]
Emeric Brunf099e792010-09-27 12:05:28 +020010867 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +080010868 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010869 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +020010870 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010871
10872 Arguments :
10873 ip a table declared with "type ip" will only store IPv4 addresses.
10874 This form is very compact (about 50 bytes per entry) and allows
10875 very fast entry lookup and stores with almost no overhead. This
10876 is mainly used to store client source IP addresses.
10877
David du Colombier9a6d3c92011-03-17 10:40:24 +010010878 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
10879 This form is very compact (about 60 bytes per entry) and allows
10880 very fast entry lookup and stores with almost no overhead. This
10881 is mainly used to store client source IP addresses.
10882
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010883 integer a table declared with "type integer" will store 32bit integers
10884 which can represent a client identifier found in a request for
10885 instance.
10886
10887 string a table declared with "type string" will store substrings of up
10888 to <len> characters. If the string provided by the pattern
10889 extractor is larger than <len>, it will be truncated before
10890 being stored. During matching, at most <len> characters will be
10891 compared between the string in the table and the extracted
10892 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010893 to 32 characters.
10894
10895 binary a table declared with "type binary" will store binary blocks
10896 of <len> bytes. If the block provided by the pattern
10897 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +020010898 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010899 is shorter than <len>, it will be padded by 0. When not
10900 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010901
10902 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010903 "string" type table (See type "string" above). Or the number
10904 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010905 changing this parameter as memory usage will proportionally
10906 increase.
10907
10908 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +010010909 value directly impacts memory usage. Count approximately
10910 50 bytes per entry, plus the size of a string if any. The size
10911 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010912
10913 [nopurge] indicates that we refuse to purge older entries when the table
10914 is full. When not specified and the table is full when haproxy
10915 wants to store an entry in it, it will flush a few of the oldest
10916 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010917 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010918 be desirable to refuse new entries instead of purging the older
10919 ones. That may be the case when the amount of data to store is
10920 far above the hardware limits and we prefer not to offer access
10921 to new clients than to reject the ones already connected. When
10922 using this parameter, be sure to properly set the "expire"
10923 parameter (see below).
10924
Emeric Brunf099e792010-09-27 12:05:28 +020010925 <peersect> is the name of the peers section to use for replication. Entries
10926 which associate keys to server IDs are kept synchronized with
10927 the remote peers declared in this section. All entries are also
10928 automatically learned from the local peer (old process) during a
10929 soft restart.
10930
Willy Tarreau1abc6732015-05-01 19:21:02 +020010931 NOTE : each peers section may be referenced only by tables
10932 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010933
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010934 <expire> defines the maximum duration of an entry in the table since it
10935 was last created, refreshed or matched. The expiration delay is
10936 defined using the standard time format, similarly as the various
10937 timeouts. The maximum duration is slightly above 24 days. See
Willy Tarreau4b103022021-02-12 17:59:10 +010010938 section 2.5 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +020010939 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010940 be removed once full. Be sure not to use the "nopurge" parameter
10941 if not expiration delay is specified.
10942
Thayne McCombs92149f92020-11-20 01:28:26 -070010943 <srvkey> specifies how each server is identified for the purposes of the
10944 stick table. The valid values are "name" and "addr". If "name" is
10945 given, then <name> argument for the server (may be generated by
10946 a template). If "addr" is given, then the server is identified
10947 by its current network address, including the port. "addr" is
10948 especially useful if you are using service discovery to generate
10949 the addresses for servers with peered stick-tables and want
10950 to consistently use the same host across peers for a stickiness
10951 token.
10952
Willy Tarreau08d5f982010-06-06 13:34:54 +020010953 <data_type> is used to store additional information in the stick-table. This
10954 may be used by ACLs in order to control various criteria related
10955 to the activity of the client matching the stick-table. For each
10956 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +020010957 that the additional data can fit. Several data types may be
10958 stored with an entry. Multiple data types may be specified after
10959 the "store" keyword, as a comma-separated list. Alternatively,
10960 it is possible to repeat the "store" keyword followed by one or
10961 several data types. Except for the "server_id" type which is
10962 automatically detected and enabled, all data types must be
10963 explicitly declared to be stored. If an ACL references a data
10964 type which is not stored, the ACL will simply not match. Some
10965 data types require an argument which must be passed just after
10966 the type between parenthesis. See below for the supported data
10967 types and their arguments.
10968
10969 The data types that can be stored with an entry are the following :
10970 - server_id : this is an integer which holds the numeric ID of the server a
10971 request was assigned to. It is used by the "stick match", "stick store",
10972 and "stick on" rules. It is automatically enabled when referenced.
10973
10974 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
10975 integer which may be used for anything. Most of the time it will be used
10976 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010977 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010978
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010979 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
10980 over a period. It is a positive 32-bit integer integer which may be used
10981 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +010010982 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010983 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +010010984 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010985
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010986 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
10987 integer which may be used for anything. Most of the time it will be used
10988 to put a special tag on some entries, for instance to note that a
10989 specific behavior was detected and must be known for future matches.
10990
10991 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
10992 over a period. It is a positive 32-bit integer integer which may be used
10993 for anything. Just like <gpc1>, it counts events, but instead of keeping
10994 a cumulative number, it maintains the rate at which the counter is
10995 incremented. Most of the time it will be used to measure the frequency of
10996 occurrence of certain events (e.g. requests to a specific URL).
10997
Willy Tarreauc9705a12010-07-27 20:05:50 +020010998 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
10999 the absolute number of connections received from clients which matched
11000 this entry. It does not mean the connections were accepted, just that
11001 they were received.
11002
11003 - conn_cur : Current Connections. It is a positive 32-bit integer which
11004 stores the concurrent connection counts for the entry. It is incremented
11005 once an incoming connection matches the entry, and decremented once the
11006 connection leaves. That way it is possible to know at any time the exact
11007 number of concurrent connections for an entry.
11008
11009 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11010 integer parameter <period> which indicates in milliseconds the length
11011 of the period over which the average is measured. It reports the average
11012 incoming connection rate over that period, in connections per period. The
11013 result is an integer which can be matched using ACLs.
11014
11015 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
11016 the absolute number of sessions received from clients which matched this
11017 entry. A session is a connection that was accepted by the layer 4 rules.
11018
11019 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11020 integer parameter <period> which indicates in milliseconds the length
11021 of the period over which the average is measured. It reports the average
11022 incoming session rate over that period, in sessions per period. The
11023 result is an integer which can be matched using ACLs.
11024
11025 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
11026 counts the absolute number of HTTP requests received from clients which
11027 matched this entry. It does not matter whether they are valid requests or
11028 not. Note that this is different from sessions when keep-alive is used on
11029 the client side.
11030
11031 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11032 integer parameter <period> which indicates in milliseconds the length
11033 of the period over which the average is measured. It reports the average
11034 HTTP request rate over that period, in requests per period. The result is
11035 an integer which can be matched using ACLs. It does not matter whether
11036 they are valid requests or not. Note that this is different from sessions
11037 when keep-alive is used on the client side.
11038
11039 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
11040 counts the absolute number of HTTP requests errors induced by clients
11041 which matched this entry. Errors are counted on invalid and truncated
11042 requests, as well as on denied or tarpitted requests, and on failed
11043 authentications. If the server responds with 4xx, then the request is
11044 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +010011045 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +020011046
11047 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11048 integer parameter <period> which indicates in milliseconds the length
11049 of the period over which the average is measured. It reports the average
11050 HTTP request error rate over that period, in requests per period (see
11051 http_err_cnt above for what is accounted as an error). The result is an
11052 integer which can be matched using ACLs.
11053
Willy Tarreau826f3ab2021-02-10 12:07:15 +010011054 - http_fail_cnt : HTTP Failure Count. It is a positive 32-bit integer which
11055 counts the absolute number of HTTP response failures induced by servers
11056 which matched this entry. Errors are counted on invalid and truncated
11057 responses, as well as any 5xx response other than 501 or 505. It aims at
11058 being used combined with path or URI to detect service failures.
11059
11060 - http_fail_rate(<period>) : frequency counter (takes 12 bytes). It takes
11061 an integer parameter <period> which indicates in milliseconds the length
11062 of the period over which the average is measured. It reports the average
11063 HTTP response failure rate over that period, in requests per period (see
11064 http_fail_cnt above for what is accounted as a failure). The result is an
11065 integer which can be matched using ACLs.
11066
Willy Tarreauc9705a12010-07-27 20:05:50 +020011067 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010011068 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +020011069 which matched this entry. Headers are included in the count. This may be
11070 used to limit abuse of upload features on photo or video servers.
11071
11072 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
11073 integer parameter <period> which indicates in milliseconds the length
11074 of the period over which the average is measured. It reports the average
11075 incoming bytes rate over that period, in bytes per period. It may be used
11076 to detect users which upload too much and too fast. Warning: with large
11077 uploads, it is possible that the amount of uploaded data will be counted
11078 once upon termination, thus causing spikes in the average transfer speed
11079 instead of having a smooth one. This may partially be smoothed with
11080 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
11081 recommended for better fairness.
11082
11083 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010011084 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +020011085 matched this entry. Headers are included in the count. This may be used
11086 to limit abuse of bots sucking the whole site.
11087
11088 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
11089 an integer parameter <period> which indicates in milliseconds the length
11090 of the period over which the average is measured. It reports the average
11091 outgoing bytes rate over that period, in bytes per period. It may be used
11092 to detect users which download too much and too fast. Warning: with large
11093 transfers, it is possible that the amount of transferred data will be
11094 counted once upon termination, thus causing spikes in the average
11095 transfer speed instead of having a smooth one. This may partially be
11096 smoothed with "option contstats" though this is not perfect yet. Use of
11097 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +020011098
Willy Tarreauc00cdc22010-06-06 16:48:26 +020011099 There is only one stick-table per proxy. At the moment of writing this doc,
11100 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011101 to be required, simply create a dummy backend with a stick-table in it and
11102 reference it.
11103
11104 It is important to understand that stickiness based on learning information
11105 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +010011106 lost upon restart unless peers are properly configured to transfer such
11107 information upon restart (recommended). In general it can be good as a
11108 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011109
Willy Tarreauc9705a12010-07-27 20:05:50 +020011110 Last, memory requirements may be important when storing many data types.
11111 Indeed, storing all indicators above at once in each entry requires 116 bytes
11112 per entry, or 116 MB for a 1-million entries table. This is definitely not
11113 something that can be ignored.
11114
11115 Example:
11116 # Keep track of counters of up to 1 million IP addresses over 5 minutes
11117 # and store a general purpose counter and the average connection rate
11118 # computed over a sliding window of 30 seconds.
11119 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
11120
Willy Tarreau4b103022021-02-12 17:59:10 +010011121 See also : "stick match", "stick on", "stick store-request", section 2.5
David du Colombiera13d1b92011-03-17 10:40:22 +010011122 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010011123
11124
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011125stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +010011126 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011127 May be used in sections : defaults | frontend | listen | backend
11128 no | no | yes | yes
11129
11130 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011131 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011132 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +010011133 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011134 server is selected.
11135
11136 <table> is an optional stickiness table name. If unspecified, the same
11137 backend's table is used. A stickiness table is declared using
11138 the "stick-table" statement.
11139
11140 <cond> is an optional storage condition. It makes it possible to store
11141 certain criteria only when some conditions are met (or not met).
11142 For instance, it could be used to store the SSL session ID only
11143 when the response is a SSL server hello.
11144
11145 Some protocols or applications require complex stickiness rules and cannot
11146 always simply rely on cookies nor hashing. The "stick store-response"
11147 statement describes a rule to decide what to extract from the response and
11148 when to do it, in order to store it into a stickiness table for further
11149 requests to match it using the "stick match" statement. Obviously the
11150 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011151 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011152 See section 7 for a complete list of possible patterns and transformation
11153 rules.
11154
11155 The table has to be declared using the "stick-table" statement. It must be of
11156 a type compatible with the pattern. By default it is the one which is present
11157 in the same backend. It is possible to share a table with other backends by
11158 referencing it using the "table" keyword. If another table is referenced,
11159 the server's ID inside the backends are used. By default, all server IDs
11160 start at 1 in each backend, so the server ordering is enough. But in case of
11161 doubt, it is highly recommended to force server IDs using their "id" setting.
11162
11163 It is possible to restrict the conditions where a "stick store-response"
11164 statement will apply, using "if" or "unless" followed by a condition. This
11165 condition will be evaluated while parsing the response, so any criteria can
11166 be used. See section 7 for ACL based conditions.
11167
11168 There is no limit on the number of "stick store-response" statements, but
11169 there is a limit of 8 simultaneous stores per request or response. This
11170 makes it possible to store up to 8 criteria, all extracted from either the
11171 request or the response, regardless of the number of rules. Only the 8 first
11172 ones which match will be kept. Using this, it is possible to feed multiple
11173 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010011174 another protocol or access method. Using multiple store-response rules with
11175 the same table is possible and may be used to find the best criterion to rely
11176 on, by arranging the rules by decreasing preference order. Only the first
11177 extracted criterion for a given table will be stored. All subsequent store-
11178 response rules referencing the same table will be skipped and their ACLs will
11179 not be evaluated. However, even if a store-request rule references a table, a
11180 store-response rule may also use the same table. This means that each table
11181 may learn exactly one element from the request and one element from the
11182 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011183
11184 The table will contain the real server that processed the request.
11185
11186 Example :
11187 # Learn SSL session ID from both request and response and create affinity.
11188 backend https
11189 mode tcp
11190 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +020011191 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011192 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011193
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011194 acl clienthello req_ssl_hello_type 1
11195 acl serverhello rep_ssl_hello_type 2
11196
11197 # use tcp content accepts to detects ssl client and server hello.
11198 tcp-request inspect-delay 5s
11199 tcp-request content accept if clienthello
11200
11201 # no timeout on response inspect delay by default.
11202 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011203
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011204 # SSL session ID (SSLID) may be present on a client or server hello.
11205 # Its length is coded on 1 byte at offset 43 and its value starts
11206 # at offset 44.
11207
11208 # Match and learn on request if client hello.
11209 stick on payload_lv(43,1) if clienthello
11210
11211 # Learn on response if server hello.
11212 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +020011213
Emeric Brun6a1cefa2010-09-24 18:15:17 +020011214 server s1 192.168.1.1:443
11215 server s2 192.168.1.1:443
11216
11217 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
11218 extraction.
11219
11220
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011221tcp-check comment <string>
11222 Defines a comment for the following the tcp-check rule, reported in logs if
11223 it fails.
11224 May be used in sections : defaults | frontend | listen | backend
11225 yes | no | yes | yes
11226
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011227 Arguments :
11228 <string> is the comment message to add in logs if the following tcp-check
11229 rule fails.
11230
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011231 It only works for connect, send and expect rules. It is useful to make
11232 user-friendly error reporting.
11233
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011234 See also : "option tcp-check", "tcp-check connect", "tcp-check send" and
11235 "tcp-check expect".
11236
11237
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011238tcp-check connect [default] [port <expr>] [addr <ip>] [send-proxy] [via-socks4]
11239 [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +020011240 [proto <name>] [comment <msg>]
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011241 Opens a new connection
11242 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011243 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011244
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011245 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011246 comment <msg> defines a message to report if the rule evaluation fails.
11247
Christopher Faulet4dce5922020-03-30 13:54:42 +020011248 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -040011249 checks. The server options are used only if not redefined.
Christopher Faulet4dce5922020-03-30 13:54:42 +020011250
Christopher Fauletb7d30092020-03-30 15:19:03 +020011251 port <expr> if not set, check port or server port is used.
Christopher Faulet5c288742020-03-31 08:15:58 +020011252 It tells HAProxy where to open the connection to.
11253 <port> must be a valid TCP port source integer, from 1 to
Christopher Fauletb7d30092020-03-30 15:19:03 +020011254 65535 or an sample-fetch expression.
Christopher Faulet5c288742020-03-31 08:15:58 +020011255
11256 addr <ip> defines the IP address to do the health check.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011257
11258 send-proxy send a PROXY protocol string
11259
Christopher Faulet085426a2020-03-30 13:07:02 +020011260 via-socks4 enables outgoing health checks using upstream socks4 proxy.
11261
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011262 ssl opens a ciphered connection
11263
Christopher Faulet79b31d42020-03-30 13:00:05 +020011264 sni <sni> specifies the SNI to use to do health checks over SSL.
11265
Christopher Faulet98572322020-03-30 13:16:44 +020011266 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
11267 list consists in a comma-delimited list of protocol names,
11268 for instance: "http/1.1,http/1.0" (without quotes).
11269 If it is not set, the server ALPN is used.
11270
Christopher Fauletedc6ed92020-04-23 16:27:59 +020011271 proto <name> forces the multiplexer's protocol to use for this connection.
11272 It must be a TCP mux protocol and it must be usable on the
11273 backend side. The list of available protocols is reported in
11274 haproxy -vv.
11275
Christopher Faulet5c288742020-03-31 08:15:58 +020011276 linger cleanly close the connection instead of using a single RST.
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010011277
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011278 When an application lies on more than a single TCP port or when HAProxy
11279 load-balance many services in a single backend, it makes sense to probe all
11280 the services individually before considering a server as operational.
11281
11282 When there are no TCP port configured on the server line neither server port
11283 directive, then the 'tcp-check connect port <port>' must be the first step
11284 of the sequence.
11285
11286 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
11287 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
11288 do.
11289
11290 When a connect must start the ruleset, if may still be preceded by set-var,
11291 unset-var or comment rules.
11292
11293 Examples :
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011294 # check HTTP and HTTPs services on a server.
11295 # first open port 80 thanks to server line port directive, then
11296 # tcp-check opens port 443, ciphered and run a request on it:
11297 option tcp-check
11298 tcp-check connect
11299 tcp-check send GET\ /\ HTTP/1.0\r\n
11300 tcp-check send Host:\ haproxy.1wt.eu\r\n
11301 tcp-check send \r\n
11302 tcp-check expect rstring (2..|3..)
11303 tcp-check connect port 443 ssl
11304 tcp-check send GET\ /\ HTTP/1.0\r\n
11305 tcp-check send Host:\ haproxy.1wt.eu\r\n
11306 tcp-check send \r\n
11307 tcp-check expect rstring (2..|3..)
11308 server www 10.0.0.1 check port 80
11309
11310 # check both POP and IMAP from a single server:
11311 option tcp-check
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010011312 tcp-check connect port 110 linger
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011313 tcp-check expect string +OK\ POP3\ ready
11314 tcp-check connect port 143
11315 tcp-check expect string *\ OK\ IMAP4\ ready
11316 server mail 10.0.0.1 check
11317
11318 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
11319
11320
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011321tcp-check expect [min-recv <int>] [comment <msg>]
Christopher Fauletec07e382020-04-07 14:56:26 +020011322 [ok-status <st>] [error-status <st>] [tout-status <st>]
Christopher Faulet98cc57c2020-04-01 20:52:31 +020011323 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011324 [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +010011325 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011326 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011327 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011328
11329 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011330 comment <msg> defines a message to report if the rule evaluation fails.
11331
Gaetan Rivet1afd8262020-02-07 15:37:17 +010011332 min-recv is optional and can define the minimum amount of data required to
11333 evaluate the current expect rule. If the number of received bytes
11334 is under this limit, the check will wait for more data. This
11335 option can be used to resolve some ambiguous matching rules or to
11336 avoid executing costly regex matches on content known to be still
11337 incomplete. If an exact string (string or binary) is used, the
11338 minimum between the string length and this parameter is used.
11339 This parameter is ignored if it is set to -1. If the expect rule
11340 does not match, the check will wait for more data. If set to 0,
11341 the evaluation result is always conclusive.
11342
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011343 <match> is a keyword indicating how to look for a specific pattern in the
Gaetan Rivetefab6c62020-02-07 15:37:17 +010011344 response. The keyword may be one of "string", "rstring", "binary" or
11345 "rbinary".
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011346 The keyword may be preceded by an exclamation mark ("!") to negate
11347 the match. Spaces are allowed between the exclamation mark and the
11348 keyword. See below for more details on the supported keywords.
11349
Christopher Fauletec07e382020-04-07 14:56:26 +020011350 ok-status <st> is optional and can be used to set the check status if
11351 the expect rule is successfully evaluated and if it is
11352 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011353 "L6OK" and "L4OK" are supported :
11354 - L7OK : check passed on layer 7
Christopher Faulet83662b52020-11-20 17:47:47 +010011355 - L7OKC : check conditionally passed on layer 7, set
11356 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011357 - L6OK : check passed on layer 6
11358 - L4OK : check passed on layer 4
Christopher Fauletec07e382020-04-07 14:56:26 +020011359 By default "L7OK" is used.
11360
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011361 error-status <st> is optional and can be used to set the check status if
11362 an error occurred during the expect rule evaluation.
Christopher Faulet83662b52020-11-20 17:47:47 +010011363 "L7OKC", "L7RSP", "L7STS", "L6RSP" and "L4CON" are
11364 supported :
11365 - L7OKC : check conditionally passed on layer 7, set
11366 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011367 - L7RSP : layer 7 invalid response - protocol error
11368 - L7STS : layer 7 response error, for example HTTP 5xx
11369 - L6RSP : layer 6 invalid response - protocol error
11370 - L4CON : layer 1-4 connection problem
11371 By default "L7RSP" is used.
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011372
Christopher Fauletec07e382020-04-07 14:56:26 +020011373 tout-status <st> is optional and can be used to set the check status if
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011374 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011375 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
11376 - L7TOUT : layer 7 (HTTP/SMTP) timeout
11377 - L6TOUT : layer 6 (SSL) timeout
11378 - L4TOUT : layer 1-4 timeout
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011379 By default "L7TOUT" is used.
11380
Christopher Fauletbe52b4d2020-04-01 16:30:22 +020011381 on-success <fmt> is optional and can be used to customize the
11382 informational message reported in logs if the expect
11383 rule is successfully evaluated and if it is the last rule
11384 in the tcp-check ruleset. <fmt> is a log-format string.
11385
11386 on-error <fmt> is optional and can be used to customize the
11387 informational message reported in logs if an error
11388 occurred during the expect rule evaluation. <fmt> is a
11389 log-format string.
11390
Christopher Faulet98cc57c2020-04-01 20:52:31 +020011391 status-code <expr> is optional and can be used to set the check status code
11392 reported in logs, on success or on error. <expr> is a
11393 standard HAProxy expression formed by a sample-fetch
11394 followed by some converters.
11395
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011396 <pattern> is the pattern to look for. It may be a string or a regular
11397 expression. If the pattern contains spaces, they must be escaped
11398 with the usual backslash ('\').
11399 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +010011400 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011401 two digits will represent a byte. The hexadecimal digits may be
11402 used upper or lower case.
11403
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011404 The available matches are intentionally similar to their http-check cousins :
11405
11406 string <string> : test the exact string matches in the response buffer.
11407 A health check response will be considered valid if the
11408 response's buffer contains this exact string. If the
11409 "string" keyword is prefixed with "!", then the response
11410 will be considered invalid if the body contains this
11411 string. This can be used to look for a mandatory pattern
11412 in a protocol response, or to detect a failure when a
11413 specific error appears in a protocol banner.
11414
11415 rstring <regex> : test a regular expression on the response buffer.
11416 A health check response will be considered valid if the
11417 response's buffer matches this expression. If the
11418 "rstring" keyword is prefixed with "!", then the response
11419 will be considered invalid if the body matches the
11420 expression.
11421
Christopher Fauletaaab0832020-05-05 15:54:22 +020011422 string-lf <fmt> : test a log-format string match in the response's buffer.
11423 A health check response will be considered valid if the
11424 response's buffer contains the string resulting of the
11425 evaluation of <fmt>, which follows the log-format rules.
11426 If prefixed with "!", then the response will be
11427 considered invalid if the buffer contains the string.
11428
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011429 binary <hexstring> : test the exact string in its hexadecimal form matches
11430 in the response buffer. A health check response will
11431 be considered valid if the response's buffer contains
11432 this exact hexadecimal string.
11433 Purpose is to match data on binary protocols.
11434
Gaetan Rivetefab6c62020-02-07 15:37:17 +010011435 rbinary <regex> : test a regular expression on the response buffer, like
11436 "rstring". However, the response buffer is transformed
11437 into its hexadecimal form, including NUL-bytes. This
11438 allows using all regex engines to match any binary
11439 content. The hexadecimal transformation takes twice the
11440 size of the original response. As such, the expected
11441 pattern should work on at-most half the response buffer
11442 size.
11443
Christopher Fauletaaab0832020-05-05 15:54:22 +020011444 binary-lf <hexfmt> : test a log-format string in its hexadecimal form
11445 match in the response's buffer. A health check response
11446 will be considered valid if the response's buffer
11447 contains the hexadecimal string resulting of the
11448 evaluation of <fmt>, which follows the log-format
11449 rules. If prefixed with "!", then the response will be
11450 considered invalid if the buffer contains the
11451 hexadecimal string. The hexadecimal string is converted
11452 in a binary string before matching the response's
11453 buffer.
11454
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011455 It is important to note that the responses will be limited to a certain size
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011456 defined by the global "tune.bufsize" option, which defaults to 16384 bytes.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011457 Thus, too large responses may not contain the mandatory pattern when using
11458 "string", "rstring" or binary. If a large response is absolutely required, it
11459 is possible to change the default max size by setting the global variable.
11460 However, it is worth keeping in mind that parsing very large responses can
11461 waste some CPU cycles, especially when regular expressions are used, and that
11462 it is always better to focus the checks on smaller resources. Also, in its
11463 current state, the check will not find any string nor regex past a null
11464 character in the response. Similarly it is not possible to request matching
11465 the null character.
11466
11467 Examples :
11468 # perform a POP check
11469 option tcp-check
11470 tcp-check expect string +OK\ POP3\ ready
11471
11472 # perform an IMAP check
11473 option tcp-check
11474 tcp-check expect string *\ OK\ IMAP4\ ready
11475
11476 # look for the redis master server
11477 option tcp-check
11478 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +020011479 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011480 tcp-check send info\ replication\r\n
11481 tcp-check expect string role:master
11482 tcp-check send QUIT\r\n
11483 tcp-check expect string +OK
11484
11485
11486 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011487 "tcp-check send-binary", "http-check expect", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011488
11489
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011490tcp-check send <data> [comment <msg>]
11491tcp-check send-lf <fmt> [comment <msg>]
11492 Specify a string or a log-format string to be sent as a question during a
11493 generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011494 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011495 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011496
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011497 Arguments :
11498 comment <msg> defines a message to report if the rule evaluation fails.
11499
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011500 <data> is the string that will be sent during a generic health
11501 check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020011502
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011503 <fmt> is the log-format string that will be sent, once evaluated,
11504 during a generic health check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011505
11506 Examples :
11507 # look for the redis master server
11508 option tcp-check
11509 tcp-check send info\ replication\r\n
11510 tcp-check expect string role:master
11511
11512 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011513 "tcp-check send-binary", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011514
11515
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011516tcp-check send-binary <hexstring> [comment <msg>]
11517tcp-check send-binary-lf <hexfmt> [comment <msg>]
11518 Specify an hex digits string or an hex digits log-format string to be sent as
11519 a binary question during a raw tcp health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011520 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011521 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011522
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011523 Arguments :
11524 comment <msg> defines a message to report if the rule evaluation fails.
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011525
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011526 <hexstring> is the hexadecimal string that will be send, once converted
11527 to binary, during a generic health check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020011528
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011529 <hexfmt> is the hexadecimal log-format string that will be send, once
11530 evaluated and converted to binary, during a generic health
11531 check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011532
11533 Examples :
11534 # redis check in binary
11535 option tcp-check
11536 tcp-check send-binary 50494e470d0a # PING\r\n
11537 tcp-check expect binary 2b504F4e47 # +PONG
11538
11539
11540 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011541 "tcp-check send", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011542
11543
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011544tcp-check set-var(<var-name>) <expr>
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011545 This operation sets the content of a variable. The variable is declared inline.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011546 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011547 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011548
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011549 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011550 <var-name> The name of the variable starts with an indication about its
11551 scope. The scopes allowed for tcp-check are:
11552 "proc" : the variable is shared with the whole process.
11553 "sess" : the variable is shared with the tcp-check session.
11554 "check": the variable is declared for the lifetime of the tcp-check.
11555 This prefix is followed by a name. The separator is a '.'.
11556 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11557 and '-'.
11558
11559 <expr> Is a sample-fetch expression potentially followed by converters.
11560
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011561 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011562 tcp-check set-var(check.port) int(1234)
11563
11564
11565tcp-check unset-var(<var-name>)
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011566 Free a reference to a variable within its scope.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011567 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011568 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011569
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011570 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011571 <var-name> The name of the variable starts with an indication about its
11572 scope. The scopes allowed for tcp-check are:
11573 "proc" : the variable is shared with the whole process.
11574 "sess" : the variable is shared with the tcp-check session.
11575 "check": the variable is declared for the lifetime of the tcp-check.
11576 This prefix is followed by a name. The separator is a '.'.
11577 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11578 and '-'.
11579
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011580 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011581 tcp-check unset-var(check.port)
11582
11583
Willy Tarreaue9656522010-08-17 15:40:09 +020011584tcp-request connection <action> [{if | unless} <condition>]
11585 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +020011586 May be used in sections : defaults | frontend | listen | backend
11587 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +020011588 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011589 <action> defines the action to perform if the condition applies. See
11590 below.
Willy Tarreau1a687942010-05-23 22:40:30 +020011591
Willy Tarreaue9656522010-08-17 15:40:09 +020011592 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011593
11594 Immediately after acceptance of a new incoming connection, it is possible to
11595 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +020011596 or dropped or have its counters tracked. Those conditions cannot make use of
11597 any data contents because the connection has not been read from yet, and the
11598 buffers are not yet allocated. This is used to selectively and very quickly
11599 accept or drop connections from various sources with a very low overhead. If
11600 some contents need to be inspected in order to take the decision, the
11601 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011602
Willy Tarreaue9656522010-08-17 15:40:09 +020011603 The "tcp-request connection" rules are evaluated in their exact declaration
11604 order. If no rule matches or if there is no rule, the default action is to
11605 accept the incoming connection. There is no specific limit to the number of
11606 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011607
Willy Tarreaua9083d02015-05-08 15:27:59 +020011608 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +020011609 - accept :
11610 accepts the connection if the condition is true (when used with "if")
11611 or false (when used with "unless"). The first such rule executed ends
11612 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011613
Willy Tarreaue9656522010-08-17 15:40:09 +020011614 - reject :
11615 rejects the connection if the condition is true (when used with "if")
11616 or false (when used with "unless"). The first such rule executed ends
11617 the rules evaluation. Rejected connections do not even become a
11618 session, which is why they are accounted separately for in the stats,
11619 as "denied connections". They are not considered for the session
11620 rate-limit and are not logged either. The reason is that these rules
11621 should only be used to filter extremely high connection rates such as
11622 the ones encountered during a massive DDoS attack. Under these extreme
11623 conditions, the simple action of logging each event would make the
11624 system collapse and would considerably lower the filtering capacity. If
11625 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +020011626 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011627
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011628 - expect-proxy layer4 :
11629 configures the client-facing connection to receive a PROXY protocol
11630 header before any byte is read from the socket. This is equivalent to
11631 having the "accept-proxy" keyword on the "bind" line, except that using
11632 the TCP rule allows the PROXY protocol to be accepted only for certain
11633 IP address ranges using an ACL. This is convenient when multiple layers
11634 of load balancers are passed through by traffic coming from public
11635 hosts.
11636
Bertrand Jacquin90759682016-06-06 15:35:39 +010011637 - expect-netscaler-cip layer4 :
11638 configures the client-facing connection to receive a NetScaler Client
11639 IP insertion protocol header before any byte is read from the socket.
11640 This is equivalent to having the "accept-netscaler-cip" keyword on the
11641 "bind" line, except that using the TCP rule allows the PROXY protocol
11642 to be accepted only for certain IP address ranges using an ACL. This
11643 is convenient when multiple layers of load balancers are passed
11644 through by traffic coming from public hosts.
11645
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011646 - capture <sample> len <length> :
11647 This only applies to "tcp-request content" rules. It captures sample
11648 expression <sample> from the request buffer, and converts it to a
11649 string of at most <len> characters. The resulting string is stored into
11650 the next request "capture" slot, so it will possibly appear next to
11651 some captured HTTP headers. It will then automatically appear in the
11652 logs, and it will be possible to extract it using sample fetch rules to
11653 feed it into headers or anything. The length should be limited given
11654 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +020011655 session life. Please check section 7.3 (Fetching samples) and "capture
11656 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011657
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011658 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +020011659 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011660 rules do not stop evaluation and do not change default action. The
11661 number of counters that may be simultaneously tracked by the same
11662 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -050011663 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Matteo Contrini1857b8c2020-10-16 17:35:54 +020011664 and (MAX_SESS_STKCTR-1). The first "track-sc0" rule executed enables
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011665 tracking of the counters of the specified table as the first set. The
11666 first "track-sc1" rule executed enables tracking of the counters of the
11667 specified table as the second set. The first "track-sc2" rule executed
11668 enables tracking of the counters of the specified table as the third
11669 set. It is a recommended practice to use the first set of counters for
11670 the per-frontend counters and the second set for the per-backend ones.
11671 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011672
Willy Tarreaue9656522010-08-17 15:40:09 +020011673 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011674 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +020011675 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +010011676 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011677 and used to select which table entry to update the counters.
11678 Note that "tcp-request connection" cannot use content-based
11679 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011680
Willy Tarreaue9656522010-08-17 15:40:09 +020011681 <table> is an optional table to be used instead of the default one,
11682 which is the stick-table declared in the current proxy. All
11683 the counters for the matches and updates for the key will
11684 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011685
Willy Tarreaue9656522010-08-17 15:40:09 +020011686 Once a "track-sc*" rule is executed, the key is looked up in the table
11687 and if it is not found, an entry is allocated for it. Then a pointer to
11688 that entry is kept during all the session's life, and this entry's
11689 counters are updated as often as possible, every time the session's
11690 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011691 Counters are only updated for events that happen after the tracking has
11692 been started. For example, connection counters will not be updated when
11693 tracking layer 7 information, since the connection event happens before
11694 layer7 information is extracted.
11695
Willy Tarreaue9656522010-08-17 15:40:09 +020011696 If the entry tracks concurrent connection counters, one connection is
11697 counted for as long as the entry is tracked, and the entry will not
11698 expire during that time. Tracking counters also provides a performance
11699 advantage over just checking the keys, because only one table lookup is
11700 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011701
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011702 - sc-inc-gpc0(<sc-id>):
11703 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
11704 counter designated by <sc-id>. If an error occurs, this action silently
11705 fails and the actions evaluation continues.
11706
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011707 - sc-inc-gpc1(<sc-id>):
11708 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
11709 counter designated by <sc-id>. If an error occurs, this action silently
11710 fails and the actions evaluation continues.
11711
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011712 - sc-set-gpt0(<sc-id>) { <int> | <expr> }:
11713 This action sets the 32-bit unsigned GPT0 tag according to the sticky
11714 counter designated by <sc-id> and the value of <int>/<expr>. The
11715 expected result is a boolean. If an error occurs, this action silently
11716 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011717
William Lallemand2e785f22016-05-25 01:48:42 +020011718 - set-src <expr> :
11719 Is used to set the source IP address to the value of specified
11720 expression. Useful if you want to mask source IP for privacy.
11721 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011722 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +020011723
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011724 Arguments:
11725 <expr> Is a standard HAProxy expression formed by a sample-fetch
11726 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +020011727
11728 Example:
William Lallemand2e785f22016-05-25 01:48:42 +020011729 tcp-request connection set-src src,ipmask(24)
11730
Willy Tarreau0c630532016-10-21 17:52:58 +020011731 When possible, set-src preserves the original source port as long as the
11732 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +020011733
William Lallemand44be6402016-05-25 01:51:35 +020011734 - set-src-port <expr> :
11735 Is used to set the source port address to the value of specified
11736 expression.
11737
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011738 Arguments:
11739 <expr> Is a standard HAProxy expression formed by a sample-fetch
11740 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +020011741
11742 Example:
William Lallemand44be6402016-05-25 01:51:35 +020011743 tcp-request connection set-src-port int(4000)
11744
Willy Tarreau0c630532016-10-21 17:52:58 +020011745 When possible, set-src-port preserves the original source address as long
11746 as the address family supports a port, otherwise it forces the source
11747 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +020011748
William Lallemand13e9b0c2016-05-25 02:34:07 +020011749 - set-dst <expr> :
11750 Is used to set the destination IP address to the value of specified
11751 expression. Useful if you want to mask IP for privacy in log.
11752 If you want to provide an IP from a HTTP header use "http-request
11753 set-dst". If you want to connect to the new address/port, use
11754 '0.0.0.0:0' as a server address in the backend.
11755
11756 <expr> Is a standard HAProxy expression formed by a sample-fetch
11757 followed by some converters.
11758
11759 Example:
11760
11761 tcp-request connection set-dst dst,ipmask(24)
11762 tcp-request connection set-dst ipv4(10.0.0.1)
11763
Willy Tarreau0c630532016-10-21 17:52:58 +020011764 When possible, set-dst preserves the original destination port as long as
11765 the address family allows it, otherwise the destination port is set to 0.
11766
William Lallemand13e9b0c2016-05-25 02:34:07 +020011767 - set-dst-port <expr> :
11768 Is used to set the destination port address to the value of specified
11769 expression. If you want to connect to the new address/port, use
11770 '0.0.0.0:0' as a server address in the backend.
11771
11772
11773 <expr> Is a standard HAProxy expression formed by a sample-fetch
11774 followed by some converters.
11775
11776 Example:
11777
11778 tcp-request connection set-dst-port int(4000)
11779
Willy Tarreau0c630532016-10-21 17:52:58 +020011780 When possible, set-dst-port preserves the original destination address as
11781 long as the address family supports a port, otherwise it forces the
11782 destination address to IPv4 "0.0.0.0" before rewriting the port.
11783
Willy Tarreau2d392c22015-08-24 01:43:45 +020011784 - "silent-drop" :
11785 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011786 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020011787 to prevent the client from being notified. The effect it then that the
11788 client still sees an established connection while there's none on
11789 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
11790 except that it doesn't use any local resource at all on the machine
11791 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010011792 slow down stronger attackers. It is important to understand the impact
11793 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011794 client and HAProxy (firewalls, proxies, load balancers) will also keep
11795 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011796 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011797 TCP_REPAIR socket option is used to block the emission of a TCP
11798 reset. On other systems, the socket's TTL is reduced to 1 so that the
11799 TCP reset doesn't pass the first router, though it's still delivered to
11800 local networks. Do not use it unless you fully understand how it works.
11801
Willy Tarreaue9656522010-08-17 15:40:09 +020011802 Note that the "if/unless" condition is optional. If no condition is set on
11803 the action, it is simply performed unconditionally. That can be useful for
11804 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011805
Willy Tarreaue9656522010-08-17 15:40:09 +020011806 Example: accept all connections from white-listed hosts, reject too fast
11807 connection without counting them, and track accepted connections.
11808 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011809
Willy Tarreaue9656522010-08-17 15:40:09 +020011810 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011811 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011812 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011813
Willy Tarreaue9656522010-08-17 15:40:09 +020011814 Example: accept all connections from white-listed hosts, count all other
11815 connections and reject too fast ones. This results in abusive ones
11816 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011817
Willy Tarreaue9656522010-08-17 15:40:09 +020011818 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011819 tcp-request connection track-sc0 src
11820 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011821
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011822 Example: enable the PROXY protocol for traffic coming from all known proxies.
11823
11824 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
11825
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011826 See section 7 about ACL usage.
11827
Willy Tarreau4f614292016-10-21 17:49:36 +020011828 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011829
11830
Willy Tarreaue9656522010-08-17 15:40:09 +020011831tcp-request content <action> [{if | unless} <condition>]
11832 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +020011833 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020011834 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +020011835 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011836 <action> defines the action to perform if the condition applies. See
11837 below.
Willy Tarreau62644772008-07-16 18:36:06 +020011838
Willy Tarreaue9656522010-08-17 15:40:09 +020011839 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +020011840
Davor Ocelice9ed2812017-12-25 17:49:28 +010011841 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +020011842 called "TCP content inspection". During this stage, ACL-based rules are
11843 evaluated every time the request contents are updated, until either an
Christopher Fauletae863c62021-03-15 12:03:44 +010011844 "accept", a "reject" or a "switch-mode" rule matches, or the TCP request
11845 inspection delay expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +020011846
Willy Tarreaue9656522010-08-17 15:40:09 +020011847 The first difference between these rules and "tcp-request connection" rules
11848 is that "tcp-request content" rules can make use of contents to take a
11849 decision. Most often, these decisions will consider a protocol recognition or
11850 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +010011851 both frontends and backends. In case of HTTP keep-alive with the client, all
11852 tcp-request content rules are evaluated again, so haproxy keeps a record of
11853 what sticky counters were assigned by a "tcp-request connection" versus a
11854 "tcp-request content" rule, and flushes all the content-related ones after
11855 processing an HTTP request, so that they may be evaluated again by the rules
11856 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011857 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +010011858 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +020011859
Willy Tarreaue9656522010-08-17 15:40:09 +020011860 Content-based rules are evaluated in their exact declaration order. If no
11861 rule matches or if there is no rule, the default action is to accept the
11862 contents. There is no specific limit to the number of rules which may be
11863 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +020011864
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011865 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011866 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +010011867 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011868 - reject : the request is rejected and the connection is closed
11869 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -040011870 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011871 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011872 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011873 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011874 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011875 - set-dst <expr>
11876 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011877 - set-var(<var-name>) <expr>
Christopher Fauletae863c62021-03-15 12:03:44 +010011878 - switch-mode http [ proto <name> ]
Christopher Faulet85d79c92016-11-09 16:54:56 +010011879 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +020011880 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +010011881 - send-spoe-group <engine-name> <group-name>
Christopher Faulet579d83b2019-11-22 15:34:17 +010011882 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +020011883
Willy Tarreaue9656522010-08-17 15:40:09 +020011884 They have the same meaning as their counter-parts in "tcp-request connection"
11885 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +010011886 For "do-resolve" action, please check the "http-request do-resolve"
11887 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +020011888
Willy Tarreauf3338342014-01-28 21:40:28 +010011889 While there is nothing mandatory about it, it is recommended to use the
11890 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
11891 content" rules in the frontend, and track-sc2 for "tcp-request content"
11892 rules in the backend, because that makes the configuration more readable
11893 and easier to troubleshoot, but this is just a guideline and all counters
11894 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +020011895
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011896 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +020011897 the action, it is simply performed unconditionally. That can be useful for
11898 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +020011899
Christopher Faulet2079a4a2020-10-02 11:48:57 +020011900 Note also that it is recommended to use a "tcp-request session" rule to track
11901 information that does *not* depend on Layer 7 contents, especially for HTTP
11902 frontends. Some HTTP processing are performed at the session level and may
11903 lead to an early rejection of the requests. Thus, the tracking at the content
11904 level may be disturbed in such case. A warning is emitted during startup to
11905 prevent, as far as possible, such unreliable usage.
11906
Willy Tarreaue9656522010-08-17 15:40:09 +020011907 It is perfectly possible to match layer 7 contents with "tcp-request content"
Christopher Faulet7ea509e2020-10-02 11:38:46 +020011908 rules from a TCP proxy, since HTTP-specific ACL matches are able to
11909 preliminarily parse the contents of a buffer before extracting the required
11910 data. If the buffered contents do not parse as a valid HTTP message, then the
11911 ACL does not match. The parser which is involved there is exactly the same
11912 as for all other HTTP processing, so there is no risk of parsing something
11913 differently. In an HTTP frontend or an HTTP backend, it is guaranteed that
11914 HTTP contents will always be immediately present when the rule is evaluated
11915 first because the HTTP parsing is performed in the early stages of the
11916 connection processing, at the session level. But for such proxies, using
11917 "http-request" rules is much more natural and recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020011918
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011919 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011920 are present when the rule is processed. The rule processing engine is able to
11921 wait until the inspect delay expires when the data to be tracked is not yet
11922 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011923
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011924 The "set-dst" and "set-dst-port" are used to set respectively the destination
11925 IP and port. More information on how to use it at "http-request set-dst".
11926
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011927 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +020011928 declared inline. For "tcp-request session" rules, only session-level
11929 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011930
Daniel Schneller0b547052016-03-21 20:46:57 +010011931 <var-name> The name of the variable starts with an indication about
11932 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010011933 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010011934 "sess" : the variable is shared with the whole session
11935 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011936 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010011937 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011938 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010011939 "res" : the variable is shared only during response
11940 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011941 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010011942 The name may only contain characters 'a-z', 'A-Z', '0-9',
11943 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011944
11945 <expr> Is a standard HAProxy expression formed by a sample-fetch
11946 followed by some converters.
11947
Christopher Fauletae863c62021-03-15 12:03:44 +010011948 The "switch-mode" is used to perform a conntection upgrade. Only HTTP
11949 upgrades are supported for now. The protocol may optionally be
11950 specified. This action is only available for a proxy with the frontend
11951 capability. The connection upgrade is immediately performed, following
11952 "tcp-request content" rules are not evaluated. This upgrade method should be
11953 preferred to the implicit one consisting to rely on the backend mode. When
11954 used, it is possible to set HTTP directives in a frontend without any
11955 warning. These directives will be conditionnaly evaluated if the HTTP upgrade
11956 is performed. However, an HTTP backend must still be selected. It remains
11957 unsupported to route an HTTP connection (upgraded or not) to a TCP server.
11958
Christopher Faulet4d37e532021-03-26 14:44:00 +010011959 See section 4 about Proxies for more details on HTTP upgrades.
11960
Christopher Faulet85d79c92016-11-09 16:54:56 +010011961 The "unset-var" is used to unset a variable. See above for details about
11962 <var-name>.
11963
Patrick Hemmer268a7072018-05-11 12:52:31 -040011964 The "set-priority-class" is used to set the queue priority class of the
11965 current request. The value must be a sample expression which converts to an
11966 integer in the range -2047..2047. Results outside this range will be
11967 truncated. The priority class determines the order in which queued requests
11968 are processed. Lower values have higher priority.
11969
11970 The "set-priority-offset" is used to set the queue priority timestamp offset
11971 of the current request. The value must be a sample expression which converts
11972 to an integer in the range -524287..524287. Results outside this range will be
11973 truncated. When a request is queued, it is ordered first by the priority
11974 class, then by the current timestamp adjusted by the given offset in
11975 milliseconds. Lower values have higher priority.
11976 Note that the resulting timestamp is is only tracked with enough precision for
11977 524,287ms (8m44s287ms). If the request is queued long enough to where the
11978 adjusted timestamp exceeds this value, it will be misidentified as highest
11979 priority. Thus it is important to set "timeout queue" to a value, where when
11980 combined with the offset, does not exceed this limit.
11981
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011982 The "send-spoe-group" is used to trigger sending of a group of SPOE
11983 messages. To do so, the SPOE engine used to send messages must be defined, as
11984 well as the SPOE group to send. Of course, the SPOE engine must refer to an
11985 existing SPOE filter. If not engine name is provided on the SPOE filter line,
11986 the SPOE agent name must be used.
11987
11988 <engine-name> The SPOE engine name.
11989
11990 <group-name> The SPOE group name as specified in the engine configuration.
11991
Christopher Faulet579d83b2019-11-22 15:34:17 +010011992 The "use-service" is used to executes a TCP service which will reply to the
11993 request and stop the evaluation of the rules. This service may choose to
11994 reply by sending any valid response or it may immediately close the
11995 connection without sending anything. Outside natives services, it is possible
11996 to write your own services in Lua. No further "tcp-request" rules are
11997 evaluated.
11998
11999 Example:
12000 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
12001
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012002 Example:
12003
12004 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +010012005 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012006
Willy Tarreau62644772008-07-16 18:36:06 +020012007 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +020012008 # Accept HTTP requests containing a Host header saying "example.com"
Christopher Fauletae863c62021-03-15 12:03:44 +010012009 # and reject everything else. (Only works for HTTP/1 connections)
Willy Tarreaue9656522010-08-17 15:40:09 +020012010 acl is_host_com hdr(Host) -i example.com
12011 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +020012012 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +020012013 tcp-request content reject
12014
Christopher Fauletae863c62021-03-15 12:03:44 +010012015 # Accept HTTP requests containing a Host header saying "example.com"
12016 # and reject everything else. (works for HTTP/1 and HTTP/2 connections)
12017 acl is_host_com hdr(Host) -i example.com
12018 tcp-request inspect-delay 5s
12019 tcp-request switch-mode http if HTTP
12020 tcp-request reject # non-HTTP traffic is implicit here
12021 ...
12022 http-request reject unless is_host_com
12023
Willy Tarreaue9656522010-08-17 15:40:09 +020012024 Example:
Willy Tarreau62644772008-07-16 18:36:06 +020012025 # reject SMTP connection if client speaks first
12026 tcp-request inspect-delay 30s
12027 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012028 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +020012029
12030 # Forward HTTPS connection only if client speaks
12031 tcp-request inspect-delay 30s
12032 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020012033 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +020012034 tcp-request content reject
12035
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012036 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012037 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012038 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020012039 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012040 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
12041 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012042
12043 Example:
12044 # track request counts per "base" (concatenation of Host+URL)
12045 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020012046 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010012047
Willy Tarreaue9656522010-08-17 15:40:09 +020012048 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012049 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +020012050
12051 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +010012052 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +020012053 # protecting all our sites
12054 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012055 tcp-request connection track-sc0 src
12056 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +020012057 ...
12058 use_backend http_dynamic if { path_end .php }
12059
12060 backend http_dynamic
12061 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012062 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +020012063 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012064 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030012065 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020012066 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +020012067 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +020012068
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012069 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +020012070
Jarno Huuskonen95b012b2017-04-06 13:59:14 +030012071 See also : "tcp-request connection", "tcp-request session",
12072 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +020012073
12074
12075tcp-request inspect-delay <timeout>
12076 Set the maximum allowed time to wait for data during content inspection
12077 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020012078 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +020012079 Arguments :
12080 <timeout> is the timeout value specified in milliseconds by default, but
12081 can be in any other unit if the number is suffixed by the unit,
12082 as explained at the top of this document.
12083
12084 People using haproxy primarily as a TCP relay are often worried about the
12085 risk of passing any type of protocol to a server without any analysis. In
12086 order to be able to analyze the request contents, we must first withhold
12087 the data then analyze them. This statement simply enables withholding of
12088 data for at most the specified amount of time.
12089
Willy Tarreaufb356202010-08-03 14:02:05 +020012090 TCP content inspection applies very early when a connection reaches a
12091 frontend, then very early when the connection is forwarded to a backend. This
12092 means that a connection may experience a first delay in the frontend and a
12093 second delay in the backend if both have tcp-request rules.
12094
Willy Tarreau62644772008-07-16 18:36:06 +020012095 Note that when performing content inspection, haproxy will evaluate the whole
12096 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012097 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +020012098 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +010012099 contents are definitive. If no delay is set, haproxy will not wait at all
12100 and will immediately apply a verdict based on the available information.
12101 Obviously this is unlikely to be very useful and might even be racy, so such
12102 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020012103
12104 As soon as a rule matches, the request is released and continues as usual. If
12105 the timeout is reached and no rule matches, the default policy will be to let
12106 it pass through unaffected.
12107
12108 For most protocols, it is enough to set it to a few seconds, as most clients
12109 send the full request immediately upon connection. Add 3 or more seconds to
12110 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +010012111 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +010012112 before the server (e.g. SMTP), or to wait for a client to talk before passing
12113 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +020012114 least the inspection delay, otherwise it will expire first. If the client
12115 closes the connection or if the buffer is full, the delay immediately expires
12116 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +020012117
Willy Tarreau55165fe2009-05-10 12:02:55 +020012118 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020012119 "timeout client".
12120
12121
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012122tcp-response content <action> [{if | unless} <condition>]
12123 Perform an action on a session response depending on a layer 4-7 condition
12124 May be used in sections : defaults | frontend | listen | backend
12125 no | no | yes | yes
12126 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020012127 <action> defines the action to perform if the condition applies. See
12128 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012129
12130 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
12131
Davor Ocelice9ed2812017-12-25 17:49:28 +010012132 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012133 called "TCP content inspection". During this stage, ACL-based rules are
12134 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020012135 "accept", "close" or a "reject" rule matches, or a TCP response inspection
12136 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012137
12138 Most often, these decisions will consider a protocol recognition or validity.
12139
12140 Content-based rules are evaluated in their exact declaration order. If no
12141 rule matches or if there is no rule, the default action is to accept the
12142 contents. There is no specific limit to the number of rules which may be
12143 inserted.
12144
Thierry FOURNIER236657b2015-08-19 08:25:14 +020012145 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012146 - accept :
12147 accepts the response if the condition is true (when used with "if")
12148 or false (when used with "unless"). The first such rule executed ends
12149 the rules evaluation.
12150
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020012151 - close :
12152 immediately closes the connection with the server if the condition is
12153 true (when used with "if"), or false (when used with "unless"). The
12154 first such rule executed ends the rules evaluation. The main purpose of
12155 this action is to force a connection to be finished between a client
12156 and a server after an exchange when the application protocol expects
12157 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012158 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020012159 protocols.
12160
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012161 - reject :
12162 rejects the response if the condition is true (when used with "if")
12163 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012164 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012165
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012166 - set-var(<var-name>) <expr>
12167 Sets a variable.
12168
Christopher Faulet85d79c92016-11-09 16:54:56 +010012169 - unset-var(<var-name>)
12170 Unsets a variable.
12171
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020012172 - sc-inc-gpc0(<sc-id>):
12173 This action increments the GPC0 counter according to the sticky
12174 counter designated by <sc-id>. If an error occurs, this action fails
12175 silently and the actions evaluation continues.
12176
Frédéric Lécaille6778b272018-01-29 15:22:53 +010012177 - sc-inc-gpc1(<sc-id>):
12178 This action increments the GPC1 counter according to the sticky
12179 counter designated by <sc-id>. If an error occurs, this action fails
12180 silently and the actions evaluation continues.
12181
Cédric Dufour0d7712d2019-11-06 18:38:53 +010012182 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
12183 This action sets the 32-bit unsigned GPT0 tag according to the sticky
12184 counter designated by <sc-id> and the value of <int>/<expr>. The
12185 expected result is a boolean. If an error occurs, this action silently
12186 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020012187
Willy Tarreau2d392c22015-08-24 01:43:45 +020012188 - "silent-drop" :
12189 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012190 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020012191 to prevent the client from being notified. The effect it then that the
12192 client still sees an established connection while there's none on
12193 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
12194 except that it doesn't use any local resource at all on the machine
12195 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010012196 slow down stronger attackers. It is important to understand the impact
12197 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020012198 client and HAProxy (firewalls, proxies, load balancers) will also keep
12199 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012200 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020012201 TCP_REPAIR socket option is used to block the emission of a TCP
12202 reset. On other systems, the socket's TTL is reduced to 1 so that the
12203 TCP reset doesn't pass the first router, though it's still delivered to
12204 local networks. Do not use it unless you fully understand how it works.
