blob: 899bdf553a854babb28c771dc9c94cc02e899f68 [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 Tarreau421ed392021-01-06 17:41:32 +01007 2021/01/06
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
442.4. Time format
452.5. Examples
Willy Tarreauc57f0e22009-05-10 13:12:33 +020046
473. Global parameters
483.1. Process management and security
493.2. Performance tuning
503.3. Debugging
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +0100513.4. Userlists
Cyril Bontédc4d9032012-04-08 21:57:39 +0200523.5. Peers
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200533.6. Mailers
William Lallemandc9515522019-06-12 16:32:11 +0200543.7. Programs
Christopher Faulet76edc0f2020-01-13 15:52:01 +0100553.8. HTTP-errors
Emeric Brun99c453d2020-05-25 15:01:04 +0200563.9. Rings
William Lallemand0217b7b2020-11-18 10:41:24 +0100573.10. Log forwarding
Willy Tarreauc57f0e22009-05-10 13:12:33 +020058
594. Proxies
604.1. Proxy keywords matrix
614.2. Alphabetically sorted keywords reference
62
Davor Ocelice9ed2812017-12-25 17:49:28 +0100635. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200645.1. Bind options
655.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200665.3. Server DNS resolution
675.3.1. Global overview
685.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020069
Julien Pivotto6ccee412019-11-27 15:49:54 +0100706. Cache
716.1. Limitation
726.2. Setup
736.2.1. Cache section
746.2.2. Proxy section
75
Willy Tarreau74ca5042013-06-11 23:12:07 +0200767. Using ACLs and fetching samples
777.1. ACL basics
787.1.1. Matching booleans
797.1.2. Matching integers
807.1.3. Matching strings
817.1.4. Matching regular expressions (regexes)
827.1.5. Matching arbitrary data blocks
837.1.6. Matching IPv4 and IPv6 addresses
847.2. Using ACLs to form conditions
857.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200867.3.1. Converters
877.3.2. Fetching samples from internal states
887.3.3. Fetching samples at Layer 4
897.3.4. Fetching samples at Layer 5
907.3.5. Fetching samples from buffer contents (Layer 6)
917.3.6. Fetching HTTP samples (Layer 7)
Christopher Faulete596d182020-05-05 17:46:34 +0200927.3.7. Fetching samples for developers
Willy Tarreau74ca5042013-06-11 23:12:07 +0200937.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020094
958. Logging
968.1. Log levels
978.2. Log formats
988.2.1. Default log format
998.2.2. TCP log format
1008.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +01001018.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +01001028.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +02001038.3. Advanced logging options
1048.3.1. Disabling logging of external tests
1058.3.2. Logging before waiting for the session to terminate
1068.3.3. Raising log level upon errors
1078.3.4. Disabling logging of successful connections
1088.4. Timing events
1098.5. Session state at disconnection
1108.6. Non-printable characters
1118.7. Capturing HTTP cookies
1128.8. Capturing HTTP headers
1138.9. Examples of logs
114
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001159. Supported filters
1169.1. Trace
1179.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001189.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001199.4. Cache
Christopher Fauletb30b3102019-09-12 23:03:09 +02001209.5. fcgi-app
Miroslav Zagoracdc32cd92020-12-13 18:32:57 +01001219.6. OpenTracing
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200122
Christopher Fauletb30b3102019-09-12 23:03:09 +020012310. FastCGI applications
12410.1. Setup
12510.1.1. Fcgi-app section
12610.1.2. Proxy section
12710.1.3. Example
12810.2. Default parameters
12910.3. Limitations
130
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200131
1321. Quick reminder about HTTP
133----------------------------
134
Davor Ocelice9ed2812017-12-25 17:49:28 +0100135When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200136fully analyzed and indexed, thus it becomes possible to build matching criteria
137on almost anything found in the contents.
138
139However, it is important to understand how HTTP requests and responses are
140formed, and how HAProxy decomposes them. It will then become easier to write
141correct rules and to debug existing configurations.
142
143
1441.1. The HTTP transaction model
145-------------------------------
146
147The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100148to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100149from the client to the server, a request is sent by the client through the
150connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200151will involve a new connection :
152
153 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
154
155In this mode, called the "HTTP close" mode, there are as many connection
156establishments as there are HTTP transactions. Since the connection is closed
157by the server after the response, the client does not need to know the content
158length.