12205
Christopher Faulet76c09ef2017-09-21 11:03:52 +020012206 - send-spoe-group <engine-name> <group-name>
12207 Send a group of SPOE messages.
12208
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012209 Note that the "if/unless" condition is optional. If no condition is set on
12210 the action, it is simply performed unconditionally. That can be useful for
12211 for changing the default action to a reject.
12212
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012213 It is perfectly possible to match layer 7 contents with "tcp-response
12214 content" rules, but then it is important to ensure that a full response has
12215 been buffered, otherwise no contents will match. In order to achieve this,
12216 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012217 period.
12218
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012219 The "set-var" is used to set the content of a variable. The variable is
12220 declared inline.
12221
Daniel Schneller0b547052016-03-21 20:46:57 +010012222 <var-name> The name of the variable starts with an indication about
12223 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010012224 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010012225 "sess" : the variable is shared with the whole session
12226 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012227 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010012228 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012229 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010012230 "res" : the variable is shared only during response
12231 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012232 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010012233 The name may only contain characters 'a-z', 'A-Z', '0-9',
12234 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020012235
12236 <expr> Is a standard HAProxy expression formed by a sample-fetch
12237 followed by some converters.
12238
12239 Example:
12240
12241 tcp-request content set-var(sess.my_var) src
12242
Christopher Faulet85d79c92016-11-09 16:54:56 +010012243 The "unset-var" is used to unset a variable. See above for details about
12244 <var-name>.
12245
12246 Example:
12247
12248 tcp-request content unset-var(sess.my_var)
12249
Christopher Faulet76c09ef2017-09-21 11:03:52 +020012250 The "send-spoe-group" is used to trigger sending of a group of SPOE
12251 messages. To do so, the SPOE engine used to send messages must be defined, as
12252 well as the SPOE group to send. Of course, the SPOE engine must refer to an
12253 existing SPOE filter. If not engine name is provided on the SPOE filter line,
12254 the SPOE agent name must be used.
12255
12256 <engine-name> The SPOE engine name.
12257
12258 <group-name> The SPOE group name as specified in the engine configuration.
12259
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012260 See section 7 about ACL usage.
12261
12262 See also : "tcp-request content", "tcp-response inspect-delay"
12263
12264
Willy Tarreau4f614292016-10-21 17:49:36 +020012265tcp-request session <action> [{if | unless} <condition>]
12266 Perform an action on a validated session depending on a layer 5 condition
12267 May be used in sections : defaults | frontend | listen | backend
12268 no | yes | yes | no
12269 Arguments :
12270 <action> defines the action to perform if the condition applies. See
12271 below.
12272
12273 <condition> is a standard layer5-only ACL-based condition (see section 7).
12274
Davor Ocelice9ed2812017-12-25 17:49:28 +010012275 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020012276 it is possible to evaluate some conditions to decide whether this session
12277 must be accepted or dropped or have its counters tracked. Those conditions
12278 cannot make use of any data contents because no buffers are allocated yet and
12279 the processing cannot wait at this stage. The main use case it to copy some
12280 early information into variables (since variables are accessible in the
12281 session), or to keep track of some information collected after the handshake,
12282 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010012283 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020012284 extracted information can thus be copied to a variable or tracked using
12285 "track-sc" rules. Of course it is also possible to decide to accept/reject as
12286 with other rulesets. Most operations performed here could also be performed
12287 in "tcp-request content" rules, except that in HTTP these rules are evaluated
12288 for each new request, and that might not always be acceptable. For example a
12289 rule might increment a counter on each evaluation. It would also be possible
12290 that a country is resolved by geolocation from the source IP address,
12291 assigned to a session-wide variable, then the source address rewritten from
12292 an HTTP header for all requests. If some contents need to be inspected in
12293 order to take the decision, the "tcp-request content" statements must be used
12294 instead.
12295
12296 The "tcp-request session" rules are evaluated in their exact declaration
12297 order. If no rule matches or if there is no rule, the default action is to
12298 accept the incoming session. There is no specific limit to the number of
12299 rules which may be inserted.
12300
12301 Several types of actions are supported :
12302 - accept : the request is accepted
12303 - reject : the request is rejected and the connection is closed
12304 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
12305 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010012306 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010012307 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Willy Tarreau4f614292016-10-21 17:49:36 +020012308 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010012309 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020012310 - silent-drop
12311
12312 These actions have the same meaning as their respective counter-parts in
12313 "tcp-request connection" and "tcp-request content", so please refer to these
12314 sections for a complete description.
12315
12316 Note that the "if/unless" condition is optional. If no condition is set on
12317 the action, it is simply performed unconditionally. That can be useful for
12318 "track-sc*" actions as well as for changing the default action to a reject.
12319
12320 Example: track the original source address by default, or the one advertised
12321 in the PROXY protocol header for connection coming from the local
12322 proxies. The first connection-level rule enables receipt of the
12323 PROXY protocol for these ones, the second rule tracks whatever
12324 address we decide to keep after optional decoding.
12325
12326 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
12327 tcp-request session track-sc0 src
12328
12329 Example: accept all sessions from white-listed hosts, reject too fast
12330 sessions without counting them, and track accepted sessions.
12331 This results in session rate being capped from abusive sources.
12332
12333 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
12334 tcp-request session reject if { src_sess_rate gt 10 }
12335 tcp-request session track-sc0 src
12336
12337 Example: accept all sessions from white-listed hosts, count all other
12338 sessions and reject too fast ones. This results in abusive ones
12339 being blocked as long as they don't slow down.
12340
12341 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
12342 tcp-request session track-sc0 src
12343 tcp-request session reject if { sc0_sess_rate gt 10 }
12344
12345 See section 7 about ACL usage.
12346
12347 See also : "tcp-request connection", "tcp-request content", "stick-table"
12348
12349
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012350tcp-response inspect-delay <timeout>
12351 Set the maximum allowed time to wait for a response during content inspection
12352 May be used in sections : defaults | frontend | listen | backend
12353 no | no | yes | yes
12354 Arguments :
12355 <timeout> is the timeout value specified in milliseconds by default, but
12356 can be in any other unit if the number is suffixed by the unit,
12357 as explained at the top of this document.
12358
12359 See also : "tcp-response content", "tcp-request inspect-delay".
12360
12361
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012362timeout check <timeout>
12363 Set additional check timeout, but only after a connection has been already
12364 established.
12365
12366 May be used in sections: defaults | frontend | listen | backend
12367 yes | no | yes | yes
12368 Arguments:
12369 <timeout> is the timeout value specified in milliseconds by default, but
12370 can be in any other unit if the number is suffixed by the unit,
12371 as explained at the top of this document.
12372
12373 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
12374 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010012375 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012376 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010012377 (Please also note that there is no valid reason to have such long connect
12378 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
12379 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012380
12381 If "timeout check" is not set haproxy uses "inter" for complete check
12382 timeout (connect + read) exactly like all <1.3.15 version.
12383
12384 In most cases check request is much simpler and faster to handle than normal
12385 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010012386 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012387
12388 This parameter is specific to backends, but can be specified once for all in
12389 "defaults" sections. This is in fact one of the easiest solutions not to
12390 forget about it.
12391
Willy Tarreau41a340d2008-01-22 12:25:31 +010012392 See also: "timeout connect", "timeout queue", "timeout server",
12393 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012394
12395
Willy Tarreau0ba27502007-12-24 16:55:16 +010012396timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010012397 Set the maximum inactivity time on the client side.
12398 May be used in sections : defaults | frontend | listen | backend
12399 yes | yes | yes | no
12400 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012401 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010012402 can be in any other unit if the number is suffixed by the unit,
12403 as explained at the top of this document.
12404
12405 The inactivity timeout applies when the client is expected to acknowledge or
12406 send data. In HTTP mode, this timeout is particularly important to consider
12407 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010012408 response while it is reading data sent by the server. That said, for the
12409 first phase, it is preferable to set the "timeout http-request" to better
12410 protect HAProxy from Slowloris like attacks. The value is specified in
12411 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010012412 suffixed by the unit, as specified at the top of this document. In TCP mode
12413 (and to a lesser extent, in HTTP mode), it is highly recommended that the
12414 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012415 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010012416 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010012417 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
12418 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020012419 which overrides "timeout client" and "timeout server" for tunnels, as well as
12420 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012421
12422 This parameter is specific to frontends, but can be specified once for all in
12423 "defaults" sections. This is in fact one of the easiest solutions not to
12424 forget about it. An unspecified timeout results in an infinite timeout, which
12425 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012426 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010012427 the system if the system's timeouts are not configured either.
12428
Willy Tarreau95c4e142017-11-26 12:18:55 +010012429 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010012430
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012431 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012432
Willy Tarreau0ba27502007-12-24 16:55:16 +010012433
Willy Tarreau05cdd962014-05-10 14:30:07 +020012434timeout client-fin <timeout>
12435 Set the inactivity timeout on the client side for half-closed connections.
12436 May be used in sections : defaults | frontend | listen | backend
12437 yes | yes | yes | no
12438 Arguments :
12439 <timeout> is the timeout value specified in milliseconds by default, but
12440 can be in any other unit if the number is suffixed by the unit,
12441 as explained at the top of this document.
12442
12443 The inactivity timeout applies when the client is expected to acknowledge or
12444 send data while one direction is already shut down. This timeout is different
12445 from "timeout client" in that it only applies to connections which are closed
12446 in one direction. This is particularly useful to avoid keeping connections in
12447 FIN_WAIT state for too long when clients do not disconnect cleanly. This
12448 problem is particularly common long connections such as RDP or WebSocket.
12449 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010012450 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
12451 frame was sent, often indicating an expectation that the connection quickly
12452 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020012453
12454 This parameter is specific to frontends, but can be specified once for all in
12455 "defaults" sections. By default it is not set, so half-closed connections
12456 will use the other timeouts (timeout.client or timeout.tunnel).
12457
12458 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
12459
12460
Willy Tarreau0ba27502007-12-24 16:55:16 +010012461timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010012462 Set the maximum time to wait for a connection attempt to a server to succeed.
12463 May be used in sections : defaults | frontend | listen | backend
12464 yes | no | yes | yes
12465 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012466 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010012467 can be in any other unit if the number is suffixed by the unit,
12468 as explained at the top of this document.
12469
12470 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012471 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010012472 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010012473 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012474 connect timeout also presets both queue and tarpit timeouts to the same value
12475 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012476
12477 This parameter is specific to backends, but can be specified once for all in
12478 "defaults" sections. This is in fact one of the easiest solutions not to
12479 forget about it. An unspecified timeout results in an infinite timeout, which
12480 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012481 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010012482 the system if the system's timeouts are not configured either.
12483
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012484 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012485
Willy Tarreau0ba27502007-12-24 16:55:16 +010012486
Willy Tarreaub16a5742010-01-10 14:46:16 +010012487timeout http-keep-alive <timeout>
12488 Set the maximum allowed time to wait for a new HTTP request to appear
12489 May be used in sections : defaults | frontend | listen | backend
12490 yes | yes | yes | yes
12491 Arguments :
12492 <timeout> is the timeout value specified in milliseconds by default, but
12493 can be in any other unit if the number is suffixed by the unit,
12494 as explained at the top of this document.
12495
12496 By default, the time to wait for a new request in case of keep-alive is set
12497 by "timeout http-request". However this is not always convenient because some
12498 people want very short keep-alive timeouts in order to release connections
12499 faster, and others prefer to have larger ones but still have short timeouts
12500 once the request has started to present itself.
12501
12502 The "http-keep-alive" timeout covers these needs. It will define how long to
12503 wait for a new HTTP request to start coming after a response was sent. Once
12504 the first byte of request has been seen, the "http-request" timeout is used
12505 to wait for the complete request to come. Note that empty lines prior to a
12506 new request do not refresh the timeout and are not counted as a new request.
12507
12508 There is also another difference between the two timeouts : when a connection
12509 expires during timeout http-keep-alive, no error is returned, the connection
12510 just closes. If the connection expires in "http-request" while waiting for a
12511 connection to complete, a HTTP 408 error is returned.
12512
12513 In general it is optimal to set this value to a few tens to hundreds of
12514 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010012515 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010012516 1 millisecond) it will probably only accept pipelined requests but not the
12517 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020012518 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010012519
12520 If this parameter is not set, the "http-request" timeout applies, and if both
12521 are not set, "timeout client" still applies at the lower level. It should be
12522 set in the frontend to take effect, unless the frontend is in TCP mode, in
12523 which case the HTTP backend's timeout will be used.
12524
Willy Tarreau95c4e142017-11-26 12:18:55 +010012525 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
12526 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010012527 (where we only have one connection per client and a connection setup).
12528
Willy Tarreaub16a5742010-01-10 14:46:16 +010012529 See also : "timeout http-request", "timeout client".
12530
12531
Willy Tarreau036fae02008-01-06 13:24:40 +010012532timeout http-request <timeout>
12533 Set the maximum allowed time to wait for a complete HTTP request
12534 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012535 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010012536 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012537 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010012538 can be in any other unit if the number is suffixed by the unit,
12539 as explained at the top of this document.
12540
12541 In order to offer DoS protection, it may be required to lower the maximum
12542 accepted time to receive a complete HTTP request without affecting the client
12543 timeout. This helps protecting against established connections on which
12544 nothing is sent. The client timeout cannot offer a good protection against
12545 this abuse because it is an inactivity timeout, which means that if the
12546 attacker sends one character every now and then, the timeout will not
12547 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020012548 types, the request will be aborted if it does not complete in time. When the
12549 timeout expires, an HTTP 408 response is sent to the client to inform it
12550 about the problem, and the connection is closed. The logs will report
12551 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012552 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020012553 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
12554 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010012555
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012556 By default, this timeout only applies to the header part of the request,
12557 and not to any data. As soon as the empty line is received, this timeout is
12558 not used anymore. When combined with "option http-buffer-request", this
12559 timeout also applies to the body of the request..
12560 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010012561 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010012562
12563 Generally it is enough to set it to a few seconds, as most clients send the
12564 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010012565 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010012566 generally work on local networks as long as there are no packet losses. This
12567 will prevent people from sending bare HTTP requests using telnet.
12568
12569 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012570 chunk of the incoming request. It should be set in the frontend to take
12571 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
12572 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010012573
Willy Tarreau0f228a02015-05-01 15:37:53 +020012574 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012575 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010012576
Willy Tarreau844e3c52008-01-11 16:28:18 +010012577
12578timeout queue <timeout>
12579 Set the maximum time to wait in the queue for a connection slot to be free
12580 May be used in sections : defaults | frontend | listen | backend
12581 yes | no | yes | yes
12582 Arguments :
12583 <timeout> is the timeout value specified in milliseconds by default, but
12584 can be in any other unit if the number is suffixed by the unit,
12585 as explained at the top of this document.
12586
12587 When a server's maxconn is reached, connections are left pending in a queue
12588 which may be server-specific or global to the backend. In order not to wait
12589 indefinitely, a timeout is applied to requests pending in the queue. If the
12590 timeout is reached, it is considered that the request will almost never be
12591 served, so it is dropped and a 503 error is returned to the client.
12592
12593 The "timeout queue" statement allows to fix the maximum time for a request to
12594 be left pending in a queue. If unspecified, the same value as the backend's
12595 connection timeout ("timeout connect") is used, for backwards compatibility
12596 with older versions with no "timeout queue" parameter.
12597
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012598 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012599
12600
12601timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010012602 Set the maximum inactivity time on the server side.
12603 May be used in sections : defaults | frontend | listen | backend
12604 yes | no | yes | yes
12605 Arguments :
12606 <timeout> is the timeout value specified in milliseconds by default, but
12607 can be in any other unit if the number is suffixed by the unit,
12608 as explained at the top of this document.
12609
12610 The inactivity timeout applies when the server is expected to acknowledge or
12611 send data. In HTTP mode, this timeout is particularly important to consider
12612 during the first phase of the server's response, when it has to send the
12613 headers, as it directly represents the server's processing time for the
12614 request. To find out what value to put there, it's often good to start with
12615 what would be considered as unacceptable response times, then check the logs
12616 to observe the response time distribution, and adjust the value accordingly.
12617
12618 The value is specified in milliseconds by default, but can be in any other
12619 unit if the number is suffixed by the unit, as specified at the top of this
12620 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
12621 recommended that the client timeout remains equal to the server timeout in
12622 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012623 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010012624 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010012625 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
12626 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020012627 "timeout tunnel", which overrides "timeout client" and "timeout server" for
12628 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012629
12630 This parameter is specific to backends, but can be specified once for all in
12631 "defaults" sections. This is in fact one of the easiest solutions not to
12632 forget about it. An unspecified timeout results in an infinite timeout, which
12633 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012634 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010012635 the system if the system's timeouts are not configured either.
12636
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012637 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012638
Willy Tarreau05cdd962014-05-10 14:30:07 +020012639
12640timeout server-fin <timeout>
12641 Set the inactivity timeout on the server side for half-closed connections.
12642 May be used in sections : defaults | frontend | listen | backend
12643 yes | no | yes | yes
12644 Arguments :
12645 <timeout> is the timeout value specified in milliseconds by default, but
12646 can be in any other unit if the number is suffixed by the unit,
12647 as explained at the top of this document.
12648
12649 The inactivity timeout applies when the server is expected to acknowledge or
12650 send data while one direction is already shut down. This timeout is different
12651 from "timeout server" in that it only applies to connections which are closed
12652 in one direction. This is particularly useful to avoid keeping connections in
12653 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
12654 This problem is particularly common long connections such as RDP or WebSocket.
12655 Note that this timeout can override "timeout tunnel" when a connection shuts
12656 down in one direction. This setting was provided for completeness, but in most
12657 situations, it should not be needed.
12658
12659 This parameter is specific to backends, but can be specified once for all in
12660 "defaults" sections. By default it is not set, so half-closed connections
12661 will use the other timeouts (timeout.server or timeout.tunnel).
12662
12663 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
12664
Willy Tarreau844e3c52008-01-11 16:28:18 +010012665
12666timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010012667 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010012668 May be used in sections : defaults | frontend | listen | backend
12669 yes | yes | yes | yes
12670 Arguments :
12671 <timeout> is the tarpit duration specified in milliseconds by default, but
12672 can be in any other unit if the number is suffixed by the unit,
12673 as explained at the top of this document.
12674
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020012675 When a connection is tarpitted using "http-request tarpit", it is maintained
12676 open with no activity for a certain amount of time, then closed. "timeout
12677 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012678
12679 The value is specified in milliseconds by default, but can be in any other
12680 unit if the number is suffixed by the unit, as specified at the top of this
12681 document. If unspecified, the same value as the backend's connection timeout
12682 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010012683 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012684
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012685 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012686
12687
Willy Tarreauce887fd2012-05-12 12:50:00 +020012688timeout tunnel <timeout>
12689 Set the maximum inactivity time on the client and server side for tunnels.
12690 May be used in sections : defaults | frontend | listen | backend
12691 yes | no | yes | yes
12692 Arguments :
12693 <timeout> is the timeout value specified in milliseconds by default, but
12694 can be in any other unit if the number is suffixed by the unit,
12695 as explained at the top of this document.
12696
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012697 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020012698 between a client and a server, and the connection remains inactive in both
12699 directions. This timeout supersedes both the client and server timeouts once
12700 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012701 analyzer remains attached to either connection (e.g. tcp content rules are
12702 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020012703 when switching to the WebSocket protocol, or forwarding a CONNECT request
12704 to a proxy), or after the first response when no keepalive/close option is
12705 specified.
12706
Willy Tarreau05cdd962014-05-10 14:30:07 +020012707 Since this timeout is usually used in conjunction with long-lived connections,
12708 it usually is a good idea to also set "timeout client-fin" to handle the
12709 situation where a client suddenly disappears from the net and does not
12710 acknowledge a close, or sends a shutdown and does not acknowledge pending
12711 data anymore. This can happen in lossy networks where firewalls are present,
12712 and is detected by the presence of large amounts of sessions in a FIN_WAIT
12713 state.
12714
Willy Tarreauce887fd2012-05-12 12:50:00 +020012715 The value is specified in milliseconds by default, but can be in any other
12716 unit if the number is suffixed by the unit, as specified at the top of this
12717 document. Whatever the expected normal idle time, it is a good practice to
12718 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010012719 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020012720
12721 This parameter is specific to backends, but can be specified once for all in
12722 "defaults" sections. This is in fact one of the easiest solutions not to
12723 forget about it.
12724
12725 Example :
12726 defaults http
12727 option http-server-close
12728 timeout connect 5s
12729 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020012730 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020012731 timeout server 30s
12732 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
12733
Willy Tarreau05cdd962014-05-10 14:30:07 +020012734 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020012735
12736
Willy Tarreau844e3c52008-01-11 16:28:18 +010012737transparent (deprecated)
12738 Enable client-side transparent proxying
12739 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010012740 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010012741 Arguments : none
12742
12743 This keyword was introduced in order to provide layer 7 persistence to layer
12744 3 load balancers. The idea is to use the OS's ability to redirect an incoming
12745 connection for a remote address to a local process (here HAProxy), and let
12746 this process know what address was initially requested. When this option is
12747 used, sessions without cookies will be forwarded to the original destination
12748 IP address of the incoming request (which should match that of another
12749 equipment), while requests with cookies will still be forwarded to the
12750 appropriate server.
12751
12752 The "transparent" keyword is deprecated, use "option transparent" instead.
12753
12754 Note that contrary to a common belief, this option does NOT make HAProxy
12755 present the client's IP to the server when establishing the connection.
12756
Willy Tarreau844e3c52008-01-11 16:28:18 +010012757 See also: "option transparent"
12758
William Lallemanda73203e2012-03-12 12:48:57 +010012759unique-id-format <string>
12760 Generate a unique ID for each request.
12761 May be used in sections : defaults | frontend | listen | backend
12762 yes | yes | yes | no
12763 Arguments :
12764 <string> is a log-format string.
12765
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012766 This keyword creates a ID for each request using the custom log format. A
12767 unique ID is useful to trace a request passing through many components of
12768 a complex infrastructure. The newly created ID may also be logged using the
12769 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010012770
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012771 The format should be composed from elements that are guaranteed to be
12772 unique when combined together. For instance, if multiple haproxy instances
12773 are involved, it might be important to include the node name. It is often
12774 needed to log the incoming connection's source and destination addresses
12775 and ports. Note that since multiple requests may be performed over the same
12776 connection, including a request counter may help differentiate them.
12777 Similarly, a timestamp may protect against a rollover of the counter.
12778 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010012779
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012780 It is recommended to use hexadecimal notation for many fields since it
12781 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010012782
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012783 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012784
Julien Vehentf21be322014-03-07 08:27:34 -050012785 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012786
12787 will generate:
12788
12789 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12790
12791 See also: "unique-id-header"
12792
12793unique-id-header <name>
12794 Add a unique ID header in the HTTP request.
12795 May be used in sections : defaults | frontend | listen | backend
12796 yes | yes | yes | no
12797 Arguments :
12798 <name> is the name of the header.
12799
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012800 Add a unique-id header in the HTTP request sent to the server, using the
12801 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010012802
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012803 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012804
Julien Vehentf21be322014-03-07 08:27:34 -050012805 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012806 unique-id-header X-Unique-ID
12807
12808 will generate:
12809
12810 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12811
12812 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010012813
Willy Tarreauf51658d2014-04-23 01:21:56 +020012814use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012815 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012816 May be used in sections : defaults | frontend | listen | backend
12817 no | yes | yes | no
12818 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012819 <backend> is the name of a valid backend or "listen" section, or a
12820 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012821
Willy Tarreauf51658d2014-04-23 01:21:56 +020012822 <condition> is a condition composed of ACLs, as described in section 7. If
12823 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012824
12825 When doing content-switching, connections arrive on a frontend and are then
12826 dispatched to various backends depending on a number of conditions. The
12827 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012828 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010012829 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012830 source address validation) or combined with a "tcp-request" rule to wait for
12831 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012832
12833 There may be as many "use_backend" rules as desired. All of these rules are
12834 evaluated in their declaration order, and the first one which matches will
12835 assign the backend.
12836
12837 In the first form, the backend will be used if the condition is met. In the
12838 second form, the backend will be used if the condition is not met. If no
12839 condition is valid, the backend defined with "default_backend" will be used.
12840 If no default backend is defined, either the servers in the same section are
12841 used (in case of a "listen" section) or, in case of a frontend, no server is
12842 used and a 503 service unavailable response is returned.
12843
Willy Tarreau51aecc72009-07-12 09:47:04 +020012844 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012845 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020012846 and backend processing will immediately follow, or the backend will wait for
12847 a complete HTTP request to get in. This feature is useful when a frontend
12848 must decode several protocols on a unique port, one of them being HTTP.
12849
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012850 When <backend> is a simple name, it is resolved at configuration time, and an
12851 error is reported if the specified backend does not exist. If <backend> is
12852 a log-format string instead, no check may be done at configuration time, so
12853 the backend name is resolved dynamically at run time. If the resulting
12854 backend name does not correspond to any valid backend, no other rule is
12855 evaluated, and the default_backend directive is applied instead. Note that
12856 when using dynamic backend names, it is highly recommended to use a prefix
12857 that no other backend uses in order to ensure that an unauthorized backend
12858 cannot be forced from the request.
12859
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012860 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012861 used to detect the association between frontends and backends to compute the
12862 backend's "fullconn" setting. This cannot be done for dynamic names.
12863
12864 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
12865 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010012866
Christopher Fauletb30b3102019-09-12 23:03:09 +020012867use-fcgi-app <name>
12868 Defines the FastCGI application to use for the backend.
12869 May be used in sections : defaults | frontend | listen | backend
12870 no | no | yes | yes
12871 Arguments :
12872 <name> is the name of the FastCGI application to use.
12873
12874 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010012875
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012876use-server <server> if <condition>
12877use-server <server> unless <condition>
12878 Only use a specific server if/unless an ACL-based condition is matched.
12879 May be used in sections : defaults | frontend | listen | backend
12880 no | no | yes | yes
12881 Arguments :
Jerome Magnin824186b2020-03-29 09:37:12 +020012882 <server> is the name of a valid server in the same backend section
12883 or a "log-format" string resolving to a server name.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012884
12885 <condition> is a condition composed of ACLs, as described in section 7.
12886
12887 By default, connections which arrive to a backend are load-balanced across
12888 the available servers according to the configured algorithm, unless a
12889 persistence mechanism such as a cookie is used and found in the request.
12890
12891 Sometimes it is desirable to forward a particular request to a specific
12892 server without having to declare a dedicated backend for this server. This
12893 can be achieved using the "use-server" rules. These rules are evaluated after
12894 the "redirect" rules and before evaluating cookies, and they have precedence
12895 on them. There may be as many "use-server" rules as desired. All of these
12896 rules are evaluated in their declaration order, and the first one which
12897 matches will assign the server.
12898
12899 If a rule designates a server which is down, and "option persist" is not used
12900 and no force-persist rule was validated, it is ignored and evaluation goes on
12901 with the next rules until one matches.
12902
12903 In the first form, the server will be used if the condition is met. In the
12904 second form, the server will be used if the condition is not met. If no
12905 condition is valid, the processing continues and the server will be assigned
12906 according to other persistence mechanisms.
12907
12908 Note that even if a rule is matched, cookie processing is still performed but
12909 does not assign the server. This allows prefixed cookies to have their prefix
12910 stripped.
12911
12912 The "use-server" statement works both in HTTP and TCP mode. This makes it
12913 suitable for use with content-based inspection. For instance, a server could
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012914 be selected in a farm according to the TLS SNI field when using protocols with
12915 implicit TLS (also see "req_ssl_sni"). And if these servers have their weight
12916 set to zero, they will not be used for other traffic.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012917
12918 Example :
12919 # intercept incoming TLS requests based on the SNI field
12920 use-server www if { req_ssl_sni -i www.example.com }
12921 server www 192.168.0.1:443 weight 0
12922 use-server mail if { req_ssl_sni -i mail.example.com }
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012923 server mail 192.168.0.1:465 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012924 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000012925 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012926 # all the rest is forwarded to this server
12927 server default 192.168.0.2:443 check
12928
Jerome Magnin824186b2020-03-29 09:37:12 +020012929 When <server> is a simple name, it is checked against existing servers in the
12930 configuration and an error is reported if the specified server does not exist.
12931 If it is a log-format, no check is performed when parsing the configuration,
12932 and if we can't resolve a valid server name at runtime but the use-server rule
Ilya Shipitsin11057a32020-06-21 21:18:27 +050012933 was conditioned by an ACL returning true, no other use-server rule is applied
Jerome Magnin824186b2020-03-29 09:37:12 +020012934 and we fall back to load balancing.
12935
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012936 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012937
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012938
Davor Ocelice9ed2812017-12-25 17:49:28 +0100129395. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012940--------------------------
12941
12942The "bind", "server" and "default-server" keywords support a number of settings
12943depending on some build options and on the system HAProxy was built on. These
12944settings generally each consist in one word sometimes followed by a value,
12945written on the same line as the "bind" or "server" line. All these options are
12946described in this section.
12947
12948
129495.1. Bind options
12950-----------------
12951
12952The "bind" keyword supports a certain number of settings which are all passed
12953as arguments on the same line. The order in which those arguments appear makes
12954no importance, provided that they appear after the bind address. All of these
12955parameters are optional. Some of them consist in a single words (booleans),
12956while other ones expect a value after them. In this case, the value must be
12957provided immediately after the setting name.
12958
12959The currently supported settings are the following ones.
12960
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012961accept-netscaler-cip <magic number>
12962 Enforces the use of the NetScaler Client IP insertion protocol over any
12963 connection accepted by any of the TCP sockets declared on the same line. The
12964 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
12965 the incoming connection to be used everywhere an address is used, with the
12966 only exception of "tcp-request connection" rules which will only see the
12967 real connection address. Logs will reflect the addresses indicated in the
12968 protocol, unless it is violated, in which case the real address will still
12969 be used. This keyword combined with support from external components can be
12970 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010012971 mechanism which is not always reliable and not even always usable. See also
12972 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
12973 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012974
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012975accept-proxy
12976 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020012977 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
12978 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012979 3/4 addresses of the incoming connection to be used everywhere an address is
12980 used, with the only exception of "tcp-request connection" rules which will
12981 only see the real connection address. Logs will reflect the addresses
12982 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010012983 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012984 components can be used as an efficient and reliable alternative to the
12985 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020012986 usable. See also "tcp-request connection expect-proxy" for a finer-grained
12987 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012988
Olivier Houchardc2aae742017-09-22 18:26:28 +020012989allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010012990 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010012991 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050012992 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010012993 that are idempotent. You can use the "wait-for-handshake" action for any
12994 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020012995
Willy Tarreauab861d32013-04-02 02:30:41 +020012996alpn <protocols>
12997 This enables the TLS ALPN extension and advertises the specified protocol
12998 list as supported on top of ALPN. The protocol list consists in a comma-
12999 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050013000 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020013001 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010013002 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
13003 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
13004 now obsolete NPN extension. At the time of writing this, most browsers still
13005 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
13006 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
13007 are expected to be supported, both versions can be advertised, in order of
13008 preference, like below :
13009
13010 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020013011
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013012backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010013013 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013014 backlog is used instead, which generally defaults to the maxconn value.
13015
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010013016curves <curves>
13017 This setting is only available when support for OpenSSL was built in. It sets
13018 the string describing the list of elliptic curves algorithms ("curve suite")
13019 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
13020 string is a colon-delimited list of curve name.
13021 Example: "X25519:P-256" (without quote)
13022 When "curves" is set, "ecdhe" parameter is ignored.
13023
Emeric Brun7fb34422012-09-28 15:26:15 +020013024ecdhe <named curve>
13025 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010013026 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
13027 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020013028
Emeric Brunfd33a262012-10-11 16:28:27 +020013029ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020013030 This setting is only available when support for OpenSSL was built in. It
13031 designates a PEM file from which to load CA certificates used to verify
13032 client's certificate.
13033
Emeric Brunb6dc9342012-09-28 17:55:37 +020013034ca-ignore-err [all|<errorID>,...]
13035 This setting is only available when support for OpenSSL was built in.
13036 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
13037 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
13038 error is ignored.
13039
Christopher Faulet31af49d2015-06-09 17:29:50 +020013040ca-sign-file <cafile>
13041 This setting is only available when support for OpenSSL was built in. It
13042 designates a PEM file containing both the CA certificate and the CA private
13043 key used to create and sign server's certificates. This is a mandatory
13044 setting when the dynamic generation of certificates is enabled. See
13045 'generate-certificates' for details.
13046
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000013047ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020013048 This setting is only available when support for OpenSSL was built in. It is
13049 the CA private key passphrase. This setting is optional and used only when
13050 the dynamic generation of certificates is enabled. See
13051 'generate-certificates' for details.
13052
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010013053ca-verify-file <cafile>
13054 This setting designates a PEM file from which to load CA certificates used to
13055 verify client's certificate. It designates CA certificates which must not be
13056 included in CA names sent in server hello message. Typically, "ca-file" must
13057 be defined with intermediate certificates, and "ca-verify-file" with
13058 certificates to ending the chain, like root CA.
13059
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013060ciphers <ciphers>
13061 This setting is only available when support for OpenSSL was built in. It sets
13062 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000013063 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013064 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013065 information and recommendations see e.g.
13066 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
13067 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
13068 cipher configuration, please check the "ciphersuites" keyword.
13069
13070ciphersuites <ciphersuites>
13071 This setting is only available when support for OpenSSL was built in and
13072 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
13073 the list of cipher algorithms ("cipher suite") that are negotiated during the
13074 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013075 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
13076 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013077
Emeric Brunfd33a262012-10-11 16:28:27 +020013078crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020013079 This setting is only available when support for OpenSSL was built in. It
13080 designates a PEM file from which to load certificate revocation list used
13081 to verify client's certificate.
13082
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013083crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000013084 This setting is only available when support for OpenSSL was built in. It
13085 designates a PEM file containing both the required certificates and any
13086 associated private keys. This file can be built by concatenating multiple
13087 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
13088 requires an intermediate certificate, this can also be concatenated into this
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +010013089 file. Intermediate certificate can also be shared in a directory via
13090 "issuers-chain-path" directive.
Alex Davies0fbf0162013-03-02 16:04:50 +000013091
William Lallemand4c5adbf2020-02-24 14:23:22 +010013092 If the file does not contain a private key, HAProxy will try to load
13093 the key at the same path suffixed by a ".key".
13094
Alex Davies0fbf0162013-03-02 16:04:50 +000013095 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
13096 are loaded.
13097
13098 If a directory name is used instead of a PEM file, then all files found in
William Lallemand3f25ae32020-02-24 16:30:12 +010013099 that directory will be loaded in alphabetic order unless their name ends
13100 with '.key', '.issuer', '.ocsp' or '.sctl' (reserved extensions). This
13101 directive may be specified multiple times in order to load certificates from
13102 multiple files or directories. The certificates will be presented to clients
13103 who provide a valid TLS Server Name Indication field matching one of their
13104 CN or alt subjects. Wildcards are supported, where a wildcard character '*'
13105 is used instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010013106 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000013107
13108 If no SNI is provided by the client or if the SSL library does not support
13109 TLS extensions, or if the client provides an SNI hostname which does not
13110 match any certificate, then the first loaded certificate will be presented.
13111 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010013112 recommended to load the default one first as a file or to ensure that it will
13113 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000013114
Emeric Brune032bfa2012-09-28 13:01:45 +020013115 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013116
Davor Ocelice9ed2812017-12-25 17:49:28 +010013117 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000013118 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010013119 choose a web server that the CA believes requires an intermediate CA (for
13120 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000013121 others, e.g. nginx, result in a wrong bundle that will not work for some
13122 clients).
13123
Emeric Brun4147b2e2014-06-16 18:36:30 +020013124 For each PEM file, haproxy checks for the presence of file at the same path
13125 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
13126 Status Request extension (also known as "OCSP stapling") is automatically
13127 enabled. The content of this file is optional. If not empty, it must contain
13128 a valid OCSP Response in DER format. In order to be valid an OCSP Response
13129 must comply with the following rules: it has to indicate a good status,
13130 it has to be a single response for the certificate of the PEM file, and it
13131 has to be valid at the moment of addition. If these rules are not respected
13132 the OCSP Response is ignored and a warning is emitted. In order to identify
13133 which certificate an OCSP Response applies to, the issuer's certificate is
13134 necessary. If the issuer's certificate is not found in the PEM file, it will
13135 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
13136 if it exists otherwise it will fail with an error.
13137
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010013138 For each PEM file, haproxy also checks for the presence of file at the same
13139 path suffixed by ".sctl". If such file is found, support for Certificate
13140 Transparency (RFC6962) TLS extension is enabled. The file must contain a
13141 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
13142 to check basic syntax, but no signatures are verified.
13143
yanbzhu6c25e9e2016-01-05 12:52:02 -050013144 There are cases where it is desirable to support multiple key types, e.g. RSA
13145 and ECDSA in the cipher suites offered to the clients. This allows clients
13146 that support EC certificates to be able to use EC ciphers, while
13147 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050013148
William Lallemandf9ff3ec2020-10-02 17:57:44 +020013149 To achieve this, OpenSSL 1.1.1 is required, you can configure this behavior
13150 by providing one crt entry per certificate type, or by configuring a "cert
13151 bundle" like it was required before HAProxy 1.8. See "ssl-load-extra-files".
yanbzhud19630c2015-12-14 15:10:25 -050013152
Emeric Brunb6dc9342012-09-28 17:55:37 +020013153crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000013154 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010013155 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013156 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000013157 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020013158
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013159crt-list <file>
13160 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013161 designates a list of PEM file with an optional ssl configuration and a SNI
13162 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013163
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013164 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
13165
William Lallemand5d036392020-06-30 16:11:36 +020013166 sslbindconf supports "allow-0rtt", "alpn", "ca-file", "ca-verify-file",
13167 "ciphers", "ciphersuites", "crl-file", "curves", "ecdhe", "no-ca-names",
13168 "npn", "verify" configuration. With BoringSSL and Openssl >= 1.1.1
13169 "ssl-min-ver" and "ssl-max-ver" are also supported. It overrides the
13170 configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013171
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020013172 Wildcards are supported in the SNI filter. Negative filter are also supported,
Joao Moraise51fab02020-11-21 07:42:20 -030013173 useful in combination with a wildcard filter to exclude a particular SNI, or
13174 after the first certificate to exclude a pattern from its CN or Subject Alt
13175 Name (SAN). The certificates will be presented to clients who provide a valid
13176 TLS Server Name Indication field matching one of the SNI filters. If no SNI
13177 filter is specified, the CN and SAN are used. This directive may be specified
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020013178 multiple times. See the "crt" option for more information. The default
13179 certificate is still needed to meet OpenSSL expectations. If it is not used,
13180 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010013181
William Lallemandf9ff3ec2020-10-02 17:57:44 +020013182 Multi-cert bundling (see "ssl-load-extra-files") is supported with crt-list,
13183 as long as only the base name is given in the crt-list. SNI filter will do
13184 the same work on all bundled certificates.
yanbzhud19630c2015-12-14 15:10:25 -050013185
William Lallemand7c26ed72020-06-03 17:34:48 +020013186 Empty lines as well as lines beginning with a hash ('#') will be ignored.
13187
Joao Moraisaa8fcc42020-11-24 08:24:30 -030013188 The first declared certificate of a bind line is used as the default
13189 certificate, either from crt or crt-list option, which haproxy should use in
13190 the TLS handshake if no other certificate matches. This certificate will also
13191 be used if the provided SNI matches its CN or SAN, even if a matching SNI
13192 filter is found on any crt-list. The SNI filter !* can be used after the first
13193 declared certificate to not include its CN and SAN in the SNI tree, so it will
13194 never match except if no other certificate matches. This way the first
13195 declared certificate act as a fallback.
Joao Moraise51fab02020-11-21 07:42:20 -030013196
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013197 crt-list file example:
Joao Moraise51fab02020-11-21 07:42:20 -030013198 cert1.pem !*
William Lallemand7c26ed72020-06-03 17:34:48 +020013199 # comment
Emmanuel Hocdet05942112017-02-20 16:11:50 +010013200 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013201 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010013202 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010013203
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013204defer-accept
13205 Is an optional keyword which is supported only on certain Linux kernels. It
13206 states that a connection will only be accepted once some data arrive on it,
13207 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010013208 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013209 performance by ensuring that most of the request is already available when
13210 the connection is accepted. On the other hand, it will not be able to detect
13211 connections which don't talk. It is important to note that this option is
13212 broken in all kernels up to 2.6.31, as the connection is never accepted until
13213 the client talks. This can cause issues with front firewalls which would see
13214 an established connection while the proxy will only see it in SYN_RECV. This
13215 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
13216
William Lallemandf6975e92017-05-26 17:42:10 +020013217expose-fd listeners
13218 This option is only usable with the stats socket. It gives your stats socket
13219 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020013220 During a reload with the master-worker mode, the process is automatically
13221 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013222 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020013223
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013224force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013225 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013226 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013227 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013228 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013229
13230force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013231 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013232 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013233 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013234
13235force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013236 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013237 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013238 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013239
13240force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013241 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013242 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013243 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013244
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013245force-tlsv13
13246 This option enforces use of TLSv1.3 only on SSL connections instantiated from
13247 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013248 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013249
Christopher Faulet31af49d2015-06-09 17:29:50 +020013250generate-certificates
13251 This setting is only available when support for OpenSSL was built in. It
13252 enables the dynamic SSL certificates generation. A CA certificate and its
13253 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
13254 a transparent forward proxy, SSL requests generate errors because of a common
13255 name mismatch on the certificate presented to the client. With this option
13256 enabled, HAProxy will try to forge a certificate using the SNI hostname
13257 indicated by the client. This is done only if no certificate matches the SNI
13258 hostname (see 'crt-list'). If an error occurs, the default certificate is
13259 used, else the 'strict-sni' option is set.
13260 It can also be used when HAProxy is configured as a reverse proxy to ease the
13261 deployment of an architecture with many backends.
13262
13263 Creating a SSL certificate is an expensive operation, so a LRU cache is used
13264 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010013265 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020013266 certificate is used many times.
13267
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013268gid <gid>
13269 Sets the group of the UNIX sockets to the designated system gid. It can also
13270 be set by default in the global section's "unix-bind" statement. Note that
13271 some platforms simply ignore this. This setting is equivalent to the "group"
13272 setting except that the group ID is used instead of its name. This setting is
13273 ignored by non UNIX sockets.
13274
13275group <group>
13276 Sets the group of the UNIX sockets to the designated system group. It can
13277 also be set by default in the global section's "unix-bind" statement. Note
13278 that some platforms simply ignore this. This setting is equivalent to the
13279 "gid" setting except that the group name is used instead of its gid. This
13280 setting is ignored by non UNIX sockets.
13281
13282id <id>
13283 Fixes the socket ID. By default, socket IDs are automatically assigned, but
13284 sometimes it is more convenient to fix them to ease monitoring. This value
13285 must be strictly positive and unique within the listener/frontend. This
13286 option can only be used when defining only a single socket.
13287
13288interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010013289 Restricts the socket to a specific interface. When specified, only packets
13290 received from that particular interface are processed by the socket. This is
13291 currently only supported on Linux. The interface must be a primary system
13292 interface, not an aliased interface. It is also possible to bind multiple
13293 frontends to the same address if they are bound to different interfaces. Note
13294 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010013295 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
13296 uses the same interface as inbound traffic, and its associated routing table,
13297 even if there are explicit routes through different interfaces configured.
13298 This can prove useful to address asymmetric routing issues when the same
13299 client IP addresses need to be able to reach frontends hosted on different
13300 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013301
Willy Tarreauabb175f2012-09-24 12:43:26 +020013302level <level>
13303 This setting is used with the stats sockets only to restrict the nature of
13304 the commands that can be issued on the socket. It is ignored by other
13305 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013306 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020013307 read, and no change is allowed. It would make sense on systems where it
13308 is not easy to restrict access to the socket.
13309 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010013310 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020013311 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010013312 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020013313 all counters).
13314
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020013315severity-output <format>
13316 This setting is used with the stats sockets only to configure severity
13317 level output prepended to informational feedback messages. Severity
13318 level of messages can range between 0 and 7, conforming to syslog
13319 rfc5424. Valid and successful socket commands requesting data
13320 (i.e. "show map", "get acl foo" etc.) will never have a severity level
13321 prepended. It is ignored by other sockets. <format> can be one of :
13322 - "none" (default) no severity level is prepended to feedback messages.
13323 - "number" severity level is prepended as a number.
13324 - "string" severity level is prepended as a string following the
13325 rfc5424 convention.
13326
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013327maxconn <maxconn>
13328 Limits the sockets to this number of concurrent connections. Extraneous
13329 connections will remain in the system's backlog until a connection is
13330 released. If unspecified, the limit will be the same as the frontend's
13331 maxconn. Note that in case of port ranges or multiple addresses, the same
13332 value will be applied to each socket. This setting enables different
13333 limitations on expensive sockets, for instance SSL entries which may easily
13334 eat all memory.
13335
13336mode <mode>
13337 Sets the octal mode used to define access permissions on the UNIX socket. It
13338 can also be set by default in the global section's "unix-bind" statement.
13339 Note that some platforms simply ignore this. This setting is ignored by non
13340 UNIX sockets.
13341
13342mss <maxseg>
13343 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
13344 connections. This can be used to force a lower MSS for certain specific
13345 ports, for instance for connections passing through a VPN. Note that this
13346 relies on a kernel feature which is theoretically supported under Linux but
13347 was buggy in all versions prior to 2.6.28. It may or may not work on other
13348 operating systems. It may also not change the advertised value but change the
13349 effective size of outgoing segments. The commonly advertised value for TCPv4
13350 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
13351 positive, it will be used as the advertised MSS. If it is negative, it will
13352 indicate by how much to reduce the incoming connection's advertised MSS for
13353 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
13354
13355name <name>
13356 Sets an optional name for these sockets, which will be reported on the stats
13357 page.
13358
Willy Tarreaud72f0f32015-10-13 14:50:22 +020013359namespace <name>
13360 On Linux, it is possible to specify which network namespace a socket will
13361 belong to. This directive makes it possible to explicitly bind a listener to
13362 a namespace different from the default one. Please refer to your operating
13363 system's documentation to find more details about network namespaces.
13364
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013365nice <nice>
13366 Sets the 'niceness' of connections initiated from the socket. Value must be
13367 in the range -1024..1024 inclusive, and defaults to zero. Positive values
13368 means that such connections are more friendly to others and easily offer
13369 their place in the scheduler. On the opposite, negative values mean that
13370 connections want to run with a higher priority than others. The difference
13371 only happens under high loads when the system is close to saturation.
13372 Negative values are appropriate for low-latency or administration services,
13373 and high values are generally recommended for CPU intensive tasks such as SSL
13374 processing or bulk transfers which are less sensible to latency. For example,
13375 it may make sense to use a positive value for an SMTP socket and a negative
13376 one for an RDP socket.
13377
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020013378no-ca-names
13379 This setting is only available when support for OpenSSL was built in. It
13380 prevents from send CA names in server hello message when ca-file is used.
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010013381 Use "ca-verify-file" instead of "ca-file" with "no-ca-names".
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020013382
Emeric Brun9b3009b2012-10-05 11:55:06 +020013383no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013384 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013385 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013386 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013387 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013388 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
13389 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013390
Emeric Brun90ad8722012-10-02 14:00:59 +020013391no-tls-tickets
13392 This setting is only available when support for OpenSSL was built in. It
13393 disables the stateless session resumption (RFC 5077 TLS Ticket
13394 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013395 session resumption is more expensive in CPU usage. This option is also
13396 available on global statement "ssl-default-bind-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010013397 The TLS ticket mechanism is only used up to TLS 1.2.
13398 Forward Secrecy is compromised with TLS tickets, unless ticket keys
13399 are periodically rotated (via reload or by using "tls-ticket-keys").
Emeric Brun90ad8722012-10-02 14:00:59 +020013400
Emeric Brun9b3009b2012-10-05 11:55:06 +020013401no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013402 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013403 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013404 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013405 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013406 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13407 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013408
Emeric Brun9b3009b2012-10-05 11:55:06 +020013409no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020013410 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013411 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013412 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013413 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013414 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13415 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020013416
Emeric Brun9b3009b2012-10-05 11:55:06 +020013417no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020013418 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013419 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013420 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013421 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013422 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13423 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020013424
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013425no-tlsv13
13426 This setting is only available when support for OpenSSL was built in. It
13427 disables support for TLSv1.3 on any sockets instantiated from the listener
13428 when SSL is supported. Note that SSLv2 is forced disabled in the code and
13429 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013430 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13431 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013432
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020013433npn <protocols>
13434 This enables the NPN TLS extension and advertises the specified protocol list
13435 as supported on top of NPN. The protocol list consists in a comma-delimited
13436 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050013437 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020013438 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010013439 replaced with the ALPN extension (see the "alpn" keyword), though this one is
13440 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
13441 version of OpenSSL, NPN might still be used as most clients still support it
13442 at the time of writing this. It is possible to enable both NPN and ALPN
13443 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020013444
Lukas Tribus53ae85c2017-05-04 15:45:40 +000013445prefer-client-ciphers
13446 Use the client's preference when selecting the cipher suite, by default
13447 the server's preference is enforced. This option is also available on
13448 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020013449 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
13450 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
13451 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000013452
Christopher Fauletc644fa92017-11-23 22:44:11 +010013453process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010013454 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010013455 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010013456 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010013457 between the two is applied. If in the end the listener is not allowed to run
13458 on any remaining process, a warning is emitted, and the listener will either
13459 run on the first process of the listener if a single process was specified,
13460 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010013461 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010013462 connections for this listener, for the the process set. If multiple processes
13463 and threads are configured, a warning is emitted, as it either results from a
13464 configuration error or a misunderstanding of these models. For the unlikely
13465 case where several ranges are needed, this directive may be repeated.
13466 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010013467
13468 all | odd | even | number[-[number]]
13469
13470 Ranges can be partially defined. The higher bound can be omitted. In such
13471 case, it is replaced by the corresponding maximum value. The main purpose of
13472 this directive is to be used with the stats sockets and have one different
13473 socket per process. The second purpose is to have multiple bind lines sharing
13474 the same IP:port but not the same process in a listener, so that the system
13475 can distribute the incoming connections into multiple queues and allow a
13476 smoother inter-process load balancing. Currently Linux 3.9 and above is known
13477 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020013478
Christopher Fauleta717b992018-04-10 14:43:00 +020013479proto <name>
13480 Forces the multiplexer's protocol to use for the incoming connections. It
13481 must be compatible with the mode of the frontend (TCP or HTTP). It must also
13482 be usable on the frontend side. The list of available protocols is reported
Christopher Faulet982e17d2021-03-26 14:44:18 +010013483 in haproxy -vv. The protocols properties are reported : the mode (TCP/HTTP),
13484 the side (FE/BE), the mux name and its flags.
13485
13486 Some protocols report errors on aborts (flag=CLEAN_ABRT). Some others are
13487 subject to the head-of-line blocking on server side (flag=HOL_RISK). Finally
13488 some protocols don't support upgrades (flag=NO_UPG). The HTX compatibility is
13489 also reported (flag=HTX).
13490
13491 Here are the protocols that may be used as argument to a "proto" directive on
13492 a bind line :
13493
13494 h2 : mode=HTTP side=FE|BE mux=H2 flags=HTX|CLEAN_ABRT|HOL_RISK|NO_UPG
13495 h1 : mode=HTTP side=FE|BE mux=H1 flags=HTX|NO_UPG
13496 none : mode=TCP side=FE|BE mux=PASS flags=NO_UPG
13497
Daniel Corbett67a82712020-07-06 23:01:19 -040013498 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauleta717b992018-04-10 14:43:00 +020013499 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080013500 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020013501 h2" on the bind line.
13502
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013503ssl
13504 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013505 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013506 certificate is necessary (see "crt" above). All contents in the buffers will
13507 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020013508 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
13509 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013510
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013511ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
13512 This option enforces use of <version> or lower on SSL connections instantiated
William Lallemand50df1cb2020-06-02 10:52:24 +020013513 from this listener. Using this setting without "ssl-min-ver" can be
13514 ambiguous because the default ssl-min-ver value could change in future HAProxy
13515 versions. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013516 "ssl-default-bind-options". See also "ssl-min-ver".
13517
13518ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
William Lallemand50df1cb2020-06-02 10:52:24 +020013519 This option enforces use of <version> or upper on SSL connections
13520 instantiated from this listener. The default value is "TLSv1.2". This option
13521 is also available on global statement "ssl-default-bind-options".
13522 See also "ssl-max-ver".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013523
Emmanuel Hocdet65623372013-01-24 17:17:15 +010013524strict-sni
13525 This setting is only available when support for OpenSSL was built in. The
13526 SSL/TLS negotiation is allow only if the client provided an SNI which match
13527 a certificate. The default certificate is not used.
13528 See the "crt" option for more information.
13529
Willy Tarreau2af207a2015-02-04 00:45:58 +010013530tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010013531 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010013532 listening socket. This option is available on Linux since version 2.6.37. It
13533 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010013534 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010013535 useful on long-lived connections experiencing long idle periods such as
13536 remote terminals or database connection pools, where the client and server
13537 timeouts must remain high to allow a long period of idle, but where it is
13538 important to detect that the client has disappeared in order to release all
13539 resources associated with its connection (and the server's session). The
13540 argument is a delay expressed in milliseconds by default. This only works
13541 for regular TCP connections, and is ignored for other protocols.
13542
Willy Tarreau1c862c52012-10-05 16:21:00 +020013543tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010013544 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020013545 enables TCP Fast Open on the listening socket, which means that clients which
13546 support this feature will be able to send a request and receive a response
13547 during the 3-way handshake starting from second connection, thus saving one
13548 round-trip after the first connection. This only makes sense with protocols
13549 that use high connection rates and where each round trip matters. This can
13550 possibly cause issues with many firewalls which do not accept data on SYN
13551 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020013552 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
13553 need to build HAProxy with USE_TFO=1 if your libc doesn't define
13554 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020013555
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010013556tls-ticket-keys <keyfile>
13557 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010013558 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
13559 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
13560 The first key determines the key length used for next keys: you can't mix
13561 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
13562 build option (default 3) and at least as many keys need to be present in
13563 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
13564 penultimate one for encryption. This enables easy key rotation by just
13565 appending new key to the file and reloading the process. Keys must be
13566 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
13567 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010013568 storage such as hard drives (hint: use tmpfs and don't swap those files).
13569 Lifetime hint can be changed using tune.ssl.timeout.
13570
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013571transparent
13572 Is an optional keyword which is supported only on certain Linux kernels. It
13573 indicates that the addresses will be bound even if they do not belong to the
13574 local machine, and that packets targeting any of these addresses will be
13575 intercepted just as if the addresses were locally configured. This normally
13576 requires that IP forwarding is enabled. Caution! do not use this with the
13577 default address '*', as it would redirect any traffic for the specified port.
13578 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
13579 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
13580 kernel version. Some distribution kernels include backports of the feature,
13581 so check for support with your vendor.
13582
Willy Tarreau77e3af92012-11-24 15:07:23 +010013583v4v6
13584 Is an optional keyword which is supported only on most recent systems
13585 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
13586 and IPv6 when it uses the default address. Doing so is sometimes necessary
13587 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013588 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010013589
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013590v6only
13591 Is an optional keyword which is supported only on most recent systems
13592 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
13593 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010013594 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
13595 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013596
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013597uid <uid>
13598 Sets the owner of the UNIX sockets to the designated system uid. It can also
13599 be set by default in the global section's "unix-bind" statement. Note that
13600 some platforms simply ignore this. This setting is equivalent to the "user"
13601 setting except that the user numeric ID is used instead of its name. This
13602 setting is ignored by non UNIX sockets.
13603
13604user <user>
13605 Sets the owner of the UNIX sockets to the designated system user. It can also
13606 be set by default in the global section's "unix-bind" statement. Note that
13607 some platforms simply ignore this. This setting is equivalent to the "uid"
13608 setting except that the user name is used instead of its uid. This setting is
13609 ignored by non UNIX sockets.
13610
Emeric Brun1a073b42012-09-28 17:07:34 +020013611verify [none|optional|required]
13612 This setting is only available when support for OpenSSL was built in. If set
13613 to 'none', client certificate is not requested. This is the default. In other
13614 cases, a client certificate is requested. If the client does not provide a
13615 certificate after the request and if 'verify' is set to 'required', then the
13616 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020013617 certificate provided by the client is always verified using CAs from
13618 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
13619 is aborted, regardless of the 'verify' option, unless the error code exactly
13620 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013621
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200136225.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010013623------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013624
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013625The "server" and "default-server" keywords support a certain number of settings
13626which are all passed as arguments on the server line. The order in which those
13627arguments appear does not count, and they are all optional. Some of those
13628settings are single words (booleans) while others expect one or several values
13629after them. In this case, the values must immediately follow the setting name.
13630Except default-server, all those settings must be specified after the server's
13631address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013632
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013633 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013634 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020013635
Frédéric Lécailled2376272017-03-21 18:52:12 +010013636Note that all these settings are supported both by "server" and "default-server"
13637keywords, except "id" which is only supported by "server".
13638
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013639The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013640
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013641addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013642 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010013643 to send health-checks or to probe the agent-check. On some servers, it may be
13644 desirable to dedicate an IP address to specific component able to perform
13645 complex tests which are more suitable to health-checks than the application.
13646 This parameter is ignored if the "check" parameter is not set. See also the
13647 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013648
Simon Hormand60d6912013-11-25 10:46:36 +090013649agent-check
13650 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010013651 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010013652 to the port set by the "agent-port" parameter and reading an ASCII string
13653 terminated by the first '\r' or '\n' met. The string is made of a series of
13654 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090013655
Willy Tarreau81f5d942013-12-09 20:51:51 +010013656 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090013657 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020013658 weight of a server as configured when haproxy starts. Note that a zero
13659 weight is reported on the stats page as "DRAIN" since it has the same
13660 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090013661
Davor Ocelice9ed2812017-12-25 17:49:28 +010013662 - The string "maxconn:" followed by an integer (no space between). Values
13663 in this format will set the maxconn of a server. The maximum number of
13664 connections advertised needs to be multiplied by the number of load
13665 balancers and different backends that use this health check to get the
13666 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020013667
Willy Tarreau81f5d942013-12-09 20:51:51 +010013668 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013669 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090013670
Willy Tarreau81f5d942013-12-09 20:51:51 +010013671 - The word "drain". This will turn the server's administrative state to the
13672 DRAIN mode, thus it will not accept any new connections other than those
13673 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090013674
Willy Tarreau81f5d942013-12-09 20:51:51 +010013675 - The word "maint". This will turn the server's administrative state to the
13676 MAINT mode, thus it will not accept any new connections at all, and health
13677 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090013678
William Dauchyf8e795c2020-09-26 13:35:51 +020013679 - The words "down", "fail", or "stopped", optionally followed by a
Willy Tarreau81f5d942013-12-09 20:51:51 +010013680 description string after a sharp ('#'). All of these mark the server's
13681 operating state as DOWN, but since the word itself is reported on the stats
13682 page, the difference allows an administrator to know if the situation was
13683 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010013684 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010013685 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090013686
Willy Tarreau81f5d942013-12-09 20:51:51 +010013687 - The word "up" sets back the server's operating state as UP if health checks
13688 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090013689
Willy Tarreau81f5d942013-12-09 20:51:51 +010013690 Parameters which are not advertised by the agent are not changed. For
13691 example, an agent might be designed to monitor CPU usage and only report a
13692 relative weight and never interact with the operating status. Similarly, an
13693 agent could be designed as an end-user interface with 3 radio buttons
13694 allowing an administrator to change only the administrative state. However,
13695 it is important to consider that only the agent may revert its own actions,
13696 so if a server is set to DRAIN mode or to DOWN state using the agent, the
13697 agent must implement the other equivalent actions to bring the service into
13698 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090013699
Simon Horman2f1f9552013-11-25 10:46:37 +090013700 Failure to connect to the agent is not considered an error as connectivity
13701 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010013702 parameter. Warning though, it is not a good idea to stop an agent after it
13703 reports "down", since only an agent reporting "up" will be able to turn the
13704 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010013705 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090013706
Willy Tarreau81f5d942013-12-09 20:51:51 +010013707 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010013708 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090013709
James Brown55f9ff12015-10-21 18:19:05 -070013710agent-send <string>
13711 If this option is specified, haproxy will send the given string (verbatim)
13712 to the agent server upon connection. You could, for example, encode
13713 the backend name into this string, which would enable your agent to send
13714 different responses based on the backend. Make sure to include a '\n' if
13715 you want to terminate your request with a newline.
13716
Simon Hormand60d6912013-11-25 10:46:36 +090013717agent-inter <delay>
13718 The "agent-inter" parameter sets the interval between two agent checks
13719 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
13720
13721 Just as with every other time-based parameter, it may be entered in any
13722 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
13723 parameter also serves as a timeout for agent checks "timeout check" is
13724 not set. In order to reduce "resonance" effects when multiple servers are
13725 hosted on the same hardware, the agent and health checks of all servers
13726 are started with a small time offset between them. It is also possible to
13727 add some random noise in the agent and health checks interval using the
13728 global "spread-checks" keyword. This makes sense for instance when a lot
13729 of backends use the same servers.
13730
13731 See also the "agent-check" and "agent-port" parameters.
13732
Misiek768d8602017-01-09 09:52:43 +010013733agent-addr <addr>
13734 The "agent-addr" parameter sets address for agent check.
13735
13736 You can offload agent-check to another target, so you can make single place
13737 managing status and weights of servers defined in haproxy in case you can't
13738 make self-aware and self-managing services. You can specify both IP or
13739 hostname, it will be resolved.
13740
Simon Hormand60d6912013-11-25 10:46:36 +090013741agent-port <port>
13742 The "agent-port" parameter sets the TCP port used for agent checks.
13743
13744 See also the "agent-check" and "agent-inter" parameters.
13745
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013746allow-0rtt
13747 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020013748 Note that early data will be sent only if the client used early data, or
13749 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013750
Olivier Houchardc7566002018-11-20 23:33:50 +010013751alpn <protocols>
13752 This enables the TLS ALPN extension and advertises the specified protocol
13753 list as supported on top of ALPN. The protocol list consists in a comma-
13754 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050013755 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010013756 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
13757 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
13758 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
13759 now obsolete NPN extension.
13760 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
13761 be advertised, in order of preference, like below :
13762
13763 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
13764
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013765backup
13766 When "backup" is present on a server line, the server is only used in load
13767 balancing when all other non-backup servers are unavailable. Requests coming
13768 with a persistence cookie referencing the server will always be served
13769 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010013770 the "allbackups" option is set in the backend. See also the "no-backup" and
13771 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013772
Emeric Brunef42d922012-10-11 16:11:36 +020013773ca-file <cafile>
13774 This setting is only available when support for OpenSSL was built in. It
13775 designates a PEM file from which to load CA certificates used to verify
13776 server's certificate.
13777
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013778check
Jerome Magnin90702bc2020-04-26 14:23:04 +020013779 This option enables health checks on a server:
13780 - when not set, no health checking is performed, and the server is always
13781 considered available.
13782 - when set and no other check method is configured, the server is considered
13783 available when a connection can be established at the highest configured
13784 transport layer. This means TCP by default, or SSL/TLS when "ssl" or
13785 "check-ssl" are set, both possibly combined with connection prefixes such
13786 as a PROXY protocol header when "send-proxy" or "check-send-proxy" are
13787 set.
13788 - when set and an application-level health check is defined, the
13789 application-level exchanges are performed on top of the configured
13790 transport layer and the server is considered available if all of the
13791 exchanges succeed.
13792
13793 By default, health checks are performed on the same address and port as
13794 configured on the server, using the same encapsulation parameters (SSL/TLS,
13795 proxy-protocol header, etc... ). It is possible to change the destination
13796 address using "addr" and the port using "port". When done, it is assumed the
13797 server isn't checked on the service port, and configured encapsulation
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +050013798 parameters are not reused. One must explicitly set "check-send-proxy" to send
Jerome Magnin90702bc2020-04-26 14:23:04 +020013799 connection headers, "check-ssl" to use SSL/TLS.
13800
13801 When "sni" or "alpn" are set on the server line, their value is not used for
13802 health checks and one must use "check-sni" or "check-alpn".
13803
13804 The default source address for health check traffic is the same as the one
13805 defined in the backend. It can be changed with the "source" keyword.
13806
13807 The interval between checks can be set using the "inter" keyword, and the
13808 "rise" and "fall" keywords can be used to define how many successful or
13809 failed health checks are required to flag a server available or not
13810 available.
13811
13812 Optional application-level health checks can be configured with "option
13813 httpchk", "option mysql-check" "option smtpchk", "option pgsql-check",
13814 "option ldap-check", or "option redis-check".
13815
13816 Example:
13817 # simple tcp check
13818 backend foo
13819 server s1 192.168.0.1:80 check
13820 # this does a tcp connect + tls handshake
13821 backend foo
13822 server s1 192.168.0.1:443 ssl check
13823 # simple tcp check is enough for check success
13824 backend foo
13825 option tcp-check
13826 tcp-check connect
13827 server s1 192.168.0.1:443 ssl check
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013828
Willy Tarreau6c16adc2012-10-05 00:04:16 +020013829check-send-proxy
13830 This option forces emission of a PROXY protocol line with outgoing health
13831 checks, regardless of whether the server uses send-proxy or not for the
13832 normal traffic. By default, the PROXY protocol is enabled for health checks
13833 if it is already enabled for normal traffic and if no "port" nor "addr"
13834 directive is present. However, if such a directive is present, the
13835 "check-send-proxy" option needs to be used to force the use of the
13836 protocol. See also the "send-proxy" option for more information.
13837
Olivier Houchard92150142018-12-21 19:47:01 +010013838check-alpn <protocols>
13839 Defines which protocols to advertise with ALPN. The protocol list consists in
13840 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
13841 (without quotes). If it is not set, the server ALPN is used.
13842
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013843check-proto <name>
13844 Forces the multiplexer's protocol to use for the server's health-check
13845 connections. It must be compatible with the health-check type (TCP or
13846 HTTP). It must also be usable on the backend side. The list of available
Christopher Faulet982e17d2021-03-26 14:44:18 +010013847 protocols is reported in haproxy -vv. The protocols properties are
13848 reported : the mode (TCP/HTTP), the side (FE/BE), the mux name and its flags.
13849
13850 Some protocols report errors on aborts (flag=CLEAN_ABRT). Some others are
13851 subject to the head-of-line blocking on server side (flag=HOL_RISK). Finally
13852 some protocols don't support upgrades (flag=NO_UPG). The HTX compatibility is
13853 also reported (flag=HTX).
13854
13855 Here are the protocols that may be used as argument to a "check-proto"
13856 directive on a server line:
13857
13858 h2 : mode=HTTP side=FE|BE mux=H2 flags=HTX|CLEAN_ABRT|HOL_RISK|NO_UPG
13859 fcgi : mode=HTTP side=BE mux=FCGI flags=HTX|HOL_RISK|NO_UPG
13860 h1 : mode=HTTP side=FE|BE mux=H1 flags=HTX|NO_UPG
13861 none : mode=TCP side=FE|BE mux=PASS flags=NO_UPG
13862
Daniel Corbett67a82712020-07-06 23:01:19 -040013863 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013864 protocol for health-check connections established to this server.
13865 If not defined, the server one will be used, if set.
13866
Jérôme Magninae9bb762018-12-09 16:08:26 +010013867check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020013868 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010013869 over SSL. It is only possible to use a string to set <sni>. If you want to
13870 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020013871
Willy Tarreau763a95b2012-10-04 23:15:39 +020013872check-ssl
13873 This option forces encryption of all health checks over SSL, regardless of
13874 whether the server uses SSL or not for the normal traffic. This is generally
13875 used when an explicit "port" or "addr" directive is specified and SSL health
13876 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013877 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020013878 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
13879 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013880 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010013881 See the "ssl" option for more information and "no-check-ssl" to disable
13882 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013883
Alexander Liu2a54bb72019-05-22 19:44:48 +080013884check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050013885 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080013886 default, the health checks won't go through socks tunnel even it was enabled
13887 for normal traffic.
13888
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013889ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013890 This setting is only available when support for OpenSSL was built in. This
13891 option sets the string describing the list of cipher algorithms that is
13892 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013893 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
13894 information and recommendations see e.g.
13895 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
13896 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
13897 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013898
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013899ciphersuites <ciphersuites>
13900 This setting is only available when support for OpenSSL was built in and
13901 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
13902 describing the list of cipher algorithms that is negotiated during the TLS
13903 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013904 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
13905 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
13906 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013907
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013908cookie <value>
13909 The "cookie" parameter sets the cookie value assigned to the server to
13910 <value>. This value will be checked in incoming requests, and the first
13911 operational server possessing the same value will be selected. In return, in
13912 cookie insertion or rewrite modes, this value will be assigned to the cookie
13913 sent to the client. There is nothing wrong in having several servers sharing
13914 the same cookie value, and it is in fact somewhat common between normal and
13915 backup servers. See also the "cookie" keyword in backend section.
13916
Emeric Brunef42d922012-10-11 16:11:36 +020013917crl-file <crlfile>
13918 This setting is only available when support for OpenSSL was built in. It
13919 designates a PEM file from which to load certificate revocation list used
13920 to verify server's certificate.
13921
Emeric Bruna7aa3092012-10-26 12:58:00 +020013922crt <cert>
13923 This setting is only available when support for OpenSSL was built in.
13924 It designates a PEM file from which to load both a certificate and the
13925 associated private key. This file can be built by concatenating both PEM
13926 files into one. This certificate will be sent if the server send a client
13927 certificate request.
13928
Willy Tarreau96839092010-03-29 10:02:24 +020013929disabled
13930 The "disabled" keyword starts the server in the "disabled" state. That means
13931 that it is marked down in maintenance mode, and no connection other than the
13932 ones allowed by persist mode will reach it. It is very well suited to setup
13933 new servers, because normal traffic will never reach them, while it is still
13934 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013935 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020013936
Frédéric Lécailled2376272017-03-21 18:52:12 +010013937enabled
13938 This option may be used as 'server' setting to reset any 'disabled'
13939 setting which would have been inherited from 'default-server' directive as
13940 default value.
13941 It may also be used as 'default-server' setting to reset any previous
13942 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020013943
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013944error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010013945 If health observing is enabled, the "error-limit" parameter specifies the
13946 number of consecutive errors that triggers event selected by the "on-error"
13947 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013948
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013949 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013950
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013951fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013952 The "fall" parameter states that a server will be considered as dead after
13953 <count> consecutive unsuccessful health checks. This value defaults to 3 if
13954 unspecified. See also the "check", "inter" and "rise" parameters.
13955
Emeric Brun8694b9a2012-10-05 14:39:07 +020013956force-sslv3
13957 This option enforces use of SSLv3 only when SSL is used to communicate with
13958 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013959 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013960 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013961
13962force-tlsv10
13963 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013964 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013965 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013966
13967force-tlsv11
13968 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013969 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013970 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013971
13972force-tlsv12
13973 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013974 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013975 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013976
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013977force-tlsv13
13978 This option enforces use of TLSv1.3 only when SSL is used to communicate with
13979 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013980 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013981
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013982id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020013983 Set a persistent ID for the server. This ID must be positive and unique for
13984 the proxy. An unused ID will automatically be assigned if unset. The first
13985 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013986
Willy Tarreau6a031d12016-11-07 19:42:35 +010013987init-addr {last | libc | none | <ip>},[...]*
13988 Indicate in what order the server's address should be resolved upon startup
13989 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010013990 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010013991 method which succeeds is used. If the end of the list is reached without
13992 finding a working method, an error is thrown. Method "last" suggests to pick
13993 the address which appears in the state file (see "server-state-file"). Method
13994 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
13995 depending on the operating system and build options). Method "none"
13996 specifically indicates that the server should start without any valid IP
13997 address in a down state. It can be useful to ignore some DNS issues upon
13998 startup, waiting for the situation to get fixed later. Finally, an IP address
13999 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014000 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010014001 server used to catch old sessions and present them with a decent error
14002 message for example. When the "first" load balancing algorithm is used, this
14003 IP address could point to a fake server used to trigger the creation of new
14004 instances on the fly. This option defaults to "last,libc" indicating that the
14005 previous address found in the state file (if any) is used first, otherwise
14006 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014007 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010014008
14009 Example:
14010 defaults
14011 # never fail on address resolution
14012 default-server init-addr last,libc,none
14013
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014014inter <delay>
14015fastinter <delay>
14016downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014017 The "inter" parameter sets the interval between two consecutive health checks
14018 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
14019 It is also possible to use "fastinter" and "downinter" to optimize delays
14020 between checks depending on the server state :
14021
Pieter Baauw44fc9df2015-09-17 21:30:46 +020014022 Server state | Interval used
14023 ----------------------------------------+----------------------------------
14024 UP 100% (non-transitional) | "inter"
14025 ----------------------------------------+----------------------------------
14026 Transitionally UP (going down "fall"), | "fastinter" if set,
14027 Transitionally DOWN (going up "rise"), | "inter" otherwise.
14028 or yet unchecked. |
14029 ----------------------------------------+----------------------------------
14030 DOWN 100% (non-transitional) | "downinter" if set,
14031 | "inter" otherwise.
14032 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010014033
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014034 Just as with every other time-based parameter, they can be entered in any
14035 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
14036 serves as a timeout for health checks sent to servers if "timeout check" is
14037 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090014038 hosted on the same hardware, the agent and health checks of all servers
14039 are started with a small time offset between them. It is also possible to
14040 add some random noise in the agent and health checks interval using the
14041 global "spread-checks" keyword. This makes sense for instance when a lot
14042 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014043
Emeric Brun97556472020-05-30 01:42:45 +020014044log-proto <logproto>
14045 The "log-proto" specifies the protocol used to forward event messages to
14046 a server configured in a ring section. Possible values are "legacy"
14047 and "octet-count" corresponding respectively to "Non-transparent-framing"
14048 and "Octet counting" in rfc6587. "legacy" is the default.
14049
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014050maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014051 The "maxconn" parameter specifies the maximal number of concurrent
14052 connections that will be sent to this server. If the number of incoming
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010014053 concurrent connections goes higher than this value, they will be queued,
14054 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014055 save fragile servers from going down under extreme loads. If a "minconn"
14056 parameter is specified, the limit becomes dynamic. The default value is "0"
14057 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
14058 the backend's "fullconn" keyword.
14059
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010014060 In HTTP mode this parameter limits the number of concurrent requests instead
14061 of the number of connections. Multiple requests might be multiplexed over a
14062 single TCP connection to the server. As an example if you specify a maxconn
14063 of 50 you might see between 1 and 50 actual server connections, but no more
14064 than 50 concurrent requests.
14065
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014066maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014067 The "maxqueue" parameter specifies the maximal number of connections which
14068 will wait in the queue for this server. If this limit is reached, next
14069 requests will be redispatched to other servers instead of indefinitely
14070 waiting to be served. This will break persistence but may allow people to
Willy Tarreau8ae8c482020-10-22 17:19:07 +020014071 quickly re-log in when the server they try to connect to is dying. Some load
14072 balancing algorithms such as leastconn take this into account and accept to
14073 add requests into a server's queue up to this value if it is explicitly set
14074 to a value greater than zero, which often allows to better smooth the load
14075 when dealing with single-digit maxconn values. The default value is "0" which
14076 means the queue is unlimited. See also the "maxconn" and "minconn" parameters
14077 and "balance leastconn".
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014078
Willy Tarreau9c538e02019-01-23 10:21:49 +010014079max-reuse <count>
14080 The "max-reuse" argument indicates the HTTP connection processors that they
14081 should not reuse a server connection more than this number of times to send
14082 new requests. Permitted values are -1 (the default), which disables this
14083 limit, or any positive value. Value zero will effectively disable keep-alive.
14084 This is only used to work around certain server bugs which cause them to leak
14085 resources over time. The argument is not necessarily respected by the lower
14086 layers as there might be technical limitations making it impossible to
14087 enforce. At least HTTP/2 connections to servers will respect it.
14088
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014089minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014090 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
14091 limit following the backend's load. The server will always accept at least
14092 <minconn> connections, never more than <maxconn>, and the limit will be on
14093 the ramp between both values when the backend has less than <fullconn>
14094 concurrent connections. This makes it possible to limit the load on the
14095 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010014096 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014097 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014098
Willy Tarreaud72f0f32015-10-13 14:50:22 +020014099namespace <name>
14100 On Linux, it is possible to specify which network namespace a socket will
14101 belong to. This directive makes it possible to explicitly bind a server to
14102 a namespace different from the default one. Please refer to your operating
14103 system's documentation to find more details about network namespaces.
14104
Frédéric Lécailled2376272017-03-21 18:52:12 +010014105no-agent-check
14106 This option may be used as "server" setting to reset any "agent-check"
14107 setting which would have been inherited from "default-server" directive as
14108 default value.
14109 It may also be used as "default-server" setting to reset any previous
14110 "default-server" "agent-check" setting.
14111
14112no-backup
14113 This option may be used as "server" setting to reset any "backup"
14114 setting which would have been inherited from "default-server" directive as
14115 default value.
14116 It may also be used as "default-server" setting to reset any previous
14117 "default-server" "backup" setting.
14118
14119no-check
14120 This option may be used as "server" setting to reset any "check"
14121 setting which would have been inherited from "default-server" directive as
14122 default value.
14123 It may also be used as "default-server" setting to reset any previous
14124 "default-server" "check" setting.
14125
14126no-check-ssl
14127 This option may be used as "server" setting to reset any "check-ssl"
14128 setting which would have been inherited from "default-server" directive as
14129 default value.
14130 It may also be used as "default-server" setting to reset any previous
14131 "default-server" "check-ssl" setting.
14132
Frédéric Lécailled2376272017-03-21 18:52:12 +010014133no-send-proxy
14134 This option may be used as "server" setting to reset any "send-proxy"
14135 setting which would have been inherited from "default-server" directive as
14136 default value.
14137 It may also be used as "default-server" setting to reset any previous
14138 "default-server" "send-proxy" setting.
14139
14140no-send-proxy-v2
14141 This option may be used as "server" setting to reset any "send-proxy-v2"
14142 setting which would have been inherited from "default-server" directive as
14143 default value.
14144 It may also be used as "default-server" setting to reset any previous
14145 "default-server" "send-proxy-v2" setting.
14146
14147no-send-proxy-v2-ssl
14148 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
14149 setting which would have been inherited from "default-server" directive as
14150 default value.
14151 It may also be used as "default-server" setting to reset any previous
14152 "default-server" "send-proxy-v2-ssl" setting.
14153
14154no-send-proxy-v2-ssl-cn
14155 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
14156 setting which would have been inherited from "default-server" directive as
14157 default value.
14158 It may also be used as "default-server" setting to reset any previous
14159 "default-server" "send-proxy-v2-ssl-cn" setting.
14160
14161no-ssl
14162 This option may be used as "server" setting to reset any "ssl"
14163 setting which would have been inherited from "default-server" directive as
14164 default value.
14165 It may also be used as "default-server" setting to reset any previous
14166 "default-server" "ssl" setting.
14167
William Dauchyf6370442020-11-14 19:25:33 +010014168 Note that using `default-server ssl` setting and `no-ssl` on server will
14169 however init SSL connection, so it can be later be enabled through the
14170 runtime API: see `set server` commands in management doc.
14171
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010014172no-ssl-reuse
14173 This option disables SSL session reuse when SSL is used to communicate with
14174 the server. It will force the server to perform a full handshake for every
14175 new connection. It's probably only useful for benchmarking, troubleshooting,
14176 and for paranoid users.
14177
Emeric Brun9b3009b2012-10-05 11:55:06 +020014178no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014179 This option disables support for SSLv3 when SSL is used to communicate with
14180 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014181 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014182
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014183 Supported in default-server: No
14184
Emeric Brunf9c5c472012-10-11 15:28:34 +020014185no-tls-tickets
14186 This setting is only available when support for OpenSSL was built in. It
14187 disables the stateless session resumption (RFC 5077 TLS Ticket
14188 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014189 session resumption is more expensive in CPU usage for servers. This option
14190 is also available on global statement "ssl-default-server-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010014191 The TLS ticket mechanism is only used up to TLS 1.2.
14192 Forward Secrecy is compromised with TLS tickets, unless ticket keys
14193 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010014194 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020014195
Emeric Brun9b3009b2012-10-05 11:55:06 +020014196no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020014197 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020014198 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14199 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014200 often makes sense to disable it when communicating with local servers. This
14201 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014202 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014203
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014204 Supported in default-server: No
14205
Emeric Brun9b3009b2012-10-05 11:55:06 +020014206no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020014207 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020014208 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14209 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014210 often makes sense to disable it when communicating with local servers. This
14211 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014212 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014213
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014214 Supported in default-server: No
14215
Emeric Brun9b3009b2012-10-05 11:55:06 +020014216no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020014217 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014218 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14219 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010014220 often makes sense to disable it when communicating with local servers. This
14221 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014222 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020014223
14224 Supported in default-server: No
14225
14226no-tlsv13
14227 This option disables support for TLSv1.3 when SSL is used to communicate with
14228 the server. Note that SSLv2 is disabled in the code and cannot be enabled
14229 using any configuration option. TLSv1 is more expensive than SSLv3 so it
14230 often makes sense to disable it when communicating with local servers. This
14231 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014232 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014233
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020014234 Supported in default-server: No
14235
Frédéric Lécailled2376272017-03-21 18:52:12 +010014236no-verifyhost
14237 This option may be used as "server" setting to reset any "verifyhost"
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" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014242
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020014243no-tfo
14244 This option may be used as "server" setting to reset any "tfo"
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" "tfo" setting.
14249
Simon Hormanfa461682011-06-25 09:39:49 +090014250non-stick
14251 Never add connections allocated to this sever to a stick-table.
14252 This may be used in conjunction with backup to ensure that
14253 stick-table persistence is disabled for backup servers.
14254
Olivier Houchardc7566002018-11-20 23:33:50 +010014255npn <protocols>
14256 This enables the NPN TLS extension and advertises the specified protocol list
14257 as supported on top of NPN. The protocol list consists in a comma-delimited
14258 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050014259 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010014260 enabled (check with haproxy -vv). Note that the NPN extension has been
14261 replaced with the ALPN extension (see the "alpn" keyword), though this one is
14262 only available starting with OpenSSL 1.0.2.
14263
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014264observe <mode>
14265 This option enables health adjusting based on observing communication with
14266 the server. By default this functionality is disabled and enabling it also
14267 requires to enable health checks. There are two supported modes: "layer4" and
14268 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
14269 significant. In layer7, which is only allowed for http proxies, responses
14270 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010014271 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014272
14273 See also the "check", "on-error" and "error-limit".
14274
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014275on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010014276 Select what should happen when enough consecutive errors are detected.
14277 Currently, four modes are available:
14278 - fastinter: force fastinter
14279 - fail-check: simulate a failed check, also forces fastinter (default)
14280 - sudden-death: simulate a pre-fatal failed health check, one more failed
14281 check will mark a server down, forces fastinter
14282 - mark-down: mark the server immediately down and force fastinter
14283
14284 See also the "check", "observe" and "error-limit".
14285
Simon Hormane0d1bfb2011-06-21 14:34:58 +090014286on-marked-down <action>
14287 Modify what occurs when a server is marked down.
14288 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014289 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
14290 all connections to the server are immediately terminated when the server
14291 goes down. It might be used if the health check detects more complex cases
14292 than a simple connection status, and long timeouts would cause the service
14293 to remain unresponsive for too long a time. For instance, a health check
14294 might detect that a database is stuck and that there's no chance to reuse
14295 existing connections anymore. Connections killed this way are logged with
14296 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090014297
14298 Actions are disabled by default
14299
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014300on-marked-up <action>
14301 Modify what occurs when a server is marked up.
14302 Currently one action is available:
14303 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
14304 done only if the server is not in backup state and if it is not disabled
14305 (it must have an effective weight > 0). This can be used sometimes to force
14306 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014307 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
14308 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014309 with extreme care. Sessions killed because a server comes up are logged
14310 with an 'U' termination code (for "Up").
14311
14312 Actions are disabled by default
14313
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020014314pool-low-conn <max>
14315 Set a low threshold on the number of idling connections for a server, below
14316 which a thread will not try to steal a connection from another thread. This
14317 can be useful to improve CPU usage patterns in scenarios involving many very
14318 fast servers, in order to ensure all threads will keep a few idle connections
14319 all the time instead of letting them accumulate over one thread and migrating
14320 them from thread to thread. Typical values of twice the number of threads
14321 seem to show very good performance already with sub-millisecond response
14322 times. The default is zero, indicating that any idle connection can be used
14323 at any time. It is the recommended setting for normal use. This only applies
14324 to connections that can be shared according to the same principles as those
Willy Tarreau0784db82021-02-19 11:45:22 +010014325 applying to "http-reuse". In case connection sharing between threads would
14326 be disabled via "tune.idle-pool.shared", it can become very important to use
14327 this setting to make sure each thread always has a few connections, or the
14328 connection reuse rate will decrease as thread count increases.
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020014329
Olivier Houchard006e3102018-12-10 18:30:32 +010014330pool-max-conn <max>
14331 Set the maximum number of idling connections for a server. -1 means unlimited
14332 connections, 0 means no idle connections. The default is -1. When idle
14333 connections are enabled, orphaned idle connections which do not belong to any
14334 client session anymore are moved to a dedicated pool so that they remain
14335 usable by future clients. This only applies to connections that can be shared
14336 according to the same principles as those applying to "http-reuse".
14337
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010014338pool-purge-delay <delay>
14339 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010014340 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020014341 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010014342
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014343port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014344 Using the "port" parameter, it becomes possible to use a different port to
William Dauchy4858fb22021-02-03 22:30:09 +010014345 send health-checks or to probe the agent-check. On some servers, it may be
14346 desirable to dedicate a port to a specific component able to perform complex
14347 tests which are more suitable to health-checks than the application. It is
14348 common to run a simple script in inetd for instance. This parameter is
14349 ignored if the "check" parameter is not set. See also the "addr" parameter.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014350
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014351proto <name>
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014352 Forces the multiplexer's protocol to use for the outgoing connections to this
14353 server. It must be compatible with the mode of the backend (TCP or HTTP). It
14354 must also be usable on the backend side. The list of available protocols is
Christopher Faulet982e17d2021-03-26 14:44:18 +010014355 reported in haproxy -vv.The protocols properties are reported : the mode
14356 (TCP/HTTP), the side (FE/BE), the mux name and its flags.
14357
14358 Some protocols report errors on aborts (flag=CLEAN_ABRT). Some others are
14359 subject to the head-of-line blocking on server side (flag=HOL_RISK). Finally
14360 some protocols don't support upgrades (flag=NO_UPG). The HTX compatibility is
14361 also reported (flag=HTX).
14362
14363 Here are the protocols that may be used as argument to a "proto" directive on
14364 a server line :
14365
14366 h2 : mode=HTTP side=FE|BE mux=H2 flags=HTX|CLEAN_ABRT|HOL_RISK|NO_UPG
14367 fcgi : mode=HTTP side=BE mux=FCGI flags=HTX|HOL_RISK|NO_UPG
14368 h1 : mode=HTTP side=FE|BE mux=H1 flags=HTX|NO_UPG
14369 none : mode=TCP side=FE|BE mux=PASS flags=NO_UPG
14370
Daniel Corbett67a82712020-07-06 23:01:19 -040014371 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014372 protocol for all connections established to this server.
14373
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014374redir <prefix>
14375 The "redir" parameter enables the redirection mode for all GET and HEAD
14376 requests addressing this server. This means that instead of having HAProxy
14377 forward the request to the server, it will send an "HTTP 302" response with
14378 the "Location" header composed of this prefix immediately followed by the
14379 requested URI beginning at the leading '/' of the path component. That means
14380 that no trailing slash should be used after <prefix>. All invalid requests
14381 will be rejected, and all non-GET or HEAD requests will be normally served by
14382 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010014383 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014384 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014385 users to a remote location in case of local disaster. Main use consists in
14386 increasing bandwidth for static servers by having the clients directly
14387 connect to them. Note: never use a relative location here, it would cause a
14388 loop between the client and HAProxy!
14389
14390 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
14391
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014392rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014393 The "rise" parameter states that a server will be considered as operational
14394 after <count> consecutive successful health checks. This value defaults to 2
14395 if unspecified. See also the "check", "inter" and "fall" parameters.
14396
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020014397resolve-opts <option>,<option>,...
14398 Comma separated list of options to apply to DNS resolution linked to this
14399 server.
14400
14401 Available options:
14402
14403 * allow-dup-ip
14404 By default, HAProxy prevents IP address duplication in a backend when DNS
14405 resolution at runtime is in operation.
14406 That said, for some cases, it makes sense that two servers (in the same
14407 backend, being resolved by the same FQDN) have the same IP address.
14408 For such case, simply enable this option.
14409 This is the opposite of prevent-dup-ip.
14410
Daniel Corbettf8716912019-11-17 09:48:56 -050014411 * ignore-weight
14412 Ignore any weight that is set within an SRV record. This is useful when
14413 you would like to control the weights using an alternate method, such as
14414 using an "agent-check" or through the runtime api.
14415
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020014416 * prevent-dup-ip
14417 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
14418 an IP address already set to a server in the same backend and sharing the
14419 same fqdn.
14420 This is the opposite of allow-dup-ip.
14421
14422 Example:
14423 backend b_myapp
14424 default-server init-addr none resolvers dns
14425 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
14426 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
14427
14428 With the option allow-dup-ip set:
14429 * if the nameserver returns a single IP address, then both servers will use
14430 it
14431 * If the nameserver returns 2 IP addresses, then each server will pick up a
14432 different address
14433
14434 Default value: not set
14435
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014436resolve-prefer <family>
14437 When DNS resolution is enabled for a server and multiple IP addresses from
14438 different families are returned, HAProxy will prefer using an IP address
14439 from the family mentioned in the "resolve-prefer" parameter.
14440 Available families: "ipv4" and "ipv6"
14441
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020014442 Default value: ipv6
14443
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014444 Example:
14445
14446 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014447
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014448resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050014449 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014450 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010014451 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010014452 different datacenters. The latency between datacenter is not negligible, so
14453 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014454 configured network, another address is selected.
14455
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014456 Example:
14457
14458 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014459
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014460resolvers <id>
14461 Points to an existing "resolvers" section to resolve current server's
14462 hostname.
14463
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014464 Example:
14465
14466 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014467
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014468 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014469
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010014470send-proxy
14471 The "send-proxy" parameter enforces use of the PROXY protocol over any
14472 connection established to this server. The PROXY protocol informs the other
14473 end about the layer 3/4 addresses of the incoming connection, so that it can
14474 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014475 upper layer protocol. For connections accepted by an "accept-proxy" or
14476 "accept-netscaler-cip" listener, the advertised address will be used. Only
14477 TCPv4 and TCPv6 address families are supported. Other families such as
14478 Unix sockets, will report an UNKNOWN family. Servers using this option can
14479 fully be chained to another instance of haproxy listening with an
14480 "accept-proxy" setting. This setting must not be used if the server isn't
14481 aware of the protocol. When health checks are sent to the server, the PROXY
14482 protocol is automatically used when this option is set, unless there is an
14483 explicit "port" or "addr" directive, in which case an explicit
14484 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014485 See also the "no-send-proxy" option of this section and "accept-proxy" and
14486 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010014487
David Safb76832014-05-08 23:42:08 -040014488send-proxy-v2
14489 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
14490 over any connection established to this server. The PROXY protocol informs
14491 the other end about the layer 3/4 addresses of the incoming connection, so
14492 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020014493 whatever the upper layer protocol. It also send ALPN information if an alpn
14494 have been negotiated. This setting must not be used if the server isn't aware
14495 of this version of the protocol. See also the "no-send-proxy-v2" option of
14496 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014497
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010014498proxy-v2-options <option>[,<option>]*
Tim Duesterhuscf6e0c82020-03-13 12:34:24 +010014499 The "proxy-v2-options" parameter add options to send in PROXY protocol
14500 version 2 when "send-proxy-v2" is used. Options available are:
14501
14502 - ssl : See also "send-proxy-v2-ssl".
14503 - cert-cn : See also "send-proxy-v2-ssl-cn".
14504 - ssl-cipher: Name of the used cipher.
14505 - cert-sig : Signature algorithm of the used certificate.
14506 - cert-key : Key algorithm of the used certificate
14507 - authority : Host name value passed by the client (only SNI from a TLS
14508 connection is supported).
14509 - crc32c : Checksum of the PROXYv2 header.
14510 - unique-id : Send a unique ID generated using the frontend's
14511 "unique-id-format" within the PROXYv2 header.
14512 This unique-id is primarily meant for "mode tcp". It can
14513 lead to unexpected results in "mode http", because the
14514 generated unique ID is also used for the first HTTP request
14515 within a Keep-Alive connection.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010014516
David Safb76832014-05-08 23:42:08 -040014517send-proxy-v2-ssl
14518 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
14519 2 over any connection established to this server. The PROXY protocol informs
14520 the other end about the layer 3/4 addresses of the incoming connection, so
14521 that it can know the client's address or the public address it accessed to,
14522 whatever the upper layer protocol. In addition, the SSL information extension
14523 of the PROXY protocol is added to the PROXY protocol header. This setting
14524 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014525 See also the "no-send-proxy-v2-ssl" option of this section and the
14526 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014527
14528send-proxy-v2-ssl-cn
14529 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
14530 2 over any connection established to this server. The PROXY protocol informs
14531 the other end about the layer 3/4 addresses of the incoming connection, so
14532 that it can know the client's address or the public address it accessed to,
14533 whatever the upper layer protocol. In addition, the SSL information extension
14534 of the PROXY protocol, along along with the Common Name from the subject of
14535 the client certificate (if any), is added to the PROXY protocol header. This
14536 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014537 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
14538 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014539
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014540slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014541 The "slowstart" parameter for a server accepts a value in milliseconds which
14542 indicates after how long a server which has just come back up will run at
14543 full speed. Just as with every other time-based parameter, it can be entered
14544 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
14545 linearly from 0 to 100% during this time. The limitation applies to two
14546 parameters :
14547
14548 - maxconn: the number of connections accepted by the server will grow from 1
14549 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
14550
14551 - weight: when the backend uses a dynamic weighted algorithm, the weight
14552 grows linearly from 1 to 100%. In this case, the weight is updated at every
14553 health-check. For this reason, it is important that the "inter" parameter
14554 is smaller than the "slowstart", in order to maximize the number of steps.
14555
14556 The slowstart never applies when haproxy starts, otherwise it would cause
14557 trouble to running servers. It only applies when a server has been previously
14558 seen as failed.
14559
Willy Tarreau732eac42015-07-09 11:40:25 +020014560sni <expression>
14561 The "sni" parameter evaluates the sample fetch expression, converts it to a
14562 string and uses the result as the host name sent in the SNI TLS extension to
14563 the server. A typical use case is to send the SNI received from the client in
14564 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020014565 expression, though alternatives such as req.hdr(host) can also make sense. If
14566 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014567 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010014568 "verify" directive for more details. If you want to set a SNI for health
14569 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020014570
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014571source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020014572source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014573source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014574 The "source" parameter sets the source address which will be used when
14575 connecting to the server. It follows the exact same parameters and principle
14576 as the backend "source" keyword, except that it only applies to the server
14577 referencing it. Please consult the "source" keyword for details.
14578
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014579 Additionally, the "source" statement on a server line allows one to specify a
14580 source port range by indicating the lower and higher bounds delimited by a
14581 dash ('-'). Some operating systems might require a valid IP address when a
14582 source port range is specified. It is permitted to have the same IP/range for
14583 several servers. Doing so makes it possible to bypass the maximum of 64k
14584 total concurrent connections. The limit will then reach 64k connections per
14585 server.
14586
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000014587 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
14588 specifying the source address without port(s).
14589
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014590ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020014591 This option enables SSL ciphering on outgoing connections to the server. It
14592 is critical to verify server certificates using "verify" when using SSL to
14593 connect to servers, otherwise the communication is prone to trivial man in
14594 the-middle attacks rendering SSL useless. When this option is used, health
14595 checks are automatically sent in SSL too unless there is a "port" or an
14596 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014597 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
14598 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014599
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014600ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
14601 This option enforces use of <version> or lower when SSL is used to communicate
14602 with the server. This option is also available on global statement
14603 "ssl-default-server-options". See also "ssl-min-ver".
14604
14605ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
14606 This option enforces use of <version> or upper when SSL is used to communicate
14607 with the server. This option is also available on global statement
14608 "ssl-default-server-options". See also "ssl-max-ver".
14609
Frédéric Lécailled2376272017-03-21 18:52:12 +010014610ssl-reuse
14611 This option may be used as "server" setting to reset any "no-ssl-reuse"
14612 setting which would have been inherited from "default-server" directive as
14613 default value.
14614 It may also be used as "default-server" setting to reset any previous
14615 "default-server" "no-ssl-reuse" setting.
14616
14617stick
14618 This option may be used as "server" setting to reset any "non-stick"
14619 setting which would have been inherited from "default-server" directive as
14620 default value.
14621 It may also be used as "default-server" setting to reset any previous
14622 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014623
Alexander Liu2a54bb72019-05-22 19:44:48 +080014624socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050014625 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080014626 server. Using this option won't force the health check to go via socks4 by
14627 default. You will have to use the keyword "check-via-socks4" to enable it.
14628
Willy Tarreau163d4622015-10-13 16:16:41 +020014629tcp-ut <delay>
14630 Sets the TCP User Timeout for all outgoing connections to this server. This
14631 option is available on Linux since version 2.6.37. It allows haproxy to
14632 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010014633 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020014634 long-lived connections experiencing long idle periods such as remote
14635 terminals or database connection pools, where the client and server timeouts
14636 must remain high to allow a long period of idle, but where it is important to
14637 detect that the server has disappeared in order to release all resources
14638 associated with its connection (and the client's session). One typical use
14639 case is also to force dead server connections to die when health checks are
14640 too slow or during a soft reload since health checks are then disabled. The
14641 argument is a delay expressed in milliseconds by default. This only works for
14642 regular TCP connections, and is ignored for other protocols.
14643
Willy Tarreau034c88c2017-01-23 23:36:45 +010014644tfo
14645 This option enables using TCP fast open when connecting to servers, on
14646 systems that support it (currently only the Linux kernel >= 4.11).
14647 See the "tfo" bind option for more information about TCP fast open.
14648 Please note that when using tfo, you should also use the "conn-failure",
14649 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020014650 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010014651
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014652track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020014653 This option enables ability to set the current state of the server by tracking
14654 another one. It is possible to track a server which itself tracks another
14655 server, provided that at the end of the chain, a server has health checks
14656 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014657 used, it has to be enabled on both proxies.
14658
Frédéric Lécailled2376272017-03-21 18:52:12 +010014659tls-tickets
14660 This option may be used as "server" setting to reset any "no-tls-tickets"
14661 setting which would have been inherited from "default-server" directive as
14662 default value.
Lukas Tribusbdb386d2020-03-10 00:56:09 +010014663 The TLS ticket mechanism is only used up to TLS 1.2.
14664 Forward Secrecy is compromised with TLS tickets, unless ticket keys
14665 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010014666 It may also be used as "default-server" setting to reset any previous
Bjoern Jacke5ab7eb62020-02-13 14:16:16 +010014667 "default-server" "no-tls-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014668
Emeric Brunef42d922012-10-11 16:11:36 +020014669verify [none|required]
14670 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010014671 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014672 certificate provided by the server is verified using CAs from 'ca-file' and
14673 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014674 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014675 the name passed using the "sni" directive, or if not provided, the static
14676 host name passed using the "verifyhost" directive. When no name is found, the
14677 certificate's names are ignored. For this reason, without SNI it's important
14678 to use "verifyhost". On verification failure the handshake is aborted. It is
14679 critically important to verify server certificates when using SSL to connect
14680 to servers, otherwise the communication is prone to trivial man-in-the-middle
14681 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
14682 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020014683
Evan Broderbe554312013-06-27 00:05:25 -070014684verifyhost <hostname>
14685 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014686 only takes effect if 'verify required' is also specified. This directive sets
14687 a default static hostname to check the server's certificate against when no
14688 SNI was used to connect to the server. If SNI is not used, this is the only
14689 way to enable hostname verification. This static hostname, when set, will
14690 also be used for health checks (which cannot provide an SNI value). If none
14691 of the hostnames in the certificate match the specified hostname, the
14692 handshake is aborted. The hostnames in the server-provided certificate may
14693 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070014694
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014695weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014696 The "weight" parameter is used to adjust the server's weight relative to
14697 other servers. All servers will receive a load proportional to their weight
14698 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020014699 load. The default weight is 1, and the maximal value is 256. A value of 0
14700 means the server will not participate in load-balancing but will still accept
14701 persistent connections. If this parameter is used to distribute the load
14702 according to server's capacity, it is recommended to start with values which
14703 can both grow and shrink, for instance between 10 and 100 to leave enough
14704 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014705
14706
Cyril Bonté46175dd2015-07-02 22:45:32 +0200147075.3. Server IP address resolution using DNS
14708-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014709
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014710HAProxy allows using a host name on the server line to retrieve its IP address
14711using name servers. By default, HAProxy resolves the name when parsing the
Thayne McCombscdbcca92021-01-07 21:24:41 -070014712configuration file, at startup and cache the result for the process's life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014713This is not sufficient in some cases, such as in Amazon where a server's IP
14714can change after a reboot or an ELB Virtual IP can change based on current
14715workload.
14716This chapter describes how HAProxy can be configured to process server's name
14717resolution at run time.
14718Whether run time server name resolution has been enable or not, HAProxy will
14719carry on doing the first resolution when parsing the configuration.
14720
14721
Cyril Bonté46175dd2015-07-02 22:45:32 +0200147225.3.1. Global overview
14723----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014724
14725As we've seen in introduction, name resolution in HAProxy occurs at two
14726different steps of the process life:
14727
14728 1. when starting up, HAProxy parses the server line definition and matches a
14729 host name. It uses libc functions to get the host name resolved. This
14730 resolution relies on /etc/resolv.conf file.
14731
Christopher Faulet67957bd2017-09-27 11:00:59 +020014732 2. at run time, HAProxy performs periodically name resolutions for servers
14733 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014734
14735A few other events can trigger a name resolution at run time:
14736 - when a server's health check ends up in a connection timeout: this may be
14737 because the server has a new IP address. So we need to trigger a name
14738 resolution to know this new IP.
14739
Christopher Faulet67957bd2017-09-27 11:00:59 +020014740When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010014741HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020014742SRV label is specified, then the corresponding SRV records will be retrieved
14743from the DNS server, and the provided hostnames will be used. The SRV label
14744will be checked periodically, and if any server are added or removed, haproxy
14745will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020014746
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014747A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050014748 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014749 first valid response.
14750
14751 - a resolution is considered as invalid (NX, timeout, refused), when all the
14752 servers return an error.
14753
14754
Cyril Bonté46175dd2015-07-02 22:45:32 +0200147555.3.2. The resolvers section
14756----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014757
14758This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020014759HAProxy. There can be as many as resolvers section as needed. Each section can
14760contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014761
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014762When multiple name servers are configured in a resolvers section, then HAProxy
14763uses the first valid response. In case of invalid responses, only the last one
14764is treated. Purpose is to give the chance to a slow server to deliver a valid
14765answer after a fast faulty or outdated server.
14766
14767When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020014768used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014769
Christopher Faulet67957bd2017-09-27 11:00:59 +020014770 1. HAProxy retries the same DNS query with a new query type. The A queries are
14771 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
14772 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014773
Christopher Faulet67957bd2017-09-27 11:00:59 +020014774 2. When the fallback on the query type was done (or not applicable), HAProxy
14775 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014776
Christopher Faulet67957bd2017-09-27 11:00:59 +020014777 3. HAProxy retries previous steps <resolve_retires> times. If no valid
14778 response is received after that, it stops the DNS resolution and reports
14779 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014780
Christopher Faulet67957bd2017-09-27 11:00:59 +020014781For example, with 2 name servers configured in a resolvers section, the
14782following scenarios are possible:
14783
14784 - First response is valid and is applied directly, second response is
14785 ignored
14786
14787 - First response is invalid and second one is valid, then second response is
14788 applied
14789
14790 - First response is a NX domain and second one a truncated response, then
14791 HAProxy retries the query with a new type
14792
14793 - First response is a NX domain and second one is a timeout, then HAProxy
14794 retries the query with a new type
14795
14796 - Query timed out for both name servers, then HAProxy retries it with the
14797 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014798
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014799As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
14800a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020014801<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014802
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014803
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014804resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014805 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014806
14807A resolvers section accept the following parameters:
14808
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014809accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014810 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020014811 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014812 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
14813 by RFC 6891)
14814
Emeric Brun4c751952021-03-08 16:41:29 +010014815 Note: the maximum allowed value is 65535. Recommended value for UDP is
14816 4096 and it is not recommended to exceed 8192 except if you are sure
14817 that your system and network can handle this (over 65507 makes no sense
14818 since is the maximum UDP payload size). If you are using only TCP
14819 nameservers to handle huge DNS responses, you should put this value
14820 to the max: 65535.
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020014821
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014822nameserver <id> <ip>:<port>
Emeric Brun56fc5d92021-02-12 20:05:45 +010014823 UDP DNS server description:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014824 <id> : label of the server, should be unique
14825 <ip> : IP address of the server
14826 <port> : port where the DNS service actually runs
14827
Emeric Brun56fc5d92021-02-12 20:05:45 +010014828server <name> <address> [param*]
14829 Used to configure a DNS TCP or stream server. This supports for all
14830 "server" parameters found in 5.2 paragraph. Some of these parameters
14831 are irrelevant for DNS resolving. Note: currently 4 queries are pipelined
14832 on the same connections. A batch of idle connections are removed every
14833 5 seconds. "maxconn" can be configured to limit the amount of those
14834 concurrent connections and TLS should also usable if the server supports.
14835
Ben Draut44e609b2018-05-29 15:40:08 -060014836parse-resolv-conf
14837 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
14838 list. Ordered as if each nameserver in /etc/resolv.conf was individually
14839 placed in the resolvers section in place of this directive.
14840
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014841hold <status> <period>
14842 Defines <period> during which the last name resolution should be kept based
14843 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014844 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014845 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014846 <period> : interval between two successive name resolution when the last
14847 answer was in <status>. It follows the HAProxy time format.
14848 <period> is in milliseconds by default.
14849
Baptiste Assmann686408b2017-08-18 10:15:42 +020014850 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014851
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014852resolve_retries <nb>
14853 Defines the number <nb> of queries to send to resolve a server name before
14854 giving up.
14855 Default value: 3
14856
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014857 A retry occurs on name server timeout or when the full sequence of DNS query
14858 type failover is over and we need to start up from the default ANY query
14859 type.
14860
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014861timeout <event> <time>
14862 Defines timeouts related to name resolution
14863 <event> : the event on which the <time> timeout period applies to.
14864 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014865 - resolve : default time to trigger name resolutions when no
14866 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014867 Default value: 1s
14868 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014869 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014870 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014871 <time> : time related to the event. It follows the HAProxy time format.
14872 <time> is expressed in milliseconds.
14873
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014874 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014875
14876 resolvers mydns
14877 nameserver dns1 10.0.0.1:53
14878 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060014879 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014880 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020014881 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014882 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014883 hold other 30s
14884 hold refused 30s
14885 hold nx 30s
14886 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014887 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014888 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014889
14890
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200148916. Cache
14892---------
14893
14894HAProxy provides a cache, which was designed to perform cache on small objects
14895(favicon, css...). This is a minimalist low-maintenance cache which runs in
14896RAM.
14897
14898The cache is based on a memory which is shared between processes and threads,
14899this memory is split in blocks of 1k.
14900
14901If an object is not used anymore, it can be deleted to store a new object
14902independently of its expiration date. The oldest objects are deleted first
14903when we try to allocate a new one.
14904
14905The cache uses a hash of the host header and the URI as the key.
14906
14907It's possible to view the status of a cache using the Unix socket command
14908"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
14909for more details.
14910
14911When an object is delivered from the cache, the server name in the log is
14912replaced by "<CACHE>".
14913
14914
149156.1. Limitation
14916----------------
14917
14918The cache won't store and won't deliver objects in these cases:
14919
14920- If the response is not a 200
Remi Tricot-Le Breton4f730832020-11-26 15:51:50 +010014921- If the response contains a Vary header and either the process-vary option is
14922 disabled, or a currently unmanaged header is specified in the Vary value (only
14923 accept-encoding and referer are managed for now)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020014924- If the Content-Length + the headers size is greater than "max-object-size"
14925- If the response is not cacheable
Remi Tricot-Le Bretond493bc82020-11-26 15:51:29 +010014926- If the response does not have an explicit expiration time (s-maxage or max-age
14927 Cache-Control directives or Expires header) or a validator (ETag or Last-Modified
14928 headers)
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010014929- If the process-vary option is enabled and there are already max-secondary-entries
14930 entries with the same primary key as the current response
Remi Tricot-Le Breton6ca89162021-01-07 14:50:51 +010014931- If the process-vary option is enabled and the response has an unknown encoding (not
14932 mentioned in https://www.iana.org/assignments/http-parameters/http-parameters.xhtml)
14933 while varying on the accept-encoding client header
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020014934
14935- If the request is not a GET
14936- If the HTTP version of the request is smaller than 1.1
14937- If the request contains an Authorization header
14938
14939
149406.2. Setup
14941-----------
14942
14943To setup a cache, you must define a cache section and use it in a proxy with
14944the corresponding http-request and response actions.
14945
14946
149476.2.1. Cache section
14948---------------------
14949
14950cache <name>
14951 Declare a cache section, allocate a shared cache memory named <name>, the
14952 size of cache is mandatory.
14953
14954total-max-size <megabytes>
14955 Define the size in RAM of the cache in megabytes. This size is split in
14956 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
14957
14958max-object-size <bytes>
14959 Define the maximum size of the objects to be cached. Must not be greater than
14960 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
14961 All objects with sizes larger than "max-object-size" will not be cached.
14962
14963max-age <seconds>
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010014964 Define the maximum expiration duration. The expiration is set as the lowest
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020014965 value between the s-maxage or max-age (in this order) directive in the
14966 Cache-Control response header and this value. The default value is 60
14967 seconds, which means that you can't cache an object more than 60 seconds by
14968 default.
14969
Remi Tricot-Le Bretone6cc5b52020-12-23 18:13:53 +010014970process-vary <on/off>
14971 Enable or disable the processing of the Vary header. When disabled, a response
Remi Tricot-Le Breton754b2422020-11-16 15:56:10 +010014972 containing such a header will never be cached. When enabled, we need to calculate
14973 a preliminary hash for a subset of request headers on all the incoming requests
14974 (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 +010014975 for a given request (see RFC 7234#4.1). The default value is off (disabled).
Remi Tricot-Le Breton754b2422020-11-16 15:56:10 +010014976
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010014977max-secondary-entries <number>
14978 Define the maximum number of simultaneous secondary entries with the same primary
14979 key in the cache. This needs the vary support to be enabled. Its default value is 10
14980 and should be passed a strictly positive integer.
14981
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020014982
149836.2.2. Proxy section
14984---------------------
14985
14986http-request cache-use <name> [ { if | unless } <condition> ]
14987 Try to deliver a cached object from the cache <name>. This directive is also
14988 mandatory to store the cache as it calculates the cache hash. If you want to
14989 use a condition for both storage and delivering that's a good idea to put it
14990 after this one.
14991
14992http-response cache-store <name> [ { if | unless } <condition> ]
14993 Store an http-response within the cache. The storage of the response headers
14994 is done at this step, which means you can use others http-response actions
14995 to modify headers before or after the storage of the response. This action
14996 is responsible for the setup of the cache storage filter.
14997
14998
14999Example:
15000
15001 backend bck1
15002 mode http
15003
15004 http-request cache-use foobar
15005 http-response cache-store foobar
15006 server srv1 127.0.0.1:80
15007
15008 cache foobar
15009 total-max-size 4
15010 max-age 240
15011
15012
Willy Tarreau74ca5042013-06-11 23:12:07 +0200150137. Using ACLs and fetching samples
15014----------------------------------
15015
Davor Ocelice9ed2812017-12-25 17:49:28 +010015016HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020015017client or server information, from tables, environmental information etc...
15018The action of extracting such data is called fetching a sample. Once retrieved,
15019these samples may be used for various purposes such as a key to a stick-table,
15020but most common usages consist in matching them against predefined constant
15021data called patterns.
15022
15023
150247.1. ACL basics
15025---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015026
15027The use of Access Control Lists (ACL) provides a flexible solution to perform
15028content switching and generally to take decisions based on content extracted
15029from the request, the response or any environmental status. The principle is
15030simple :
15031
Willy Tarreau74ca5042013-06-11 23:12:07 +020015032 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015033 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020015034 - apply one or multiple pattern matching methods on this sample
15035 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015036
Willy Tarreau74ca5042013-06-11 23:12:07 +020015037The actions generally consist in blocking a request, selecting a backend, or
15038adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015039
15040In order to define a test, the "acl" keyword is used. The syntax is :
15041
Willy Tarreau74ca5042013-06-11 23:12:07 +020015042 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015043
15044This creates a new ACL <aclname> or completes an existing one with new tests.
15045Those tests apply to the portion of request/response specified in <criterion>
15046and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015047an operator which may be specified before the set of values. Optionally some
15048conversion operators may be applied to the sample, and they will be specified
15049as a comma-delimited list of keywords just after the first keyword. The values
15050are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015051
15052ACL names must be formed from upper and lower case letters, digits, '-' (dash),
15053'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
15054which means that "my_acl" and "My_Acl" are two different ACLs.
15055
15056There is no enforced limit to the number of ACLs. The unused ones do not affect
15057performance, they just consume a small amount of memory.
15058
Willy Tarreau74ca5042013-06-11 23:12:07 +020015059The criterion generally is the name of a sample fetch method, or one of its ACL
15060specific declinations. The default test method is implied by the output type of
15061this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015062methods of a same sample fetch method. The sample fetch methods are the only
15063ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015064
15065Sample fetch methods return data which can be of the following types :
15066 - boolean
15067 - integer (signed or unsigned)
15068 - IPv4 or IPv6 address
15069 - string
15070 - data block
15071
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015072Converters transform any of these data into any of these. For example, some
15073converters might convert a string to a lower-case string while other ones
15074would turn a string to an IPv4 address, or apply a netmask to an IP address.
15075The resulting sample is of the type of the last converter applied to the list,
15076which defaults to the type of the sample fetch method.
15077
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015078Each sample or converter returns data of a specific type, specified with its
15079keyword in this documentation. When an ACL is declared using a standard sample
15080fetch method, certain types automatically involved a default matching method
15081which are summarized in the table below :
15082
15083 +---------------------+-----------------+
15084 | Sample or converter | Default |
15085 | output type | matching method |
15086 +---------------------+-----------------+
15087 | boolean | bool |
15088 +---------------------+-----------------+
15089 | integer | int |
15090 +---------------------+-----------------+
15091 | ip | ip |
15092 +---------------------+-----------------+
15093 | string | str |
15094 +---------------------+-----------------+
15095 | binary | none, use "-m" |
15096 +---------------------+-----------------+
15097
15098Note that in order to match a binary samples, it is mandatory to specify a
15099matching method, see below.
15100
Willy Tarreau74ca5042013-06-11 23:12:07 +020015101The ACL engine can match these types against patterns of the following types :
15102 - boolean
15103 - integer or integer range
15104 - IP address / network
15105 - string (exact, substring, suffix, prefix, subdir, domain)
15106 - regular expression
15107 - hex block
15108
Willy Tarreauc57f0e22009-05-10 13:12:33 +020015109The following ACL flags are currently supported :
15110
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015111 -i : ignore case during matching of all subsequent patterns.
15112 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015113 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010015114 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010015115 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010015116 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020015117 -- : force end of flags. Useful when a string looks like one of the flags.
15118
Willy Tarreau74ca5042013-06-11 23:12:07 +020015119The "-f" flag is followed by the name of a file from which all lines will be
15120read as individual values. It is even possible to pass multiple "-f" arguments
15121if the patterns are to be loaded from multiple files. Empty lines as well as
15122lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
15123will be stripped. If it is absolutely necessary to insert a valid pattern
15124beginning with a sharp, just prefix it with a space so that it is not taken for
15125a comment. Depending on the data type and match method, haproxy may load the
15126lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
15127exact string matching. In this case, duplicates will automatically be removed.
15128
Thierry FOURNIER9860c412014-01-29 14:23:29 +010015129The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
15130parsed as two column file. The first column contains the patterns used by the
15131ACL, and the second column contain the samples. The sample can be used later by
15132a map. This can be useful in some rare cases where an ACL would just be used to
15133check for the existence of a pattern in a map before a mapping is applied.
15134
Thierry FOURNIER3534d882014-01-20 17:01:44 +010015135The "-u" flag forces the unique id of the ACL. This unique id is used with the
15136socket interface to identify ACL and dynamically change its values. Note that a
15137file is always identified by its name even if an id is set.
15138
Willy Tarreau74ca5042013-06-11 23:12:07 +020015139Also, note that the "-i" flag applies to subsequent entries and not to entries
15140loaded from files preceding it. For instance :
15141
15142 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
15143
15144In this example, each line of "exact-ua.lst" will be exactly matched against
15145the "user-agent" header of the request. Then each line of "generic-ua" will be
15146case-insensitively matched. Then the word "test" will be insensitively matched
15147as well.
15148
15149The "-m" flag is used to select a specific pattern matching method on the input
15150sample. All ACL-specific criteria imply a pattern matching method and generally
15151do not need this flag. However, this flag is useful with generic sample fetch
15152methods to describe how they're going to be matched against the patterns. This
15153is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010015154obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020015155followed by a pattern matching method name, this method is used instead of the
15156default one for the criterion. This makes it possible to match contents in ways
15157that were not initially planned, or with sample fetch methods which return a
15158string. The matching method also affects the way the patterns are parsed.
15159
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010015160The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
15161By default, if the parser cannot parse ip address it considers that the parsed
15162string is maybe a domain name and try dns resolution. The flag "-n" disable this
15163resolution. It is useful for detecting malformed ip lists. Note that if the DNS
15164server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050015165waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010015166flag "-n" disable this behavior. Note also that during the runtime, this
15167function is disabled for the dynamic acl modifications.
15168
Willy Tarreau74ca5042013-06-11 23:12:07 +020015169There are some restrictions however. Not all methods can be used with all
15170sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
15171be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020015172
15173 - "found" : only check if the requested sample could be found in the stream,
15174 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020015175 to pass any pattern to avoid confusion. This matching method is
15176 particularly useful to detect presence of certain contents such
15177 as headers, cookies, etc... even if they are empty and without
15178 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015179
15180 - "bool" : check the value as a boolean. It can only be applied to fetches
15181 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015182 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015183
15184 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020015185 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015186
15187 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020015188 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015189
Davor Ocelice9ed2812017-12-25 17:49:28 +010015190 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020015191 binary sequence. This may be used with binary or string samples.
15192
15193 - "len" : match the sample's length as an integer. This may be used with
15194 binary or string samples.
15195
Willy Tarreau74ca5042013-06-11 23:12:07 +020015196 - "str" : exact match : match the contents against a string. This may be
15197 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015198
Willy Tarreau74ca5042013-06-11 23:12:07 +020015199 - "sub" : substring match : check that the contents contain at least one of
15200 the provided string patterns. This may be used with binary or
15201 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015202
Willy Tarreau74ca5042013-06-11 23:12:07 +020015203 - "reg" : regex match : match the contents against a list of regular
15204 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015205
Willy Tarreau74ca5042013-06-11 23:12:07 +020015206 - "beg" : prefix match : check that the contents begin like the provided
15207 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015208
Willy Tarreau74ca5042013-06-11 23:12:07 +020015209 - "end" : suffix match : check that the contents end like the provided
15210 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015211
Willy Tarreau74ca5042013-06-11 23:12:07 +020015212 - "dir" : subdir match : check that a slash-delimited portion of the
15213 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015214 This may be used with binary or string samples.
15215
Willy Tarreau74ca5042013-06-11 23:12:07 +020015216 - "dom" : domain match : check that a dot-delimited portion of the contents
15217 exactly match one of the provided string patterns. This may be
15218 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020015219
15220For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
15221request, it is possible to do :
15222
15223 acl jsess_present cook(JSESSIONID) -m found
15224
15225In order to apply a regular expression on the 500 first bytes of data in the
15226buffer, one would use the following acl :
15227
15228 acl script_tag payload(0,500) -m reg -i <script>
15229
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015230On systems where the regex library is much slower when using "-i", it is
15231possible to convert the sample to lowercase before matching, like this :
15232
15233 acl script_tag payload(0,500),lower -m reg <script>
15234
Willy Tarreau74ca5042013-06-11 23:12:07 +020015235All ACL-specific criteria imply a default matching method. Most often, these
15236criteria are composed by concatenating the name of the original sample fetch
15237method and the matching method. For example, "hdr_beg" applies the "beg" match
15238to samples retrieved using the "hdr" fetch method. Since all ACL-specific
15239criteria rely on a sample fetch method, it is always possible instead to use
15240the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015241
Willy Tarreau74ca5042013-06-11 23:12:07 +020015242If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015243the matching method is simply applied to the underlying sample fetch method.
15244For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015245
Willy Tarreau74ca5042013-06-11 23:12:07 +020015246 acl short_form hdr_beg(host) www.
15247 acl alternate1 hdr_beg(host) -m beg www.
15248 acl alternate2 hdr_dom(host) -m beg www.
15249 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015250
Willy Tarreau2b5285d2010-05-09 23:45:24 +020015251
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015252The table below summarizes the compatibility matrix between sample or converter
15253types and the pattern types to fetch against. It indicates for each compatible
15254combination the name of the matching method to be used, surrounded with angle
15255brackets ">" and "<" when the method is the default one and will work by
15256default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015257
Willy Tarreau74ca5042013-06-11 23:12:07 +020015258 +-------------------------------------------------+
15259 | Input sample type |
15260 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015261 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015262 +----------------------+---------+---------+---------+---------+---------+
15263 | none (presence only) | found | found | found | found | found |
15264 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015265 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015266 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015267 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015268 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015269 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015270 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015271 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015272 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020015273 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015274 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015275 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015276 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015277 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015278 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015279 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015280 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015281 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015282 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015283 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015284 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010015285 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020015286 +----------------------+---------+---------+---------+---------+---------+
15287 | hex block | | | | bin | bin |
15288 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020015289
15290
Willy Tarreau74ca5042013-06-11 23:12:07 +0200152917.1.1. Matching booleans
15292------------------------
15293
15294In order to match a boolean, no value is needed and all values are ignored.
15295Boolean matching is used by default for all fetch methods of type "boolean".
15296When boolean matching is used, the fetched value is returned as-is, which means
15297that a boolean "true" will always match and a boolean "false" will never match.
15298
15299Boolean matching may also be enforced using "-m bool" on fetch methods which
15300return an integer value. Then, integer value 0 is converted to the boolean
15301"false" and all other values are converted to "true".
15302
Willy Tarreau6a06a402007-07-15 20:15:28 +020015303
Willy Tarreau74ca5042013-06-11 23:12:07 +0200153047.1.2. Matching integers
15305------------------------
15306
15307Integer matching applies by default to integer fetch methods. It can also be
15308enforced on boolean fetches using "-m int". In this case, "false" is converted
15309to the integer 0, and "true" is converted to the integer 1.
15310
15311Integer matching also supports integer ranges and operators. Note that integer
15312matching only applies to positive values. A range is a value expressed with a
15313lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015314
15315For instance, "1024:65535" is a valid range to represent a range of
15316unprivileged ports, and "1024:" would also work. "0:1023" is a valid
15317representation of privileged ports, and ":1023" would also work.
15318
Willy Tarreau62644772008-07-16 18:36:06 +020015319As a special case, some ACL functions support decimal numbers which are in fact
15320two integers separated by a dot. This is used with some version checks for
15321instance. All integer properties apply to those decimal numbers, including
15322ranges and operators.
15323
Willy Tarreau6a06a402007-07-15 20:15:28 +020015324For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010015325operators with ranges does not make much sense and is strongly discouraged.
15326Similarly, it does not make much sense to perform order comparisons with a set
15327of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015328
Willy Tarreau0ba27502007-12-24 16:55:16 +010015329Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020015330
15331 eq : true if the tested value equals at least one value
15332 ge : true if the tested value is greater than or equal to at least one value
15333 gt : true if the tested value is greater than at least one value
15334 le : true if the tested value is less than or equal to at least one value
15335 lt : true if the tested value is less than at least one value
15336
Willy Tarreau0ba27502007-12-24 16:55:16 +010015337For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020015338
15339 acl negative-length hdr_val(content-length) lt 0
15340
Willy Tarreau62644772008-07-16 18:36:06 +020015341This one matches SSL versions between 3.0 and 3.1 (inclusive) :
15342
15343 acl sslv3 req_ssl_ver 3:3.1
15344
Willy Tarreau6a06a402007-07-15 20:15:28 +020015345
Willy Tarreau74ca5042013-06-11 23:12:07 +0200153467.1.3. Matching strings
15347-----------------------
15348
15349String matching applies to string or binary fetch methods, and exists in 6
15350different forms :
15351
15352 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015353 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020015354
15355 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015356 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020015357
15358 - prefix match (-m beg) : the patterns are compared with the beginning of
15359 the extracted string, and the ACL matches if any of them matches.
15360
15361 - suffix match (-m end) : the patterns are compared with the end of the
15362 extracted string, and the ACL matches if any of them matches.
15363
Baptiste Assmann33db6002016-03-06 23:32:10 +010015364 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020015365 string, delimited with slashes ("/"), and the ACL matches if any of them
15366 matches.
15367
15368 - domain match (-m dom) : the patterns are looked up inside the extracted
15369 string, delimited with dots ("."), and the ACL matches if any of them
15370 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015371
15372String matching applies to verbatim strings as they are passed, with the
15373exception of the backslash ("\") which makes it possible to escape some
15374characters such as the space. If the "-i" flag is passed before the first
15375string, then the matching will be performed ignoring the case. In order
15376to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010015377before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020015378
Mathias Weiersmuellercb250fc2019-12-02 09:43:40 +010015379Do not use string matches for binary fetches which might contain null bytes
15380(0x00), as the comparison stops at the occurrence of the first null byte.
15381Instead, convert the binary fetch to a hex string with the hex converter first.
15382
15383Example:
15384 # matches if the string <tag> is present in the binary sample
15385 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
15386
Willy Tarreau6a06a402007-07-15 20:15:28 +020015387
Willy Tarreau74ca5042013-06-11 23:12:07 +0200153887.1.4. Matching regular expressions (regexes)
15389---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020015390
15391Just like with string matching, regex matching applies to verbatim strings as
15392they are passed, with the exception of the backslash ("\") which makes it
15393possible to escape some characters such as the space. If the "-i" flag is
15394passed before the first regex, then the matching will be performed ignoring
15395the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010015396the "--" flag before the first string. Same principle applies of course to
15397match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020015398
15399
Willy Tarreau74ca5042013-06-11 23:12:07 +0200154007.1.5. Matching arbitrary data blocks
15401-------------------------------------
15402
15403It is possible to match some extracted samples against a binary block which may
15404not safely be represented as a string. For this, the patterns must be passed as
15405a series of hexadecimal digits in an even number, when the match method is set
15406to binary. Each sequence of two digits will represent a byte. The hexadecimal
15407digits may be used upper or lower case.
15408
15409Example :
15410 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
15411 acl hello payload(0,6) -m bin 48656c6c6f0a
15412
15413
154147.1.6. Matching IPv4 and IPv6 addresses
15415---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020015416
15417IPv4 addresses values can be specified either as plain addresses or with a
15418netmask appended, in which case the IPv4 address matches whenever it is
15419within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010015420host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010015421difficult to read and debug configurations. If hostnames are used, you should
15422at least ensure that they are present in /etc/hosts so that the configuration
15423does not depend on any random DNS match at the moment the configuration is
15424parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015425
Daniel Schnellereba56342016-04-13 00:26:52 +020015426The dotted IPv4 address notation is supported in both regular as well as the
15427abbreviated form with all-0-octets omitted:
15428
15429 +------------------+------------------+------------------+
15430 | Example 1 | Example 2 | Example 3 |
15431 +------------------+------------------+------------------+
15432 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
15433 | 192.168.1 | 10.12 | 127.1 |
15434 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
15435 | 192.168.1/22 | 10.12/8 | 127.1/8 |
15436 +------------------+------------------+------------------+
15437
15438Notice that this is different from RFC 4632 CIDR address notation in which
15439192.168.42/24 would be equivalent to 192.168.42.0/24.
15440
Willy Tarreauceb4ac92012-04-28 00:41:46 +020015441IPv6 may be entered in their usual form, with or without a netmask appended.
15442Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
15443trouble with randomly resolved IP addresses, host names are never allowed in
15444IPv6 patterns.
15445
15446HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
15447following situations :
15448 - tested address is IPv4, pattern address is IPv4, the match applies
15449 in IPv4 using the supplied mask if any.
15450 - tested address is IPv6, pattern address is IPv6, the match applies
15451 in IPv6 using the supplied mask if any.
15452 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
15453 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
15454 ::IPV4 or ::ffff:IPV4, otherwise it fails.
15455 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
15456 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
15457 applied in IPv6 using the supplied IPv6 mask.
15458
Willy Tarreau74ca5042013-06-11 23:12:07 +020015459
154607.2. Using ACLs to form conditions
15461----------------------------------
15462
15463Some actions are only performed upon a valid condition. A condition is a
15464combination of ACLs with operators. 3 operators are supported :
15465
15466 - AND (implicit)
15467 - OR (explicit with the "or" keyword or the "||" operator)
15468 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020015469
Willy Tarreau74ca5042013-06-11 23:12:07 +020015470A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020015471
Willy Tarreau74ca5042013-06-11 23:12:07 +020015472 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020015473
Willy Tarreau74ca5042013-06-11 23:12:07 +020015474Such conditions are generally used after an "if" or "unless" statement,
15475indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020015476
Willy Tarreau74ca5042013-06-11 23:12:07 +020015477For instance, to block HTTP requests to the "*" URL with methods other than
15478"OPTIONS", as well as POST requests without content-length, and GET or HEAD
15479requests with a content-length greater than 0, and finally every request which
15480is not either GET/HEAD/POST/OPTIONS !
15481
15482 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015483 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
15484 http-request deny if METH_GET HTTP_CONTENT
15485 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020015486
15487To select a different backend for requests to static contents on the "www" site
15488and to every request on the "img", "video", "download" and "ftp" hosts :
15489
15490 acl url_static path_beg /static /images /img /css
15491 acl url_static path_end .gif .png .jpg .css .js
15492 acl host_www hdr_beg(host) -i www
15493 acl host_static hdr_beg(host) -i img. video. download. ftp.
15494
Davor Ocelice9ed2812017-12-25 17:49:28 +010015495 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020015496 # of host "www". Use backend "www" for the rest.
15497 use_backend static if host_static or host_www url_static
15498 use_backend www if host_www
15499
15500It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
15501expressions that are built on the fly without needing to be declared. They must
15502be enclosed between braces, with a space before and after each brace (because
15503the braces must be seen as independent words). Example :
15504
15505 The following rule :
15506
15507 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015508 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020015509
15510 Can also be written that way :
15511
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015512 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020015513
15514It is generally not recommended to use this construct because it's a lot easier
15515to leave errors in the configuration when written that way. However, for very
15516simple rules matching only one source IP address for instance, it can make more
15517sense to use them than to declare ACLs with random names. Another example of
15518good use is the following :
15519
15520 With named ACLs :
15521
15522 acl site_dead nbsrv(dynamic) lt 2
15523 acl site_dead nbsrv(static) lt 2
15524 monitor fail if site_dead
15525
15526 With anonymous ACLs :
15527
15528 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
15529
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015530See section 4.2 for detailed help on the "http-request deny" and "use_backend"
15531keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015532
15533
155347.3. Fetching samples
15535---------------------
15536
15537Historically, sample fetch methods were only used to retrieve data to match
15538against patterns using ACLs. With the arrival of stick-tables, a new class of
15539sample fetch methods was created, most often sharing the same syntax as their
15540ACL counterpart. These sample fetch methods are also known as "fetches". As
15541of now, ACLs and fetches have converged. All ACL fetch methods have been made
15542available as fetch methods, and ACLs may use any sample fetch method as well.
15543
15544This section details all available sample fetch methods and their output type.
15545Some sample fetch methods have deprecated aliases that are used to maintain
15546compatibility with existing configurations. They are then explicitly marked as
15547deprecated and should not be used in new setups.
15548
15549The ACL derivatives are also indicated when available, with their respective
15550matching methods. These ones all have a well defined default pattern matching
15551method, so it is never necessary (though allowed) to pass the "-m" option to
15552indicate how the sample will be matched using ACLs.
15553
15554As indicated in the sample type versus matching compatibility matrix above,
15555when using a generic sample fetch method in an ACL, the "-m" option is
15556mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
15557the same keyword exists as an ACL keyword and as a standard fetch method, the
15558ACL engine will automatically pick the ACL-only one by default.
15559
15560Some of these keywords support one or multiple mandatory arguments, and one or
15561multiple optional arguments. These arguments are strongly typed and are checked
15562when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010015563incorrect argument (e.g. an unresolved backend name). Fetch function arguments
15564are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020015565is optional, it will be indicated below between square brackets ('[ ]'). When
15566all arguments are optional, the parenthesis may be omitted.
15567
15568Thus, the syntax of a standard sample fetch method is one of the following :
15569 - name
15570 - name(arg1)
15571 - name(arg1,arg2)
15572
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015573
155747.3.1. Converters
15575-----------------
15576
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015577Sample fetch methods may be combined with transformations to be applied on top
15578of the fetched sample (also called "converters"). These combinations form what
15579is called "sample expressions" and the result is a "sample". Initially this
15580was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010015581has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015582unique-id-format, add-header, ...).
15583
15584These transformations are enumerated as a series of specific keywords after the
15585sample fetch method. These keywords may equally be appended immediately after
15586the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010015587support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015588
Willy Tarreau97707872015-01-27 15:12:13 +010015589A certain category of converters are bitwise and arithmetic operators which
15590support performing basic operations on integers. Some bitwise operations are
15591supported (and, or, xor, cpl) and some arithmetic operations are supported
15592(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
15593bool) which make it possible to report a match without having to write an ACL.
15594
Willy Tarreau74ca5042013-06-11 23:12:07 +020015595The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010015596
Ben Shillitof25e8e52016-12-02 14:25:37 +00001559751d.single(<prop>[,<prop>*])
15598 Returns values for the properties requested as a string, where values are
15599 separated by the delimiter specified with "51degrees-property-separator".
15600 The device is identified using the User-Agent header passed to the
15601 converter. The function can be passed up to five property names, and if a
15602 property name can't be found, the value "NoData" is returned.
15603
15604 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010015605 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
15606 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000015607 # User-Agent passed to the converter.
15608 frontend http-in
15609 bind *:8081
15610 default_backend servers
15611 http-request set-header X-51D-DeviceTypeMobileTablet \
15612 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
15613
Willy Tarreau97707872015-01-27 15:12:13 +010015614add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015615 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015616 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015617 name. The name of the variable starts with an indication about its scope. The
15618 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015619 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015620 "sess" : the variable is shared with the whole session
15621 "txn" : the variable is shared with the transaction (request and response)
15622 "req" : the variable is shared only during request processing
15623 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015624 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015625 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015626
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010015627aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
15628 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
15629 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
15630 need to be base64 encoded and the returned result is in raw byte format.
15631 If the <aead_tag> validation fails, the converter doesn't return any data.
15632 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
15633 converter requires at least OpenSSL 1.0.1.
15634
15635 Example:
15636 http-response set-header X-Decrypted-Text %[var(txn.enc),\
15637 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
15638
Willy Tarreau97707872015-01-27 15:12:13 +010015639and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015640 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015641 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010015642 numeric value or a variable name. The name of the variable starts with an
15643 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015644 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015645 "sess" : the variable is shared with the whole session
15646 "txn" : the variable is shared with the transaction (request and response)
15647 "req" : the variable is shared only during request processing
15648 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015649 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015650 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015651
Holger Just1bfc24b2017-05-06 00:56:53 +020015652b64dec
15653 Converts (decodes) a base64 encoded input string to its binary
15654 representation. It performs the inverse operation of base64().
15655
Emeric Brun53d1a982014-04-30 18:21:37 +020015656base64
15657 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010015658 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020015659 an SSL ID can be copied in a header).
15660
Willy Tarreau97707872015-01-27 15:12:13 +010015661bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015662 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010015663 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010015664 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010015665 presence of a flag).
15666
Emeric Brun54c4ac82014-11-03 15:32:43 +010015667bytes(<offset>[,<length>])
15668 Extracts some bytes from an input binary sample. The result is a binary
15669 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010015670 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010015671
Willy Tarreau280f42b2018-02-19 15:34:12 +010015672concat([<start>],[<var>],[<end>])
15673 Concatenates up to 3 fields after the current sample which is then turned to
15674 a string. The first one, <start>, is a constant string, that will be appended
15675 immediately after the existing sample. It may be omitted if not used. The
15676 second one, <var>, is a variable name. The variable will be looked up, its
15677 contents converted to a string, and it will be appended immediately after the
15678 <first> part. If the variable is not found, nothing is appended. It may be
15679 omitted as well. The third field, <end> is a constant string that will be
15680 appended after the variable. It may also be omitted. Together, these elements
15681 allow to concatenate variables with delimiters to an existing set of
15682 variables. This can be used to build new variables made of a succession of
Willy Tarreauef21fac2020-02-14 13:37:20 +010015683 other variables, such as colon-delimited values. If commas or closing
Daniel Corbett67a82712020-07-06 23:01:19 -040015684 parenthesis are needed as delimiters, they must be protected by quotes or
Willy Tarreauef21fac2020-02-14 13:37:20 +010015685 backslashes, themselves protected so that they are not stripped by the first
15686 level parser. See examples below.
Willy Tarreau280f42b2018-02-19 15:34:12 +010015687
15688 Example:
15689 tcp-request session set-var(sess.src) src
15690 tcp-request session set-var(sess.dn) ssl_c_s_dn
15691 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
Willy Tarreauef21fac2020-02-14 13:37:20 +010015692 tcp-request session set-var(txn.ipport) "str(),concat('addr=(',sess.ip),concat(',',sess.port,')')"
Willy Tarreau280f42b2018-02-19 15:34:12 +010015693 http-request set-header x-hap-sig %[var(txn.sig)]
15694
Willy Tarreau97707872015-01-27 15:12:13 +010015695cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015696 Takes the input value of type signed integer, applies a ones-complement
15697 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010015698
Willy Tarreau80599772015-01-20 19:35:24 +010015699crc32([<avalanche>])
15700 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
15701 hash function. Optionally, it is possible to apply a full avalanche hash
15702 function to the output if the optional <avalanche> argument equals 1. This
15703 converter uses the same functions as used by the various hash-based load
15704 balancing algorithms, so it will provide exactly the same results. It is
15705 provided for compatibility with other software which want a CRC32 to be
15706 computed on some input keys, so it follows the most common implementation as
15707 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
15708 but may provide a better or at least less predictable distribution. It must
15709 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015710 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
15711
15712crc32c([<avalanche>])
15713 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
15714 hash function. Optionally, it is possible to apply a full avalanche hash
15715 function to the output if the optional <avalanche> argument equals 1. This
15716 converter uses the same functions as described in RFC4960, Appendix B [8].
15717 It is provided for compatibility with other software which want a CRC32C to be
15718 computed on some input keys. It is slower than the other algorithms and it must
15719 not be used for security purposes as a 32-bit hash is trivial to break. See
15720 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010015721
Christopher Fauletea159d62020-04-01 16:21:44 +020015722cut_crlf
15723 Cuts the string representation of the input sample on the first carriage
15724 return ('\r') or newline ('\n') character found. Only the string length is
15725 updated.
15726
David Carlier29b3ca32015-09-25 14:09:21 +010015727da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020015728 Asks the DeviceAtlas converter to identify the User Agent string passed on
15729 input, and to emit a string made of the concatenation of the properties
15730 enumerated in argument, delimited by the separator defined by the global
15731 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000015732 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020015733 configuration language.
15734
15735 Example:
15736 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015737 bind *:8881
15738 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000015739 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 +020015740
Willy Tarreau0851fd52019-12-17 10:07:25 +010015741debug([<prefix][,<destination>])
15742 This converter is used as debug tool. It takes a capture of the input sample
15743 and sends it to event sink <destination>, which may designate a ring buffer
15744 such as "buf0", as well as "stdout", or "stderr". Available sinks may be
15745 checked at run time by issuing "show events" on the CLI. When not specified,
15746 the output will be "buf0", which may be consulted via the CLI's "show events"
15747 command. An optional prefix <prefix> may be passed to help distinguish
15748 outputs from multiple expressions. It will then appear before the colon in
15749 the output message. The input sample is passed as-is on the output, so that
15750 it is safe to insert the debug converter anywhere in a chain, even with non-
15751 printable sample types.
15752
15753 Example:
15754 tcp-request connection track-sc0 src,debug(track-sc)
Thierry FOURNIER9687c772015-05-07 15:46:29 +020015755
Patrick Gansterer8e366512020-04-22 16:47:57 +020015756digest(<algorithm>)
15757 Converts a binary input sample to a message digest. The result is a binary
15758 sample. The <algorithm> must be an OpenSSL message digest name (e.g. sha256).
15759
15760 Please note that this converter is only available when haproxy has been
15761 compiled with USE_OPENSSL.
15762
Willy Tarreau97707872015-01-27 15:12:13 +010015763div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015764 Divides the input value of type signed integer by <value>, and returns the
15765 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015766 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010015767 variable name. The name of the variable starts with an indication about its
15768 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015769 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015770 "sess" : the variable is shared with the whole session
15771 "txn" : the variable is shared with the transaction (request and response)
15772 "req" : the variable is shared only during request processing
15773 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015774 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015775 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015776
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015777djb2([<avalanche>])
15778 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
15779 hash function. Optionally, it is possible to apply a full avalanche hash
15780 function to the output if the optional <avalanche> argument equals 1. This
15781 converter uses the same functions as used by the various hash-based load
15782 balancing algorithms, so it will provide exactly the same results. It is
15783 mostly intended for debugging, but can be used as a stick-table entry to
15784 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015785 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
15786 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015787
Willy Tarreau97707872015-01-27 15:12:13 +010015788even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015789 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010015790 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
15791
Marcin Deranek9631a282018-04-16 14:30:46 +020015792field(<index>,<delimiters>[,<count>])
15793 Extracts the substring at the given index counting from the beginning
15794 (positive index) or from the end (negative index) considering given delimiters
15795 from an input string. Indexes start at 1 or -1 and delimiters are a string
15796 formatted list of chars. Optionally you can specify <count> of fields to
15797 extract (default: 1). Value of 0 indicates extraction of all remaining
15798 fields.
15799
15800 Example :
15801 str(f1_f2_f3__f5),field(5,_) # f5
15802 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
15803 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
15804 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
15805 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010015806
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015807fix_is_valid
15808 Parses a binary payload and performs sanity checks regarding FIX (Financial
15809 Information eXchange):
15810
15811 - checks that all tag IDs and values are not empty and the tags IDs are well
15812 numeric
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050015813 - checks the BeginString tag is the first tag with a valid FIX version
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015814 - checks the BodyLength tag is the second one with the right body length
Christopher Fauleted4bef72021-03-18 17:40:56 +010015815 - checks the MsgType tag is the third tag.
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015816 - checks that last tag in the message is the CheckSum tag with a valid
15817 checksum
15818
15819 Due to current HAProxy design, only the first message sent by the client and
15820 the server can be parsed.
15821
15822 This converter returns a boolean, true if the payload contains a valid FIX
15823 message, false if not.
15824
15825 See also the fix_tag_value converter.
15826
15827 Example:
15828 tcp-request inspect-delay 10s
15829 tcp-request content reject unless { req.payload(0,0),fix_is_valid }
15830
15831fix_tag_value(<tag>)
15832 Parses a FIX (Financial Information eXchange) message and extracts the value
15833 from the tag <tag>. <tag> can be a string or an integer pointing to the
15834 desired tag. Any integer value is accepted, but only the following strings
15835 are translated into their integer equivalent: BeginString, BodyLength,
Daniel Corbettbefef702021-03-09 23:00:34 -050015836 MsgType, SenderCompID, TargetCompID, CheckSum. More tag names can be easily
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015837 added.
15838
15839 Due to current HAProxy design, only the first message sent by the client and
15840 the server can be parsed. No message validation is performed by this
15841 converter. It is highly recommended to validate the message first using
15842 fix_is_valid converter.
15843
15844 See also the fix_is_valid converter.
15845
15846 Example:
15847 tcp-request inspect-delay 10s
15848 tcp-request content reject unless { req.payload(0,0),fix_is_valid }
15849 # MsgType tag ID is 35, so both lines below will return the same content
15850 tcp-request content set-var(txn.foo) req.payload(0,0),fix_tag_value(35)
15851 tcp-request content set-var(txn.bar) req.payload(0,0),fix_tag_value(MsgType)
15852
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015853hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010015854 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015855 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010015856 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 +020015857 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010015858
Dragan Dosen3f957b22017-10-24 09:27:34 +020015859hex2i
15860 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050015861 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020015862
Christopher Faulet4ccc12f2020-04-01 09:08:32 +020015863htonl
15864 Converts the input integer value to its 32-bit binary representation in the
15865 network byte order. Because sample fetches own signed 64-bit integer, when
15866 this converter is used, the input integer value is first casted to an
15867 unsigned 32-bit integer.
15868
Tim Duesterhusa3082092021-01-21 17:40:49 +010015869hmac(<algorithm>,<key>)
Patrick Gansterer8e366512020-04-22 16:47:57 +020015870 Converts a binary input sample to a message authentication code with the given
15871 key. The result is a binary sample. The <algorithm> must be one of the
15872 registered OpenSSL message digest names (e.g. sha256). The <key> parameter must
15873 be base64 encoded and can either be a string or a variable.
15874
15875 Please note that this converter is only available when haproxy has been
15876 compiled with USE_OPENSSL.
15877
Cyril Bonté6bcd1822019-11-05 23:13:59 +010015878http_date([<offset],[<unit>])
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015879 Converts an integer supposed to contain a date since epoch to a string
15880 representing this date in a format suitable for use in HTTP header fields. If
Damien Claisseae6f1252019-10-30 15:57:28 +000015881 an offset value is specified, then it is added to the date before the
15882 conversion is operated. This is particularly useful to emit Date header fields,
15883 Expires values in responses when combined with a positive offset, or
15884 Last-Modified values when the offset is negative.
15885 If a unit value is specified, then consider the timestamp as either
15886 "s" for seconds (default behavior), "ms" for milliseconds, or "us" for
15887 microseconds since epoch. Offset is assumed to have the same unit as
15888 input timestamp.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015889
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015890iif(<true>,<false>)
15891 Returns the <true> string if the input value is true. Returns the <false>
15892 string otherwise.
15893
15894 Example:
Tim Duesterhus870713b2020-09-11 17:13:12 +020015895 http-request set-header x-forwarded-proto %[ssl_fc,iif(https,http)]
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015896
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015897in_table(<table>)
15898 Uses the string representation of the input sample to perform a look up in
15899 the specified table. If the key is not found in the table, a boolean false
15900 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010015901 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015902 or not a source IP address or an Authorization header was already seen).
15903
Tim Duesterhusa3082092021-01-21 17:40:49 +010015904ipmask(<mask4>,[<mask6>])
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015905 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015906 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015907 table entries and as such use the same server. The mask4 can be passed in
15908 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
15909 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
15910 If no mask6 is given IPv6 addresses will fail to convert for backwards
15911 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015912
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015913json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010015914 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015915 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020015916 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015917 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
15918 of errors:
15919 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
15920 bytes, ...)
15921 - invalid range (the decoded value is within a UTF-8 prohibited range),
15922 - code overlong (the value is encoded with more bytes than necessary).
15923
15924 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
15925 character is greater than 0xffff because the JSON string escape specification
15926 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
15927 in 4 variants designated by a combination of two suffix letters : "p" for
15928 "permissive" and "s" for "silently ignore". The behaviors of the decoders
15929 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010015930 - "ascii" : never fails;
15931 - "utf8" : fails on any detected errors;
15932 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015933 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010015934 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015935 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
15936 characters corresponding to the other errors.
15937
15938 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010015939 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015940
15941 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015942 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020015943 capture request header user-agent len 150
15944 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015945
15946 Input request from client 127.0.0.1:
15947 GET / HTTP/1.0
15948 User-Agent: Very "Ugly" UA 1/2
15949
15950 Output log:
15951 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
15952
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015953language(<value>[,<default>])
15954 Returns the value with the highest q-factor from a list as extracted from the
15955 "accept-language" header using "req.fhdr". Values with no q-factor have a
15956 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
15957 belong to the list of semi-colon delimited <values> will be considered. The
15958 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
15959 given list and a default value is provided, it is returned. Note that language
15960 names may have a variant after a dash ('-'). If this variant is present in the
15961 list, it will be matched, but if it is not, only the base language is checked.
15962 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010015963 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015964 ordering of the values in the request counts, as the first value among
15965 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020015966
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015967 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020015968
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015969 # this configuration switches to the backend matching a
15970 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020015971
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015972 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
15973 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
15974 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
15975 use_backend spanish if es
15976 use_backend french if fr
15977 use_backend english if en
15978 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020015979
Willy Tarreau60a2ee72017-12-15 07:13:48 +010015980length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010015981 Get the length of the string. This can only be placed after a string
15982 sample fetch function or after a transformation keyword returning a string
15983 type. The result is of type integer.
15984
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015985lower
15986 Convert a string sample to lower case. This can only be placed after a string
15987 sample fetch function or after a transformation keyword returning a string
15988 type. The result is of type string.
15989
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015990ltime(<format>[,<offset>])
15991 Converts an integer supposed to contain a date since epoch to a string
15992 representing this date in local time using a format defined by the <format>
15993 string using strftime(3). The purpose is to allow any date format to be used
15994 in logs. An optional <offset> in seconds may be applied to the input date
15995 (positive or negative). See the strftime() man page for the format supported
15996 by your operating system. See also the utime converter.
15997
15998 Example :
15999
16000 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010016001 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016002 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
16003
Christopher Faulet51fc9d12020-04-01 17:24:41 +020016004ltrim(<chars>)
16005 Skips any characters from <chars> from the beginning of the string
16006 representation of the input sample.
16007
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016008map(<map_file>[,<default_value>])
16009map_<match_type>(<map_file>[,<default_value>])
16010map_<match_type>_<output_type>(<map_file>[,<default_value>])
16011 Search the input value from <map_file> using the <match_type> matching method,
16012 and return the associated value converted to the type <output_type>. If the
16013 input value cannot be found in the <map_file>, the converter returns the
16014 <default_value>. If the <default_value> is not set, the converter fails and
16015 acts as if no input value could be fetched. If the <match_type> is not set, it
16016 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
16017 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
16018 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016019
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016020 It is important to avoid overlapping between the keys : IP addresses and
16021 strings are stored in trees, so the first of the finest match will be used.
16022 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016023
Tim Düsterhus4896c442016-11-29 02:15:19 +010016024 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016025 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016026
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016027 input type | match method | output type str | output type int | output type ip
16028 -----------+--------------+-----------------+-----------------+---------------
16029 str | str | map_str | map_str_int | map_str_ip
16030 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020016031 str | beg | map_beg | map_beg_int | map_end_ip
16032 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016033 str | sub | map_sub | map_sub_int | map_sub_ip
16034 -----------+--------------+-----------------+-----------------+---------------
16035 str | dir | map_dir | map_dir_int | map_dir_ip
16036 -----------+--------------+-----------------+-----------------+---------------
16037 str | dom | map_dom | map_dom_int | map_dom_ip
16038 -----------+--------------+-----------------+-----------------+---------------
16039 str | end | map_end | map_end_int | map_end_ip
16040 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080016041 str | reg | map_reg | map_reg_int | map_reg_ip
16042 -----------+--------------+-----------------+-----------------+---------------
16043 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016044 -----------+--------------+-----------------+-----------------+---------------
16045 int | int | map_int | map_int_int | map_int_ip
16046 -----------+--------------+-----------------+-----------------+---------------
16047 ip | ip | map_ip | map_ip_int | map_ip_ip
16048 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016049
Thierry Fournier8feaa662016-02-10 22:55:20 +010016050 The special map called "map_regm" expect matching zone in the regular
16051 expression and modify the output replacing back reference (like "\1") by
16052 the corresponding match text.
16053
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016054 The file contains one key + value per line. Lines which start with '#' are
16055 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
16056 is then the first "word" (series of non-space/tabs characters), and the value
16057 is what follows this series of space/tab till the end of the line excluding
16058 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016059
Thierry FOURNIER060762e2014-04-23 13:29:15 +020016060 Example :
16061
16062 # this is a comment and is ignored
16063 2.22.246.0/23 United Kingdom \n
16064 <-><-----------><--><------------><---->
16065 | | | | `- trailing spaces ignored
16066 | | | `---------- value
16067 | | `-------------------- middle spaces ignored
16068 | `---------------------------- key
16069 `------------------------------------ leading spaces ignored
16070
Willy Tarreau97707872015-01-27 15:12:13 +010016071mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016072 Divides the input value of type signed integer by <value>, and returns the
16073 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016074 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016075 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016076 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016077 "sess" : the variable is shared with the whole session
16078 "txn" : the variable is shared with the transaction (request and response)
16079 "req" : the variable is shared only during request processing
16080 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016081 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016082 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016083
Baptiste Assmanne279ca62020-10-27 18:10:06 +010016084mqtt_field_value(<packettype>,<fieldname or property ID>)
16085 Returns value of <fieldname> found in input MQTT payload of type
16086 <packettype>.
16087 <packettype> can be either a string (case insensitive matching) or a numeric
16088 value corresponding to the type of packet we're supposed to extract data
16089 from.
16090 Supported string and integers can be found here:
16091 https://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html#_Toc398718021
16092 https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html#_Toc3901022
16093
16094 <fieldname> depends on <packettype> and can be any of the following below.
16095 (note that <fieldname> matching is case insensitive).
16096 <property id> can only be found in MQTT v5.0 streams. check this table:
16097 https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html#_Toc3901029
16098
16099 - CONNECT (or 1): flags, protocol_name, protocol_version, client_identifier,
16100 will_topic, will_payload, username, password, keepalive
16101 OR any property ID as a numeric value (for MQTT v5.0
16102 packets only):
16103 17: Session Expiry Interval
16104 33: Receive Maximum
16105 39: Maximum Packet Size
16106 34: Topic Alias Maximum
16107 25: Request Response Information
16108 23: Request Problem Information
16109 21: Authentication Method
16110 22: Authentication Data
16111 18: Will Delay Interval
16112 1: Payload Format Indicator
16113 2: Message Expiry Interval
16114 3: Content Type
16115 8: Response Topic
16116 9: Correlation Data
16117 Not supported yet:
16118 38: User Property
16119
16120 - CONNACK (or 2): flags, protocol_version, reason_code
16121 OR any property ID as a numeric value (for MQTT v5.0
16122 packets only):
16123 17: Session Expiry Interval
16124 33: Receive Maximum
16125 36: Maximum QoS
16126 37: Retain Available
16127 39: Maximum Packet Size
16128 18: Assigned Client Identifier
16129 34: Topic Alias Maximum
16130 31: Reason String
16131 40; Wildcard Subscription Available
16132 41: Subscription Identifiers Available
16133 42: Shared Subscription Available
16134 19: Server Keep Alive
16135 26: Response Information
16136 28: Server Reference
16137 21: Authentication Method
16138 22: Authentication Data
16139 Not supported yet:
16140 38: User Property
16141
16142 Due to current HAProxy design, only the first message sent by the client and
16143 the server can be parsed. Thus this converter can extract data only from
16144 CONNECT and CONNACK packet types. CONNECT is the first message sent by the
16145 client and CONNACK is the first response sent by the server.
16146
16147 Example:
16148
16149 acl data_in_buffer req.len ge 4
16150 tcp-request content set-var(txn.username) \
16151 req.payload(0,0),mqtt_field_value(connect,protocol_name) \
16152 if data_in_buffer
16153 # do the same as above
16154 tcp-request content set-var(txn.username) \
16155 req.payload(0,0),mqtt_field_value(1,protocol_name) \
16156 if data_in_buffer
16157
16158mqtt_is_valid
16159 Checks that the binary input is a valid MQTT packet. It returns a boolean.
16160
16161 Due to current HAProxy design, only the first message sent by the client and
16162 the server can be parsed. Thus this converter can extract data only from
16163 CONNECT and CONNACK packet types. CONNECT is the first message sent by the
16164 client and CONNACK is the first response sent by the server.
16165
16166 Example:
16167
16168 acl data_in_buffer req.len ge 4
16169 tcp-request content reject unless req.payload(0,0),mqtt_is_valid
16170
Willy Tarreau97707872015-01-27 15:12:13 +010016171mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016172 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020016173 the product as an signed integer. In case of overflow, the largest possible
16174 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016175 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016176 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016177 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016178 "sess" : the variable is shared with the whole session
16179 "txn" : the variable is shared with the transaction (request and response)
16180 "req" : the variable is shared only during request processing
16181 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016182 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016183 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016184
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010016185nbsrv
16186 Takes an input value of type string, interprets it as a backend name and
16187 returns the number of usable servers in that backend. Can be used in places
16188 where we want to look up a backend from a dynamic name, like a result of a
16189 map lookup.
16190
Willy Tarreau97707872015-01-27 15:12:13 +010016191neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016192 Takes the input value of type signed integer, computes the opposite value,
16193 and returns the remainder as an signed integer. 0 is identity. This operator
16194 is provided for reversed subtracts : in order to subtract the input from a
16195 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010016196
16197not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016198 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010016199 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016200 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010016201 absence of a flag).
16202
16203odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016204 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010016205 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
16206
16207or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016208 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016209 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010016210 numeric value or a variable name. The name of the variable starts with an
16211 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016212 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016213 "sess" : the variable is shared with the whole session
16214 "txn" : the variable is shared with the transaction (request and response)
16215 "req" : the variable is shared only during request processing
16216 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010016217 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016218 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016219
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016220protobuf(<field_number>,[<field_type>])
16221 This extracts the protocol buffers message field in raw mode of an input binary
16222 sample representation of a protocol buffer message with <field_number> as field
16223 number (dotted notation) if <field_type> is not present, or as an integer sample
16224 if this field is present (see also "ungrpc" below).
16225 The list of the authorized types is the following one: "int32", "int64", "uint32",
16226 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
16227 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
16228 "float" for the wire type 5. Note that "string" is considered as a length-delimited
16229 type, so it does not require any <field_type> argument to be extracted.
16230 More information may be found here about the protocol buffers message field types:
16231 https://developers.google.com/protocol-buffers/docs/encoding
16232
Willy Tarreauc4dc3502015-01-23 20:39:28 +010016233regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010016234 Applies a regex-based substitution to the input string. It does the same
16235 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
16236 default it will replace in the input string the first occurrence of the
16237 largest part matching the regular expression <regex> with the substitution
16238 string <subst>. It is possible to replace all occurrences instead by adding
16239 the flag "g" in the third argument <flags>. It is also possible to make the
16240 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
16241 string, it is made up from the concatenation of all desired flags. Thus if
16242 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
Willy Tarreauef21fac2020-02-14 13:37:20 +010016243 The first use of this converter is to replace certain characters or sequence
16244 of characters with other ones.
16245
16246 It is highly recommended to enclose the regex part using protected quotes to
16247 improve clarity and never have a closing parenthesis from the regex mixed up
16248 with the parenthesis from the function. Just like in Bourne shell, the first
16249 level of quotes is processed when delimiting word groups on the line, a
16250 second level is usable for argument. It is recommended to use single quotes
16251 outside since these ones do not try to resolve backslashes nor dollar signs.
Willy Tarreau7eda8492015-01-20 19:47:06 +010016252
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010016253 Examples:
Willy Tarreau7eda8492015-01-20 19:47:06 +010016254
16255 # de-duplicate "/" in header "x-path".
16256 # input: x-path: /////a///b/c/xzxyz/
16257 # output: x-path: /a/b/c/xzxyz/
Willy Tarreauef21fac2020-02-14 13:37:20 +010016258 http-request set-header x-path "%[hdr(x-path),regsub('/+','/','g')]"
Willy Tarreau7eda8492015-01-20 19:47:06 +010016259
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010016260 # copy query string to x-query and drop all leading '?', ';' and '&'
16261 http-request set-header x-query "%[query,regsub([?;&]*,'')]"
16262
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010016263 # capture groups and backreferences
16264 # both lines do the same.
Willy Tarreau465dc7d2020-10-08 18:05:56 +020016265 http-request redirect location %[url,'regsub("(foo|bar)([0-9]+)?","\2\1",i)']
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010016266 http-request redirect location %[url,regsub(\"(foo|bar)([0-9]+)?\",\"\2\1\",i)]
16267
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020016268capture-req(<id>)
16269 Capture the string entry in the request slot <id> and returns the entry as
16270 is. If the slot doesn't exist, the capture fails silently.
16271
16272 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020016273 "http-response capture", "capture.req.hdr" and
16274 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020016275
16276capture-res(<id>)
16277 Capture the string entry in the response slot <id> and returns the entry as
16278 is. If the slot doesn't exist, the capture fails silently.
16279
16280 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020016281 "http-response capture", "capture.req.hdr" and
16282 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020016283
Christopher Faulet568415a2020-04-01 17:24:47 +020016284rtrim(<chars>)
16285 Skips any characters from <chars> from the end of the string representation
16286 of the input sample.
16287
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016288sdbm([<avalanche>])
16289 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
16290 hash function. Optionally, it is possible to apply a full avalanche hash
16291 function to the output if the optional <avalanche> argument equals 1. This
16292 converter uses the same functions as used by the various hash-based load
16293 balancing algorithms, so it will provide exactly the same results. It is
16294 mostly intended for debugging, but can be used as a stick-table entry to
16295 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010016296 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
16297 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016298
Tim Duesterhusf38175c2020-06-09 11:48:42 +020016299secure_memcmp(<var>)
16300 Compares the contents of <var> with the input value. Both values are treated
16301 as a binary string. Returns a boolean indicating whether both binary strings
16302 match.
16303
16304 If both binary strings have the same length then the comparison will be
16305 performed in constant time.
16306
16307 Please note that this converter is only available when haproxy has been
16308 compiled with USE_OPENSSL.
16309
16310 Example :
16311
16312 http-request set-var(txn.token) hdr(token)
16313 # Check whether the token sent by the client matches the secret token
16314 # value, without leaking the contents using a timing attack.
16315 acl token_given str(my_secret_token),secure_memcmp(txn.token)
16316
Tim Duesterhusef4e45c2021-01-21 17:40:50 +010016317set-var(<var>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010016318 Sets a variable with the input content and returns the content on the output
16319 as-is. The variable keeps the value and the associated input type. The name of
16320 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016321 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016322 "sess" : the variable is shared with the whole session
16323 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016324 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016325 "req" : the variable is shared only during request processing,
16326 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016327 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016328 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016329
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020016330sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020016331 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020016332 sample with length of 20 bytes.
16333
Tim Duesterhusd4376302019-06-17 12:41:44 +020016334sha2([<bits>])
16335 Converts a binary input sample to a digest in the SHA-2 family. The result
16336 is a binary sample with length of <bits>/8 bytes.
16337
16338 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
16339 SHA-<bits>. The default value is 256.
16340
16341 Please note that this converter is only available when haproxy has been
16342 compiled with USE_OPENSSL.
16343
Nenad Merdanovic177adc92019-08-27 01:58:13 +020016344srv_queue
16345 Takes an input value of type string, either a server name or <backend>/<server>
16346 format and returns the number of queued sessions on that server. Can be used
16347 in places where we want to look up queued sessions from a dynamic name, like a
16348 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
16349 persistence or direct a request elsewhere.
16350
Tim Duesterhusca097c12018-04-27 21:18:45 +020016351strcmp(<var>)
16352 Compares the contents of <var> with the input value of type string. Returns
16353 the result as a signed integer compatible with strcmp(3): 0 if both strings
16354 are identical. A value less than 0 if the left string is lexicographically
16355 smaller than the right string or if the left string is shorter. A value greater
16356 than 0 otherwise (right string greater than left string or the right string is
16357 shorter).
16358
Tim Duesterhusf38175c2020-06-09 11:48:42 +020016359 See also the secure_memcmp converter if you need to compare two binary
16360 strings in constant time.
16361
Tim Duesterhusca097c12018-04-27 21:18:45 +020016362 Example :
16363
16364 http-request set-var(txn.host) hdr(host)
16365 # Check whether the client is attempting domain fronting.
16366 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
16367
16368
Willy Tarreau97707872015-01-27 15:12:13 +010016369sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016370 Subtracts <value> from the input value of type signed integer, and returns
16371 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016372 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010016373 or a variable name. The name of the variable starts with an indication about
16374 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016375 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016376 "sess" : the variable is shared with the whole session
16377 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016378 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016379 "req" : the variable is shared only during request processing,
16380 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016381 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016382 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016383
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016384table_bytes_in_rate(<table>)
16385 Uses the string representation of the input sample to perform a look up in
16386 the specified table. If the key is not found in the table, integer value zero
16387 is returned. Otherwise the converter returns the average client-to-server
16388 bytes rate associated with the input sample in the designated table, measured
16389 in amount of bytes over the period configured in the table. See also the
16390 sc_bytes_in_rate sample fetch keyword.
16391
16392
16393table_bytes_out_rate(<table>)
16394 Uses the string representation of the input sample to perform a look up in
16395 the specified table. If the key is not found in the table, integer value zero
16396 is returned. Otherwise the converter returns the average server-to-client
16397 bytes rate associated with the input sample in the designated table, measured
16398 in amount of bytes over the period configured in the table. See also the
16399 sc_bytes_out_rate sample fetch keyword.
16400
16401table_conn_cnt(<table>)
16402 Uses the string representation of the input sample to perform a look up in
16403 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016404 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016405 connections associated with the input sample in the designated table. See
16406 also the sc_conn_cnt sample fetch keyword.
16407
16408table_conn_cur(<table>)
16409 Uses the string representation of the input sample to perform a look up in
16410 the specified table. If the key is not found in the table, integer value zero
16411 is returned. Otherwise the converter returns the current amount of concurrent
16412 tracked connections associated with the input sample in the designated table.
16413 See also the sc_conn_cur sample fetch keyword.
16414
16415table_conn_rate(<table>)
16416 Uses the string representation of the input sample to perform a look up in
16417 the specified table. If the key is not found in the table, integer value zero
16418 is returned. Otherwise the converter returns the average incoming connection
16419 rate associated with the input sample in the designated table. See also the
16420 sc_conn_rate sample fetch keyword.
16421
Thierry FOURNIER236657b2015-08-19 08:25:14 +020016422table_gpt0(<table>)
16423 Uses the string representation of the input sample to perform a look up in
16424 the specified table. If the key is not found in the table, boolean value zero
16425 is returned. Otherwise the converter returns the current value of the first
16426 general purpose tag associated with the input sample in the designated table.
16427 See also the sc_get_gpt0 sample fetch keyword.
16428
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016429table_gpc0(<table>)
16430 Uses the string representation of the input sample to perform a look up in
16431 the specified table. If the key is not found in the table, integer value zero
16432 is returned. Otherwise the converter returns the current value of the first
16433 general purpose counter associated with the input sample in the designated
16434 table. See also the sc_get_gpc0 sample fetch keyword.
16435
16436table_gpc0_rate(<table>)
16437 Uses the string representation of the input sample to perform a look up in
16438 the specified table. If the key is not found in the table, integer value zero
16439 is returned. Otherwise the converter returns the frequency which the gpc0
16440 counter was incremented over the configured period in the table, associated
16441 with the input sample in the designated table. See also the sc_get_gpc0_rate
16442 sample fetch keyword.
16443
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016444table_gpc1(<table>)
16445 Uses the string representation of the input sample to perform a look up in
16446 the specified table. If the key is not found in the table, integer value zero
16447 is returned. Otherwise the converter returns the current value of the second
16448 general purpose counter associated with the input sample in the designated
16449 table. See also the sc_get_gpc1 sample fetch keyword.
16450
16451table_gpc1_rate(<table>)
16452 Uses the string representation of the input sample to perform a look up in
16453 the specified table. If the key is not found in the table, integer value zero
16454 is returned. Otherwise the converter returns the frequency which the gpc1
16455 counter was incremented over the configured period in the table, associated
16456 with the input sample in the designated table. See also the sc_get_gpc1_rate
16457 sample fetch keyword.
16458
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016459table_http_err_cnt(<table>)
16460 Uses the string representation of the input sample to perform a look up in
16461 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016462 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016463 errors associated with the input sample in the designated table. See also the
16464 sc_http_err_cnt sample fetch keyword.
16465
16466table_http_err_rate(<table>)
16467 Uses the string representation of the input sample to perform a look up in
16468 the specified table. If the key is not found in the table, integer value zero
16469 is returned. Otherwise the average rate of HTTP errors associated with the
16470 input sample in the designated table, measured in amount of errors over the
16471 period configured in the table. See also the sc_http_err_rate sample fetch
16472 keyword.
16473
Willy Tarreau826f3ab2021-02-10 12:07:15 +010016474table_http_fail_cnt(<table>)
16475 Uses the string representation of the input sample to perform a look up in
16476 the specified table. If the key is not found in the table, integer value zero
16477 is returned. Otherwise the converter returns the cumulative number of HTTP
16478 failures associated with the input sample in the designated table. See also
16479 the sc_http_fail_cnt sample fetch keyword.
16480
16481table_http_fail_rate(<table>)
16482 Uses the string representation of the input sample to perform a look up in
16483 the specified table. If the key is not found in the table, integer value zero
16484 is returned. Otherwise the average rate of HTTP failures associated with the
16485 input sample in the designated table, measured in amount of failures over the
16486 period configured in the table. See also the sc_http_fail_rate sample fetch
16487 keyword.
16488
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016489table_http_req_cnt(<table>)
16490 Uses the string representation of the input sample to perform a look up in
16491 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016492 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016493 requests associated with the input sample in the designated table. See also
16494 the sc_http_req_cnt sample fetch keyword.
16495
16496table_http_req_rate(<table>)
16497 Uses the string representation of the input sample to perform a look up in
16498 the specified table. If the key is not found in the table, integer value zero
16499 is returned. Otherwise the average rate of HTTP requests associated with the
16500 input sample in the designated table, measured in amount of requests over the
16501 period configured in the table. See also the sc_http_req_rate sample fetch
16502 keyword.
16503
16504table_kbytes_in(<table>)
16505 Uses the string representation of the input sample to perform a look up in
16506 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016507 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016508 to-server data associated with the input sample in the designated table,
16509 measured in kilobytes. The test is currently performed on 32-bit integers,
16510 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
16511 keyword.
16512
16513table_kbytes_out(<table>)
16514 Uses the string representation of the input sample to perform a look up in
16515 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016516 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016517 to-client data associated with the input sample in the designated table,
16518 measured in kilobytes. The test is currently performed on 32-bit integers,
16519 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
16520 keyword.
16521
16522table_server_id(<table>)
16523 Uses the string representation of the input sample to perform a look up in
16524 the specified table. If the key is not found in the table, integer value zero
16525 is returned. Otherwise the converter returns the server ID associated with
16526 the input sample in the designated table. A server ID is associated to a
16527 sample by a "stick" rule when a connection to a server succeeds. A server ID
16528 zero means that no server is associated with this key.
16529
16530table_sess_cnt(<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, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016533 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016534 sessions associated with the input sample in the designated table. Note that
16535 a session here refers to an incoming connection being accepted by the
16536 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
16537 keyword.
16538
16539table_sess_rate(<table>)
16540 Uses the string representation of the input sample to perform a look up in
16541 the specified table. If the key is not found in the table, integer value zero
16542 is returned. Otherwise the converter returns the average incoming session
16543 rate associated with the input sample in the designated table. Note that a
16544 session here refers to an incoming connection being accepted by the
16545 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
16546 keyword.
16547
16548table_trackers(<table>)
16549 Uses the string representation of the input sample to perform a look up in
16550 the specified table. If the key is not found in the table, integer value zero
16551 is returned. Otherwise the converter returns the current amount of concurrent
16552 connections tracking the same key as the input sample in the designated
16553 table. It differs from table_conn_cur in that it does not rely on any stored
16554 information but on the table's reference count (the "use" value which is
16555 returned by "show table" on the CLI). This may sometimes be more suited for
16556 layer7 tracking. It can be used to tell a server how many concurrent
16557 connections there are from a given address for example. See also the
16558 sc_trackers sample fetch keyword.
16559
Willy Tarreauffcb2e42014-07-10 16:29:08 +020016560upper
16561 Convert a string sample to upper case. This can only be placed after a string
16562 sample fetch function or after a transformation keyword returning a string
16563 type. The result is of type string.
16564
Willy Tarreau62ba9ba2020-04-23 17:54:47 +020016565url_dec([<in_form>])
16566 Takes an url-encoded string provided as input and returns the decoded version
16567 as output. The input and the output are of type string. If the <in_form>
16568 argument is set to a non-zero integer value, the input string is assumed to
16569 be part of a form or query string and the '+' character will be turned into a
16570 space (' '). Otherwise this will only happen after a question mark indicating
16571 a query string ('?').
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020016572
William Dauchy888b0ae2021-01-06 23:39:50 +010016573url_enc([<enc_type>])
16574 Takes a string provided as input and returns the encoded version as output.
16575 The input and the output are of type string. By default the type of encoding
16576 is meant for `query` type. There is no other type supported for now but the
16577 optional argument is here for future changes.
16578
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016579ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016580 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016581 sample representation of a gRPC message with <field_number> as field number
16582 (dotted notation) if <field_type> is not present, or as an integer sample if this
16583 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016584 The list of the authorized types is the following one: "int32", "int64", "uint32",
16585 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
16586 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
16587 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010016588 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016589 More information may be found here about the protocol buffers message field types:
16590 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016591
16592 Example:
16593 // with such a protocol buffer .proto file content adapted from
16594 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
16595
16596 message Point {
16597 int32 latitude = 1;
16598 int32 longitude = 2;
16599 }
16600
16601 message PPoint {
16602 Point point = 59;
16603 }
16604
16605 message Rectangle {
16606 // One corner of the rectangle.
16607 PPoint lo = 48;
16608 // The other corner of the rectangle.
16609 PPoint hi = 49;
16610 }
16611
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016612 let's say a body request is made of a "Rectangle" object value (two PPoint
16613 protocol buffers messages), the four protocol buffers fields could be
16614 extracted with these "ungrpc" directives:
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016615
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016616 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
16617 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050016618 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016619 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
16620
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016621 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016622
Frédéric Lécaille93d33162019-03-06 09:35:59 +010016623 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016624
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016625 As a gRPC message is always made of a gRPC header followed by protocol buffers
16626 messages, in the previous example the "latitude" of "lo" first PPoint
16627 could be extracted with these equivalent directives:
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016628
16629 req.body,ungrpc(48.59),protobuf(1,int32)
16630 req.body,ungrpc(48),protobuf(59.1,int32)
16631 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
16632
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016633 Note that the first convert must be "ungrpc", the remaining ones must be
16634 "protobuf" and only the last one may have or not a second argument to
16635 interpret the previous binary sample.
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016636
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016637
Tim Duesterhusef4e45c2021-01-21 17:40:50 +010016638unset-var(<var>)
Christopher Faulet85d79c92016-11-09 16:54:56 +010016639 Unsets a variable if the input content is defined. The name of the variable
16640 starts with an indication about its scope. The scopes allowed are:
16641 "proc" : the variable is shared with the whole process
16642 "sess" : the variable is shared with the whole session
16643 "txn" : the variable is shared with the transaction (request and
16644 response),
16645 "req" : the variable is shared only during request processing,
16646 "res" : the variable is shared only during response processing.
16647 This prefix is followed by a name. The separator is a '.'. The name may only
16648 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
16649
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016650utime(<format>[,<offset>])
16651 Converts an integer supposed to contain a date since epoch to a string
16652 representing this date in UTC time using a format defined by the <format>
16653 string using strftime(3). The purpose is to allow any date format to be used
16654 in logs. An optional <offset> in seconds may be applied to the input date
16655 (positive or negative). See the strftime() man page for the format supported
16656 by your operating system. See also the ltime converter.
16657
16658 Example :
16659
16660 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010016661 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016662 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
16663
Marcin Deranek9631a282018-04-16 14:30:46 +020016664word(<index>,<delimiters>[,<count>])
16665 Extracts the nth word counting from the beginning (positive index) or from
16666 the end (negative index) considering given delimiters from an input string.
16667 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
Jerome Magnin88209322020-01-28 13:33:44 +010016668 Delimiters at the beginning or end of the input string are ignored.
Marcin Deranek9631a282018-04-16 14:30:46 +020016669 Optionally you can specify <count> of words to extract (default: 1).
16670 Value of 0 indicates extraction of all remaining words.
16671
16672 Example :
16673 str(f1_f2_f3__f5),word(4,_) # f5
16674 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
16675 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
16676 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
16677 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Jerome Magnin88209322020-01-28 13:33:44 +010016678 str(/f1/f2/f3/f4),word(1,/) # f1
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010016679
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016680wt6([<avalanche>])
16681 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
16682 hash function. Optionally, it is possible to apply a full avalanche hash
16683 function to the output if the optional <avalanche> argument equals 1. This
16684 converter uses the same functions as used by the various hash-based load
16685 balancing algorithms, so it will provide exactly the same results. It is
16686 mostly intended for debugging, but can be used as a stick-table entry to
16687 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010016688 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
16689 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016690
Willy Tarreau97707872015-01-27 15:12:13 +010016691xor(<value>)
16692 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016693 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016694 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016695 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016696 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016697 "sess" : the variable is shared with the whole session
16698 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016699 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016700 "req" : the variable is shared only during request processing,
16701 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016702 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016703 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016704
Dragan Dosen04bf0cc2020-12-22 21:44:33 +010016705xxh3([<seed>])
16706 Hashes a binary input sample into a signed 64-bit quantity using the XXH3
16707 64-bit variant of the XXhash hash function. This hash supports a seed which
16708 defaults to zero but a different value maybe passed as the <seed> argument.
16709 This hash is known to be very good and very fast so it can be used to hash
16710 URLs and/or URL parameters for use as stick-table keys to collect statistics
16711 with a low collision rate, though care must be taken as the algorithm is not
16712 considered as cryptographically secure.
16713
Thierry FOURNIER01e09742016-12-26 11:46:11 +010016714xxh32([<seed>])
16715 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
16716 variant of the XXHash hash function. This hash supports a seed which defaults
16717 to zero but a different value maybe passed as the <seed> argument. This hash
16718 is known to be very good and very fast so it can be used to hash URLs and/or
16719 URL parameters for use as stick-table keys to collect statistics with a low
16720 collision rate, though care must be taken as the algorithm is not considered
16721 as cryptographically secure.
16722
16723xxh64([<seed>])
16724 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
16725 variant of the XXHash hash function. This hash supports a seed which defaults
16726 to zero but a different value maybe passed as the <seed> argument. This hash
16727 is known to be very good and very fast so it can be used to hash URLs and/or
16728 URL parameters for use as stick-table keys to collect statistics with a low
16729 collision rate, though care must be taken as the algorithm is not considered
16730 as cryptographically secure.
16731
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016732
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200167337.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020016734--------------------------------------------
16735
16736A first set of sample fetch methods applies to internal information which does
16737not even relate to any client information. These ones are sometimes used with
16738"monitor-fail" directives to report an internal status to external watchers.
16739The sample fetch methods described in this section are usable anywhere.
16740
16741always_false : boolean
16742 Always returns the boolean "false" value. It may be used with ACLs as a
16743 temporary replacement for another one when adjusting configurations.
16744
16745always_true : boolean
16746 Always returns the boolean "true" value. It may be used with ACLs as a
16747 temporary replacement for another one when adjusting configurations.
16748
16749avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016750 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020016751 divided by the number of active servers. The current backend is used if no
16752 backend is specified. This is very similar to "queue" except that the size of
16753 the farm is considered, in order to give a more accurate measurement of the
16754 time it may take for a new connection to be processed. The main usage is with
16755 ACL to return a sorry page to new users when it becomes certain they will get
16756 a degraded service, or to pass to the backend servers in a header so that
16757 they decide to work in degraded mode or to disable some functions to speed up
16758 the processing a bit. Note that in the event there would not be any active
16759 server anymore, twice the number of queued connections would be considered as
16760 the measured value. This is a fair estimate, as we expect one server to get
16761 back soon anyway, but we still prefer to send new traffic to another backend
16762 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
16763 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010016764
Willy Tarreau74ca5042013-06-11 23:12:07 +020016765be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020016766 Applies to the number of currently established connections on the backend,
16767 possibly including the connection being evaluated. If no backend name is
16768 specified, the current one is used. But it is also possible to check another
16769 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040016770 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
16771
16772be_conn_free([<backend>]) : integer
16773 Returns an integer value corresponding to the number of available connections
16774 across available servers in the backend. Queue slots are not included. Backup
16775 servers are also not included, unless all other servers are down. If no
16776 backend name is specified, the current one is used. But it is also possible
16777 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040016778 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
16779 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040016780
16781 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
16782 (meaning unlimited), then this fetch clearly does not make sense, in which
16783 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016784
Willy Tarreau74ca5042013-06-11 23:12:07 +020016785be_sess_rate([<backend>]) : integer
16786 Returns an integer value corresponding to the sessions creation rate on the
16787 backend, in number of new sessions per second. This is used with ACLs to
16788 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010016789 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020016790 online dictionary). It can also be useful to add this element to logs using a
16791 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016792
16793 Example :
16794 # Redirect to an error page if the dictionary is requested too often
16795 backend dynamic
16796 mode http
16797 acl being_scanned be_sess_rate gt 100
16798 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010016799
Davor Ocelice9ed2812017-12-25 17:49:28 +010016800bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016801 Returns a binary chain. The input is the hexadecimal representation
16802 of the string.
16803
16804bool(<bool>) : bool
16805 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
16806 'false' and '0' are the same. 'true' and '1' are the same.
16807
Willy Tarreau74ca5042013-06-11 23:12:07 +020016808connslots([<backend>]) : integer
16809 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016810 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020016811 connections on all servers and the maximum queue size. This is probably only
16812 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050016813
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016814 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020016815 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016816 usage; see "use_backend" keyword) can be redirected to a different backend.
16817
Willy Tarreau55165fe2009-05-10 12:02:55 +020016818 'connslots' = number of available server connection slots, + number of
16819 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016820
Willy Tarreaua36af912009-10-10 12:02:45 +020016821 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020016822 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020016823 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020016824 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010016825 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020016826 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020016827 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016828
Willy Tarreau55165fe2009-05-10 12:02:55 +020016829 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
16830 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020016831 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020016832 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016833
Willy Tarreau70fe9442018-11-22 16:07:39 +010016834cpu_calls : integer
16835 Returns the number of calls to the task processing the stream or current
16836 request since it was allocated. This number is reset for each new request on
16837 the same connections in case of HTTP keep-alive. This value should usually be
16838 low and stable (around 2 calls for a typically simple request) but may become
16839 high if some processing (compression, caching or analysis) is performed. This
16840 is purely for performance monitoring purposes.
16841
16842cpu_ns_avg : integer
16843 Returns the average number of nanoseconds spent in each call to the task
16844 processing the stream or current request. This number is reset for each new
16845 request on the same connections in case of HTTP keep-alive. This value
16846 indicates the overall cost of processing the request or the connection for
16847 each call. There is no good nor bad value but the time spent in a call
16848 automatically causes latency for other processing (see lat_ns_avg below),
16849 and may affect other connection's apparent response time. Certain operations
16850 like compression, complex regex matching or heavy Lua operations may directly
16851 affect this value, and having it in the logs will make it easier to spot the
16852 faulty processing that needs to be fixed to recover decent performance.
16853 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
16854
16855cpu_ns_tot : integer
16856 Returns the total number of nanoseconds spent in each call to the task
16857 processing the stream or current request. This number is reset for each new
16858 request on the same connections in case of HTTP keep-alive. This value
16859 indicates the overall cost of processing the request or the connection for
16860 each call. There is no good nor bad value but the time spent in a call
16861 automatically causes latency for other processing (see lat_ns_avg below),
16862 induces CPU costs on the machine, and may affect other connection's apparent
16863 response time. Certain operations like compression, complex regex matching or
16864 heavy Lua operations may directly affect this value, and having it in the
16865 logs will make it easier to spot the faulty processing that needs to be fixed
16866 to recover decent performance. The value may be artificially high due to a
16867 high cpu_calls count, for example when processing many HTTP chunks, and for
16868 this reason it is often preferred to log cpu_ns_avg instead.
16869
Cyril Bonté6bcd1822019-11-05 23:13:59 +010016870date([<offset>],[<unit>]) : integer
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016871 Returns the current date as the epoch (number of seconds since 01/01/1970).
Damien Claisseae6f1252019-10-30 15:57:28 +000016872
16873 If an offset value is specified, then it is added to the current date before
16874 returning the value. This is particularly useful to compute relative dates,
16875 as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020016876 It is useful combined with the http_date converter.
16877
Damien Claisseae6f1252019-10-30 15:57:28 +000016878 <unit> is facultative, and can be set to "s" for seconds (default behavior),
16879 "ms" for milliseconds or "us" for microseconds.
16880 If unit is set, return value is an integer reflecting either seconds,
16881 milliseconds or microseconds since epoch, plus offset.
16882 It is useful when a time resolution of less than a second is needed.
16883
Willy Tarreau276fae92013-07-25 14:36:01 +020016884 Example :
16885
16886 # set an expires header to now+1 hour in every response
16887 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016888
Damien Claisseae6f1252019-10-30 15:57:28 +000016889 # set an expires header to now+1 hour in every response, with
16890 # millisecond granularity
16891 http-response set-header Expires %[date(3600000,ms),http_date(0,ms)]
16892
Etienne Carrierea792a0a2018-01-17 13:43:24 +010016893date_us : integer
16894 Return the microseconds part of the date (the "second" part is returned by
16895 date sample). This sample is coherent with the date sample as it is comes
16896 from the same timeval structure.
16897
Willy Tarreaud716f9b2017-10-13 11:03:15 +020016898distcc_body(<token>[,<occ>]) : binary
16899 Parses a distcc message and returns the body associated to occurrence #<occ>
16900 of the token <token>. Occurrences start at 1, and when unspecified, any may
16901 match though in practice only the first one is checked for now. This can be
16902 used to extract file names or arguments in files built using distcc through
16903 haproxy. Please refer to distcc's protocol documentation for the complete
16904 list of supported tokens.
16905
16906distcc_param(<token>[,<occ>]) : integer
16907 Parses a distcc message and returns the parameter associated to occurrence
16908 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
16909 any may match though in practice only the first one is checked for now. This
16910 can be used to extract certain information such as the protocol version, the
16911 file size or the argument in files built using distcc through haproxy.
16912 Another use case consists in waiting for the start of the preprocessed file
16913 contents before connecting to the server to avoid keeping idle connections.
16914 Please refer to distcc's protocol documentation for the complete list of
16915 supported tokens.
16916
16917 Example :
16918 # wait up to 20s for the pre-processed file to be uploaded
16919 tcp-request inspect-delay 20s
16920 tcp-request content accept if { distcc_param(DOTI) -m found }
16921 # send large files to the big farm
16922 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
16923
Willy Tarreau595ec542013-06-12 21:34:28 +020016924env(<name>) : string
16925 Returns a string containing the value of environment variable <name>. As a
16926 reminder, environment variables are per-process and are sampled when the
16927 process starts. This can be useful to pass some information to a next hop
16928 server, or with ACLs to take specific action when the process is started a
16929 certain way.
16930
16931 Examples :
16932 # Pass the Via header to next hop with the local hostname in it
16933 http-request add-header Via 1.1\ %[env(HOSTNAME)]
16934
16935 # reject cookie-less requests when the STOP environment variable is set
16936 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
16937
Willy Tarreau74ca5042013-06-11 23:12:07 +020016938fe_conn([<frontend>]) : integer
16939 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010016940 possibly including the connection being evaluated. If no frontend name is
16941 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020016942 frontend. It can be used to return a sorry page before hard-blocking, or to
16943 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010016944 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020016945 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
16946 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020016947
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020016948fe_req_rate([<frontend>]) : integer
16949 Returns an integer value corresponding to the number of HTTP requests per
16950 second sent to a frontend. This number can differ from "fe_sess_rate" in
16951 situations where client-side keep-alive is enabled.
16952
Willy Tarreau74ca5042013-06-11 23:12:07 +020016953fe_sess_rate([<frontend>]) : integer
16954 Returns an integer value corresponding to the sessions creation rate on the
16955 frontend, in number of new sessions per second. This is used with ACLs to
16956 limit the incoming session rate to an acceptable range in order to prevent
16957 abuse of service at the earliest moment, for example when combined with other
16958 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
16959 down below the limit. It can also be useful to add this element to logs using
16960 a log-format directive. See also the "rate-limit sessions" directive for use
16961 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010016962
16963 Example :
16964 # This frontend limits incoming mails to 10/s with a max of 100
16965 # concurrent connections. We accept any connection below 10/s, and
16966 # force excess clients to wait for 100 ms. Since clients are limited to
16967 # 100 max, there cannot be more than 10 incoming mails per second.
16968 frontend mail
16969 bind :25
16970 mode tcp
16971 maxconn 100
16972 acl too_fast fe_sess_rate ge 10
16973 tcp-request inspect-delay 100ms
16974 tcp-request content accept if ! too_fast
16975 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010016976
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010016977hostname : string
16978 Returns the system hostname.
16979
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016980int(<integer>) : signed integer
16981 Returns a signed integer.
16982
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016983ipv4(<ipv4>) : ipv4
16984 Returns an ipv4.
16985
16986ipv6(<ipv6>) : ipv6
16987 Returns an ipv6.
16988
Willy Tarreau70fe9442018-11-22 16:07:39 +010016989lat_ns_avg : integer
16990 Returns the average number of nanoseconds spent between the moment the task
16991 handling the stream is woken up and the moment it is effectively called. This
16992 number is reset for each new request on the same connections in case of HTTP
16993 keep-alive. This value indicates the overall latency inflicted to the current
16994 request by all other requests being processed in parallel, and is a direct
16995 indicator of perceived performance due to noisy neighbours. In order to keep
16996 the value low, it is possible to reduce the scheduler's run queue depth using
16997 "tune.runqueue-depth", to reduce the number of concurrent events processed at
16998 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020016999 the "nice" option on the "bind" lines or in the frontend, to enable low
17000 latency scheduling using "tune.sched.low-latency", or to look for other heavy
17001 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
17002 processing needs to be adjusted or fixed. Compression of large buffers could
17003 be a culprit, like heavy regex or long lists of regex. Note: this value is
17004 exactly lat_ns_tot divided by cpu_calls.
Willy Tarreau70fe9442018-11-22 16:07:39 +010017005
17006lat_ns_tot : integer
17007 Returns the total number of nanoseconds spent between the moment the task
17008 handling the stream is woken up and the moment it is effectively called. This
17009 number is reset for each new request on the same connections in case of HTTP
17010 keep-alive. This value indicates the overall latency inflicted to the current
17011 request by all other requests being processed in parallel, and is a direct
17012 indicator of perceived performance due to noisy neighbours. In order to keep
17013 the value low, it is possible to reduce the scheduler's run queue depth using
17014 "tune.runqueue-depth", to reduce the number of concurrent events processed at
17015 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020017016 the "nice" option on the "bind" lines or in the frontend, to enable low
17017 latency scheduling using "tune.sched.low-latency", or to look for other heavy
17018 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
17019 processing needs to be adjusted or fixed. Compression of large buffers could
17020 be a culprit, like heavy regex or long lists of regex. Note: while it
Willy Tarreau70fe9442018-11-22 16:07:39 +010017021 may intuitively seem that the total latency adds to a transfer time, it is
17022 almost never true because while a task waits for the CPU, network buffers
17023 continue to fill up and the next call will process more at once. The value
17024 may be artificially high due to a high cpu_calls count, for example when
17025 processing many HTTP chunks, and for this reason it is often preferred to log
17026 lat_ns_avg instead, which is a more relevant performance indicator.
17027
Thierry FOURNIERcc103292015-06-06 19:30:17 +020017028meth(<method>) : method
17029 Returns a method.
17030
Willy Tarreau0f30d262014-11-24 16:02:05 +010017031nbproc : integer
17032 Returns an integer value corresponding to the number of processes that were
17033 started (it equals the global "nbproc" setting). This is useful for logging
17034 and debugging purposes.
17035
Willy Tarreau74ca5042013-06-11 23:12:07 +020017036nbsrv([<backend>]) : integer
17037 Returns an integer value corresponding to the number of usable servers of
17038 either the current backend or the named backend. This is mostly used with
17039 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010017040 switch to an alternate backend when the number of servers is too low to
17041 to handle some load. It is useful to report a failure when combined with
17042 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010017043
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040017044prio_class : integer
17045 Returns the priority class of the current session for http mode or connection
17046 for tcp mode. The value will be that set by the last call to "http-request
17047 set-priority-class" or "tcp-request content set-priority-class".
17048
17049prio_offset : integer
17050 Returns the priority offset of the current session for http mode or
17051 connection for tcp mode. The value will be that set by the last call to
17052 "http-request set-priority-offset" or "tcp-request content
17053 set-priority-offset".
17054
Willy Tarreau0f30d262014-11-24 16:02:05 +010017055proc : integer
17056 Returns an integer value corresponding to the position of the process calling
17057 the function, between 1 and global.nbproc. This is useful for logging and
17058 debugging purposes.
17059
Willy Tarreau74ca5042013-06-11 23:12:07 +020017060queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010017061 Returns the total number of queued connections of the designated backend,
17062 including all the connections in server queues. If no backend name is
17063 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020017064 one. This is useful with ACLs or to pass statistics to backend servers. This
17065 can be used to take actions when queuing goes above a known level, generally
17066 indicating a surge of traffic or a massive slowdown on the servers. One
17067 possible action could be to reject new users but still accept old ones. See
17068 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
17069
Willy Tarreau84310e22014-02-14 11:59:04 +010017070rand([<range>]) : integer
17071 Returns a random integer value within a range of <range> possible values,
17072 starting at zero. If the range is not specified, it defaults to 2^32, which
17073 gives numbers between 0 and 4294967295. It can be useful to pass some values
17074 needed to take some routing decisions for example, or just for debugging
17075 purposes. This random must not be used for security purposes.
17076
Luca Schimweg8a694b82019-09-10 15:42:52 +020017077uuid([<version>]) : string
17078 Returns a UUID following the RFC4122 standard. If the version is not
17079 specified, a UUID version 4 (fully random) is returned.
17080 Currently, only version 4 is supported.
17081
Willy Tarreau74ca5042013-06-11 23:12:07 +020017082srv_conn([<backend>/]<server>) : integer
17083 Returns an integer value corresponding to the number of currently established
17084 connections on the designated server, possibly including the connection being
17085 evaluated. If <backend> is omitted, then the server is looked up in the
17086 current backend. It can be used to use a specific farm when one server is
17087 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040017088 connections with it. See also the "fe_conn", "be_conn", "queue", and
17089 "srv_conn_free" fetch methods.
17090
17091srv_conn_free([<backend>/]<server>) : integer
17092 Returns an integer value corresponding to the number of available connections
17093 on the designated server, possibly including the connection being evaluated.
17094 The value does not include queue slots. If <backend> is omitted, then the
17095 server is looked up in the current backend. It can be used to use a specific
17096 farm when one server is full, or to inform the server about our view of the
17097 number of active connections with it. See also the "be_conn_free" and
17098 "srv_conn" fetch methods.
17099
17100 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
17101 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017102
17103srv_is_up([<backend>/]<server>) : boolean
17104 Returns true when the designated server is UP, and false when it is either
17105 DOWN or in maintenance mode. If <backend> is omitted, then the server is
17106 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010017107 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017108 availability). Another possible use which is more of a hack consists in
17109 using dummy servers as boolean variables that can be enabled or disabled from
17110 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
17111
Willy Tarreauff2b7af2017-10-13 11:46:26 +020017112srv_queue([<backend>/]<server>) : integer
17113 Returns an integer value corresponding to the number of connections currently
17114 pending in the designated server's queue. If <backend> is omitted, then the
17115 server is looked up in the current backend. It can sometimes be used together
17116 with the "use-server" directive to force to use a known faster server when it
17117 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
17118 fetch methods.
17119
Willy Tarreau74ca5042013-06-11 23:12:07 +020017120srv_sess_rate([<backend>/]<server>) : integer
17121 Returns an integer corresponding to the sessions creation rate on the
17122 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017123 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020017124 used with ACLs but can make sense with logs too. This is used to switch to an
17125 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010017126 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020017127 overloading servers).
17128
17129 Example :
17130 # Redirect to a separate back
17131 acl srv1_full srv_sess_rate(be1/srv1) gt 50
17132 acl srv2_full srv_sess_rate(be1/srv2) gt 50
17133 use_backend be2 if srv1_full or srv2_full
17134
Christopher Faulet1bea8652020-07-10 16:03:45 +020017135srv_iweight([<backend>/]<server>): integer
17136 Returns an integer corresponding to the server's initial weight. If <backend>
17137 is omitted, then the server is looked up in the current backend. See also
17138 "srv_weight" and "srv_uweight".
17139
17140srv_uweight([<backend>/]<server>): integer
17141 Returns an integer corresponding to the user visible server's weight. If
17142 <backend> is omitted, then the server is looked up in the current
17143 backend. See also "srv_weight" and "srv_iweight".
17144
17145srv_weight([<backend>/]<server>): integer
17146 Returns an integer corresponding to the current (or effective) server's
17147 weight. If <backend> is omitted, then the server is looked up in the current
17148 backend. See also "srv_iweight" and "srv_uweight".
17149
Willy Tarreau0f30d262014-11-24 16:02:05 +010017150stopping : boolean
17151 Returns TRUE if the process calling the function is currently stopping. This
17152 can be useful for logging, or for relaxing certain checks or helping close
17153 certain connections upon graceful shutdown.
17154
Thierry FOURNIERcc103292015-06-06 19:30:17 +020017155str(<string>) : string
17156 Returns a string.
17157
Willy Tarreau74ca5042013-06-11 23:12:07 +020017158table_avl([<table>]) : integer
17159 Returns the total number of available entries in the current proxy's
17160 stick-table or in the designated stick-table. See also table_cnt.
17161
17162table_cnt([<table>]) : integer
17163 Returns the total number of entries currently in use in the current proxy's
17164 stick-table or in the designated stick-table. See also src_conn_cnt and
17165 table_avl for other entry counting methods.
17166
Christopher Faulet34adb2a2017-11-21 21:45:38 +010017167thread : integer
17168 Returns an integer value corresponding to the position of the thread calling
17169 the function, between 0 and (global.nbthread-1). This is useful for logging
17170 and debugging purposes.
17171
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017172var(<var-name>) : undefined
17173 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010017174 sample fetch fails. The name of the variable starts with an indication
17175 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010017176 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010017177 "sess" : the variable is shared with the whole session
17178 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017179 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010017180 "req" : the variable is shared only during request processing,
17181 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017182 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010017183 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020017184
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200171857.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020017186----------------------------------
17187
17188The layer 4 usually describes just the transport layer which in haproxy is
17189closest to the connection, where no content is yet made available. The fetch
17190methods described here are usable as low as the "tcp-request connection" rule
17191sets unless they require some future information. Those generally include
17192TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017193the incoming connection. For retrieving a value from a sticky counters, the
17194counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020017195"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
17196used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
17197can be specified as the first integer argument when using the "sc_" prefix.
17198Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
17199table may be specified with the "sc*" form, in which case the currently
17200tracked key will be looked up into this alternate table instead of the table
17201currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017202
Jérôme Magnin35e53a62019-01-16 14:38:37 +010017203bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010017204 Returns the backend connection's HTTP major version encoding, which may be 1
17205 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
17206 encoding and not the version present in the request header.
17207
Willy Tarreau74ca5042013-06-11 23:12:07 +020017208be_id : integer
17209 Returns an integer containing the current backend's id. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020017210 frontends with responses to check which backend processed the request. It can
17211 also be used in a tcp-check or an http-check ruleset.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017212
Marcin Deranekd2471c22016-12-12 14:08:05 +010017213be_name : string
17214 Returns a string containing the current backend's name. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020017215 frontends with responses to check which backend processed the request. It can
17216 also be used in a tcp-check or an http-check ruleset.
Marcin Deranekd2471c22016-12-12 14:08:05 +010017217
Amaury Denoyelled91d7792020-12-10 13:43:56 +010017218be_server_timeout : integer
17219 Returns the configuration value in millisecond for the server timeout of the
17220 current backend. This timeout can be overwritten by a "set-timeout" rule. See
17221 also the "cur_server_timeout".
17222
17223be_tunnel_timeout : integer
17224 Returns the configuration value in millisecond for the tunnel timeout of the
17225 current backend. This timeout can be overwritten by a "set-timeout" rule. See
17226 also the "cur_tunnel_timeout".
17227
Amaury Denoyellef7719a22020-12-10 13:43:58 +010017228cur_server_timeout : integer
17229 Returns the currently applied server timeout in millisecond for the stream.
17230 In the default case, this will be equal to be_server_timeout unless a
17231 "set-timeout" rule has been applied. See also "be_server_timeout".
17232
17233cur_tunnel_timeout : integer
17234 Returns the currently applied tunnel timeout in millisecond for the stream.
17235 In the default case, this will be equal to be_tunnel_timeout unless a
17236 "set-timeout" rule has been applied. See also "be_tunnel_timeout".
17237
Willy Tarreau74ca5042013-06-11 23:12:07 +020017238dst : ip
17239 This is the destination IPv4 address of the connection on the client side,
17240 which is the address the client connected to. It can be useful when running
17241 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
17242 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010017243 RFC 4291. When the incoming connection passed through address translation or
17244 redirection involving connection tracking, the original destination address
17245 before the redirection will be reported. On Linux systems, the source and
17246 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
17247 is set, because a late response may reopen a timed out connection and switch
17248 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017249
17250dst_conn : integer
17251 Returns an integer value corresponding to the number of currently established
17252 connections on the same socket including the one being evaluated. It is
17253 normally used with ACLs but can as well be used to pass the information to
17254 servers in an HTTP header or in logs. It can be used to either return a sorry
17255 page before hard-blocking, or to use a specific backend to drain new requests
17256 when the socket is considered saturated. This offers the ability to assign
17257 different limits to different listening ports or addresses. See also the
17258 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017259
Willy Tarreau16e01562016-08-09 16:46:18 +020017260dst_is_local : boolean
17261 Returns true if the destination address of the incoming connection is local
17262 to the system, or false if the address doesn't exist on the system, meaning
17263 that it was intercepted in transparent mode. It can be useful to apply
17264 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010017265 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020017266 be delivered only on this address, or SSH access could be locally redirected.
17267 Please note that the check involves a few system calls, so it's better to do
17268 it only once per connection.
17269
Willy Tarreau74ca5042013-06-11 23:12:07 +020017270dst_port : integer
17271 Returns an integer value corresponding to the destination TCP port of the
17272 connection on the client side, which is the port the client connected to.
17273 This might be used when running in transparent mode, when assigning dynamic
17274 ports to some clients for a whole application session, to stick all users to
17275 a same server, or to pass the destination port information to a server using
17276 an HTTP header.
17277
Willy Tarreau60ca10a2017-08-18 15:26:54 +020017278fc_http_major : integer
17279 Reports the front connection's HTTP major version encoding, which may be 1
17280 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
17281 encoding and not on the version present in the request header.
17282
Geoff Simmons7185b782019-08-27 18:31:16 +020017283fc_pp_authority : string
17284 Returns the authority TLV sent by the client in the PROXY protocol header,
17285 if any.
17286
Tim Duesterhusd1b15b62020-03-13 12:34:23 +010017287fc_pp_unique_id : string
17288 Returns the unique ID TLV sent by the client in the PROXY protocol header,
17289 if any.
17290
Emeric Brun4f603012017-01-05 15:11:44 +010017291fc_rcvd_proxy : boolean
17292 Returns true if the client initiated the connection with a PROXY protocol
17293 header.
17294
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020017295fc_rtt(<unit>) : integer
17296 Returns the Round Trip Time (RTT) measured by the kernel for the client
17297 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
17298 can be set to "ms" for milliseconds or "us" for microseconds. If the server
17299 connection is not established, if the connection is not TCP or if the
17300 operating system does not support TCP_INFO, for example Linux kernels before
17301 2.4, the sample fetch fails.
17302
17303fc_rttvar(<unit>) : integer
17304 Returns the Round Trip Time (RTT) variance measured by the kernel for the
17305 client connection. <unit> is facultative, by default the unit is milliseconds.
17306 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
17307 server connection is not established, if the connection is not TCP or if the
17308 operating system does not support TCP_INFO, for example Linux kernels before
17309 2.4, the sample fetch fails.
17310
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017311fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017312 Returns the unacked counter measured by the kernel for the client connection.
17313 If the server connection is not established, if the connection is not TCP or
17314 if the operating system does not support TCP_INFO, for example Linux kernels
17315 before 2.4, the sample fetch fails.
17316
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017317fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017318 Returns the sacked counter measured by the kernel for the client connection.
17319 If the server connection is not established, if the connection is not TCP or
17320 if the operating system does not support TCP_INFO, for example Linux kernels
17321 before 2.4, the sample fetch fails.
17322
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017323fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017324 Returns the retransmits counter measured by the kernel for the client
17325 connection. If the server connection is not established, if the connection is
17326 not TCP or if the operating system does not support TCP_INFO, for example
17327 Linux kernels before 2.4, the sample fetch fails.
17328
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017329fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017330 Returns the fack counter measured by the kernel for the client
17331 connection. If the server connection is not established, if the connection is
17332 not TCP or if the operating system does not support TCP_INFO, for example
17333 Linux kernels before 2.4, the sample fetch fails.
17334
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017335fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017336 Returns the lost counter measured by the kernel for the client
17337 connection. If the server connection is not established, if the connection is
17338 not TCP or if the operating system does not support TCP_INFO, for example
17339 Linux kernels before 2.4, the sample fetch fails.
17340
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017341fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017342 Returns the reordering counter measured by the kernel for the client
17343 connection. If the server connection is not established, if the connection is
17344 not TCP or if the operating system does not support TCP_INFO, for example
17345 Linux kernels before 2.4, the sample fetch fails.
17346
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020017347fe_defbe : string
17348 Returns a string containing the frontend's default backend name. It can be
17349 used in frontends to check which backend will handle requests by default.
17350
Willy Tarreau74ca5042013-06-11 23:12:07 +020017351fe_id : integer
17352 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010017353 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020017354 coming via a same frontend to the same server.
17355
Marcin Deranekd2471c22016-12-12 14:08:05 +010017356fe_name : string
17357 Returns a string containing the current frontend's name. It can be used in
17358 backends to check from which frontend it was called, or to stick all users
17359 coming via a same frontend to the same server.
17360
Amaury Denoyelleda184d52020-12-10 13:43:55 +010017361fe_client_timeout : integer
17362 Returns the configuration value in millisecond for the client timeout of the
17363 current frontend.
17364
Cyril Bonté62ba8702014-04-22 23:52:25 +020017365sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017366sc0_bytes_in_rate([<table>]) : integer
17367sc1_bytes_in_rate([<table>]) : integer
17368sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017369 Returns the average client-to-server bytes rate from the currently tracked
17370 counters, measured in amount of bytes over the period configured in the
17371 table. See also src_bytes_in_rate.
17372
Cyril Bonté62ba8702014-04-22 23:52:25 +020017373sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017374sc0_bytes_out_rate([<table>]) : integer
17375sc1_bytes_out_rate([<table>]) : integer
17376sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017377 Returns the average server-to-client bytes rate from the currently tracked
17378 counters, measured in amount of bytes over the period configured in the
17379 table. See also src_bytes_out_rate.
17380
Cyril Bonté62ba8702014-04-22 23:52:25 +020017381sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017382sc0_clr_gpc0([<table>]) : integer
17383sc1_clr_gpc0([<table>]) : integer
17384sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020017385 Clears the first General Purpose Counter associated to the currently tracked
17386 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010017387 stored value is zero, so first invocation will always return zero. This is
17388 typically used as a second ACL in an expression in order to mark a connection
17389 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020017390
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017391 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020017392 # block if 5 consecutive requests continue to come faster than 10 sess
17393 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017394 acl abuse sc0_http_req_rate gt 10
17395 acl kill sc0_inc_gpc0 gt 5
17396 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020017397 tcp-request connection accept if !abuse save
17398 tcp-request connection reject if abuse kill
17399
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017400sc_clr_gpc1(<ctr>[,<table>]) : integer
17401sc0_clr_gpc1([<table>]) : integer
17402sc1_clr_gpc1([<table>]) : integer
17403sc2_clr_gpc1([<table>]) : integer
17404 Clears the second General Purpose Counter associated to the currently tracked
17405 counters, and returns its previous value. Before the first invocation, the
17406 stored value is zero, so first invocation will always return zero. This is
17407 typically used as a second ACL in an expression in order to mark a connection
17408 when a first ACL was verified.
17409
Cyril Bonté62ba8702014-04-22 23:52:25 +020017410sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017411sc0_conn_cnt([<table>]) : integer
17412sc1_conn_cnt([<table>]) : integer
17413sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017414 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017415 counters. See also src_conn_cnt.
17416
Cyril Bonté62ba8702014-04-22 23:52:25 +020017417sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017418sc0_conn_cur([<table>]) : integer
17419sc1_conn_cur([<table>]) : integer
17420sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017421 Returns the current amount of concurrent connections tracking the same
17422 tracked counters. This number is automatically incremented when tracking
17423 begins and decremented when tracking stops. See also src_conn_cur.
17424
Cyril Bonté62ba8702014-04-22 23:52:25 +020017425sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017426sc0_conn_rate([<table>]) : integer
17427sc1_conn_rate([<table>]) : integer
17428sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017429 Returns the average connection rate from the currently tracked counters,
17430 measured in amount of connections over the period configured in the table.
17431 See also src_conn_rate.
17432
Cyril Bonté62ba8702014-04-22 23:52:25 +020017433sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017434sc0_get_gpc0([<table>]) : integer
17435sc1_get_gpc0([<table>]) : integer
17436sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017437 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017438 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017439
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017440sc_get_gpc1(<ctr>[,<table>]) : integer
17441sc0_get_gpc1([<table>]) : integer
17442sc1_get_gpc1([<table>]) : integer
17443sc2_get_gpc1([<table>]) : integer
17444 Returns the value of the second General Purpose Counter associated to the
17445 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
17446
Thierry FOURNIER236657b2015-08-19 08:25:14 +020017447sc_get_gpt0(<ctr>[,<table>]) : integer
17448sc0_get_gpt0([<table>]) : integer
17449sc1_get_gpt0([<table>]) : integer
17450sc2_get_gpt0([<table>]) : integer
17451 Returns the value of the first General Purpose Tag associated to the
17452 currently tracked counters. See also src_get_gpt0.
17453
Cyril Bonté62ba8702014-04-22 23:52:25 +020017454sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017455sc0_gpc0_rate([<table>]) : integer
17456sc1_gpc0_rate([<table>]) : integer
17457sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017458 Returns the average increment rate of the first General Purpose Counter
17459 associated to the currently tracked counters. It reports the frequency
17460 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017461 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
17462 that the "gpc0_rate" counter must be stored in the stick-table for a value to
17463 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020017464
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017465sc_gpc1_rate(<ctr>[,<table>]) : integer
17466sc0_gpc1_rate([<table>]) : integer
17467sc1_gpc1_rate([<table>]) : integer
17468sc2_gpc1_rate([<table>]) : integer
17469 Returns the average increment rate of the second General Purpose Counter
17470 associated to the currently tracked counters. It reports the frequency
17471 which the gpc1 counter was incremented over the configured period. See also
17472 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
17473 that the "gpc1_rate" counter must be stored in the stick-table for a value to
17474 be returned, as "gpc1" only holds the event count.
17475
Cyril Bonté62ba8702014-04-22 23:52:25 +020017476sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017477sc0_http_err_cnt([<table>]) : integer
17478sc1_http_err_cnt([<table>]) : integer
17479sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017480 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017481 counters. This includes the both request errors and 4xx error responses.
17482 See also src_http_err_cnt.
17483
Cyril Bonté62ba8702014-04-22 23:52:25 +020017484sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017485sc0_http_err_rate([<table>]) : integer
17486sc1_http_err_rate([<table>]) : integer
17487sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017488 Returns the average rate of HTTP errors from the currently tracked counters,
17489 measured in amount of errors over the period configured in the table. This
17490 includes the both request errors and 4xx error responses. See also
17491 src_http_err_rate.
17492
Willy Tarreau826f3ab2021-02-10 12:07:15 +010017493sc_http_fail_cnt(<ctr>[,<table>]) : integer
17494sc0_http_fail_cnt([<table>]) : integer
17495sc1_http_fail_cnt([<table>]) : integer
17496sc2_http_fail_cnt([<table>]) : integer
17497 Returns the cumulative number of HTTP response failures from the currently
17498 tracked counters. This includes the both response errors and 5xx status codes
17499 other than 501 and 505. See also src_http_fail_cnt.
17500
17501sc_http_fail_rate(<ctr>[,<table>]) : integer
17502sc0_http_fail_rate([<table>]) : integer
17503sc1_http_fail_rate([<table>]) : integer
17504sc2_http_fail_rate([<table>]) : integer
17505 Returns the average rate of HTTP response failures from the currently tracked
17506 counters, measured in amount of failures over the period configured in the
17507 table. This includes the both response errors and 5xx status codes other than
17508 501 and 505. See also src_http_fail_rate.
17509
Cyril Bonté62ba8702014-04-22 23:52:25 +020017510sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017511sc0_http_req_cnt([<table>]) : integer
17512sc1_http_req_cnt([<table>]) : integer
17513sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017514 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017515 counters. This includes every started request, valid or not. See also
17516 src_http_req_cnt.
17517
Cyril Bonté62ba8702014-04-22 23:52:25 +020017518sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017519sc0_http_req_rate([<table>]) : integer
17520sc1_http_req_rate([<table>]) : integer
17521sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017522 Returns the average rate of HTTP requests from the currently tracked
17523 counters, measured in amount of requests over the period configured in
17524 the table. This includes every started request, valid or not. See also
17525 src_http_req_rate.
17526
Cyril Bonté62ba8702014-04-22 23:52:25 +020017527sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017528sc0_inc_gpc0([<table>]) : integer
17529sc1_inc_gpc0([<table>]) : integer
17530sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017531 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010017532 tracked counters, and returns its new value. Before the first invocation,
17533 the stored value is zero, so first invocation will increase it to 1 and will
17534 return 1. This is typically used as a second ACL in an expression in order
17535 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020017536
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017537 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017538 acl abuse sc0_http_req_rate gt 10
17539 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020017540 tcp-request connection reject if abuse kill
17541
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017542sc_inc_gpc1(<ctr>[,<table>]) : integer
17543sc0_inc_gpc1([<table>]) : integer
17544sc1_inc_gpc1([<table>]) : integer
17545sc2_inc_gpc1([<table>]) : integer
17546 Increments the second General Purpose Counter associated to the currently
17547 tracked counters, and returns its new value. Before the first invocation,
17548 the stored value is zero, so first invocation will increase it to 1 and will
17549 return 1. This is typically used as a second ACL in an expression in order
17550 to mark a connection when a first ACL was verified.
17551
Cyril Bonté62ba8702014-04-22 23:52:25 +020017552sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017553sc0_kbytes_in([<table>]) : integer
17554sc1_kbytes_in([<table>]) : integer
17555sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017556 Returns the total amount of client-to-server data from the currently tracked
17557 counters, measured in kilobytes. The test is currently performed on 32-bit
17558 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020017559
Cyril Bonté62ba8702014-04-22 23:52:25 +020017560sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017561sc0_kbytes_out([<table>]) : integer
17562sc1_kbytes_out([<table>]) : integer
17563sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017564 Returns the total amount of server-to-client data from the currently tracked
17565 counters, measured in kilobytes. The test is currently performed on 32-bit
17566 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020017567
Cyril Bonté62ba8702014-04-22 23:52:25 +020017568sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017569sc0_sess_cnt([<table>]) : integer
17570sc1_sess_cnt([<table>]) : integer
17571sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017572 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020017573 into sessions, which means that they were accepted by a "tcp-request
17574 connection" rule, from the currently tracked counters. A backend may count
17575 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017576 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020017577 with the client. See also src_sess_cnt.
17578
Cyril Bonté62ba8702014-04-22 23:52:25 +020017579sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017580sc0_sess_rate([<table>]) : integer
17581sc1_sess_rate([<table>]) : integer
17582sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017583 Returns the average session rate from the currently tracked counters,
17584 measured in amount of sessions over the period configured in the table. A
17585 session is a connection that got past the early "tcp-request connection"
17586 rules. A backend may count more sessions than connections because each
17587 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017588 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020017589
Cyril Bonté62ba8702014-04-22 23:52:25 +020017590sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020017591sc0_tracked([<table>]) : boolean
17592sc1_tracked([<table>]) : boolean
17593sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020017594 Returns true if the designated session counter is currently being tracked by
17595 the current session. This can be useful when deciding whether or not we want
17596 to set some values in a header passed to the server.
17597
Cyril Bonté62ba8702014-04-22 23:52:25 +020017598sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017599sc0_trackers([<table>]) : integer
17600sc1_trackers([<table>]) : integer
17601sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010017602 Returns the current amount of concurrent connections tracking the same
17603 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017604 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010017605 that it does not rely on any stored information but on the table's reference
17606 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020017607 may sometimes be more suited for layer7 tracking. It can be used to tell a
17608 server how many concurrent connections there are from a given address for
17609 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010017610
Willy Tarreau74ca5042013-06-11 23:12:07 +020017611so_id : integer
17612 Returns an integer containing the current listening socket's id. It is useful
17613 in frontends involving many "bind" lines, or to stick all users coming via a
17614 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017615
Jerome Magnineb421b22020-03-27 22:08:40 +010017616so_name : string
17617 Returns a string containing the current listening socket's name, as defined
17618 with name on a "bind" line. It can serve the same purposes as so_id but with
17619 strings instead of integers.
17620
Willy Tarreau74ca5042013-06-11 23:12:07 +020017621src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010017622 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020017623 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
17624 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
17625 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017626 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
17627 directive is used, it can be the address of a client behind another
17628 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010017629 "tcp-request connection" which sees the real address. When the incoming
17630 connection passed through address translation or redirection involving
17631 connection tracking, the original destination address before the redirection
17632 will be reported. On Linux systems, the source and destination may seldom
17633 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
17634 response may reopen a timed out connection and switch what is believed to be
17635 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017636
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010017637 Example:
17638 # add an HTTP header in requests with the originating address' country
17639 http-request set-header X-Country %[src,map_ip(geoip.lst)]
17640
Willy Tarreau74ca5042013-06-11 23:12:07 +020017641src_bytes_in_rate([<table>]) : integer
17642 Returns the average bytes rate from the incoming connection's source address
17643 in the current proxy's stick-table or in the designated stick-table, measured
17644 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017645 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017646
Willy Tarreau74ca5042013-06-11 23:12:07 +020017647src_bytes_out_rate([<table>]) : integer
17648 Returns the average bytes rate to the incoming connection's source address in
17649 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017650 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017651 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017652
Willy Tarreau74ca5042013-06-11 23:12:07 +020017653src_clr_gpc0([<table>]) : integer
17654 Clears the first General Purpose Counter associated to the incoming
17655 connection's source address in the current proxy's stick-table or in the
17656 designated stick-table, and returns its previous value. If the address is not
17657 found, an entry is created and 0 is returned. This is typically used as a
17658 second ACL in an expression in order to mark a connection when a first ACL
17659 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020017660
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017661 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020017662 # block if 5 consecutive requests continue to come faster than 10 sess
17663 # per second, and reset the counter as soon as the traffic slows down.
17664 acl abuse src_http_req_rate gt 10
17665 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010017666 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020017667 tcp-request connection accept if !abuse save
17668 tcp-request connection reject if abuse kill
17669
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017670src_clr_gpc1([<table>]) : integer
17671 Clears the second General Purpose Counter associated to the incoming
17672 connection's source address in the current proxy's stick-table or in the
17673 designated stick-table, and returns its previous value. If the address is not
17674 found, an entry is created and 0 is returned. This is typically used as a
17675 second ACL in an expression in order to mark a connection when a first ACL
17676 was verified.
17677
Willy Tarreau74ca5042013-06-11 23:12:07 +020017678src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017679 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020017680 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017681 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017682 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017683
Willy Tarreau74ca5042013-06-11 23:12:07 +020017684src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020017685 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017686 current incoming connection's source address in the current proxy's
17687 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017688 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017689
Willy Tarreau74ca5042013-06-11 23:12:07 +020017690src_conn_rate([<table>]) : integer
17691 Returns the average connection rate from the incoming connection's source
17692 address in the current proxy's stick-table or in the designated stick-table,
17693 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017694 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017695
Willy Tarreau74ca5042013-06-11 23:12:07 +020017696src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020017697 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017698 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017699 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017700 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017701
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017702src_get_gpc1([<table>]) : integer
17703 Returns the value of the second General Purpose Counter associated to the
17704 incoming connection's source address in the current proxy's stick-table or in
17705 the designated stick-table. If the address is not found, zero is returned.
17706 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
17707
Thierry FOURNIER236657b2015-08-19 08:25:14 +020017708src_get_gpt0([<table>]) : integer
17709 Returns the value of the first General Purpose Tag associated to the
17710 incoming connection's source address in the current proxy's stick-table or in
17711 the designated stick-table. If the address is not found, zero is returned.
17712 See also sc/sc0/sc1/sc2_get_gpt0.
17713
Willy Tarreau74ca5042013-06-11 23:12:07 +020017714src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017715 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020017716 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017717 stick-table or in the designated stick-table. It reports the frequency
17718 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017719 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
17720 that the "gpc0_rate" counter must be stored in the stick-table for a value to
17721 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017722
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017723src_gpc1_rate([<table>]) : integer
17724 Returns the average increment rate of the second General Purpose Counter
17725 associated to the incoming connection's source address in the current proxy's
17726 stick-table or in the designated stick-table. It reports the frequency
17727 which the gpc1 counter was incremented over the configured period. See also
17728 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
17729 that the "gpc1_rate" counter must be stored in the stick-table for a value to
17730 be returned, as "gpc1" only holds the event count.
17731
Willy Tarreau74ca5042013-06-11 23:12:07 +020017732src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017733 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017734 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020017735 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017736 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020017737 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017738
Willy Tarreau74ca5042013-06-11 23:12:07 +020017739src_http_err_rate([<table>]) : integer
17740 Returns the average rate of HTTP errors from the incoming connection's source
17741 address in the current proxy's stick-table or in the designated stick-table,
17742 measured in amount of errors over the period configured in the table. This
17743 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017744 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017745
Willy Tarreau826f3ab2021-02-10 12:07:15 +010017746src_http_fail_cnt([<table>]) : integer
17747 Returns the cumulative number of HTTP response failures triggered by the
17748 incoming connection's source address in the current proxy's stick-table or in
Ilya Shipitsin0de36ad2021-02-20 00:23:36 +050017749 the designated stick-table. This includes the both response errors and 5xx
Willy Tarreau826f3ab2021-02-10 12:07:15 +010017750 status codes other than 501 and 505. See also sc/sc0/sc1/sc2_http_fail_cnt.
17751 If the address is not found, zero is returned.
17752
17753src_http_fail_rate([<table>]) : integer
17754 Returns the average rate of HTTP response failures triggered by the incoming
17755 connection's source address in the current proxy's stick-table or in the
17756 designated stick-table, measured in amount of failures over the period
17757 configured in the table. This includes the both response errors and 5xx
17758 status codes other than 501 and 505. If the address is not found, zero is
17759 returned. See also sc/sc0/sc1/sc2_http_fail_rate.
17760
Willy Tarreau74ca5042013-06-11 23:12:07 +020017761src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017762 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017763 source address in the current proxy's stick-table or in the designated stick-
17764 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017765 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017766
Willy Tarreau74ca5042013-06-11 23:12:07 +020017767src_http_req_rate([<table>]) : integer
17768 Returns the average rate of HTTP requests from the incoming connection's
17769 source address in the current proxy's stick-table or in the designated stick-
17770 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020017771 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017772 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017773
Willy Tarreau74ca5042013-06-11 23:12:07 +020017774src_inc_gpc0([<table>]) : integer
17775 Increments the first General Purpose Counter associated to the incoming
17776 connection's source address in the current proxy's stick-table or in the
17777 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017778 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017779 This is typically used as a second ACL in an expression in order to mark a
17780 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020017781
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017782 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020017783 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010017784 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020017785 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020017786
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017787src_inc_gpc1([<table>]) : integer
17788 Increments the second General Purpose Counter associated to the incoming
17789 connection's source address in the current proxy's stick-table or in the
17790 designated stick-table, and returns its new value. If the address is not
17791 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
17792 This is typically used as a second ACL in an expression in order to mark a
17793 connection when a first ACL was verified.
17794
Willy Tarreau16e01562016-08-09 16:46:18 +020017795src_is_local : boolean
17796 Returns true if the source address of the incoming connection is local to the
17797 system, or false if the address doesn't exist on the system, meaning that it
17798 comes from a remote machine. Note that UNIX addresses are considered local.
17799 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017800 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020017801 note that the check involves a few system calls, so it's better to do it only
17802 once per connection.
17803
Willy Tarreau74ca5042013-06-11 23:12:07 +020017804src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017805 Returns the total amount of data received from the incoming connection's
17806 source address in the current proxy's stick-table or in the designated
17807 stick-table, measured in kilobytes. If the address is not found, zero is
17808 returned. The test is currently performed on 32-bit integers, which limits
17809 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017810
Willy Tarreau74ca5042013-06-11 23:12:07 +020017811src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017812 Returns the total amount of data sent to the incoming connection's source
17813 address in the current proxy's stick-table or in the designated stick-table,
17814 measured in kilobytes. If the address is not found, zero is returned. The
17815 test is currently performed on 32-bit integers, which limits values to 4
17816 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017817
Willy Tarreau74ca5042013-06-11 23:12:07 +020017818src_port : integer
17819 Returns an integer value corresponding to the TCP source port of the
17820 connection on the client side, which is the port the client connected from.
17821 Usage of this function is very limited as modern protocols do not care much
17822 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010017823
Willy Tarreau74ca5042013-06-11 23:12:07 +020017824src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017825 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020017826 connection's source IPv4 address in the current proxy's stick-table or in the
17827 designated stick-table, that were transformed into sessions, which means that
17828 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017829 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017830
Willy Tarreau74ca5042013-06-11 23:12:07 +020017831src_sess_rate([<table>]) : integer
17832 Returns the average session rate from the incoming connection's source
17833 address in the current proxy's stick-table or in the designated stick-table,
17834 measured in amount of sessions over the period configured in the table. A
17835 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017836 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017837
Willy Tarreau74ca5042013-06-11 23:12:07 +020017838src_updt_conn_cnt([<table>]) : integer
17839 Creates or updates the entry associated to the incoming connection's source
17840 address in the current proxy's stick-table or in the designated stick-table.
17841 This table must be configured to store the "conn_cnt" data type, otherwise
17842 the match will be ignored. The current count is incremented by one, and the
17843 expiration timer refreshed. The updated count is returned, so this match
17844 can't return zero. This was used to reject service abusers based on their
17845 source address. Note: it is recommended to use the more complete "track-sc*"
17846 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017847
17848 Example :
17849 # This frontend limits incoming SSH connections to 3 per 10 second for
17850 # each source address, and rejects excess connections until a 10 second
17851 # silence is observed. At most 20 addresses are tracked.
17852 listen ssh
17853 bind :22
17854 mode tcp
17855 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020017856 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020017857 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017858 server local 127.0.0.1:22
17859
Willy Tarreau74ca5042013-06-11 23:12:07 +020017860srv_id : integer
17861 Returns an integer containing the server's id when processing the response.
17862 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020017863 debugging. It can also be used in a tcp-check or an http-check ruleset.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020017864
vkill1dfd1652019-10-30 16:58:14 +080017865srv_name : string
17866 Returns a string containing the server's name when processing the response.
17867 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020017868 debugging. It can also be used in a tcp-check or an http-check ruleset.
vkill1dfd1652019-10-30 16:58:14 +080017869
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200178707.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020017871----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020017872
Willy Tarreau74ca5042013-06-11 23:12:07 +020017873The layer 5 usually describes just the session layer which in haproxy is
17874closest to the session once all the connection handshakes are finished, but
17875when no content is yet made available. The fetch methods described here are
17876usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017877future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020017878
Ben Shillitof25e8e52016-12-02 14:25:37 +00001787951d.all(<prop>[,<prop>*]) : string
17880 Returns values for the properties requested as a string, where values are
17881 separated by the delimiter specified with "51degrees-property-separator".
17882 The device is identified using all the important HTTP headers from the
17883 request. The function can be passed up to five property names, and if a
17884 property name can't be found, the value "NoData" is returned.
17885
17886 Example :
17887 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
17888 # containing the three properties requested using all relevant headers from
17889 # the request.
17890 frontend http-in
17891 bind *:8081
17892 default_backend servers
17893 http-request set-header X-51D-DeviceTypeMobileTablet \
17894 %[51d.all(DeviceType,IsMobile,IsTablet)]
17895
Emeric Brun645ae792014-04-30 14:21:06 +020017896ssl_bc : boolean
17897 Returns true when the back connection was made via an SSL/TLS transport
17898 layer and is locally deciphered. This means the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017899 other a server with the "ssl" option. It can be used in a tcp-check or an
17900 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017901
17902ssl_bc_alg_keysize : integer
17903 Returns the symmetric cipher key size supported in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017904 connection was made over an SSL/TLS transport layer. It can be used in a
17905 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017906
Olivier Houchard6b77f492018-11-22 18:18:29 +010017907ssl_bc_alpn : string
17908 This extracts the Application Layer Protocol Negotiation field from an
17909 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020017910 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010017911 server. The SSL library must have been built with support for TLS
17912 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
17913 not advertised unless the "alpn" keyword on the "server" line specifies a
17914 protocol list. Also, nothing forces the server to pick a protocol from this
17915 list, any other one may be requested. The TLS ALPN extension is meant to
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017916 replace the TLS NPN extension. See also "ssl_bc_npn". It can be used in a
17917 tcp-check or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010017918
Emeric Brun645ae792014-04-30 14:21:06 +020017919ssl_bc_cipher : string
17920 Returns the name of the used cipher when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017921 over an SSL/TLS transport layer. It can be used in a tcp-check or an
17922 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017923
Patrick Hemmer65674662019-06-04 08:13:03 -040017924ssl_bc_client_random : binary
17925 Returns the client random of the back connection when the incoming connection
17926 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17927 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017928 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040017929
Emeric Brun74f7ffa2018-02-19 16:14:12 +010017930ssl_bc_is_resumed : boolean
17931 Returns true when the back connection was made over an SSL/TLS transport
17932 layer and the newly created SSL session was resumed using a cached
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017933 session or a TLS ticket. It can be used in a tcp-check or an http-check
17934 ruleset.
Emeric Brun74f7ffa2018-02-19 16:14:12 +010017935
Olivier Houchard6b77f492018-11-22 18:18:29 +010017936ssl_bc_npn : string
17937 This extracts the Next Protocol Negotiation field from an outgoing connection
17938 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020017939 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010017940 built with support for TLS extensions enabled (check haproxy -vv). Note that
17941 the TLS NPN extension is not advertised unless the "npn" keyword on the
17942 "server" line specifies a protocol list. Also, nothing forces the server to
17943 pick a protocol from this list, any other one may be used. Please note that
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017944 the TLS NPN extension was replaced with ALPN. It can be used in a tcp-check
17945 or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010017946
Emeric Brun645ae792014-04-30 14:21:06 +020017947ssl_bc_protocol : string
17948 Returns the name of the used protocol when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017949 over an SSL/TLS transport layer. It can be used in a tcp-check or an
17950 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017951
Emeric Brunb73a9b02014-04-30 18:49:19 +020017952ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020017953 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020017954 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017955 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64". It
17956 can be used in a tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017957
Patrick Hemmer65674662019-06-04 08:13:03 -040017958ssl_bc_server_random : binary
17959 Returns the server random of the back connection when the incoming connection
17960 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17961 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017962 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040017963
Emeric Brun645ae792014-04-30 14:21:06 +020017964ssl_bc_session_id : binary
17965 Returns the SSL ID of the back connection when the outgoing connection was
17966 made over an SSL/TLS transport layer. It is useful to log if we want to know
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017967 if session was reused or not. It can be used in a tcp-check or an http-check
17968 ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017969
Patrick Hemmere0275472018-04-28 19:15:51 -040017970ssl_bc_session_key : binary
17971 Returns the SSL session master key of the back connection when the outgoing
17972 connection was made over an SSL/TLS transport layer. It is useful to decrypt
17973 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017974 BoringSSL. It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmere0275472018-04-28 19:15:51 -040017975
Emeric Brun645ae792014-04-30 14:21:06 +020017976ssl_bc_use_keysize : integer
17977 Returns the symmetric cipher key size used in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017978 connection was made over an SSL/TLS transport layer. It can be used in a
17979 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017980
Willy Tarreau74ca5042013-06-11 23:12:07 +020017981ssl_c_ca_err : integer
17982 When the incoming connection was made over an SSL/TLS transport layer,
17983 returns the ID of the first error detected during verification of the client
17984 certificate at depth > 0, or 0 if no error was encountered during this
17985 verification process. Please refer to your SSL library's documentation to
17986 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020017987
Willy Tarreau74ca5042013-06-11 23:12:07 +020017988ssl_c_ca_err_depth : integer
17989 When the incoming connection was made over an SSL/TLS transport layer,
17990 returns the depth in the CA chain of the first error detected during the
17991 verification of the client certificate. If no error is encountered, 0 is
17992 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010017993
Christopher Faulet70d10d12020-11-06 12:10:33 +010017994ssl_c_chain_der : binary
William Dauchya598b502020-08-06 18:11:38 +020017995 Returns the DER formatted chain certificate presented by the client when the
17996 incoming connection was made over an SSL/TLS transport layer. When used for
17997 an ACL, the value(s) to match against can be passed in hexadecimal form. One
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050017998 can parse the result with any lib accepting ASN.1 DER data. It currently
William Dauchya598b502020-08-06 18:11:38 +020017999 does not support resumed sessions.
18000
Christopher Faulet70d10d12020-11-06 12:10:33 +010018001ssl_c_der : binary
18002 Returns the DER formatted certificate presented by the client when the
18003 incoming connection was made over an SSL/TLS transport layer. When used for
18004 an ACL, the value(s) to match against can be passed in hexadecimal form.
18005
Willy Tarreau74ca5042013-06-11 23:12:07 +020018006ssl_c_err : integer
18007 When the incoming connection was made over an SSL/TLS transport layer,
18008 returns the ID of the first error detected during verification at depth 0, or
18009 0 if no error was encountered during this verification process. Please refer
18010 to your SSL library's documentation to find the exhaustive list of error
18011 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020018012
Elliot Otchet71f82972020-01-15 08:12:14 -050018013ssl_c_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018014 When the incoming connection was made over an SSL/TLS transport layer,
18015 returns the full distinguished name of the issuer of the certificate
18016 presented by the client when no <entry> is specified, or the value of the
18017 first given entry found from the beginning of the DN. If a positive/negative
18018 occurrence number is specified as the optional second argument, it returns
18019 the value of the nth given entry value from the beginning/end of the DN.
18020 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
18021 "ssl_c_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018022 The <format> parameter allows you to receive the DN suitable for
18023 consumption by different protocols. Currently supported is rfc2253 for
18024 LDAP v3.
18025 If you'd like to modify the format only you can specify an empty string
18026 and zero for the first two parameters. Example: ssl_c_i_dn(,0,rfc2253)
Willy Tarreau62644772008-07-16 18:36:06 +020018027
Willy Tarreau74ca5042013-06-11 23:12:07 +020018028ssl_c_key_alg : string
18029 Returns the name of the algorithm used to generate the key of the certificate
18030 presented by the client when the incoming connection was made over an SSL/TLS
18031 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020018032
Willy Tarreau74ca5042013-06-11 23:12:07 +020018033ssl_c_notafter : string
18034 Returns the end date presented by the client as a formatted string
18035 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18036 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020018037
Willy Tarreau74ca5042013-06-11 23:12:07 +020018038ssl_c_notbefore : string
18039 Returns the start date presented by the client as a formatted string
18040 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18041 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010018042
Elliot Otchet71f82972020-01-15 08:12:14 -050018043ssl_c_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018044 When the incoming connection was made over an SSL/TLS transport layer,
18045 returns the full distinguished name of the subject of the certificate
18046 presented by the client when no <entry> is specified, or the value of the
18047 first given entry found from the beginning of the DN. If a positive/negative
18048 occurrence number is specified as the optional second argument, it returns
18049 the value of the nth given entry value from the beginning/end of the DN.
18050 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
18051 "ssl_c_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018052 The <format> parameter allows you to receive the DN suitable for
18053 consumption by different protocols. Currently supported is rfc2253 for
18054 LDAP v3.
18055 If you'd like to modify the format only you can specify an empty string
18056 and zero for the first two parameters. Example: ssl_c_s_dn(,0,rfc2253)
Willy Tarreaub6672b52011-12-12 17:23:41 +010018057
Willy Tarreau74ca5042013-06-11 23:12:07 +020018058ssl_c_serial : binary
18059 Returns the serial of the certificate presented by the client when the
18060 incoming connection was made over an SSL/TLS transport layer. When used for
18061 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020018062
Willy Tarreau74ca5042013-06-11 23:12:07 +020018063ssl_c_sha1 : binary
18064 Returns the SHA-1 fingerprint of the certificate presented by the client when
18065 the incoming connection was made over an SSL/TLS transport layer. This can be
18066 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020018067 Note that the output is binary, so if you want to pass that signature to the
18068 server, you need to encode it in hex or base64, such as in the example below:
18069
Jarno Huuskonen676f6222017-03-30 09:19:45 +030018070 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020018071 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020018072
Willy Tarreau74ca5042013-06-11 23:12:07 +020018073ssl_c_sig_alg : string
18074 Returns the name of the algorithm used to sign the certificate presented by
18075 the client when the incoming connection was made over an SSL/TLS transport
18076 layer.
Emeric Brun87855892012-10-17 17:39:35 +020018077
Willy Tarreau74ca5042013-06-11 23:12:07 +020018078ssl_c_used : boolean
18079 Returns true if current SSL session uses a client certificate even if current
18080 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020018081
Willy Tarreau74ca5042013-06-11 23:12:07 +020018082ssl_c_verify : integer
18083 Returns the verify result error ID when the incoming connection was made over
18084 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
18085 refer to your SSL library's documentation for an exhaustive list of error
18086 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020018087
Willy Tarreau74ca5042013-06-11 23:12:07 +020018088ssl_c_version : integer
18089 Returns the version of the certificate presented by the client when the
18090 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020018091
Emeric Brun43e79582014-10-29 19:03:26 +010018092ssl_f_der : binary
18093 Returns the DER formatted certificate presented by the frontend when the
18094 incoming connection was made over an SSL/TLS transport layer. When used for
18095 an ACL, the value(s) to match against can be passed in hexadecimal form.
18096
Elliot Otchet71f82972020-01-15 08:12:14 -050018097ssl_f_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018098 When the incoming connection was made over an SSL/TLS transport layer,
18099 returns the full distinguished name of the issuer of the certificate
18100 presented by the frontend when no <entry> is specified, or the value of the
18101 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020018102 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020018103 the value of the nth given entry value from the beginning/end of the DN.
18104 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
18105 "ssl_f_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018106 The <format> parameter allows you to receive the DN suitable for
18107 consumption by different protocols. Currently supported is rfc2253 for
18108 LDAP v3.
18109 If you'd like to modify the format only you can specify an empty string
18110 and zero for the first two parameters. Example: ssl_f_i_dn(,0,rfc2253)
Emeric Brun87855892012-10-17 17:39:35 +020018111
Willy Tarreau74ca5042013-06-11 23:12:07 +020018112ssl_f_key_alg : string
18113 Returns the name of the algorithm used to generate the key of the certificate
18114 presented by the frontend when the incoming connection was made over an
18115 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020018116
Willy Tarreau74ca5042013-06-11 23:12:07 +020018117ssl_f_notafter : string
18118 Returns the end date presented by the frontend as a formatted string
18119 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18120 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020018121
Willy Tarreau74ca5042013-06-11 23:12:07 +020018122ssl_f_notbefore : string
18123 Returns the start date presented by the frontend as a formatted string
18124 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
18125 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020018126
Elliot Otchet71f82972020-01-15 08:12:14 -050018127ssl_f_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018128 When the incoming connection was made over an SSL/TLS transport layer,
18129 returns the full distinguished name of the subject of the certificate
18130 presented by the frontend when no <entry> is specified, or the value of the
18131 first given entry found from the beginning of the DN. If a positive/negative
18132 occurrence number is specified as the optional second argument, it returns
18133 the value of the nth given entry value from the beginning/end of the DN.
18134 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
18135 "ssl_f_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050018136 The <format> parameter allows you to receive the DN suitable for
18137 consumption by different protocols. Currently supported is rfc2253 for
18138 LDAP v3.
18139 If you'd like to modify the format only you can specify an empty string
18140 and zero for the first two parameters. Example: ssl_f_s_dn(,0,rfc2253)
Emeric Brunce5ad802012-10-22 14:11:22 +020018141
Willy Tarreau74ca5042013-06-11 23:12:07 +020018142ssl_f_serial : binary
18143 Returns the serial of the certificate presented by the frontend when the
18144 incoming connection was made over an SSL/TLS transport layer. When used for
18145 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020018146
Emeric Brun55f4fa82014-04-30 17:11:25 +020018147ssl_f_sha1 : binary
18148 Returns the SHA-1 fingerprint of the certificate presented by the frontend
18149 when the incoming connection was made over an SSL/TLS transport layer. This
18150 can be used to know which certificate was chosen using SNI.
18151
Willy Tarreau74ca5042013-06-11 23:12:07 +020018152ssl_f_sig_alg : string
18153 Returns the name of the algorithm used to sign the certificate presented by
18154 the frontend when the incoming connection was made over an SSL/TLS transport
18155 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020018156
Willy Tarreau74ca5042013-06-11 23:12:07 +020018157ssl_f_version : integer
18158 Returns the version of the certificate presented by the frontend when the
18159 incoming connection was made over an SSL/TLS transport layer.
18160
18161ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020018162 Returns true when the front connection was made via an SSL/TLS transport
18163 layer and is locally deciphered. This means it has matched a socket declared
18164 with a "bind" line having the "ssl" option.
18165
Willy Tarreau74ca5042013-06-11 23:12:07 +020018166 Example :
18167 # This passes "X-Proto: https" to servers when client connects over SSL
18168 listen http-https
18169 bind :80
18170 bind :443 ssl crt /etc/haproxy.pem
18171 http-request add-header X-Proto https if { ssl_fc }
18172
18173ssl_fc_alg_keysize : integer
18174 Returns the symmetric cipher key size supported in bits when the incoming
18175 connection was made over an SSL/TLS transport layer.
18176
18177ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018178 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020018179 incoming connection made via a TLS transport layer and locally deciphered by
18180 haproxy. The result is a string containing the protocol name advertised by
18181 the client. The SSL library must have been built with support for TLS
18182 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
18183 not advertised unless the "alpn" keyword on the "bind" line specifies a
18184 protocol list. Also, nothing forces the client to pick a protocol from this
18185 list, any other one may be requested. The TLS ALPN extension is meant to
18186 replace the TLS NPN extension. See also "ssl_fc_npn".
18187
Willy Tarreau74ca5042013-06-11 23:12:07 +020018188ssl_fc_cipher : string
18189 Returns the name of the used cipher when the incoming connection was made
18190 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020018191
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018192ssl_fc_cipherlist_bin : binary
18193 Returns the binary form of the client hello cipher list. The maximum returned
18194 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010018195 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018196
18197ssl_fc_cipherlist_hex : string
18198 Returns the binary form of the client hello cipher list encoded as
18199 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010018200 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018201
18202ssl_fc_cipherlist_str : string
18203 Returns the decoded text form of the client hello cipher list. The maximum
18204 number of ciphers returned is according with the value of
18205 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010018206 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020018207 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018208
18209ssl_fc_cipherlist_xxh : integer
18210 Returns a xxh64 of the cipher list. This hash can be return only is the value
18211 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010018212 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010018213
Patrick Hemmer65674662019-06-04 08:13:03 -040018214ssl_fc_client_random : binary
18215 Returns the client random of the front connection when the incoming connection
18216 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
18217 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
18218
William Lallemand7d42ef52020-07-06 11:41:30 +020018219ssl_fc_client_early_traffic_secret : string
18220 Return the CLIENT_EARLY_TRAFFIC_SECRET as an hexadecimal string for the
18221 front connection when the incoming connection was made over a TLS 1.3
18222 transport layer.
18223 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18224 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18225 activated with "tune.ssl.keylog on" in the global section. See also
18226 "tune.ssl.keylog"
18227
18228ssl_fc_client_handshake_traffic_secret : string
18229 Return the CLIENT_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
18230 front connection when the incoming connection was made over a TLS 1.3
18231 transport layer.
18232 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18233 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18234 activated with "tune.ssl.keylog on" in the global section. See also
18235 "tune.ssl.keylog"
18236
18237ssl_fc_client_traffic_secret_0 : string
18238 Return the CLIENT_TRAFFIC_SECRET_0 as an hexadecimal string for the
18239 front connection when the incoming connection was made over a TLS 1.3
18240 transport layer.
18241 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18242 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18243 activated with "tune.ssl.keylog on" in the global section. See also
18244 "tune.ssl.keylog"
18245
18246ssl_fc_exporter_secret : string
18247 Return the EXPORTER_SECRET as an hexadecimal string for the
18248 front connection when the incoming connection was made over a TLS 1.3
18249 transport layer.
18250 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18251 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18252 activated with "tune.ssl.keylog on" in the global section. See also
18253 "tune.ssl.keylog"
18254
18255ssl_fc_early_exporter_secret : string
18256 Return the EARLY_EXPORTER_SECRET as an hexadecimal string for the
18257 front connection when the incoming connection was made over an TLS 1.3
18258 transport layer.
18259 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18260 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18261 activated with "tune.ssl.keylog on" in the global section. See also
18262 "tune.ssl.keylog"
18263
Willy Tarreau74ca5042013-06-11 23:12:07 +020018264ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020018265 Returns true if a client certificate is present in an incoming connection over
18266 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010018267 Note: on SSL session resumption with Session ID or TLS ticket, client
18268 certificate is not present in the current connection but may be retrieved
18269 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
18270 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020018271
Olivier Houchardccaa7de2017-10-02 11:51:03 +020018272ssl_fc_has_early : boolean
18273 Returns true if early data were sent, and the handshake didn't happen yet. As
18274 it has security implications, it is useful to be able to refuse those, or
18275 wait until the handshake happened.
18276
Willy Tarreau74ca5042013-06-11 23:12:07 +020018277ssl_fc_has_sni : boolean
18278 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020018279 in an incoming connection was made over an SSL/TLS transport layer. Returns
18280 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050018281 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020018282 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020018283
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020018284ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020018285 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010018286 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
18287 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020018288
Willy Tarreau74ca5042013-06-11 23:12:07 +020018289ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018290 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020018291 made via a TLS transport layer and locally deciphered by haproxy. The result
18292 is a string containing the protocol name advertised by the client. The SSL
18293 library must have been built with support for TLS extensions enabled (check
18294 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
18295 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
18296 forces the client to pick a protocol from this list, any other one may be
18297 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020018298
Willy Tarreau74ca5042013-06-11 23:12:07 +020018299ssl_fc_protocol : string
18300 Returns the name of the used protocol when the incoming connection was made
18301 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020018302
Emeric Brunb73a9b02014-04-30 18:49:19 +020018303ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040018304 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020018305 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
18306 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040018307
William Lallemand7d42ef52020-07-06 11:41:30 +020018308ssl_fc_server_handshake_traffic_secret : string
18309 Return the SERVER_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
18310 front connection when the incoming connection was made over a TLS 1.3
18311 transport layer.
18312 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18313 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18314 activated with "tune.ssl.keylog on" in the global section. See also
18315 "tune.ssl.keylog"
18316
18317ssl_fc_server_traffic_secret_0 : string
18318 Return the SERVER_TRAFFIC_SECRET_0 as an hexadecimal string for the
18319 front connection when the incoming connection was made over an TLS 1.3
18320 transport layer.
18321 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
18322 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
18323 activated with "tune.ssl.keylog on" in the global section. See also
18324 "tune.ssl.keylog"
18325
Patrick Hemmer65674662019-06-04 08:13:03 -040018326ssl_fc_server_random : binary
18327 Returns the server random of the front connection when the incoming connection
18328 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
18329 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
18330
Willy Tarreau74ca5042013-06-11 23:12:07 +020018331ssl_fc_session_id : binary
18332 Returns the SSL ID of the front connection when the incoming connection was
18333 made over an SSL/TLS transport layer. It is useful to stick a given client to
18334 a server. It is important to note that some browsers refresh their session ID
18335 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020018336
Patrick Hemmere0275472018-04-28 19:15:51 -040018337ssl_fc_session_key : binary
18338 Returns the SSL session master key of the front connection when the incoming
18339 connection was made over an SSL/TLS transport layer. It is useful to decrypt
18340 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
18341 BoringSSL.
18342
18343
Willy Tarreau74ca5042013-06-11 23:12:07 +020018344ssl_fc_sni : string
18345 This extracts the Server Name Indication TLS extension (SNI) field from an
18346 incoming connection made via an SSL/TLS transport layer and locally
18347 deciphered by haproxy. The result (when present) typically is a string
18348 matching the HTTPS host name (253 chars or less). The SSL library must have
18349 been built with support for TLS extensions enabled (check haproxy -vv).
18350
18351 This fetch is different from "req_ssl_sni" above in that it applies to the
18352 connection being deciphered by haproxy and not to SSL contents being blindly
18353 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050018354 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020018355 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020018356
Willy Tarreau74ca5042013-06-11 23:12:07 +020018357 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020018358 ssl_fc_sni_end : suffix match
18359 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020018360
Willy Tarreau74ca5042013-06-11 23:12:07 +020018361ssl_fc_use_keysize : integer
18362 Returns the symmetric cipher key size used in bits when the incoming
18363 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018364
William Lallemandbfa3e812020-06-25 20:07:18 +020018365ssl_s_der : binary
18366 Returns the DER formatted certificate presented by the server when the
18367 outgoing connection was made over an SSL/TLS transport layer. When used for
18368 an ACL, the value(s) to match against can be passed in hexadecimal form.
18369
William Dauchya598b502020-08-06 18:11:38 +020018370ssl_s_chain_der : binary
18371 Returns the DER formatted chain certificate presented by the server when the
18372 outgoing connection was made over an SSL/TLS transport layer. When used for
18373 an ACL, the value(s) to match against can be passed in hexadecimal form. One
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050018374 can parse the result with any lib accepting ASN.1 DER data. It currently
William Dauchya598b502020-08-06 18:11:38 +020018375 does not support resumed sessions.
18376
William Lallemandbfa3e812020-06-25 20:07:18 +020018377ssl_s_key_alg : string
18378 Returns the name of the algorithm used to generate the key of the certificate
18379 presented by the server when the outgoing connection was made over an
18380 SSL/TLS transport layer.
18381
18382ssl_s_notafter : string
18383 Returns the end date presented by the server as a formatted string
18384 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
18385 transport layer.
18386
18387ssl_s_notbefore : string
18388 Returns the start date presented by the server as a formatted string
18389 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
18390 transport layer.
18391
18392ssl_s_i_dn([<entry>[,<occ>[,<format>]]]) : string
18393 When the outgoing connection was made over an SSL/TLS transport layer,
18394 returns the full distinguished name of the issuer of the certificate
18395 presented by the server when no <entry> is specified, or the value of the
18396 first given entry found from the beginning of the DN. If a positive/negative
18397 occurrence number is specified as the optional second argument, it returns
18398 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020018399 For instance, "ssl_s_i_dn(OU,2)" the second organization unit, and
18400 "ssl_s_i_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020018401 The <format> parameter allows you to receive the DN suitable for
18402 consumption by different protocols. Currently supported is rfc2253 for
18403 LDAP v3.
18404 If you'd like to modify the format only you can specify an empty string
18405 and zero for the first two parameters. Example: ssl_s_i_dn(,0,rfc2253)
18406
18407ssl_s_s_dn([<entry>[,<occ>[,<format>]]]) : string
18408 When the outgoing connection was made over an SSL/TLS transport layer,
18409 returns the full distinguished name of the subject of the certificate
18410 presented by the server when no <entry> is specified, or the value of the
18411 first given entry found from the beginning of the DN. If a positive/negative
18412 occurrence number is specified as the optional second argument, it returns
18413 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020018414 For instance, "ssl_s_s_dn(OU,2)" the second organization unit, and
18415 "ssl_s_s_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020018416 The <format> parameter allows you to receive the DN suitable for
18417 consumption by different protocols. Currently supported is rfc2253 for
18418 LDAP v3.
18419 If you'd like to modify the format only you can specify an empty string
18420 and zero for the first two parameters. Example: ssl_s_s_dn(,0,rfc2253)
18421
18422ssl_s_serial : binary
18423 Returns the serial of the certificate presented by the server when the
18424 outgoing connection was made over an SSL/TLS transport layer. When used for
18425 an ACL, the value(s) to match against can be passed in hexadecimal form.
18426
18427ssl_s_sha1 : binary
18428 Returns the SHA-1 fingerprint of the certificate presented by the server
18429 when the outgoing connection was made over an SSL/TLS transport layer. This
18430 can be used to know which certificate was chosen using SNI.
18431
18432ssl_s_sig_alg : string
18433 Returns the name of the algorithm used to sign the certificate presented by
18434 the server when the outgoing connection was made over an SSL/TLS transport
18435 layer.
18436
18437ssl_s_version : integer
18438 Returns the version of the certificate presented by the server when the
18439 outgoing connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018440
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200184417.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020018442------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018443
Willy Tarreau74ca5042013-06-11 23:12:07 +020018444Fetching samples from buffer contents is a bit different from the previous
18445sample fetches above because the sampled data are ephemeral. These data can
18446only be used when they're available and will be lost when they're forwarded.
18447For this reason, samples fetched from buffer contents during a request cannot
18448be used in a response for example. Even while the data are being fetched, they
18449can change. Sometimes it is necessary to set some delays or combine multiple
18450sample fetch methods to ensure that the expected data are complete and usable,
18451for example through TCP request content inspection. Please see the "tcp-request
18452content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020018453
Christopher Fauleta434a002021-03-25 11:58:51 +010018454Warning : Following sample fetches are ignored if used from HTTP proxies. They
18455 only deal with raw contents found in the buffers. On their side,
18456 HTTTP proxies use structured content. Thus raw representation of
18457 these data are meaningless. A warning is emitted if an ACL relies on
18458 one of the following sample fetches. But it is not possible to detect
18459 all invalid usage (for instance inside a log-format string or a
18460 sample expression). So be careful.
18461
Willy Tarreau74ca5042013-06-11 23:12:07 +020018462payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010018463 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018464 "stick on", "stick match"), and for "res.payload" when used in the context of
18465 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010018466
Willy Tarreau74ca5042013-06-11 23:12:07 +020018467payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
18468 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010018469 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020018470 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010018471
Willy Tarreau74ca5042013-06-11 23:12:07 +020018472req.len : integer
18473req_len : integer (deprecated)
18474 Returns an integer value corresponding to the number of bytes present in the
18475 request buffer. This is mostly used in ACL. It is important to understand
18476 that this test does not return false as long as the buffer is changing. This
18477 means that a check with equality to zero will almost always immediately match
18478 at the beginning of the session, while a test for more data will wait for
18479 that data to come in and return false only when haproxy is certain that no
18480 more data will come in. This test was designed to be used with TCP request
18481 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018482
Willy Tarreau74ca5042013-06-11 23:12:07 +020018483req.payload(<offset>,<length>) : binary
18484 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020018485 in the request buffer. As a special case, if the <length> argument is zero,
18486 the the whole buffer from <offset> to the end is extracted. This can be used
18487 with ACLs in order to check for the presence of some content in a buffer at
18488 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018489
Willy Tarreau74ca5042013-06-11 23:12:07 +020018490 ACL alternatives :
18491 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018492
Willy Tarreau74ca5042013-06-11 23:12:07 +020018493req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
18494 This extracts a binary block whose size is specified at <offset1> for <length>
18495 bytes, and which starts at <offset2> if specified or just after the length in
18496 the request buffer. The <offset2> parameter also supports relative offsets if
18497 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018498
Willy Tarreau74ca5042013-06-11 23:12:07 +020018499 ACL alternatives :
18500 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018501
Willy Tarreau74ca5042013-06-11 23:12:07 +020018502 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018503
Willy Tarreau74ca5042013-06-11 23:12:07 +020018504req.proto_http : boolean
18505req_proto_http : boolean (deprecated)
18506 Returns true when data in the request buffer look like HTTP and correctly
18507 parses as such. It is the same parser as the common HTTP request parser which
18508 is used so there should be no surprises. The test does not match until the
18509 request is complete, failed or timed out. This test may be used to report the
18510 protocol in TCP logs, but the biggest use is to block TCP request analysis
18511 until a complete HTTP request is present in the buffer, for example to track
18512 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018513
Willy Tarreau74ca5042013-06-11 23:12:07 +020018514 Example:
18515 # track request counts per "base" (concatenation of Host+URL)
18516 tcp-request inspect-delay 10s
18517 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020018518 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018519
Willy Tarreau74ca5042013-06-11 23:12:07 +020018520req.rdp_cookie([<name>]) : string
18521rdp_cookie([<name>]) : string (deprecated)
18522 When the request buffer looks like the RDP protocol, extracts the RDP cookie
18523 <name>, or any cookie if unspecified. The parser only checks for the first
18524 cookie, as illustrated in the RDP protocol specification. The cookie name is
18525 case insensitive. Generally the "MSTS" cookie name will be used, as it can
18526 contain the user name of the client connecting to the server if properly
18527 configured on the client. The "MSTSHASH" cookie is often used as well for
18528 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018529
Willy Tarreau74ca5042013-06-11 23:12:07 +020018530 This differs from "balance rdp-cookie" in that any balancing algorithm may be
18531 used and thus the distribution of clients to backend servers is not linked to
18532 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
18533 such as "balance roundrobin" or "balance leastconn" will lead to a more even
18534 distribution of clients to backend servers than the hash used by "balance
18535 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018536
Willy Tarreau74ca5042013-06-11 23:12:07 +020018537 ACL derivatives :
18538 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018539
Willy Tarreau74ca5042013-06-11 23:12:07 +020018540 Example :
18541 listen tse-farm
18542 bind 0.0.0.0:3389
18543 # wait up to 5s for an RDP cookie in the request
18544 tcp-request inspect-delay 5s
18545 tcp-request content accept if RDP_COOKIE
18546 # apply RDP cookie persistence
18547 persist rdp-cookie
18548 # Persist based on the mstshash cookie
18549 # This is only useful makes sense if
18550 # balance rdp-cookie is not used
18551 stick-table type string size 204800
18552 stick on req.rdp_cookie(mstshash)
18553 server srv1 1.1.1.1:3389
18554 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018555
Willy Tarreau74ca5042013-06-11 23:12:07 +020018556 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
18557 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018558
Willy Tarreau74ca5042013-06-11 23:12:07 +020018559req.rdp_cookie_cnt([name]) : integer
18560rdp_cookie_cnt([name]) : integer (deprecated)
18561 Tries to parse the request buffer as RDP protocol, then returns an integer
18562 corresponding to the number of RDP cookies found. If an optional cookie name
18563 is passed, only cookies matching this name are considered. This is mostly
18564 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018565
Willy Tarreau74ca5042013-06-11 23:12:07 +020018566 ACL derivatives :
18567 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018568
Alex Zorin4afdd132018-12-30 13:56:28 +110018569req.ssl_alpn : string
18570 Returns a string containing the values of the Application-Layer Protocol
18571 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
18572 ClientHello message. Note that this only applies to raw contents found in the
18573 request buffer and not to the contents deciphered via an SSL data layer, so
18574 this will not work with "bind" lines having the "ssl" option. This is useful
18575 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020018576 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110018577
18578 Examples :
18579 # Wait for a client hello for at most 5 seconds
18580 tcp-request inspect-delay 5s
18581 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020018582 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110018583 default_backend bk_default
18584
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020018585req.ssl_ec_ext : boolean
18586 Returns a boolean identifying if client sent the Supported Elliptic Curves
18587 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020018588 message. This can be used to present ECC compatible clients with EC
18589 certificate and to use RSA for all others, on the same IP address. Note that
18590 this only applies to raw contents found in the request buffer and not to
18591 contents deciphered via an SSL data layer, so this will not work with "bind"
18592 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020018593
Willy Tarreau74ca5042013-06-11 23:12:07 +020018594req.ssl_hello_type : integer
18595req_ssl_hello_type : integer (deprecated)
18596 Returns an integer value containing the type of the SSL hello message found
18597 in the request buffer if the buffer contains data that parse as a complete
18598 SSL (v3 or superior) client hello message. Note that this only applies to raw
18599 contents found in the request buffer and not to contents deciphered via an
18600 SSL data layer, so this will not work with "bind" lines having the "ssl"
18601 option. This is mostly used in ACL to detect presence of an SSL hello message
18602 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018603
Willy Tarreau74ca5042013-06-11 23:12:07 +020018604req.ssl_sni : string
18605req_ssl_sni : string (deprecated)
18606 Returns a string containing the value of the Server Name TLS extension sent
18607 by a client in a TLS stream passing through the request buffer if the buffer
18608 contains data that parse as a complete SSL (v3 or superior) client hello
18609 message. Note that this only applies to raw contents found in the request
18610 buffer and not to contents deciphered via an SSL data layer, so this will not
Lukas Tribusa267b5d2020-07-19 00:25:06 +020018611 work with "bind" lines having the "ssl" option. This will only work for actual
18612 implicit TLS based protocols like HTTPS (443), IMAPS (993), SMTPS (465),
18613 however it will not work for explicit TLS based protocols, like SMTP (25/587)
18614 or IMAP (143). SNI normally contains the name of the host the client tries to
18615 connect to (for recent browsers). SNI is useful for allowing or denying access
18616 to certain hosts when SSL/TLS is used by the client. This test was designed to
18617 be used with TCP request content inspection. If content switching is needed,
18618 it is recommended to first wait for a complete client hello (type 1), like in
18619 the example below. See also "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018620
Willy Tarreau74ca5042013-06-11 23:12:07 +020018621 ACL derivatives :
18622 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018623
Willy Tarreau74ca5042013-06-11 23:12:07 +020018624 Examples :
18625 # Wait for a client hello for at most 5 seconds
18626 tcp-request inspect-delay 5s
18627 tcp-request content accept if { req_ssl_hello_type 1 }
18628 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
18629 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018630
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053018631req.ssl_st_ext : integer
18632 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
18633 Returns 1 if the client sent SessionTicket TLS Extension
18634 Returns 2 if the client also sent non-zero length TLS SessionTicket
18635 Note that this only applies to raw contents found in the request buffer and
18636 not to contents deciphered via an SSL data layer, so this will not work with
18637 "bind" lines having the "ssl" option. This can for example be used to detect
18638 whether the client sent a SessionTicket or not and stick it accordingly, if
18639 no SessionTicket then stick on SessionID or don't stick as there's no server
18640 side state is there when SessionTickets are in use.
18641
Willy Tarreau74ca5042013-06-11 23:12:07 +020018642req.ssl_ver : integer
18643req_ssl_ver : integer (deprecated)
18644 Returns an integer value containing the version of the SSL/TLS protocol of a
18645 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
18646 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
18647 composed of the major version multiplied by 65536, added to the minor
18648 version. Note that this only applies to raw contents found in the request
18649 buffer and not to contents deciphered via an SSL data layer, so this will not
18650 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010018651 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020018652 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018653
Willy Tarreau74ca5042013-06-11 23:12:07 +020018654 ACL derivatives :
18655 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010018656
Willy Tarreau47e8eba2013-09-11 23:28:46 +020018657res.len : integer
18658 Returns an integer value corresponding to the number of bytes present in the
18659 response buffer. This is mostly used in ACL. It is important to understand
18660 that this test does not return false as long as the buffer is changing. This
18661 means that a check with equality to zero will almost always immediately match
18662 at the beginning of the session, while a test for more data will wait for
18663 that data to come in and return false only when haproxy is certain that no
18664 more data will come in. This test was designed to be used with TCP response
Christopher Faulete596d182020-05-05 17:46:34 +020018665 content inspection. But it may also be used in tcp-check based expect rules.
Willy Tarreau47e8eba2013-09-11 23:28:46 +020018666
Willy Tarreau74ca5042013-06-11 23:12:07 +020018667res.payload(<offset>,<length>) : binary
18668 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020018669 in the response buffer. As a special case, if the <length> argument is zero,
Christopher Faulete596d182020-05-05 17:46:34 +020018670 the whole buffer from <offset> to the end is extracted. This can be used
Willy Tarreau00f00842013-08-02 11:07:32 +020018671 with ACLs in order to check for the presence of some content in a buffer at
Christopher Faulete596d182020-05-05 17:46:34 +020018672 any location. It may also be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018673
Willy Tarreau74ca5042013-06-11 23:12:07 +020018674res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
18675 This extracts a binary block whose size is specified at <offset1> for <length>
18676 bytes, and which starts at <offset2> if specified or just after the length in
18677 the response buffer. The <offset2> parameter also supports relative offsets
Christopher Faulete596d182020-05-05 17:46:34 +020018678 if prepended with a '+' or '-' sign. It may also be used in tcp-check based
18679 expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018680
Willy Tarreau74ca5042013-06-11 23:12:07 +020018681 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018682
Willy Tarreau971f7b62015-09-29 14:06:59 +020018683res.ssl_hello_type : integer
18684rep_ssl_hello_type : integer (deprecated)
18685 Returns an integer value containing the type of the SSL hello message found
18686 in the response buffer if the buffer contains data that parses as a complete
18687 SSL (v3 or superior) hello message. Note that this only applies to raw
18688 contents found in the response buffer and not to contents deciphered via an
18689 SSL data layer, so this will not work with "server" lines having the "ssl"
18690 option. This is mostly used in ACL to detect presence of an SSL hello message
18691 that is supposed to contain an SSL session ID usable for stickiness.
18692
Willy Tarreau74ca5042013-06-11 23:12:07 +020018693wait_end : boolean
18694 This fetch either returns true when the inspection period is over, or does
18695 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018696 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020018697 actions, such as a delayed reject for some special addresses. Since it either
18698 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018699 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020018700 "WAIT_END" is always usable without prior declaration. This test was designed
18701 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018702
Willy Tarreau74ca5042013-06-11 23:12:07 +020018703 Examples :
18704 # delay every incoming request by 2 seconds
18705 tcp-request inspect-delay 2s
18706 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010018707
Willy Tarreau74ca5042013-06-11 23:12:07 +020018708 # don't immediately tell bad guys they are rejected
18709 tcp-request inspect-delay 10s
18710 acl goodguys src 10.0.0.0/24
18711 acl badguys src 10.0.1.0/24
18712 tcp-request content accept if goodguys
18713 tcp-request content reject if badguys WAIT_END
18714 tcp-request content reject
18715
18716
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200187177.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020018718--------------------------------------
18719
18720It is possible to fetch samples from HTTP contents, requests and responses.
18721This application layer is also called layer 7. It is only possible to fetch the
18722data in this section when a full HTTP request or response has been parsed from
18723its respective request or response buffer. This is always the case with all
18724HTTP specific rules and for sections running with "mode http". When using TCP
18725content inspection, it may be necessary to support an inspection delay in order
18726to let the request or response come in first. These fetches may require a bit
18727more CPU resources than the layer 4 ones, but not much since the request and
18728response are indexed.
18729
Christopher Faulet4d37e532021-03-26 14:44:00 +010018730Note : Regarding HTTP processing from the tcp-request content rules, everything
18731 will work as expected from an HTTP proxy. However, from a TCP proxy,
18732 without an HTTP upgrade, it will only work for HTTP/1 content. For
18733 HTTP/2 content, only the preface is visible. Thus, it is only possible
18734 to rely to "req.proto_http", "req.ver" and eventually "method" sample
18735 fetches. All other L7 sample fetches will fail. After an HTTP upgrade,
18736 they will work in the same manner than from an HTTP proxy.
18737
Willy Tarreau74ca5042013-06-11 23:12:07 +020018738base : string
18739 This returns the concatenation of the first Host header and the path part of
18740 the request, which starts at the first slash and ends before the question
18741 mark. It can be useful in virtual hosted environments to detect URL abuses as
18742 well as to improve shared caches efficiency. Using this with a limited size
18743 stick table also allows one to collect statistics about most commonly
18744 requested objects by host/path. With ACLs it can allow simple content
18745 switching rules involving the host and the path at the same time, such as
18746 "www.example.com/favicon.ico". See also "path" and "uri".
18747
18748 ACL derivatives :
18749 base : exact string match
18750 base_beg : prefix match
18751 base_dir : subdir match
18752 base_dom : domain match
18753 base_end : suffix match
18754 base_len : length match
18755 base_reg : regex match
18756 base_sub : substring match
18757
18758base32 : integer
18759 This returns a 32-bit hash of the value returned by the "base" fetch method
18760 above. This is useful to track per-URL activity on high traffic sites without
18761 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020018762 memory. The output type is an unsigned integer. The hash function used is
18763 SDBM with full avalanche on the output. Technically, base32 is exactly equal
18764 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020018765
18766base32+src : binary
18767 This returns the concatenation of the base32 fetch above and the src fetch
18768 below. The resulting type is of type binary, with a size of 8 or 20 bytes
18769 depending on the source address family. This can be used to track per-IP,
18770 per-URL counters.
18771
Yves Lafonb4d37082021-02-11 11:01:28 +010018772baseq : string
18773 This returns the concatenation of the first Host header and the path part of
18774 the request with the query-string, which starts at the first slash. Using this
18775 instead of "base" allows one to properly identify the target resource, for
18776 statistics or caching use cases. See also "path", "pathq" and "base".
18777
William Lallemand65ad6e12014-01-31 15:08:02 +010018778capture.req.hdr(<idx>) : string
18779 This extracts the content of the header captured by the "capture request
18780 header", idx is the position of the capture keyword in the configuration.
18781 The first entry is an index of 0. See also: "capture request header".
18782
18783capture.req.method : string
18784 This extracts the METHOD of an HTTP request. It can be used in both request
18785 and response. Unlike "method", it can be used in both request and response
18786 because it's allocated.
18787
18788capture.req.uri : string
18789 This extracts the request's URI, which starts at the first slash and ends
18790 before the first space in the request (without the host part). Unlike "path"
18791 and "url", it can be used in both request and response because it's
18792 allocated.
18793
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020018794capture.req.ver : string
18795 This extracts the request's HTTP version and returns either "HTTP/1.0" or
18796 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
18797 logs because it relies on a persistent flag.
18798
William Lallemand65ad6e12014-01-31 15:08:02 +010018799capture.res.hdr(<idx>) : string
18800 This extracts the content of the header captured by the "capture response
18801 header", idx is the position of the capture keyword in the configuration.
18802 The first entry is an index of 0.
18803 See also: "capture response header"
18804
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020018805capture.res.ver : string
18806 This extracts the response's HTTP version and returns either "HTTP/1.0" or
18807 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
18808 persistent flag.
18809
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018810req.body : binary
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018811 This returns the HTTP request's available body as a block of data. It is
18812 recommended to use "option http-buffer-request" to be sure to wait, as much
18813 as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018814
Thierry FOURNIER9826c772015-05-20 15:50:54 +020018815req.body_param([<name>) : string
18816 This fetch assumes that the body of the POST request is url-encoded. The user
18817 can check if the "content-type" contains the value
18818 "application/x-www-form-urlencoded". This extracts the first occurrence of the
18819 parameter <name> in the body, which ends before '&'. The parameter name is
18820 case-sensitive. If no name is given, any parameter will match, and the first
18821 one will be returned. The result is a string corresponding to the value of the
18822 parameter <name> as presented in the request body (no URL decoding is
18823 performed). Note that the ACL version of this fetch iterates over multiple
18824 parameters and will iteratively report all parameters values if no name is
18825 given.
18826
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018827req.body_len : integer
18828 This returns the length of the HTTP request's available body in bytes. It may
18829 be lower than the advertised length if the body is larger than the buffer. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018830 is recommended to use "option http-buffer-request" to be sure to wait, as
18831 much as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018832
18833req.body_size : integer
18834 This returns the advertised length of the HTTP request's body in bytes. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018835 will represent the advertised Content-Length header, or the size of the
18836 available data in case of chunked encoding.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018837
Willy Tarreau74ca5042013-06-11 23:12:07 +020018838req.cook([<name>]) : string
18839cook([<name>]) : string (deprecated)
18840 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18841 header line from the request, and returns its value as string. If no name is
18842 specified, the first cookie value is returned. When used with ACLs, all
18843 matching cookies are evaluated. Spaces around the name and the value are
18844 ignored as requested by the Cookie header specification (RFC6265). The cookie
18845 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
18846 well return an empty value if it is present. Use the "found" match to detect
18847 presence. Use the res.cook() variant for response cookies sent by the server.
18848
18849 ACL derivatives :
18850 cook([<name>]) : exact string match
18851 cook_beg([<name>]) : prefix match
18852 cook_dir([<name>]) : subdir match
18853 cook_dom([<name>]) : domain match
18854 cook_end([<name>]) : suffix match
18855 cook_len([<name>]) : length match
18856 cook_reg([<name>]) : regex match
18857 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010018858
Willy Tarreau74ca5042013-06-11 23:12:07 +020018859req.cook_cnt([<name>]) : integer
18860cook_cnt([<name>]) : integer (deprecated)
18861 Returns an integer value representing the number of occurrences of the cookie
18862 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018863
Willy Tarreau74ca5042013-06-11 23:12:07 +020018864req.cook_val([<name>]) : integer
18865cook_val([<name>]) : integer (deprecated)
18866 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18867 header line from the request, and converts its value to an integer which is
18868 returned. If no name is specified, the first cookie value is returned. When
18869 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020018870
Willy Tarreau74ca5042013-06-11 23:12:07 +020018871cookie([<name>]) : string (deprecated)
18872 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18873 header line from the request, or a "Set-Cookie" header from the response, and
18874 returns its value as a string. A typical use is to get multiple clients
18875 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020018876 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020018877 multi-peer synchronization and state keeping across restarts. If no name is
18878 specified, the first cookie value is returned. This fetch should not be used
18879 anymore and should be replaced by req.cook() or res.cook() instead as it
18880 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018881
Willy Tarreau74ca5042013-06-11 23:12:07 +020018882hdr([<name>[,<occ>]]) : string
18883 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
18884 used on responses. Please refer to these respective fetches for more details.
18885 In case of doubt about the fetch direction, please use the explicit ones.
18886 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018887 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018888
Willy Tarreau74ca5042013-06-11 23:12:07 +020018889req.fhdr(<name>[,<occ>]) : string
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010018890 This returns the full value of the last occurrence of header <name> in an
18891 HTTP request. It differs from req.hdr() in that any commas present in the
18892 value are returned and are not used as delimiters. This is sometimes useful
18893 with headers such as User-Agent.
18894
18895 When used from an ACL, all occurrences are iterated over until a match is
18896 found.
18897
Willy Tarreau74ca5042013-06-11 23:12:07 +020018898 Optionally, a specific occurrence might be specified as a position number.
18899 Positive values indicate a position from the first occurrence, with 1 being
18900 the first one. Negative values indicate positions relative to the last one,
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010018901 with -1 being the last one.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018902
Willy Tarreau74ca5042013-06-11 23:12:07 +020018903req.fhdr_cnt([<name>]) : integer
18904 Returns an integer value representing the number of occurrences of request
18905 header field name <name>, or the total number of header fields if <name> is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010018906 not specified. Like req.fhdr() it differs from res.hdr_cnt() by not splitting
18907 headers at commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018908
Willy Tarreau74ca5042013-06-11 23:12:07 +020018909req.hdr([<name>[,<occ>]]) : string
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010018910 This returns the last comma-separated value of the header <name> in an HTTP
18911 request. The fetch considers any comma as a delimiter for distinct values.
18912 This is useful if you need to process headers that are defined to be a list
18913 of values, such as Accept, or X-Forwarded-For. If full-line headers are
18914 desired instead, use req.fhdr(). Please carefully check RFC 7231 to know how
18915 certain headers are supposed to be parsed. Also, some of them are case
18916 insensitive (e.g. Connection).
18917
18918 When used from an ACL, all occurrences are iterated over until a match is
18919 found.
18920
Willy Tarreau74ca5042013-06-11 23:12:07 +020018921 Optionally, a specific occurrence might be specified as a position number.
18922 Positive values indicate a position from the first occurrence, with 1 being
18923 the first one. Negative values indicate positions relative to the last one,
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010018924 with -1 being the last one.
18925
18926 A typical use is with the X-Forwarded-For header once converted to IP,
18927 associated with an IP stick-table.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018928
Willy Tarreau74ca5042013-06-11 23:12:07 +020018929 ACL derivatives :
18930 hdr([<name>[,<occ>]]) : exact string match
18931 hdr_beg([<name>[,<occ>]]) : prefix match
18932 hdr_dir([<name>[,<occ>]]) : subdir match
18933 hdr_dom([<name>[,<occ>]]) : domain match
18934 hdr_end([<name>[,<occ>]]) : suffix match
18935 hdr_len([<name>[,<occ>]]) : length match
18936 hdr_reg([<name>[,<occ>]]) : regex match
18937 hdr_sub([<name>[,<occ>]]) : substring match
18938
18939req.hdr_cnt([<name>]) : integer
18940hdr_cnt([<header>]) : integer (deprecated)
18941 Returns an integer value representing the number of occurrences of request
18942 header field name <name>, or the total number of header field values if
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010018943 <name> is not specified. Like req.hdr() it counts each comma separated
18944 part of the header's value. If counting of full-line headers is desired,
18945 then req.fhdr_cnt() should be used instead.
18946
18947 With ACLs, it can be used to detect presence, absence or abuse of a specific
18948 header, as well as to block request smuggling attacks by rejecting requests
18949 which contain more than one of certain headers.
18950
18951 Refer to req.hdr() for more information on header matching.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018952
18953req.hdr_ip([<name>[,<occ>]]) : ip
18954hdr_ip([<name>[,<occ>]]) : ip (deprecated)
18955 This extracts the last occurrence of header <name> in an HTTP request,
18956 converts it to an IPv4 or IPv6 address and returns this address. When used
18957 with ACLs, all occurrences are checked, and if <name> is omitted, every value
Willy Tarreau7b0e00d2021-03-25 14:12:29 +010018958 of every header is checked. The parser strictly adheres to the format
18959 described in RFC7239, with the extension that IPv4 addresses may optionally
18960 be followed by a colon (':') and a valid decimal port number (0 to 65535),
18961 which will be silently dropped. All other forms will not match and will
18962 cause the address to be ignored.
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010018963
18964 The <occ> parameter is processed as with req.hdr().
18965
18966 A typical use is with the X-Forwarded-For and X-Client-IP headers.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018967
18968req.hdr_val([<name>[,<occ>]]) : integer
18969hdr_val([<name>[,<occ>]]) : integer (deprecated)
18970 This extracts the last occurrence of header <name> in an HTTP request, and
18971 converts it to an integer value. When used with ACLs, all occurrences are
18972 checked, and if <name> is omitted, every value of every header is checked.
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010018973
18974 The <occ> parameter is processed as with req.hdr().
18975
18976 A typical use is with the X-Forwarded-For header.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018977
Christopher Faulet687a68e2020-11-24 17:13:24 +010018978req.hdrs : string
18979 Returns the current request headers as string including the last empty line
18980 separating headers from the request body. The last empty line can be used to
18981 detect a truncated header block. This sample fetch is useful for some SPOE
18982 headers analyzers and for advanced logging.
18983
18984req.hdrs_bin : binary
18985 Returns the current request headers contained in preparsed binary form. This
18986 is useful for offloading some processing with SPOE. Each string is described
18987 by a length followed by the number of bytes indicated in the length. The
18988 length is represented using the variable integer encoding detailed in the
18989 SPOE documentation. The end of the list is marked by a couple of empty header
18990 names and values (length of 0 for both).
18991
18992 *(<str:header-name><str:header-value>)<empty string><empty string>
Frédéric Lécailleec891192019-02-26 15:02:35 +010018993
Christopher Faulet687a68e2020-11-24 17:13:24 +010018994 int: refer to the SPOE documentation for the encoding
18995 str: <int:length><bytes>
Frédéric Lécailleec891192019-02-26 15:02:35 +010018996
Willy Tarreau74ca5042013-06-11 23:12:07 +020018997http_auth(<userlist>) : boolean
18998 Returns a boolean indicating whether the authentication data received from
18999 the client match a username & password stored in the specified userlist. This
19000 fetch function is not really useful outside of ACLs. Currently only http
19001 basic auth is supported.
19002
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010019003http_auth_group(<userlist>) : string
19004 Returns a string corresponding to the user name found in the authentication
19005 data received from the client if both the user name and password are valid
19006 according to the specified userlist. The main purpose is to use it in ACLs
19007 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019008 This fetch function is not really useful outside of ACLs. Currently only http
19009 basic auth is supported.
19010
19011 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010019012 http_auth_group(<userlist>) : group ...
19013 Returns true when the user extracted from the request and whose password is
19014 valid according to the specified userlist belongs to at least one of the
19015 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020019016
Christopher Fauleta4063562019-08-02 11:51:37 +020019017http_auth_pass : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010019018 Returns the user's password found in the authentication data received from
19019 the client, as supplied in the Authorization header. Not checks are
19020 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020019021
19022http_auth_type : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010019023 Returns the authentication method found in the authentication data received from
19024 the client, as supplied in the Authorization header. Not checks are
19025 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020019026
19027http_auth_user : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010019028 Returns the user name found in the authentication data received from the
19029 client, as supplied in the Authorization header. Not checks are performed by
19030 this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020019031
Willy Tarreau74ca5042013-06-11 23:12:07 +020019032http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020019033 Returns true when the request being processed is the first one of the
19034 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020019035 from some requests when a request is not the first one, or to help grouping
19036 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020019037
Willy Tarreau74ca5042013-06-11 23:12:07 +020019038method : integer + string
19039 Returns an integer value corresponding to the method in the HTTP request. For
19040 example, "GET" equals 1 (check sources to establish the matching). Value 9
19041 means "other method" and may be converted to a string extracted from the
19042 stream. This should not be used directly as a sample, this is only meant to
19043 be used from ACLs, which transparently convert methods from patterns to these
19044 integer + string values. Some predefined ACL already check for most common
19045 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019046
Willy Tarreau74ca5042013-06-11 23:12:07 +020019047 ACL derivatives :
19048 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020019049
Willy Tarreau74ca5042013-06-11 23:12:07 +020019050 Example :
19051 # only accept GET and HEAD requests
19052 acl valid_method method GET HEAD
19053 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020019054
Willy Tarreau74ca5042013-06-11 23:12:07 +020019055path : string
19056 This extracts the request's URL path, which starts at the first slash and
19057 ends before the question mark (without the host part). A typical use is with
19058 prefetch-capable caches, and with portals which need to aggregate multiple
19059 information from databases and keep them in caches. Note that with outgoing
19060 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010019061 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020019062 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019063
Willy Tarreau74ca5042013-06-11 23:12:07 +020019064 ACL derivatives :
19065 path : exact string match
19066 path_beg : prefix match
19067 path_dir : subdir match
19068 path_dom : domain match
19069 path_end : suffix match
19070 path_len : length match
19071 path_reg : regex match
19072 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020019073
Christopher Faulete720c322020-09-02 17:25:18 +020019074pathq : string
19075 This extracts the request's URL path with the query-string, which starts at
19076 the first slash. This sample fetch is pretty handy to always retrieve a
19077 relative URI, excluding the scheme and the authority part, if any. Indeed,
19078 while it is the common representation for an HTTP/1.1 request target, in
19079 HTTP/2, an absolute URI is often used. This sample fetch will return the same
19080 result in both cases.
19081
Willy Tarreau49ad95c2015-01-19 15:06:26 +010019082query : string
19083 This extracts the request's query string, which starts after the first
19084 question mark. If no question mark is present, this fetch returns nothing. If
19085 a question mark is present but nothing follows, it returns an empty string.
19086 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010019087 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010019088 which stops before the question mark.
19089
Willy Tarreaueb27ec72015-02-20 13:55:29 +010019090req.hdr_names([<delim>]) : string
19091 This builds a string made from the concatenation of all header names as they
19092 appear in the request when the rule is evaluated. The default delimiter is
19093 the comma (',') but it may be overridden as an optional argument <delim>. In
19094 this case, only the first character of <delim> is considered.
19095
Willy Tarreau74ca5042013-06-11 23:12:07 +020019096req.ver : string
19097req_ver : string (deprecated)
19098 Returns the version string from the HTTP request, for example "1.1". This can
19099 be useful for logs, but is mostly there for ACL. Some predefined ACL already
19100 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019101
Willy Tarreau74ca5042013-06-11 23:12:07 +020019102 ACL derivatives :
19103 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020019104
Christopher Faulete596d182020-05-05 17:46:34 +020019105res.body : binary
19106 This returns the HTTP response's available body as a block of data. Unlike
19107 the request side, there is no directive to wait for the response's body. This
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019108 sample fetch is really useful (and usable) in the health-check context.
19109
19110 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019111
19112res.body_len : integer
19113 This returns the length of the HTTP response available body in bytes. Unlike
19114 the request side, there is no directive to wait for the response's body. This
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019115 sample fetch is really useful (and usable) in the health-check context.
19116
19117 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019118
19119res.body_size : integer
19120 This returns the advertised length of the HTTP response body in bytes. It
19121 will represent the advertised Content-Length header, or the size of the
19122 available data in case of chunked encoding. Unlike the request side, there is
19123 no directive to wait for the response body. This sample fetch is really
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019124 useful (and usable) in the health-check context.
19125
19126 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019127
Remi Tricot-Le Bretonbf971212020-10-27 11:55:57 +010019128res.cache_hit : boolean
19129 Returns the boolean "true" value if the response has been built out of an
19130 HTTP cache entry, otherwise returns boolean "false".
19131
19132res.cache_name : string
19133 Returns a string containing the name of the HTTP cache that was used to
19134 build the HTTP response if res.cache_hit is true, otherwise returns an
19135 empty string.
19136
Willy Tarreau74ca5042013-06-11 23:12:07 +020019137res.comp : boolean
19138 Returns the boolean "true" value if the response has been compressed by
19139 HAProxy, otherwise returns boolean "false". This may be used to add
19140 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019141
Willy Tarreau74ca5042013-06-11 23:12:07 +020019142res.comp_algo : string
19143 Returns a string containing the name of the algorithm used if the response
19144 was compressed by HAProxy, for example : "deflate". This may be used to add
19145 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019146
Willy Tarreau74ca5042013-06-11 23:12:07 +020019147res.cook([<name>]) : string
19148scook([<name>]) : string (deprecated)
19149 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
19150 header line from the response, and returns its value as string. If no name is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019151 specified, the first cookie value is returned.
19152
19153 It may be used in tcp-check based expect rules.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020019154
Willy Tarreau74ca5042013-06-11 23:12:07 +020019155 ACL derivatives :
19156 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020019157
Willy Tarreau74ca5042013-06-11 23:12:07 +020019158res.cook_cnt([<name>]) : integer
19159scook_cnt([<name>]) : integer (deprecated)
19160 Returns an integer value representing the number of occurrences of the cookie
19161 <name> in the response, or all cookies if <name> is not specified. This is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019162 mostly useful when combined with ACLs to detect suspicious responses.
19163
19164 It may be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019165
Willy Tarreau74ca5042013-06-11 23:12:07 +020019166res.cook_val([<name>]) : integer
19167scook_val([<name>]) : integer (deprecated)
19168 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
19169 header line from the response, and converts its value to an integer which is
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019170 returned. If no name is specified, the first cookie value is returned.
19171
19172 It may be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010019173
Willy Tarreau74ca5042013-06-11 23:12:07 +020019174res.fhdr([<name>[,<occ>]]) : string
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019175 This fetch works like the req.fhdr() fetch with the difference that it acts
19176 on the headers within an HTTP response.
19177
19178 Like req.fhdr() the res.fhdr() fetch returns full values. If the header is
19179 defined to be a list you should use res.hdr().
19180
19181 This fetch is sometimes useful with headers such as Date or Expires.
19182
19183 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019184
Willy Tarreau74ca5042013-06-11 23:12:07 +020019185res.fhdr_cnt([<name>]) : integer
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019186 This fetch works like the req.fhdr_cnt() fetch with the difference that it
19187 acts on the headers within an HTTP response.
19188
19189 Like req.fhdr_cnt() the res.fhdr_cnt() fetch acts on full values. If the
19190 header is defined to be a list you should use res.hdr_cnt().
19191
19192 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019193
Willy Tarreau74ca5042013-06-11 23:12:07 +020019194res.hdr([<name>[,<occ>]]) : string
19195shdr([<name>[,<occ>]]) : string (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019196 This fetch works like the req.hdr() fetch with the difference that it acts
19197 on the headers within an HTTP response.
19198
Ilya Shipitsinacf84592021-02-06 22:29:08 +050019199 Like req.hdr() the res.hdr() fetch considers the comma to be a delimiter. If
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019200 this is not desired res.fhdr() should be used.
19201
19202 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019203
Willy Tarreau74ca5042013-06-11 23:12:07 +020019204 ACL derivatives :
19205 shdr([<name>[,<occ>]]) : exact string match
19206 shdr_beg([<name>[,<occ>]]) : prefix match
19207 shdr_dir([<name>[,<occ>]]) : subdir match
19208 shdr_dom([<name>[,<occ>]]) : domain match
19209 shdr_end([<name>[,<occ>]]) : suffix match
19210 shdr_len([<name>[,<occ>]]) : length match
19211 shdr_reg([<name>[,<occ>]]) : regex match
19212 shdr_sub([<name>[,<occ>]]) : substring match
19213
19214res.hdr_cnt([<name>]) : integer
19215shdr_cnt([<name>]) : integer (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019216 This fetch works like the req.hdr_cnt() fetch with the difference that it
19217 acts on the headers within an HTTP response.
19218
19219 Like req.hdr_cnt() the res.hdr_cnt() fetch considers the comma to be a
Ilya Shipitsinacf84592021-02-06 22:29:08 +050019220 delimiter. If this is not desired res.fhdr_cnt() should be used.
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019221
19222 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019223
Willy Tarreau74ca5042013-06-11 23:12:07 +020019224res.hdr_ip([<name>[,<occ>]]) : ip
19225shdr_ip([<name>[,<occ>]]) : ip (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019226 This fetch works like the req.hdr_ip() fetch with the difference that it
19227 acts on the headers within an HTTP response.
19228
19229 This can be useful to learn some data into a stick table.
19230
19231 It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020019232
Willy Tarreaueb27ec72015-02-20 13:55:29 +010019233res.hdr_names([<delim>]) : string
19234 This builds a string made from the concatenation of all header names as they
19235 appear in the response when the rule is evaluated. The default delimiter is
19236 the comma (',') but it may be overridden as an optional argument <delim>. In
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019237 this case, only the first character of <delim> is considered.
19238
19239 It may be used in tcp-check based expect rules.
Willy Tarreaueb27ec72015-02-20 13:55:29 +010019240
Willy Tarreau74ca5042013-06-11 23:12:07 +020019241res.hdr_val([<name>[,<occ>]]) : integer
19242shdr_val([<name>[,<occ>]]) : integer (deprecated)
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019243 This fetch works like the req.hdr_val() fetch with the difference that it
19244 acts on the headers within an HTTP response.
19245
19246 This can be useful to learn some data into a stick table.
19247
19248 It may be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019249
19250res.hdrs : string
19251 Returns the current response headers as string including the last empty line
19252 separating headers from the request body. The last empty line can be used to
19253 detect a truncated header block. This sample fetch is useful for some SPOE
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019254 headers analyzers and for advanced logging.
19255
19256 It may also be used in tcp-check based expect rules.
Christopher Faulete596d182020-05-05 17:46:34 +020019257
19258res.hdrs_bin : binary
19259 Returns the current response headers contained in preparsed binary form. This
19260 is useful for offloading some processing with SPOE. It may be used in
19261 tcp-check based expect rules. Each string is described by a length followed
19262 by the number of bytes indicated in the length. The length is represented
19263 using the variable integer encoding detailed in the SPOE documentation. The
19264 end of the list is marked by a couple of empty header names and values
19265 (length of 0 for both).
19266
19267 *(<str:header-name><str:header-value>)<empty string><empty string>
19268
19269 int: refer to the SPOE documentation for the encoding
19270 str: <int:length><bytes>
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010019271
Willy Tarreau74ca5042013-06-11 23:12:07 +020019272res.ver : string
19273resp_ver : string (deprecated)
19274 Returns the version string from the HTTP response, for example "1.1". This
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019275 can be useful for logs, but is mostly there for ACL.
19276
19277 It may be used in tcp-check based expect rules.
Willy Tarreau0e698542011-09-16 08:32:32 +020019278
Willy Tarreau74ca5042013-06-11 23:12:07 +020019279 ACL derivatives :
19280 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010019281
Willy Tarreau74ca5042013-06-11 23:12:07 +020019282set-cookie([<name>]) : string (deprecated)
19283 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
19284 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020019285 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020019286 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010019287
Willy Tarreau74ca5042013-06-11 23:12:07 +020019288 This fetch function is deprecated and has been superseded by the "res.cook"
19289 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010019290
Willy Tarreau74ca5042013-06-11 23:12:07 +020019291status : integer
19292 Returns an integer containing the HTTP status code in the HTTP response, for
19293 example, 302. It is mostly used within ACLs and integer ranges, for example,
Tim Duesterhus27c70ae2021-01-23 17:50:21 +010019294 to remove any Location header if the response is not a 3xx.
19295
19296 It may be used in tcp-check based expect rules.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019297
Thierry Fournier0e00dca2016-04-07 15:47:40 +020019298unique-id : string
19299 Returns the unique-id attached to the request. The directive
19300 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
19301 fails. Note that the unique-id is usually used with HTTP requests, however this
19302 sample fetch can be used with other protocols. Obviously, if it is used with
19303 other protocols than HTTP, the unique-id-format directive must not contain
19304 HTTP parts. See: unique-id-format and unique-id-header
19305
Willy Tarreau74ca5042013-06-11 23:12:07 +020019306url : string
19307 This extracts the request's URL as presented in the request. A typical use is
19308 with prefetch-capable caches, and with portals which need to aggregate
19309 multiple information from databases and keep them in caches. With ACLs, using
19310 "path" is preferred over using "url", because clients may send a full URL as
19311 is normally done with proxies. The only real use is to match "*" which does
19312 not match in "path", and for which there is already a predefined ACL. See
19313 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019314
Willy Tarreau74ca5042013-06-11 23:12:07 +020019315 ACL derivatives :
19316 url : exact string match
19317 url_beg : prefix match
19318 url_dir : subdir match
19319 url_dom : domain match
19320 url_end : suffix match
19321 url_len : length match
19322 url_reg : regex match
19323 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019324
Willy Tarreau74ca5042013-06-11 23:12:07 +020019325url_ip : ip
19326 This extracts the IP address from the request's URL when the host part is
19327 presented as an IP address. Its use is very limited. For instance, a
19328 monitoring system might use this field as an alternative for the source IP in
19329 order to test what path a given source address would follow, or to force an
19330 entry in a table for a given source address. With ACLs it can be used to
19331 restrict access to certain systems through a proxy, for example when combined
19332 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019333
Willy Tarreau74ca5042013-06-11 23:12:07 +020019334url_port : integer
19335 This extracts the port part from the request's URL. Note that if the port is
19336 not specified in the request, port 80 is assumed. With ACLs it can be used to
19337 restrict access to certain systems through a proxy, for example when combined
19338 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019339
Willy Tarreau1ede1da2015-05-07 16:06:18 +020019340urlp([<name>[,<delim>]]) : string
19341url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020019342 This extracts the first occurrence of the parameter <name> in the query
19343 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020019344 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
19345 any parameter will match, and the first one will be returned. The result is
19346 a string corresponding to the value of the parameter <name> as presented in
19347 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020019348 stickiness based on a client ID, to extract an application cookie passed as a
19349 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020019350 this fetch iterates over multiple parameters and will iteratively report all
19351 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019352
Willy Tarreau74ca5042013-06-11 23:12:07 +020019353 ACL derivatives :
19354 urlp(<name>[,<delim>]) : exact string match
19355 urlp_beg(<name>[,<delim>]) : prefix match
19356 urlp_dir(<name>[,<delim>]) : subdir match
19357 urlp_dom(<name>[,<delim>]) : domain match
19358 urlp_end(<name>[,<delim>]) : suffix match
19359 urlp_len(<name>[,<delim>]) : length match
19360 urlp_reg(<name>[,<delim>]) : regex match
19361 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019362
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019363
Willy Tarreau74ca5042013-06-11 23:12:07 +020019364 Example :
19365 # match http://example.com/foo?PHPSESSIONID=some_id
19366 stick on urlp(PHPSESSIONID)
19367 # match http://example.com/foo;JSESSIONID=some_id
19368 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020019369
Jarno Huuskonen676f6222017-03-30 09:19:45 +030019370urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020019371 See "urlp" above. This one extracts the URL parameter <name> in the request
19372 and converts it to an integer value. This can be used for session stickiness
19373 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020019374
Dragan Dosen0070cd52016-06-16 12:19:49 +020019375url32 : integer
19376 This returns a 32-bit hash of the value obtained by concatenating the first
19377 Host header and the whole URL including parameters (not only the path part of
19378 the request, as in the "base32" fetch above). This is useful to track per-URL
19379 activity. A shorter hash is stored, saving a lot of memory. The output type
19380 is an unsigned integer.
19381
19382url32+src : binary
19383 This returns the concatenation of the "url32" fetch and the "src" fetch. The
19384 resulting type is of type binary, with a size of 8 or 20 bytes depending on
19385 the source address family. This can be used to track per-IP, per-URL counters.
19386
Christopher Faulet16032ab2020-04-30 11:30:00 +020019387
Christopher Faulete596d182020-05-05 17:46:34 +0200193887.3.7. Fetching samples for developers
Christopher Fauletd47941d2020-01-08 14:40:19 +010019389---------------------------------------
19390
19391This set of sample fetch methods is reserved to developers and must never be
19392used on a production environment, except on developer demand, for debugging
19393purposes. Moreover, no special care will be taken on backwards compatibility.
19394There is no warranty the following sample fetches will never change, be renamed
19395or simply removed. So be really careful if you should use one of them. To avoid
19396any ambiguity, these sample fetches are placed in the dedicated scope "internal",
19397for instance "internal.strm.is_htx".
19398
19399internal.htx.data : integer
19400 Returns the size in bytes used by data in the HTX message associated to a
19401 channel. The channel is chosen depending on the sample direction.
19402
19403internal.htx.free : integer
19404 Returns the free space (size - used) in bytes in the HTX message associated
19405 to a channel. The channel is chosen depending on the sample direction.
19406
19407internal.htx.free_data : integer
19408 Returns the free space for the data in bytes in the HTX message associated to
19409 a channel. The channel is chosen depending on the sample direction.
19410
19411internal.htx.has_eom : boolean
Christopher Fauletd1ac2b92020-12-02 19:12:22 +010019412 Returns true if the HTX message associated to a channel contains the
19413 end-of-message flag (EOM). Otherwise, it returns false. The channel is chosen
19414 depending on the sample direction.
Christopher Fauletd47941d2020-01-08 14:40:19 +010019415
19416internal.htx.nbblks : integer
19417 Returns the number of blocks present in the HTX message associated to a
19418 channel. The channel is chosen depending on the sample direction.
19419
19420internal.htx.size : integer
19421 Returns the total size in bytes of the HTX message associated to a
19422 channel. The channel is chosen depending on the sample direction.
19423
19424internal.htx.used : integer
19425 Returns the total size used in bytes (data + metadata) in the HTX message
19426 associated to a channel. The channel is chosen depending on the sample
19427 direction.
19428
19429internal.htx_blk.size(<idx>) : integer
19430 Returns the size of the block at the position <idx> in the HTX message
19431 associated to a channel or 0 if it does not exist. The channel is chosen
19432 depending on the sample direction. <idx> may be any positive integer or one
19433 of the special value :
19434 * head : The oldest inserted block
19435 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019436 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019437
19438internal.htx_blk.type(<idx>) : string
19439 Returns the type of the block at the position <idx> in the HTX message
19440 associated to a channel or "HTX_BLK_UNUSED" if it does not exist. The channel
19441 is chosen depending on the sample direction. <idx> may be any positive
19442 integer or one of the special value :
19443 * head : The oldest inserted block
19444 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019445 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019446
19447internal.htx_blk.data(<idx>) : binary
19448 Returns the value of the DATA block at the position <idx> in the HTX message
19449 associated to a channel or an empty string if it does not exist or if it is
19450 not a DATA block. The channel is chosen depending on the sample direction.
19451 <idx> may be any positive integer or one of the special value :
19452
19453 * head : The oldest inserted block
19454 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019455 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019456
19457internal.htx_blk.hdrname(<idx>) : string
19458 Returns the header name of the HEADER block at the position <idx> in the HTX
19459 message associated to a channel or an empty string if it does not exist or if
19460 it is not an HEADER block. The channel is chosen depending on the sample
19461 direction. <idx> may be any positive integer or one of the special value :
19462
19463 * head : The oldest inserted block
19464 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019465 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019466
19467internal.htx_blk.hdrval(<idx>) : string
19468 Returns the header value of the HEADER block at the position <idx> in the HTX
19469 message associated to a channel or an empty string if it does not exist or if
19470 it is not an HEADER block. The channel is chosen depending on the sample
19471 direction. <idx> may be any positive integer or one of the special value :
19472
19473 * head : The oldest inserted block
19474 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019475 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019476
19477internal.htx_blk.start_line(<idx>) : string
19478 Returns the value of the REQ_SL or RES_SL block at the position <idx> in the
19479 HTX message associated to a channel or an empty string if it does not exist
19480 or if it is not a SL block. The channel is chosen depending on the sample
19481 direction. <idx> may be any positive integer or one of the special value :
19482
19483 * head : The oldest inserted block
19484 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019485 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019486
19487internal.strm.is_htx : boolean
19488 Returns true if the current stream is an HTX stream. It means the data in the
19489 channels buffers are stored using the internal HTX representation. Otherwise,
19490 it returns false.
19491
19492
Willy Tarreau74ca5042013-06-11 23:12:07 +0200194937.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019494---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010019495
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019496Some predefined ACLs are hard-coded so that they do not have to be declared in
19497every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020019498order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010019499
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019500ACL name Equivalent to Usage
19501---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019502FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020019503HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019504HTTP_1.0 req_ver 1.0 match HTTP version 1.0
19505HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010019506HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
19507HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
19508HTTP_URL_SLASH url_beg / match URL beginning with "/"
19509HTTP_URL_STAR url * match URL equal to "*"
19510LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019511METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020019512METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019513METH_GET method GET HEAD match HTTP GET or HEAD method
19514METH_HEAD method HEAD match HTTP HEAD method
19515METH_OPTIONS method OPTIONS match HTTP OPTIONS method
19516METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020019517METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019518METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020019519RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019520REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010019521TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019522WAIT_END wait_end wait for end of content analysis
19523---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010019524
Willy Tarreaub937b7e2010-01-12 15:27:54 +010019525
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200195268. Logging
19527----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010019528
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019529One of HAProxy's strong points certainly lies is its precise logs. It probably
19530provides the finest level of information available for such a product, which is
19531very important for troubleshooting complex environments. Standard information
19532provided in logs include client ports, TCP/HTTP state timers, precise session
19533state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010019534to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019535headers.
19536
19537In order to improve administrators reactivity, it offers a great transparency
19538about encountered problems, both internal and external, and it is possible to
19539send logs to different sources at the same time with different level filters :
19540
19541 - global process-level logs (system errors, start/stop, etc..)
19542 - per-instance system and internal errors (lack of resource, bugs, ...)
19543 - per-instance external troubles (servers up/down, max connections)
19544 - per-instance activity (client connections), either at the establishment or
19545 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010019546 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060019547 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019548
19549The ability to distribute different levels of logs to different log servers
19550allow several production teams to interact and to fix their problems as soon
19551as possible. For example, the system team might monitor system-wide errors,
19552while the application team might be monitoring the up/down for their servers in
19553real time, and the security team might analyze the activity logs with one hour
19554delay.
19555
19556
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200195578.1. Log levels
19558---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019559
Simon Hormandf791f52011-05-29 15:01:10 +090019560TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019561source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090019562HTTP request, HTTP return code, number of bytes transmitted, conditions
19563in which the session ended, and even exchanged cookies values. For example
19564track a particular user's problems. All messages may be sent to up to two
19565syslog servers. Check the "log" keyword in section 4.2 for more information
19566about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019567
19568
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200195698.2. Log formats
19570----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019571
William Lallemand48940402012-01-30 16:47:22 +010019572HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090019573and will be detailed in the following sections. A few of them may vary
19574slightly with the configuration, due to indicators specific to certain
19575options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019576
19577 - the default format, which is very basic and very rarely used. It only
19578 provides very basic information about the incoming connection at the moment
19579 it is accepted : source IP:port, destination IP:port, and frontend-name.
19580 This mode will eventually disappear so it will not be described to great
19581 extents.
19582
19583 - the TCP format, which is more advanced. This format is enabled when "option
19584 tcplog" is set on the frontend. HAProxy will then usually wait for the
19585 connection to terminate before logging. This format provides much richer
19586 information, such as timers, connection counts, queue size, etc... This
19587 format is recommended for pure TCP proxies.
19588
19589 - the HTTP format, which is the most advanced for HTTP proxying. This format
19590 is enabled when "option httplog" is set on the frontend. It provides the
19591 same information as the TCP format with some HTTP-specific fields such as
19592 the request, the status code, and captures of headers and cookies. This
19593 format is recommended for HTTP proxies.
19594
Emeric Brun3a058f32009-06-30 18:26:00 +020019595 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
19596 fields arranged in the same order as the CLF format. In this mode, all
19597 timers, captures, flags, etc... appear one per field after the end of the
19598 common fields, in the same order they appear in the standard HTTP format.
19599
William Lallemand48940402012-01-30 16:47:22 +010019600 - the custom log format, allows you to make your own log line.
19601
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019602Next sections will go deeper into details for each of these formats. Format
19603specification will be performed on a "field" basis. Unless stated otherwise, a
19604field is a portion of text delimited by any number of spaces. Since syslog
19605servers are susceptible of inserting fields at the beginning of a line, it is
19606always assumed that the first field is the one containing the process name and
19607identifier.
19608
19609Note : Since log lines may be quite long, the log examples in sections below
19610 might be broken into multiple lines. The example log lines will be
19611 prefixed with 3 closing angle brackets ('>>>') and each time a log is
19612 broken into multiple lines, each non-final line will end with a
19613 backslash ('\') and the next line will start indented by two characters.
19614
19615
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200196168.2.1. Default log format
19617-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019618
19619This format is used when no specific option is set. The log is emitted as soon
19620as the connection is accepted. One should note that this currently is the only
19621format which logs the request's destination IP and ports.
19622
19623 Example :
19624 listen www
19625 mode http
19626 log global
19627 server srv1 127.0.0.1:8000
19628
19629 >>> Feb 6 12:12:09 localhost \
19630 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
19631 (www/HTTP)
19632
19633 Field Format Extract from the example above
19634 1 process_name '[' pid ']:' haproxy[14385]:
19635 2 'Connect from' Connect from
19636 3 source_ip ':' source_port 10.0.1.2:33312
19637 4 'to' to
19638 5 destination_ip ':' destination_port 10.0.3.31:8012
19639 6 '(' frontend_name '/' mode ')' (www/HTTP)
19640
19641Detailed fields description :
19642 - "source_ip" is the IP address of the client which initiated the connection.
19643 - "source_port" is the TCP port of the client which initiated the connection.
19644 - "destination_ip" is the IP address the client connected to.
19645 - "destination_port" is the TCP port the client connected to.
19646 - "frontend_name" is the name of the frontend (or listener) which received
19647 and processed the connection.
19648 - "mode is the mode the frontend is operating (TCP or HTTP).
19649
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019650In case of a UNIX socket, the source and destination addresses are marked as
19651"unix:" and the ports reflect the internal ID of the socket which accepted the
19652connection (the same ID as reported in the stats).
19653
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019654It is advised not to use this deprecated format for newer installations as it
19655will eventually disappear.
19656
19657
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200196588.2.2. TCP log format
19659---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019660
19661The TCP format is used when "option tcplog" is specified in the frontend, and
19662is the recommended format for pure TCP proxies. It provides a lot of precious
19663information for troubleshooting. Since this format includes timers and byte
19664counts, the log is normally emitted at the end of the session. It can be
19665emitted earlier if "option logasap" is specified, which makes sense in most
19666environments with long sessions such as remote terminals. Sessions which match
19667the "monitor" rules are never logged. It is also possible not to emit logs for
19668sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019669specifying "option dontlognull" in the frontend. Successful connections will
19670not be logged if "option dontlog-normal" is specified in the frontend. A few
19671fields may slightly vary depending on some configuration options, those are
19672marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019673
19674 Example :
19675 frontend fnt
19676 mode tcp
19677 option tcplog
19678 log global
19679 default_backend bck
19680
19681 backend bck
19682 server srv1 127.0.0.1:8000
19683
19684 >>> Feb 6 12:12:56 localhost \
19685 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
19686 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
19687
19688 Field Format Extract from the example above
19689 1 process_name '[' pid ']:' haproxy[14387]:
19690 2 client_ip ':' client_port 10.0.1.2:33313
19691 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
19692 4 frontend_name fnt
19693 5 backend_name '/' server_name bck/srv1
19694 6 Tw '/' Tc '/' Tt* 0/0/5007
19695 7 bytes_read* 212
19696 8 termination_state --
19697 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
19698 10 srv_queue '/' backend_queue 0/0
19699
19700Detailed fields description :
19701 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019702 connection to haproxy. If the connection was accepted on a UNIX socket
19703 instead, the IP address would be replaced with the word "unix". Note that
19704 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019705 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010019706 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019707 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019708
19709 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019710 If the connection was accepted on a UNIX socket instead, the port would be
19711 replaced with the ID of the accepting socket, which is also reported in the
19712 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019713
19714 - "accept_date" is the exact date when the connection was received by haproxy
19715 (which might be very slightly different from the date observed on the
19716 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020019717 the same date which may appear in any upstream firewall's log. When used in
19718 HTTP mode, the accept_date field will be reset to the first moment the
19719 connection is ready to receive a new request (end of previous response for
19720 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019721
19722 - "frontend_name" is the name of the frontend (or listener) which received
19723 and processed the connection.
19724
19725 - "backend_name" is the name of the backend (or listener) which was selected
19726 to manage the connection to the server. This will be the same as the
19727 frontend if no switching rule has been applied, which is common for TCP
19728 applications.
19729
19730 - "server_name" is the name of the last server to which the connection was
19731 sent, which might differ from the first one if there were connection errors
19732 and a redispatch occurred. Note that this server belongs to the backend
19733 which processed the request. If the connection was aborted before reaching
19734 a server, "<NOSRV>" is indicated instead of a server name.
19735
19736 - "Tw" is the total time in milliseconds spent waiting in the various queues.
19737 It can be "-1" if the connection was aborted before reaching the queue.
19738 See "Timers" below for more details.
19739
19740 - "Tc" is the total time in milliseconds spent waiting for the connection to
19741 establish to the final server, including retries. It can be "-1" if the
19742 connection was aborted before a connection could be established. See
19743 "Timers" below for more details.
19744
19745 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019746 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019747 "option logasap" was specified, then the time counting stops at the moment
19748 the log is emitted. In this case, a '+' sign is prepended before the value,
19749 indicating that the final one will be larger. See "Timers" below for more
19750 details.
19751
19752 - "bytes_read" is the total number of bytes transmitted from the server to
19753 the client when the log is emitted. If "option logasap" is specified, the
19754 this value will be prefixed with a '+' sign indicating that the final one
19755 may be larger. Please note that this value is a 64-bit counter, so log
19756 analysis tools must be able to handle it without overflowing.
19757
19758 - "termination_state" is the condition the session was in when the session
19759 ended. This indicates the session state, which side caused the end of
19760 session to happen, and for what reason (timeout, error, ...). The normal
19761 flags should be "--", indicating the session was closed by either end with
19762 no data remaining in buffers. See below "Session state at disconnection"
19763 for more details.
19764
19765 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019766 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019767 limits have been reached. For instance, if actconn is close to 512 when
19768 multiple connection errors occur, chances are high that the system limits
19769 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019770 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019771
19772 - "feconn" is the total number of concurrent connections on the frontend when
19773 the session was logged. It is useful to estimate the amount of resource
19774 required to sustain high loads, and to detect when the frontend's "maxconn"
19775 has been reached. Most often when this value increases by huge jumps, it is
19776 because there is congestion on the backend servers, but sometimes it can be
19777 caused by a denial of service attack.
19778
19779 - "beconn" is the total number of concurrent connections handled by the
19780 backend when the session was logged. It includes the total number of
19781 concurrent connections active on servers as well as the number of
19782 connections pending in queues. It is useful to estimate the amount of
19783 additional servers needed to support high loads for a given application.
19784 Most often when this value increases by huge jumps, it is because there is
19785 congestion on the backend servers, but sometimes it can be caused by a
19786 denial of service attack.
19787
19788 - "srv_conn" is the total number of concurrent connections still active on
19789 the server when the session was logged. It can never exceed the server's
19790 configured "maxconn" parameter. If this value is very often close or equal
19791 to the server's "maxconn", it means that traffic regulation is involved a
19792 lot, meaning that either the server's maxconn value is too low, or that
19793 there aren't enough servers to process the load with an optimal response
19794 time. When only one of the server's "srv_conn" is high, it usually means
19795 that this server has some trouble causing the connections to take longer to
19796 be processed than on other servers.
19797
19798 - "retries" is the number of connection retries experienced by this session
19799 when trying to connect to the server. It must normally be zero, unless a
19800 server is being stopped at the same moment the connection was attempted.
19801 Frequent retries generally indicate either a network problem between
19802 haproxy and the server, or a misconfigured system backlog on the server
19803 preventing new connections from being queued. This field may optionally be
19804 prefixed with a '+' sign, indicating that the session has experienced a
19805 redispatch after the maximal retry count has been reached on the initial
19806 server. In this case, the server name appearing in the log is the one the
19807 connection was redispatched to, and not the first one, though both may
19808 sometimes be the same in case of hashing for instance. So as a general rule
19809 of thumb, when a '+' is present in front of the retry count, this count
19810 should not be attributed to the logged server.
19811
19812 - "srv_queue" is the total number of requests which were processed before
19813 this one in the server queue. It is zero when the request has not gone
19814 through the server queue. It makes it possible to estimate the approximate
19815 server's response time by dividing the time spent in queue by the number of
19816 requests in the queue. It is worth noting that if a session experiences a
19817 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010019818 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019819 backend queue unless a redispatch occurs.
19820
19821 - "backend_queue" is the total number of requests which were processed before
19822 this one in the backend's global queue. It is zero when the request has not
19823 gone through the global queue. It makes it possible to estimate the average
19824 queue length, which easily translates into a number of missing servers when
19825 divided by a server's "maxconn" parameter. It is worth noting that if a
19826 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010019827 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019828 through both the server queue and the backend queue unless a redispatch
19829 occurs.
19830
19831
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200198328.2.3. HTTP log format
19833----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019834
19835The HTTP format is the most complete and the best suited for HTTP proxies. It
19836is enabled by when "option httplog" is specified in the frontend. It provides
19837the same level of information as the TCP format with additional features which
19838are specific to the HTTP protocol. Just like the TCP format, the log is usually
19839emitted at the end of the session, unless "option logasap" is specified, which
19840generally only makes sense for download sites. A session which matches the
19841"monitor" rules will never logged. It is also possible not to log sessions for
19842which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019843frontend. Successful connections will not be logged if "option dontlog-normal"
19844is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019845
19846Most fields are shared with the TCP log, some being different. A few fields may
19847slightly vary depending on some configuration options. Those ones are marked
19848with a star ('*') after the field name below.
19849
19850 Example :
19851 frontend http-in
19852 mode http
19853 option httplog
19854 log global
19855 default_backend bck
19856
19857 backend static
19858 server srv1 127.0.0.1:8000
19859
19860 >>> Feb 6 12:14:14 localhost \
19861 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
19862 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 +010019863 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019864
19865 Field Format Extract from the example above
19866 1 process_name '[' pid ']:' haproxy[14389]:
19867 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019868 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019869 4 frontend_name http-in
19870 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019871 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019872 7 status_code 200
19873 8 bytes_read* 2750
19874 9 captured_request_cookie -
19875 10 captured_response_cookie -
19876 11 termination_state ----
19877 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
19878 13 srv_queue '/' backend_queue 0/0
19879 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
19880 15 '{' captured_response_headers* '}' {}
19881 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010019882
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019883Detailed fields description :
19884 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019885 connection to haproxy. If the connection was accepted on a UNIX socket
19886 instead, the IP address would be replaced with the word "unix". Note that
19887 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019888 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010019889 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019890 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019891
19892 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019893 If the connection was accepted on a UNIX socket instead, the port would be
19894 replaced with the ID of the accepting socket, which is also reported in the
19895 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019896
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019897 - "request_date" is the exact date when the first byte of the HTTP request
19898 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019899
19900 - "frontend_name" is the name of the frontend (or listener) which received
19901 and processed the connection.
19902
19903 - "backend_name" is the name of the backend (or listener) which was selected
19904 to manage the connection to the server. This will be the same as the
19905 frontend if no switching rule has been applied.
19906
19907 - "server_name" is the name of the last server to which the connection was
19908 sent, which might differ from the first one if there were connection errors
19909 and a redispatch occurred. Note that this server belongs to the backend
19910 which processed the request. If the request was aborted before reaching a
19911 server, "<NOSRV>" is indicated instead of a server name. If the request was
19912 intercepted by the stats subsystem, "<STATS>" is indicated instead.
19913
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019914 - "TR" is the total time in milliseconds spent waiting for a full HTTP
19915 request from the client (not counting body) after the first byte was
19916 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050019917 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019918 always be very small because a request generally fits in one single packet.
19919 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020019920 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
19921 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019922
19923 - "Tw" is the total time in milliseconds spent waiting in the various queues.
19924 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020019925 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019926
19927 - "Tc" is the total time in milliseconds spent waiting for the connection to
19928 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020019929 request was aborted before a connection could be established. See section
19930 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019931
19932 - "Tr" is the total time in milliseconds spent waiting for the server to send
19933 a full HTTP response, not counting data. It can be "-1" if the request was
19934 aborted before a complete response could be received. It generally matches
19935 the server's processing time for the request, though it may be altered by
19936 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020019937 "GET" requests generally indicate an overloaded server. See section 8.4
19938 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019939
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019940 - "Ta" is the time the request remained active in haproxy, which is the total
19941 time in milliseconds elapsed between the first byte of the request was
19942 received and the last byte of response was sent. It covers all possible
19943 processing except the handshake (see Th) and idle time (see Ti). There is
19944 one exception, if "option logasap" was specified, then the time counting
19945 stops at the moment the log is emitted. In this case, a '+' sign is
19946 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020019947 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019948
19949 - "status_code" is the HTTP status code returned to the client. This status
19950 is generally set by the server, but it might also be set by haproxy when
19951 the server cannot be reached or when its response is blocked by haproxy.
19952
19953 - "bytes_read" is the total number of bytes transmitted to the client when
19954 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050019955 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019956 the final one may be larger. Please note that this value is a 64-bit
19957 counter, so log analysis tools must be able to handle it without
19958 overflowing.
19959
19960 - "captured_request_cookie" is an optional "name=value" entry indicating that
19961 the client had this cookie in the request. The cookie name and its maximum
19962 length are defined by the "capture cookie" statement in the frontend
19963 configuration. The field is a single dash ('-') when the option is not
19964 set. Only one cookie may be captured, it is generally used to track session
19965 ID exchanges between a client and a server to detect session crossing
19966 between clients due to application bugs. For more details, please consult
19967 the section "Capturing HTTP headers and cookies" below.
19968
19969 - "captured_response_cookie" is an optional "name=value" entry indicating
19970 that the server has returned a cookie with its response. The cookie name
19971 and its maximum length are defined by the "capture cookie" statement in the
19972 frontend configuration. The field is a single dash ('-') when the option is
19973 not set. Only one cookie may be captured, it is generally used to track
19974 session ID exchanges between a client and a server to detect session
19975 crossing between clients due to application bugs. For more details, please
19976 consult the section "Capturing HTTP headers and cookies" below.
19977
19978 - "termination_state" is the condition the session was in when the session
19979 ended. This indicates the session state, which side caused the end of
19980 session to happen, for what reason (timeout, error, ...), just like in TCP
19981 logs, and information about persistence operations on cookies in the last
19982 two characters. The normal flags should begin with "--", indicating the
19983 session was closed by either end with no data remaining in buffers. See
19984 below "Session state at disconnection" for more details.
19985
19986 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019987 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019988 limits have been reached. For instance, if actconn is close to 512 or 1024
19989 when multiple connection errors occur, chances are high that the system
19990 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019991 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019992 system.
19993
19994 - "feconn" is the total number of concurrent connections on the frontend when
19995 the session was logged. It is useful to estimate the amount of resource
19996 required to sustain high loads, and to detect when the frontend's "maxconn"
19997 has been reached. Most often when this value increases by huge jumps, it is
19998 because there is congestion on the backend servers, but sometimes it can be
19999 caused by a denial of service attack.
20000
20001 - "beconn" is the total number of concurrent connections handled by the
20002 backend when the session was logged. It includes the total number of
20003 concurrent connections active on servers as well as the number of
20004 connections pending in queues. It is useful to estimate the amount of
20005 additional servers needed to support high loads for a given application.
20006 Most often when this value increases by huge jumps, it is because there is
20007 congestion on the backend servers, but sometimes it can be caused by a
20008 denial of service attack.
20009
20010 - "srv_conn" is the total number of concurrent connections still active on
20011 the server when the session was logged. It can never exceed the server's
20012 configured "maxconn" parameter. If this value is very often close or equal
20013 to the server's "maxconn", it means that traffic regulation is involved a
20014 lot, meaning that either the server's maxconn value is too low, or that
20015 there aren't enough servers to process the load with an optimal response
20016 time. When only one of the server's "srv_conn" is high, it usually means
20017 that this server has some trouble causing the requests to take longer to be
20018 processed than on other servers.
20019
20020 - "retries" is the number of connection retries experienced by this session
20021 when trying to connect to the server. It must normally be zero, unless a
20022 server is being stopped at the same moment the connection was attempted.
20023 Frequent retries generally indicate either a network problem between
20024 haproxy and the server, or a misconfigured system backlog on the server
20025 preventing new connections from being queued. This field may optionally be
20026 prefixed with a '+' sign, indicating that the session has experienced a
20027 redispatch after the maximal retry count has been reached on the initial
20028 server. In this case, the server name appearing in the log is the one the
20029 connection was redispatched to, and not the first one, though both may
20030 sometimes be the same in case of hashing for instance. So as a general rule
20031 of thumb, when a '+' is present in front of the retry count, this count
20032 should not be attributed to the logged server.
20033
20034 - "srv_queue" is the total number of requests which were processed before
20035 this one in the server queue. It is zero when the request has not gone
20036 through the server queue. It makes it possible to estimate the approximate
20037 server's response time by dividing the time spent in queue by the number of
20038 requests in the queue. It is worth noting that if a session experiences a
20039 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010020040 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020041 backend queue unless a redispatch occurs.
20042
20043 - "backend_queue" is the total number of requests which were processed before
20044 this one in the backend's global queue. It is zero when the request has not
20045 gone through the global queue. It makes it possible to estimate the average
20046 queue length, which easily translates into a number of missing servers when
20047 divided by a server's "maxconn" parameter. It is worth noting that if a
20048 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010020049 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020050 through both the server queue and the backend queue unless a redispatch
20051 occurs.
20052
20053 - "captured_request_headers" is a list of headers captured in the request due
20054 to the presence of the "capture request header" statement in the frontend.
20055 Multiple headers can be captured, they will be delimited by a vertical bar
20056 ('|'). When no capture is enabled, the braces do not appear, causing a
20057 shift of remaining fields. It is important to note that this field may
20058 contain spaces, and that using it requires a smarter log parser than when
20059 it's not used. Please consult the section "Capturing HTTP headers and
20060 cookies" below for more details.
20061
20062 - "captured_response_headers" is a list of headers captured in the response
20063 due to the presence of the "capture response header" statement in the
20064 frontend. Multiple headers can be captured, they will be delimited by a
20065 vertical bar ('|'). When no capture is enabled, the braces do not appear,
20066 causing a shift of remaining fields. It is important to note that this
20067 field may contain spaces, and that using it requires a smarter log parser
20068 than when it's not used. Please consult the section "Capturing HTTP headers
20069 and cookies" below for more details.
20070
20071 - "http_request" is the complete HTTP request line, including the method,
20072 request and HTTP version string. Non-printable characters are encoded (see
20073 below the section "Non-printable characters"). This is always the last
20074 field, and it is always delimited by quotes and is the only one which can
20075 contain quotes. If new fields are added to the log format, they will be
20076 added before this field. This field might be truncated if the request is
20077 huge and does not fit in the standard syslog buffer (1024 characters). This
20078 is the reason why this field must always remain the last one.
20079
20080
Cyril Bontédc4d9032012-04-08 21:57:39 +0200200818.2.4. Custom log format
20082------------------------
William Lallemand48940402012-01-30 16:47:22 +010020083
Willy Tarreau2beef582012-12-20 17:22:52 +010020084The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010020085mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010020086
Davor Ocelice9ed2812017-12-25 17:49:28 +010020087HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010020088Variables can take arguments using braces ('{}'), and multiple arguments are
20089separated by commas within the braces. Flags may be added or removed by
20090prefixing them with a '+' or '-' sign.
20091
20092Special variable "%o" may be used to propagate its flags to all other
20093variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010020094("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010020095
Willy Tarreauc8368452012-12-21 00:09:23 +010020096If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020020097as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010020098less common information such as the client's SSL certificate's DN, or to log
20099the key that would be used to store an entry into a stick table.
20100
Dragan Dosen1e3b16f2020-06-23 18:16:44 +020020101Note: spaces must be escaped. In configuration directives "log-format",
20102"log-format-sd" and "unique-id-format", spaces are considered as
20103delimiters and are merged. In order to emit a verbatim '%', it must be
20104preceded by another '%' resulting in '%%'.
William Lallemand48940402012-01-30 16:47:22 +010020105
Dragan Dosen835b9212016-02-12 13:23:03 +010020106Note: when using the RFC5424 syslog message format, the characters '"',
20107'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
20108https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
20109such cases, the use of the flag "E" should be considered.
20110
William Lallemand48940402012-01-30 16:47:22 +010020111Flags are :
20112 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020113 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010020114 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
20115 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010020116
20117 Example:
20118
20119 log-format %T\ %t\ Some\ Text
20120 log-format %{+Q}o\ %t\ %s\ %{-Q}r
20121
Dragan Dosen835b9212016-02-12 13:23:03 +010020122 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
20123
William Lallemand48940402012-01-30 16:47:22 +010020124At the moment, the default HTTP format is defined this way :
20125
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020126 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
20127 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010020128
William Lallemandbddd4fd2012-02-27 11:23:10 +010020129the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010020130
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020131 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
20132 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
20133 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010020134
William Lallemandbddd4fd2012-02-27 11:23:10 +010020135and the default TCP format is defined this way :
20136
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020137 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
20138 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010020139
William Lallemand48940402012-01-30 16:47:22 +010020140Please refer to the table below for currently defined variables :
20141
William Lallemandbddd4fd2012-02-27 11:23:10 +010020142 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020143 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020144 +---+------+-----------------------------------------------+-------------+
20145 | | %o | special variable, apply flags on all next var | |
20146 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010020147 | | %B | bytes_read (from server to client) | numeric |
20148 | H | %CC | captured_request_cookie | string |
20149 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020020150 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000020151 | H | %HM | HTTP method (ex: POST) | string |
Maciej Zdeb21acc332020-11-26 10:45:52 +000020152 | H | %HP | HTTP request URI without query string | string |
Maciej Zdebfcdfd852020-11-30 18:27:47 +000020153 | H | %HPO | HTTP path only (without host nor query string)| string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000020154 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000020155 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
20156 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010020157 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020020158 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020020159 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020160 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020161 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020020162 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080020163 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020164 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
20165 | H | %Ti | idle time before the HTTP request | numeric |
20166 | H | %Tq | Th + Ti + TR | numeric |
20167 | H | %TR | time to receive the full request from 1st byte| numeric |
20168 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020020169 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020170 | | %Tt | Tt | numeric |
Damien Claisse57c8eb92020-04-28 12:09:19 +000020171 | | %Tu | Tu | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020172 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010020173 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020174 | | %ac | actconn | numeric |
20175 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020176 | | %bc | beconn (backend concurrent connections) | numeric |
20177 | | %bi | backend_source_ip (connecting address) | IP |
20178 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020179 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010020180 | | %ci | client_ip (accepted address) | IP |
20181 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020182 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020183 | | %fc | feconn (frontend concurrent connections) | numeric |
20184 | | %fi | frontend_ip (accepting address) | IP |
20185 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020020186 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020020187 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020020188 | | %hr | captured_request_headers default style | string |
20189 | | %hrl | captured_request_headers CLF style | string list |
20190 | | %hs | captured_response_headers default style | string |
20191 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020020192 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020020193 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020194 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020195 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010020196 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020197 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020198 | | %sc | srv_conn (server concurrent connections) | numeric |
20199 | | %si | server_IP (target address) | IP |
20200 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020201 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020202 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
20203 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010020204 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020205 | H | %tr | date_time of HTTP request | date |
20206 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020020207 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020208 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020209 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010020210 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010020211
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020020212 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010020213
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010020214
202158.2.5. Error log format
20216-----------------------
20217
20218When an incoming connection fails due to an SSL handshake or an invalid PROXY
20219protocol header, haproxy will log the event using a shorter, fixed line format.
20220By default, logs are emitted at the LOG_INFO level, unless the option
20221"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010020222will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010020223logged if the "dontlognull" option is set.
20224
20225The format looks like this :
20226
20227 >>> Dec 3 18:27:14 localhost \
20228 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
20229 Connection error during SSL handshake
20230
20231 Field Format Extract from the example above
20232 1 process_name '[' pid ']:' haproxy[6103]:
20233 2 client_ip ':' client_port 127.0.0.1:56059
20234 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
20235 4 frontend_name "/" bind_name ":" frt/f1:
20236 5 message Connection error during SSL handshake
20237
20238These fields just provide minimal information to help debugging connection
20239failures.
20240
20241
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200202428.3. Advanced logging options
20243-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020244
20245Some advanced logging options are often looked for but are not easy to find out
20246just by looking at the various options. Here is an entry point for the few
20247options which can enable better logging. Please refer to the keywords reference
20248for more information about their usage.
20249
20250
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200202518.3.1. Disabling logging of external tests
20252------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020253
20254It is quite common to have some monitoring tools perform health checks on
20255haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
20256commercial load-balancer, and sometimes it will simply be a more complete
20257monitoring system such as Nagios. When the tests are very frequent, users often
20258ask how to disable logging for those checks. There are three possibilities :
20259
20260 - if connections come from everywhere and are just TCP probes, it is often
20261 desired to simply disable logging of connections without data exchange, by
20262 setting "option dontlognull" in the frontend. It also disables logging of
20263 port scans, which may or may not be desired.
20264
Willy Tarreau9e9919d2020-10-14 15:55:23 +020020265 - it is possible to use the "http-request set-log-level silent" action using
20266 a variety of conditions (source networks, paths, user-agents, etc).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020267
20268 - if the tests are performed on a known URI, use "monitor-uri" to declare
20269 this URI as dedicated to monitoring. Any host sending this request will
20270 only get the result of a health-check, and the request will not be logged.
20271
20272
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200202738.3.2. Logging before waiting for the session to terminate
20274----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020275
20276The problem with logging at end of connection is that you have no clue about
20277what is happening during very long sessions, such as remote terminal sessions
20278or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010020279"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020280just before data transfer begins. This means that in case of TCP, it will still
20281log the connection status to the server, and in case of HTTP, it will log just
20282after processing the server headers. In this case, the number of bytes reported
20283is the number of header bytes sent to the client. In order to avoid confusion
20284with normal logs, the total time field and the number of bytes are prefixed
20285with a '+' sign which means that real numbers are certainly larger.
20286
20287
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200202888.3.3. Raising log level upon errors
20289------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020020290
20291Sometimes it is more convenient to separate normal traffic from errors logs,
20292for instance in order to ease error monitoring from log files. When the option
20293"log-separate-errors" is used, connections which experience errors, timeouts,
20294retries, redispatches or HTTP status codes 5xx will see their syslog level
20295raised from "info" to "err". This will help a syslog daemon store the log in
20296a separate file. It is very important to keep the errors in the normal traffic
20297file too, so that log ordering is not altered. You should also be careful if
20298you already have configured your syslog daemon to store all logs higher than
20299"notice" in an "admin" file, because the "err" level is higher than "notice".
20300
20301
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200203028.3.4. Disabling logging of successful connections
20303--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020020304
20305Although this may sound strange at first, some large sites have to deal with
20306multiple thousands of logs per second and are experiencing difficulties keeping
20307them intact for a long time or detecting errors within them. If the option
20308"dontlog-normal" is set on the frontend, all normal connections will not be
20309logged. In this regard, a normal connection is defined as one without any
20310error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
20311and a response with a status 5xx is not considered normal and will be logged
20312too. Of course, doing is is really discouraged as it will remove most of the
20313useful information from the logs. Do this only if you have no other
20314alternative.
20315
20316
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200203178.4. Timing events
20318------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020319
20320Timers provide a great help in troubleshooting network problems. All values are
20321reported in milliseconds (ms). These timers should be used in conjunction with
20322the session termination flags. In TCP mode with "option tcplog" set on the
20323frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020324mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
20325addition, three other measures are provided, "Th", "Ti", and "Tq".
20326
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010020327Timings events in HTTP mode:
20328
20329 first request 2nd request
20330 |<-------------------------------->|<-------------- ...
20331 t tr t tr ...
20332 ---|----|----|----|----|----|----|----|----|--
20333 : Th Ti TR Tw Tc Tr Td : Ti ...
20334 :<---- Tq ---->: :
20335 :<-------------- Tt -------------->:
Damien Claisse57c8eb92020-04-28 12:09:19 +000020336 :<-- -----Tu--------------->:
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010020337 :<--------- Ta --------->:
20338
20339Timings events in TCP mode:
20340
20341 TCP session
20342 |<----------------->|
20343 t t
20344 ---|----|----|----|----|---
20345 | Th Tw Tc Td |
20346 |<------ Tt ------->|
20347
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020348 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010020349 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020350 only happen once during the whole connection's lifetime. A large time here
20351 may indicate that the client only pre-established the connection without
20352 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010020353 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020020354 SSL handshake was very expensive to compute. Please note that this time is
20355 reported only before the first request, so it is safe to average it over
20356 all request to calculate the amortized value. The second and subsequent
20357 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020358
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020359 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
20360 counts between the end of the handshakes and the first byte of the HTTP
20361 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020020362 to count after the end of the transmission the previous response. When a
20363 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
20364 after the previous request. Some browsers pre-establish connections to a
20365 server in order to reduce the latency of a future request, and keep them
20366 pending until they need it. This delay will be reported as the idle time. A
20367 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020368
20369 - TR: total time to get the client request (HTTP mode only). It's the time
20370 elapsed between the first bytes received and the moment the proxy received
20371 the empty line marking the end of the HTTP headers. The value "-1"
20372 indicates that the end of headers has never been seen. This happens when
20373 the client closes prematurely or times out. This time is usually very short
20374 since most requests fit in a single packet. A large time may indicate a
20375 request typed by hand during a test.
20376
20377 - Tq: total time to get the client request from the accept date or since the
20378 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010020379 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 +020020380 returns -1 as well. This timer used to be very useful before the arrival of
20381 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
20382 it in favor of TR nowadays, as the idle time adds a lot of noise to the
20383 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020384
20385 - Tw: total time spent in the queues waiting for a connection slot. It
20386 accounts for backend queue as well as the server queues, and depends on the
20387 queue size, and the time needed for the server to complete previous
20388 requests. The value "-1" means that the request was killed before reaching
20389 the queue, which is generally what happens with invalid or denied requests.
20390
20391 - Tc: total time to establish the TCP connection to the server. It's the time
20392 elapsed between the moment the proxy sent the connection request, and the
20393 moment it was acknowledged by the server, or between the TCP SYN packet and
20394 the matching SYN/ACK packet in return. The value "-1" means that the
20395 connection never established.
20396
20397 - Tr: server response time (HTTP mode only). It's the time elapsed between
20398 the moment the TCP connection was established to the server and the moment
20399 the server sent its complete response headers. It purely shows its request
20400 processing time, without the network overhead due to the data transmission.
20401 It is worth noting that when the client has data to send to the server, for
20402 instance during a POST request, the time already runs, and this can distort
20403 apparent response time. For this reason, it's generally wise not to trust
20404 too much this field for POST requests initiated from clients behind an
20405 untrusted network. A value of "-1" here means that the last the response
20406 header (empty line) was never seen, most likely because the server timeout
20407 stroke before the server managed to process the request.
20408
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020409 - Ta: total active time for the HTTP request, between the moment the proxy
20410 received the first byte of the request header and the emission of the last
20411 byte of the response body. The exception is when the "logasap" option is
20412 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
20413 a '+' sign. From this field, we can deduce "Td", the data transmission time,
20414 by subtracting other timers when valid :
20415
20416 Td = Ta - (TR + Tw + Tc + Tr)
20417
20418 Timers with "-1" values have to be excluded from this equation. Note that
20419 "Ta" can never be negative.
20420
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020421 - Tt: total session duration time, between the moment the proxy accepted it
20422 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020423 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
20424 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020425 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020426
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020427 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020428
20429 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020430 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
20431 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020432
Damien Claisse57c8eb92020-04-28 12:09:19 +000020433 - Tu: total estimated time as seen from client, between the moment the proxy
20434 accepted it and the moment both ends were closed, without idle time.
20435 This is useful to roughly measure end-to-end time as a user would see it,
20436 without idle time pollution from keep-alive time between requests. This
20437 timer in only an estimation of time seen by user as it assumes network
20438 latency is the same in both directions. The exception is when the "logasap"
20439 option is specified. In this case, it only equals (Th+TR+Tw+Tc+Tr), and is
20440 prefixed with a '+' sign.
20441
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020442These timers provide precious indications on trouble causes. Since the TCP
20443protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
20444that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020445due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
20446"Tt" is close to a timeout value specified in the configuration, it often means
20447that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020448
20449Most common cases :
20450
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020451 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
20452 the client and the proxy. This is very rare on local networks but might
20453 happen when clients are on far remote networks and send large requests. It
20454 may happen that values larger than usual appear here without any network
20455 cause. Sometimes, during an attack or just after a resource starvation has
20456 ended, haproxy may accept thousands of connections in a few milliseconds.
20457 The time spent accepting these connections will inevitably slightly delay
20458 processing of other connections, and it can happen that request times in the
20459 order of a few tens of milliseconds are measured after a few thousands of
20460 new connections have been accepted at once. Using one of the keep-alive
20461 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020020462 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020463
20464 - If "Tc" is close to 3000, a packet has probably been lost between the
20465 server and the proxy during the server connection phase. This value should
20466 always be very low, such as 1 ms on local networks and less than a few tens
20467 of ms on remote networks.
20468
Willy Tarreau55165fe2009-05-10 12:02:55 +020020469 - If "Tr" is nearly always lower than 3000 except some rare values which seem
20470 to be the average majored by 3000, there are probably some packets lost
20471 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020472
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020473 - If "Ta" is large even for small byte counts, it generally is because
20474 neither the client nor the server decides to close the connection while
20475 haproxy is running in tunnel mode and both have agreed on a keep-alive
20476 connection mode. In order to solve this issue, it will be needed to specify
20477 one of the HTTP options to manipulate keep-alive or close options on either
20478 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
20479 important when connection regulation is used with the "maxconn" option on
20480 the servers, since no new connection will be sent to the server until
20481 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020482
20483Other noticeable HTTP log cases ('xx' means any value to be ignored) :
20484
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020485 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020486 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020487 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020488
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020489 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020490 or it aborted too early. Check the session termination flags
20491 then "timeout http-request" and "timeout client" settings.
20492
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020493 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020494 servers were out of order, because the request was invalid
20495 or forbidden by ACL rules. Check the session termination
20496 flags.
20497
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020498 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
20499 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020500 Check the session termination flags, then check the
20501 "timeout connect" setting. Note that the tarpit action might
20502 return similar-looking patterns, with "Tw" equal to the time
20503 the client connection was maintained open.
20504
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020505 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020506 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020507 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020508 termination flags, then check the "timeout server" setting.
20509
20510
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200205118.5. Session state at disconnection
20512-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020513
20514TCP and HTTP logs provide a session termination indicator in the
20515"termination_state" field, just before the number of active connections. It is
205162-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
20517each of which has a special meaning :
20518
20519 - On the first character, a code reporting the first event which caused the
20520 session to terminate :
20521
20522 C : the TCP session was unexpectedly aborted by the client.
20523
20524 S : the TCP session was unexpectedly aborted by the server, or the
20525 server explicitly refused it.
20526
20527 P : the session was prematurely aborted by the proxy, because of a
20528 connection limit enforcement, because a DENY filter was matched,
20529 because of a security check which detected and blocked a dangerous
20530 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010020531 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020020532
20533 L : the session was locally processed by haproxy and was not passed to
20534 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020535
20536 R : a resource on the proxy has been exhausted (memory, sockets, source
20537 ports, ...). Usually, this appears during the connection phase, and
20538 system logs should contain a copy of the precise error. If this
20539 happens, it must be considered as a very serious anomaly which
20540 should be fixed as soon as possible by any means.
20541
20542 I : an internal error was identified by the proxy during a self-check.
20543 This should NEVER happen, and you are encouraged to report any log
20544 containing this, because this would almost certainly be a bug. It
20545 would be wise to preventively restart the process after such an
20546 event too, in case it would be caused by memory corruption.
20547
Simon Horman752dc4a2011-06-21 14:34:59 +090020548 D : the session was killed by haproxy because the server was detected
20549 as down and was configured to kill all connections when going down.
20550
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070020551 U : the session was killed by haproxy on this backup server because an
20552 active server was detected as up and was configured to kill all
20553 backup connections when going up.
20554
Willy Tarreaua2a64e92011-09-07 23:01:56 +020020555 K : the session was actively killed by an admin operating on haproxy.
20556
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020557 c : the client-side timeout expired while waiting for the client to
20558 send or receive data.
20559
20560 s : the server-side timeout expired while waiting for the server to
20561 send or receive data.
20562
20563 - : normal session completion, both the client and the server closed
20564 with nothing left in the buffers.
20565
20566 - on the second character, the TCP or HTTP session state when it was closed :
20567
Willy Tarreauf7b30a92010-12-06 22:59:17 +010020568 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020569 (HTTP mode only). Nothing was sent to any server.
20570
20571 Q : the proxy was waiting in the QUEUE for a connection slot. This can
20572 only happen when servers have a 'maxconn' parameter set. It can
20573 also happen in the global queue after a redispatch consecutive to
20574 a failed attempt to connect to a dying server. If no redispatch is
20575 reported, then no connection attempt was made to any server.
20576
20577 C : the proxy was waiting for the CONNECTION to establish on the
20578 server. The server might at most have noticed a connection attempt.
20579
20580 H : the proxy was waiting for complete, valid response HEADERS from the
20581 server (HTTP only).
20582
20583 D : the session was in the DATA phase.
20584
20585 L : the proxy was still transmitting LAST data to the client while the
20586 server had already finished. This one is very rare as it can only
20587 happen when the client dies while receiving the last packets.
20588
20589 T : the request was tarpitted. It has been held open with the client
20590 during the whole "timeout tarpit" duration or until the client
20591 closed, both of which will be reported in the "Tw" timer.
20592
20593 - : normal session completion after end of data transfer.
20594
20595 - the third character tells whether the persistence cookie was provided by
20596 the client (only in HTTP mode) :
20597
20598 N : the client provided NO cookie. This is usually the case for new
20599 visitors, so counting the number of occurrences of this flag in the
20600 logs generally indicate a valid trend for the site frequentation.
20601
20602 I : the client provided an INVALID cookie matching no known server.
20603 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020020604 cookies between HTTP/HTTPS sites, persistence conditionally
20605 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020606
20607 D : the client provided a cookie designating a server which was DOWN,
20608 so either "option persist" was used and the client was sent to
20609 this server, or it was not set and the client was redispatched to
20610 another server.
20611
Willy Tarreau996a92c2010-10-13 19:30:47 +020020612 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020613 server.
20614
Willy Tarreau996a92c2010-10-13 19:30:47 +020020615 E : the client provided a valid cookie, but with a last date which was
20616 older than what is allowed by the "maxidle" cookie parameter, so
20617 the cookie is consider EXPIRED and is ignored. The request will be
20618 redispatched just as if there was no cookie.
20619
20620 O : the client provided a valid cookie, but with a first date which was
20621 older than what is allowed by the "maxlife" cookie parameter, so
20622 the cookie is consider too OLD and is ignored. The request will be
20623 redispatched just as if there was no cookie.
20624
Willy Tarreauc89ccb62012-04-05 21:18:22 +020020625 U : a cookie was present but was not used to select the server because
20626 some other server selection mechanism was used instead (typically a
20627 "use-server" rule).
20628
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020629 - : does not apply (no cookie set in configuration).
20630
20631 - the last character reports what operations were performed on the persistence
20632 cookie returned by the server (only in HTTP mode) :
20633
20634 N : NO cookie was provided by the server, and none was inserted either.
20635
20636 I : no cookie was provided by the server, and the proxy INSERTED one.
20637 Note that in "cookie insert" mode, if the server provides a cookie,
20638 it will still be overwritten and reported as "I" here.
20639
Willy Tarreau996a92c2010-10-13 19:30:47 +020020640 U : the proxy UPDATED the last date in the cookie that was presented by
20641 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020642 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020020643 date indicated in the cookie. If any other change happens, such as
20644 a redispatch, then the cookie will be marked as inserted instead.
20645
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020646 P : a cookie was PROVIDED by the server and transmitted as-is.
20647
20648 R : the cookie provided by the server was REWRITTEN by the proxy, which
20649 happens in "cookie rewrite" or "cookie prefix" modes.
20650
20651 D : the cookie provided by the server was DELETED by the proxy.
20652
20653 - : does not apply (no cookie set in configuration).
20654
Willy Tarreau996a92c2010-10-13 19:30:47 +020020655The combination of the two first flags gives a lot of information about what
20656was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020657helpful to detect server saturation, network troubles, local system resource
20658starvation, attacks, etc...
20659
20660The most common termination flags combinations are indicated below. They are
20661alphabetically sorted, with the lowercase set just after the upper case for
20662easier finding and understanding.
20663
20664 Flags Reason
20665
20666 -- Normal termination.
20667
20668 CC The client aborted before the connection could be established to the
20669 server. This can happen when haproxy tries to connect to a recently
20670 dead (or unchecked) server, and the client aborts while haproxy is
20671 waiting for the server to respond or for "timeout connect" to expire.
20672
20673 CD The client unexpectedly aborted during data transfer. This can be
20674 caused by a browser crash, by an intermediate equipment between the
20675 client and haproxy which decided to actively break the connection,
20676 by network routing issues between the client and haproxy, or by a
20677 keep-alive session between the server and the client terminated first
20678 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010020679
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020680 cD The client did not send nor acknowledge any data for as long as the
20681 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020020682 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020683
20684 CH The client aborted while waiting for the server to start responding.
20685 It might be the server taking too long to respond or the client
20686 clicking the 'Stop' button too fast.
20687
20688 cH The "timeout client" stroke while waiting for client data during a
20689 POST request. This is sometimes caused by too large TCP MSS values
20690 for PPPoE networks which cannot transport full-sized packets. It can
20691 also happen when client timeout is smaller than server timeout and
20692 the server takes too long to respond.
20693
20694 CQ The client aborted while its session was queued, waiting for a server
20695 with enough empty slots to accept it. It might be that either all the
20696 servers were saturated or that the assigned server was taking too
20697 long a time to respond.
20698
20699 CR The client aborted before sending a full HTTP request. Most likely
20700 the request was typed by hand using a telnet client, and aborted
20701 too early. The HTTP status code is likely a 400 here. Sometimes this
20702 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020020703 and the client. "option http-ignore-probes" can be used to ignore
20704 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020705
20706 cR The "timeout http-request" stroke before the client sent a full HTTP
20707 request. This is sometimes caused by too large TCP MSS values on the
20708 client side for PPPoE networks which cannot transport full-sized
20709 packets, or by clients sending requests by hand and not typing fast
20710 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020020711 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020020712 some browsers started to implement a "pre-connect" feature consisting
20713 in speculatively connecting to some recently visited web sites just
20714 in case the user would like to visit them. This results in many
20715 connections being established to web sites, which end up in 408
20716 Request Timeout if the timeout strikes first, or 400 Bad Request when
20717 the browser decides to close them first. These ones pollute the log
20718 and feed the error counters. Some versions of some browsers have even
20719 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010020720 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020020721 http-ignore-probes" in the frontend, resulting in connections with
20722 zero data transfer to be totally ignored. This will definitely hide
20723 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020724
20725 CT The client aborted while its session was tarpitted. It is important to
20726 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020020727 wrong tarpit rules have been written. If a lot of them happen, it
20728 might make sense to lower the "timeout tarpit" value to something
20729 closer to the average reported "Tw" timer, in order not to consume
20730 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020731
Willy Tarreau570f2212013-06-10 16:42:09 +020020732 LR The request was intercepted and locally handled by haproxy. Generally
20733 it means that this was a redirect or a stats request.
20734
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020735 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020736 the TCP connection (the proxy received a TCP RST or an ICMP message
20737 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010020738 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020739 or no ARP response on local network). When this happens in HTTP mode,
20740 the status code is likely a 502 or 503 here.
20741
20742 sC The "timeout connect" stroke before a connection to the server could
20743 complete. When this happens in HTTP mode, the status code is likely a
20744 503 or 504 here.
20745
20746 SD The connection to the server died with an error during the data
20747 transfer. This usually means that haproxy has received an RST from
20748 the server or an ICMP message from an intermediate equipment while
20749 exchanging data with the server. This can be caused by a server crash
20750 or by a network issue on an intermediate equipment.
20751
20752 sD The server did not send nor acknowledge any data for as long as the
20753 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010020754 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020755 load-balancers, ...), as well as keep-alive sessions maintained
20756 between the client and the server expiring first on haproxy.
20757
20758 SH The server aborted before sending its full HTTP response headers, or
20759 it crashed while processing the request. Since a server aborting at
20760 this moment is very rare, it would be wise to inspect its logs to
20761 control whether it crashed and why. The logged request may indicate a
20762 small set of faulty requests, demonstrating bugs in the application.
20763 Sometimes this might also be caused by an IDS killing the connection
20764 between haproxy and the server.
20765
20766 sH The "timeout server" stroke before the server could return its
20767 response headers. This is the most common anomaly, indicating too
20768 long transactions, probably caused by server or database saturation.
20769 The immediate workaround consists in increasing the "timeout server"
20770 setting, but it is important to keep in mind that the user experience
20771 will suffer from these long response times. The only long term
20772 solution is to fix the application.
20773
20774 sQ The session spent too much time in queue and has been expired. See
20775 the "timeout queue" and "timeout connect" settings to find out how to
20776 fix this if it happens too often. If it often happens massively in
20777 short periods, it may indicate general problems on the affected
20778 servers due to I/O or database congestion, or saturation caused by
20779 external attacks.
20780
20781 PC The proxy refused to establish a connection to the server because the
Thayne McCombscdbcca92021-01-07 21:24:41 -070020782 process's socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020020783 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020784 so that it does not happen anymore. This status is very rare and
20785 might happen when the global "ulimit-n" parameter is forced by hand.
20786
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020787 PD The proxy blocked an incorrectly formatted chunked encoded message in
20788 a request or a response, after the server has emitted its headers. In
20789 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010020790 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020020791 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020792
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020793 PH The proxy blocked the server's response, because it was invalid,
20794 incomplete, dangerous (cache control), or matched a security filter.
20795 In any case, an HTTP 502 error is sent to the client. One possible
20796 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020797 containing unauthorized characters. It is also possible but quite
20798 rare, that the proxy blocked a chunked-encoding request from the
20799 client due to an invalid syntax, before the server responded. In this
20800 case, an HTTP 400 error is sent to the client and reported in the
20801 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020802
20803 PR The proxy blocked the client's HTTP request, either because of an
20804 invalid HTTP syntax, in which case it returned an HTTP 400 error to
20805 the client, or because a deny filter matched, in which case it
20806 returned an HTTP 403 error.
20807
20808 PT The proxy blocked the client's request and has tarpitted its
20809 connection before returning it a 500 server error. Nothing was sent
20810 to the server. The connection was maintained open for as long as
20811 reported by the "Tw" timer field.
20812
20813 RC A local resource has been exhausted (memory, sockets, source ports)
20814 preventing the connection to the server from establishing. The error
20815 logs will tell precisely what was missing. This is very rare and can
20816 only be solved by proper system tuning.
20817
Willy Tarreau996a92c2010-10-13 19:30:47 +020020818The combination of the two last flags gives a lot of information about how
20819persistence was handled by the client, the server and by haproxy. This is very
20820important to troubleshoot disconnections, when users complain they have to
20821re-authenticate. The commonly encountered flags are :
20822
20823 -- Persistence cookie is not enabled.
20824
20825 NN No cookie was provided by the client, none was inserted in the
20826 response. For instance, this can be in insert mode with "postonly"
20827 set on a GET request.
20828
20829 II A cookie designating an invalid server was provided by the client,
20830 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020831 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020020832 value can be presented by a client when no other server knows it.
20833
20834 NI No cookie was provided by the client, one was inserted in the
20835 response. This typically happens for first requests from every user
20836 in "insert" mode, which makes it an easy way to count real users.
20837
20838 VN A cookie was provided by the client, none was inserted in the
20839 response. This happens for most responses for which the client has
20840 already got a cookie.
20841
20842 VU A cookie was provided by the client, with a last visit date which is
20843 not completely up-to-date, so an updated cookie was provided in
20844 response. This can also happen if there was no date at all, or if
20845 there was a date but the "maxidle" parameter was not set, so that the
20846 cookie can be switched to unlimited time.
20847
20848 EI A cookie was provided by the client, with a last visit date which is
20849 too old for the "maxidle" parameter, so the cookie was ignored and a
20850 new cookie was inserted in the response.
20851
20852 OI A cookie was provided by the client, with a first visit date which is
20853 too old for the "maxlife" parameter, so the cookie was ignored and a
20854 new cookie was inserted in the response.
20855
20856 DI The server designated by the cookie was down, a new server was
20857 selected and a new cookie was emitted in the response.
20858
20859 VI The server designated by the cookie was not marked dead but could not
20860 be reached. A redispatch happened and selected another one, which was
20861 then advertised in the response.
20862
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020863
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200208648.6. Non-printable characters
20865-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020866
20867In order not to cause trouble to log analysis tools or terminals during log
20868consulting, non-printable characters are not sent as-is into log files, but are
20869converted to the two-digits hexadecimal representation of their ASCII code,
20870prefixed by the character '#'. The only characters that can be logged without
20871being escaped are comprised between 32 and 126 (inclusive). Obviously, the
20872escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
20873is the same for the character '"' which becomes "#22", as well as '{', '|' and
20874'}' when logging headers.
20875
20876Note that the space character (' ') is not encoded in headers, which can cause
20877issues for tools relying on space count to locate fields. A typical header
20878containing spaces is "User-Agent".
20879
20880Last, it has been observed that some syslog daemons such as syslog-ng escape
20881the quote ('"') with a backslash ('\'). The reverse operation can safely be
20882performed since no quote may appear anywhere else in the logs.
20883
20884
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200208858.7. Capturing HTTP cookies
20886---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020887
20888Cookie capture simplifies the tracking a complete user session. This can be
20889achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020890section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020891cookie will simultaneously be checked in the request ("Cookie:" header) and in
20892the response ("Set-Cookie:" header). The respective values will be reported in
20893the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020894locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020895not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
20896user switches to a new session for example, because the server will reassign it
20897a new cookie. It is also possible to detect if a server unexpectedly sets a
20898wrong cookie to a client, leading to session crossing.
20899
20900 Examples :
20901 # capture the first cookie whose name starts with "ASPSESSION"
20902 capture cookie ASPSESSION len 32
20903
20904 # capture the first cookie whose name is exactly "vgnvisitor"
20905 capture cookie vgnvisitor= len 32
20906
20907
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200209088.8. Capturing HTTP headers
20909---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020910
20911Header captures are useful to track unique request identifiers set by an upper
20912proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
20913the response, one can search for information about the response length, how the
20914server asked the cache to behave, or an object location during a redirection.
20915
20916Header captures are performed using the "capture request header" and "capture
20917response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020918section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020919
20920It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020921time. Non-existent headers are logged as empty strings, and if one header
20922appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020923are grouped within braces '{' and '}' in the same order as they were declared,
20924and delimited with a vertical bar '|' without any space. Response headers
20925follow the same representation, but are displayed after a space following the
20926request headers block. These blocks are displayed just before the HTTP request
20927in the logs.
20928
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020020929As a special case, it is possible to specify an HTTP header capture in a TCP
20930frontend. The purpose is to enable logging of headers which will be parsed in
20931an HTTP backend if the request is then switched to this HTTP backend.
20932
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020933 Example :
20934 # This instance chains to the outgoing proxy
20935 listen proxy-out
20936 mode http
20937 option httplog
20938 option logasap
20939 log global
20940 server cache1 192.168.1.1:3128
20941
20942 # log the name of the virtual server
20943 capture request header Host len 20
20944
20945 # log the amount of data uploaded during a POST
20946 capture request header Content-Length len 10
20947
20948 # log the beginning of the referrer
20949 capture request header Referer len 20
20950
20951 # server name (useful for outgoing proxies only)
20952 capture response header Server len 20
20953
20954 # logging the content-length is useful with "option logasap"
20955 capture response header Content-Length len 10
20956
Davor Ocelice9ed2812017-12-25 17:49:28 +010020957 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020958 capture response header Cache-Control len 8
20959
20960 # the Via header will report the next proxy's name
20961 capture response header Via len 20
20962
20963 # log the URL location during a redirection
20964 capture response header Location len 20
20965
20966 >>> Aug 9 20:26:09 localhost \
20967 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
20968 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
20969 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
20970 "GET http://fr.adserver.yahoo.com/"
20971
20972 >>> Aug 9 20:30:46 localhost \
20973 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
20974 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
20975 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020976 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020977
20978 >>> Aug 9 20:30:46 localhost \
20979 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
20980 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
20981 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
20982 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020983 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020984
20985
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200209868.9. Examples of logs
20987---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020988
20989These are real-world examples of logs accompanied with an explanation. Some of
20990them have been made up by hand. The syslog part has been removed for better
20991reading. Their sole purpose is to explain how to decipher them.
20992
20993 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
20994 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
20995 "HEAD / HTTP/1.0"
20996
20997 => long request (6.5s) entered by hand through 'telnet'. The server replied
20998 in 147 ms, and the session ended normally ('----')
20999
21000 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
21001 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
21002 0/9 "HEAD / HTTP/1.0"
21003
21004 => Idem, but the request was queued in the global queue behind 9 other
21005 requests, and waited there for 1230 ms.
21006
21007 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
21008 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
21009 "GET /image.iso HTTP/1.0"
21010
21011 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010021012 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021013 14 ms, 243 bytes of headers were sent to the client, and total time from
21014 accept to first data byte is 30 ms.
21015
21016 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
21017 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
21018 "GET /cgi-bin/bug.cgi? HTTP/1.0"
21019
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020021020 => the proxy blocked a server response either because of an "http-response
21021 deny" rule, or because the response was improperly formatted and not
21022 HTTP-compliant, or because it blocked sensitive information which risked
21023 being cached. In this case, the response is replaced with a "502 bad
21024 gateway". The flags ("PH--") tell us that it was haproxy who decided to
21025 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021026
21027 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010021028 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 +010021029
21030 => the client never completed its request and aborted itself ("C---") after
21031 8.5s, while the proxy was waiting for the request headers ("-R--").
21032 Nothing was sent to any server.
21033
21034 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
21035 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
21036
21037 => The client never completed its request, which was aborted by the
21038 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010021039 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021040 send a 408 return code to the client.
21041
21042 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
21043 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
21044
21045 => This log was produced with "option tcplog". The client timed out after
21046 5 seconds ("c----").
21047
21048 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
21049 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010021050 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021051
21052 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020021053 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010021054 (config says 'retries 3'), and no redispatch (otherwise we would have
21055 seen "/+3"). Status code 503 was returned to the client. There were 115
21056 connections on this server, 202 connections on this proxy, and 205 on
21057 the global process. It is possible that the server refused the
21058 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010021059
Willy Tarreau52b2d222011-09-07 23:48:48 +020021060
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200210619. Supported filters
21062--------------------
21063
21064Here are listed officially supported filters with the list of parameters they
21065accept. Depending on compile options, some of these filters might be
21066unavailable. The list of available filters is reported in haproxy -vv.
21067
21068See also : "filter"
21069
210709.1. Trace
21071----------
21072
Christopher Fauletc41d8bd2020-11-17 10:43:26 +010021073filter trace [name <name>] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021074
21075 Arguments:
21076 <name> is an arbitrary name that will be reported in
21077 messages. If no name is provided, "TRACE" is used.
21078
Christopher Faulet96a577a2020-11-17 10:45:05 +010021079 <quiet> inhibits trace messages.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021080
Davor Ocelice9ed2812017-12-25 17:49:28 +010021081 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021082 default, this filter forwards all previously parsed
21083 data. With this parameter, it only forwards a random
21084 amount of the parsed data.
21085
Davor Ocelice9ed2812017-12-25 17:49:28 +010021086 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010021087
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021088This filter can be used as a base to develop new filters. It defines all
21089callbacks and print a message on the standard error stream (stderr) with useful
21090information for all of them. It may be useful to debug the activity of other
21091filters or, quite simply, HAProxy's activity.
21092
21093Using <random-parsing> and/or <random-forwarding> parameters is a good way to
21094tests the behavior of a filter that parses data exchanged between a client and
21095a server by adding some latencies in the processing.
21096
21097
210989.2. HTTP compression
21099---------------------
21100
21101filter compression
21102
21103The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
21104keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020021105when no other filter is used, it is enough. When used with the cache or the
21106fcgi-app enabled, it is also enough. In this case, the compression is always
21107done after the response is stored in the cache. But it is mandatory to
21108explicitly use a filter line to enable the HTTP compression when at least one
21109filter other than the cache or the fcgi-app is used for the same
21110listener/frontend/backend. This is important to know the filters evaluation
21111order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021112
Christopher Fauletb30b3102019-09-12 23:03:09 +020021113See also : "compression", section 9.4 about the cache filter and section 9.5
21114 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020021115
21116
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200211179.3. Stream Processing Offload Engine (SPOE)
21118--------------------------------------------
21119
21120filter spoe [engine <name>] config <file>
21121
21122 Arguments :
21123
21124 <name> is the engine name that will be used to find the right scope in
21125 the configuration file. If not provided, all the file will be
21126 parsed.
21127
21128 <file> is the path of the engine configuration file. This file can
21129 contain configuration of several engines. In this case, each
21130 part must be placed in its own scope.
21131
21132The Stream Processing Offload Engine (SPOE) is a filter communicating with
21133external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010021134streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020021135exchanged with them are configured in dedicated files, for the main part. It
21136also requires dedicated backends, defined in HAProxy configuration.
21137
21138SPOE communicates with external components using an in-house binary protocol,
21139the Stream Processing Offload Protocol (SPOP).
21140
Tim Düsterhus4896c442016-11-29 02:15:19 +010021141For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020021142"doc/SPOE.txt".
21143
Christopher Faulet99a17a22018-12-11 09:18:27 +0100211449.4. Cache
21145----------
21146
21147filter cache <name>
21148
21149 Arguments :
21150
21151 <name> is name of the cache section this filter will use.
21152
21153The cache uses a filter to store cacheable responses. The HTTP rules
21154"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050021155cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020021156other filters than fcgi-app or compression are used, it is enough. In such
21157case, the compression filter is always evaluated after the cache filter. But it
21158is mandatory to explicitly use a filter line to use a cache when at least one
21159filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010021160listener/frontend/backend. This is important to know the filters evaluation
21161order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010021162
Christopher Fauletb30b3102019-09-12 23:03:09 +020021163See also : section 9.2 about the compression filter, section 9.5 about the
21164 fcgi-app filter and section 6 about cache.
21165
21166
211679.5. Fcgi-app
21168-------------
21169
Daniel Corbett67a82712020-07-06 23:01:19 -040021170filter fcgi-app <name>
Christopher Fauletb30b3102019-09-12 23:03:09 +020021171
21172 Arguments :
21173
21174 <name> is name of the fcgi-app section this filter will use.
21175
21176The FastCGI application uses a filter to evaluate all custom parameters on the
21177request path, and to process the headers on the response path. the <name> must
21178reference an existing fcgi-app section. The directive "use-fcgi-app" should be
21179used to define the application to use. By default the corresponding filter is
21180implicitly defined. And when no other filters than cache or compression are
21181used, it is enough. But it is mandatory to explicitly use a filter line to a
21182fcgi-app when at least one filter other than the compression or the cache is
21183used for the same backend. This is important to know the filters evaluation
21184order.
21185
21186See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
21187 about the cache filter and section 10 about FastCGI application.
21188
21189
Miroslav Zagoracdc32cd92020-12-13 18:32:57 +0100211909.6. OpenTracing
21191----------------
21192
21193The OpenTracing filter adds native support for using distributed tracing in
21194HAProxy. This is enabled by sending an OpenTracing compliant request to one
21195of the supported tracers such as Datadog, Jaeger, Lightstep and Zipkin tracers.
21196Please note: tracers are not listed by any preference, but alphabetically.
21197
21198This feature is only enabled when haproxy was built with USE_OT=1.
21199
21200The OpenTracing filter activation is done explicitly by specifying it in the
21201HAProxy configuration. If this is not done, the OpenTracing filter in no way
21202participates in the work of HAProxy.
21203
21204filter opentracing [id <id>] config <file>
21205
21206 Arguments :
21207
21208 <id> is the OpenTracing filter id that will be used to find the
21209 right scope in the configuration file. If no filter id is
21210 specified, 'ot-filter' is used as default. If scope is not
21211 specified in the configuration file, it applies to all defined
21212 OpenTracing filters.
21213
21214 <file> is the path of the OpenTracing configuration file. The same
21215 file can contain configurations for multiple OpenTracing
21216 filters simultaneously. In that case we do not need to define
21217 scope so the same configuration applies to all filters or each
21218 filter must have its own scope defined.
21219
21220More detailed documentation related to the operation, configuration and use
21221of the filter can be found in the contrib/opentracing directory.
21222
21223
Christopher Fauletb30b3102019-09-12 23:03:09 +02002122410. FastCGI applications
21225-------------------------
21226
21227HAProxy is able to send HTTP requests to Responder FastCGI applications. This
21228feature was added in HAProxy 2.1. To do so, servers must be configured to use
21229the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
21230FastCGI application must be configured and used by the backend managing these
21231servers (using the keyword "use-fcgi-app" into the proxy section). Several
21232FastCGI applications may be defined, but only one can be used at a time by a
21233backend.
21234
21235HAProxy implements all features of the FastCGI specification for Responder
21236application. Especially it is able to multiplex several requests on a simple
21237connection.
21238
2123910.1. Setup
21240-----------
21241
2124210.1.1. Fcgi-app section
21243--------------------------
21244
21245fcgi-app <name>
21246 Declare a FastCGI application named <name>. To be valid, at least the
21247 document root must be defined.
21248
21249acl <aclname> <criterion> [flags] [operator] <value> ...
21250 Declare or complete an access list.
21251
21252 See "acl" keyword in section 4.2 and section 7 about ACL usage for
21253 details. ACLs defined for a FastCGI application are private. They cannot be
21254 used by any other application or by any proxy. In the same way, ACLs defined
21255 in any other section are not usable by a FastCGI application. However,
21256 Pre-defined ACLs are available.
21257
21258docroot <path>
21259 Define the document root on the remote host. <path> will be used to build
21260 the default value of FastCGI parameters SCRIPT_FILENAME and
21261 PATH_TRANSLATED. It is a mandatory setting.
21262
21263index <script-name>
21264 Define the script name that will be appended after an URI that ends with a
21265 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
21266 is an optional setting.
21267
21268 Example :
21269 index index.php
21270
21271log-stderr global
21272log-stderr <address> [len <length>] [format <format>]
Jan Wagner3e678602020-12-17 22:22:32 +010021273 [sample <ranges>:<sample_size>] <facility> [<level> [<minlevel>]]
Christopher Fauletb30b3102019-09-12 23:03:09 +020021274 Enable logging of STDERR messages reported by the FastCGI application.
21275
21276 See "log" keyword in section 4.2 for details. It is an optional setting. By
21277 default STDERR messages are ignored.
21278
21279pass-header <name> [ { if | unless } <condition> ]
21280 Specify the name of a request header which will be passed to the FastCGI
21281 application. It may optionally be followed by an ACL-based condition, in
21282 which case it will only be evaluated if the condition is true.
21283
21284 Most request headers are already available to the FastCGI application,
21285 prefixed with "HTTP_". Thus, this directive is only required to pass headers
21286 that are purposefully omitted. Currently, the headers "Authorization",
21287 "Proxy-Authorization" and hop-by-hop headers are omitted.
21288
21289 Note that the headers "Content-type" and "Content-length" are never passed to
21290 the FastCGI application because they are already converted into parameters.
21291
21292path-info <regex>
Christopher Faulet28cb3662020-02-14 14:47:37 +010021293 Define a regular expression to extract the script-name and the path-info from
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010021294 the URL-decoded path. Thus, <regex> may have two captures: the first one to
21295 capture the script name and the second one to capture the path-info. The
21296 first one is mandatory, the second one is optional. This way, it is possible
21297 to extract the script-name from the path ignoring the path-info. It is an
21298 optional setting. If it is not defined, no matching is performed on the
21299 path. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not
21300 filled.
Christopher Faulet28cb3662020-02-14 14:47:37 +010021301
21302 For security reason, when this regular expression is defined, the newline and
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050021303 the null characters are forbidden from the path, once URL-decoded. The reason
Christopher Faulet28cb3662020-02-14 14:47:37 +010021304 to such limitation is because otherwise the matching always fails (due to a
21305 limitation one the way regular expression are executed in HAProxy). So if one
21306 of these two characters is found in the URL-decoded path, an error is
21307 returned to the client. The principle of least astonishment is applied here.
Christopher Fauletb30b3102019-09-12 23:03:09 +020021308
21309 Example :
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010021310 path-info ^(/.+\.php)(/.*)?$ # both script-name and path-info may be set
21311 path-info ^(/.+\.php) # the path-info is ignored
Christopher Fauletb30b3102019-09-12 23:03:09 +020021312
21313option get-values
21314no option get-values
21315 Enable or disable the retrieve of variables about connection management.
21316
Daniel Corbett67a82712020-07-06 23:01:19 -040021317 HAProxy is able to send the record FCGI_GET_VALUES on connection
Christopher Fauletb30b3102019-09-12 23:03:09 +020021318 establishment to retrieve the value for following variables:
21319
21320 * FCGI_MAX_REQS The maximum number of concurrent requests this
21321 application will accept.
21322
William Lallemand93e548e2019-09-30 13:54:02 +020021323 * FCGI_MPXS_CONNS "0" if this application does not multiplex connections,
21324 "1" otherwise.
Christopher Fauletb30b3102019-09-12 23:03:09 +020021325
21326 Some FastCGI applications does not support this feature. Some others close
Ilya Shipitsin11057a32020-06-21 21:18:27 +050021327 the connection immediately after sending their response. So, by default, this
Christopher Fauletb30b3102019-09-12 23:03:09 +020021328 option is disabled.
21329
21330 Note that the maximum number of concurrent requests accepted by a FastCGI
21331 application is a connection variable. It only limits the number of streams
21332 per connection. If the global load must be limited on the application, the
21333 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
21334 an application does not support connection multiplexing, the maximum number
21335 of concurrent requests is automatically set to 1.
21336
21337option keep-conn
21338no option keep-conn
21339 Instruct the FastCGI application to keep the connection open or not after
21340 sending a response.
21341
21342 If disabled, the FastCGI application closes the connection after responding
21343 to this request. By default, this option is enabled.
21344
21345option max-reqs <reqs>
21346 Define the maximum number of concurrent requests this application will
21347 accept.
21348
21349 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
21350 during connection establishment. Furthermore, if the application does not
21351 support connection multiplexing, this option will be ignored. By default set
21352 to 1.
21353
21354option mpxs-conns
21355no option mpxs-conns
21356 Enable or disable the support of connection multiplexing.
21357
21358 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
21359 during connection establishment. It is disabled by default.
21360
21361set-param <name> <fmt> [ { if | unless } <condition> ]
21362 Set a FastCGI parameter that should be passed to this application. Its
21363 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
21364 "Custom Log format"). It may optionally be followed by an ACL-based
21365 condition, in which case it will only be evaluated if the condition is true.
21366
21367 With this directive, it is possible to overwrite the value of default FastCGI
21368 parameters. If the value is evaluated to an empty string, the rule is
21369 ignored. These directives are evaluated in their declaration order.
21370
21371 Example :
21372 # PHP only, required if PHP was built with --enable-force-cgi-redirect
21373 set-param REDIRECT_STATUS 200
21374
21375 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
21376
21377
2137810.1.2. Proxy section
21379---------------------
21380
21381use-fcgi-app <name>
21382 Define the FastCGI application to use for the backend.
21383
21384 Arguments :
21385 <name> is the name of the FastCGI application to use.
21386
21387 This keyword is only available for HTTP proxies with the backend capability
21388 and with at least one FastCGI server. However, FastCGI servers can be mixed
21389 with HTTP servers. But except there is a good reason to do so, it is not
21390 recommended (see section 10.3 about the limitations for details). Only one
21391 application may be defined at a time per backend.
21392
21393 Note that, once a FastCGI application is referenced for a backend, depending
21394 on the configuration some processing may be done even if the request is not
21395 sent to a FastCGI server. Rules to set parameters or pass headers to an
21396 application are evaluated.
21397
21398
2139910.1.3. Example
21400---------------
21401
21402 frontend front-http
21403 mode http
21404 bind *:80
21405 bind *:
21406
21407 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
21408 default_backend back-static
21409
21410 backend back-static
21411 mode http
21412 server www A.B.C.D:80
21413
21414 backend back-dynamic
21415 mode http
21416 use-fcgi-app php-fpm
21417 server php-fpm A.B.C.D:9000 proto fcgi
21418
21419 fcgi-app php-fpm
21420 log-stderr global
21421 option keep-conn
21422
21423 docroot /var/www/my-app
21424 index index.php
21425 path-info ^(/.+\.php)(/.*)?$
21426
21427
2142810.2. Default parameters
21429------------------------
21430
21431A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
21432the CGI/1.1 specification (RFC3875), several variables must be passed to the
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050021433script. So HAProxy set them and some others commonly used by FastCGI
Christopher Fauletb30b3102019-09-12 23:03:09 +020021434applications. All these variables may be overwritten, with caution though.
21435
21436 +-------------------+-----------------------------------------------------+
21437 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
21438 | | to authenticate the user. Concretely, only the |
21439 | | BASIC authentication mechanism is supported. |
21440 | | |
21441 +-------------------+-----------------------------------------------------+
21442 | CONTENT_LENGTH | Contains the size of the message-body attached to |
21443 | | the request. It means only requests with a known |
21444 | | size are considered as valid and sent to the |
21445 | | application. |
21446 | | |
21447 +-------------------+-----------------------------------------------------+
21448 | CONTENT_TYPE | Contains the type of the message-body attached to |
21449 | | the request. It may not be set. |
21450 | | |
21451 +-------------------+-----------------------------------------------------+
21452 | DOCUMENT_ROOT | Contains the document root on the remote host under |
21453 | | which the script should be executed, as defined in |
21454 | | the application's configuration. |
21455 | | |
21456 +-------------------+-----------------------------------------------------+
21457 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
21458 | | to communicate with the FastCGI application. |
21459 | | Concretely, it is set to "CGI/1.1". |
21460 | | |
21461 +-------------------+-----------------------------------------------------+
21462 | PATH_INFO | Contains the portion of the URI path hierarchy |
21463 | | following the part that identifies the script |
21464 | | itself. To be set, the directive "path-info" must |
21465 | | be defined. |
21466 | | |
21467 +-------------------+-----------------------------------------------------+
21468 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
21469 | | It is the concatenation of DOCUMENT_ROOT and |
21470 | | PATH_INFO. If PATH_INFO is not set, this parameters |
21471 | | is not set too. |
21472 | | |
21473 +-------------------+-----------------------------------------------------+
21474 | QUERY_STRING | Contains the request's query string. It may not be |
21475 | | set. |
21476 | | |
21477 +-------------------+-----------------------------------------------------+
21478 | REMOTE_ADDR | Contains the network address of the client sending |
21479 | | the request. |
21480 | | |
21481 +-------------------+-----------------------------------------------------+
21482 | REMOTE_USER | Contains the user identification string supplied by |
21483 | | client as part of user authentication. |
21484 | | |
21485 +-------------------+-----------------------------------------------------+
21486 | REQUEST_METHOD | Contains the method which should be used by the |
21487 | | script to process the request. |
21488 | | |
21489 +-------------------+-----------------------------------------------------+
21490 | REQUEST_URI | Contains the request's URI. |
21491 | | |
21492 +-------------------+-----------------------------------------------------+
21493 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
21494 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
21495 | | |
21496 +-------------------+-----------------------------------------------------+
21497 | SCRIPT_NAME | Contains the name of the script. If the directive |
21498 | | "path-info" is defined, it is the first part of the |
21499 | | URI path hierarchy, ending with the script name. |
21500 | | Otherwise, it is the entire URI path. |
21501 | | |
21502 +-------------------+-----------------------------------------------------+
21503 | SERVER_NAME | Contains the name of the server host to which the |
21504 | | client request is directed. It is the value of the |
21505 | | header "Host", if defined. Otherwise, the |
21506 | | destination address of the connection on the client |
21507 | | side. |
21508 | | |
21509 +-------------------+-----------------------------------------------------+
21510 | SERVER_PORT | Contains the destination TCP port of the connection |
21511 | | on the client side, which is the port the client |
21512 | | connected to. |
21513 | | |
21514 +-------------------+-----------------------------------------------------+
21515 | SERVER_PROTOCOL | Contains the request's protocol. |
21516 | | |
21517 +-------------------+-----------------------------------------------------+
21518 | HTTPS | Set to a non-empty value ("on") if the script was |
21519 | | queried through the HTTPS protocol. |
21520 | | |
21521 +-------------------+-----------------------------------------------------+
21522
21523
2152410.3. Limitations
21525------------------
21526
21527The current implementation have some limitations. The first one is about the
21528way some request headers are hidden to the FastCGI applications. This happens
21529during the headers analysis, on the backend side, before the connection
21530establishment. At this stage, HAProxy know the backend is using a FastCGI
21531application but it don't know if the request will be routed to a FastCGI server
21532or not. But to hide request headers, it simply removes them from the HTX
21533message. So, if the request is finally routed to an HTTP server, it never see
21534these headers. For this reason, it is not recommended to mix FastCGI servers
21535and HTTP servers under the same backend.
21536
21537Similarly, the rules "set-param" and "pass-header" are evaluated during the
21538request headers analysis. So the evaluation is always performed, even if the
21539requests is finally forwarded to an HTTP server.
21540
21541About the rules "set-param", when a rule is applied, a pseudo header is added
21542into the HTX message. So, the same way than for HTTP header rewrites, it may
21543fail if the buffer is full. The rules "set-param" will compete with
21544"http-request" ones.
21545
21546Finally, all FastCGI params and HTTP headers are sent into a unique record
21547FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
21548processing error is returned. It means the record FCGI_PARAM, once encoded,
21549must not exceeds the size of a buffer. However, there is no reserve to respect
21550here.
William Lallemand86d0df02017-11-24 21:36:45 +010021551
Willy Tarreau0ba27502007-12-24 16:55:16 +010021552/*
21553 * Local variables:
21554 * fill-column: 79
21555 * End:
21556 */