159
160Due to the transactional nature of the protocol, it was possible to improve it
161to avoid closing a connection between two subsequent transactions. In this mode
162however, it is mandatory that the server indicates the content length for each
163response so that the client does not wait indefinitely. For this, a special
164header is used: "Content-length". This mode is called the "keep-alive" mode :
165
166 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
167
168Its advantages are a reduced latency between transactions, and less processing
169power required on the server side. It is generally better than the close mode,
170but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200171a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172
Willy Tarreau95c4e142017-11-26 12:18:55 +0100173Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200174keep-alive, but the client does not wait for the first response to send the
175second request. This is useful for fetching large number of images composing a
176page :
177
178 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
179
180This can obviously have a tremendous benefit on performance because the network
181latency is eliminated between subsequent requests. Many HTTP agents do not
182correctly support pipelining since there is no way to associate a response with
183the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100184server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200185
Willy Tarreau95c4e142017-11-26 12:18:55 +0100186The next improvement is the multiplexed mode, as implemented in HTTP/2. This
187time, each transaction is assigned a single stream identifier, and all streams
188are multiplexed over an existing connection. Many requests can be sent in
189parallel by the client, and responses can arrive in any order since they also
190carry the stream identifier.
191
Willy Tarreau70dffda2014-01-30 03:07:23 +0100192By default HAProxy operates in keep-alive mode with regards to persistent
193connections: for each connection it processes each request and response, and
194leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100195start of a new request. When it receives HTTP/2 connections from a client, it
196processes all the requests in parallel and leaves the connection idling,
197waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200198
Christopher Faulet315b39c2018-09-21 16:26:19 +0200199HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100200 - keep alive : all requests and responses are processed (default)
201 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100202 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100203 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200204 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100205
Willy Tarreau95c4e142017-11-26 12:18:55 +0100206
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200207
2081.2. HTTP request
209-----------------
210
211First, let's consider this HTTP request :
212
213 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100214 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200215 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
216 2 Host: www.mydomain.com
217 3 User-agent: my small browser
218 4 Accept: image/jpeg, image/gif
219 5 Accept: image/png
220
221
2221.2.1. The Request line
223-----------------------
224
225Line 1 is the "request line". It is always composed of 3 fields :
226
227 - a METHOD : GET
228 - a URI : /serv/login.php?lang=en&profile=2
229 - a version tag : HTTP/1.1
230
231All of them are delimited by what the standard calls LWS (linear white spaces),
232which are commonly spaces, but can also be tabs or line feeds/carriage returns
233followed by spaces/tabs. The method itself cannot contain any colon (':') and
234is limited to alphabetic letters. All those various combinations make it
235desirable that HAProxy performs the splitting itself rather than leaving it to
236the user to write a complex or inaccurate regular expression.
237
238The URI itself can have several forms :
239
240 - A "relative URI" :
241
242 /serv/login.php?lang=en&profile=2
243
244 It is a complete URL without the host part. This is generally what is
245 received by servers, reverse proxies and transparent proxies.
246
247 - An "absolute URI", also called a "URL" :
248
249 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
250
251 It is composed of a "scheme" (the protocol name followed by '://'), a host
252 name or address, optionally a colon (':') followed by a port number, then
253 a relative URI beginning at the first slash ('/') after the address part.
254 This is generally what proxies receive, but a server supporting HTTP/1.1
255 must accept this form too.
256
257 - a star ('*') : this form is only accepted in association with the OPTIONS
258 method and is not relayable. It is used to inquiry a next hop's
259 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100260
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200261 - an address:port combination : 192.168.0.12:80
262 This is used with the CONNECT method, which is used to establish TCP
263 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
264 other protocols too.
265
266In a relative URI, two sub-parts are identified. The part before the question
267mark is called the "path". It is typically the relative path to static objects
268on the server. The part after the question mark is called the "query string".
269It is mostly used with GET requests sent to dynamic scripts and is very
270specific to the language, framework or application in use.
271
Willy Tarreau95c4e142017-11-26 12:18:55 +0100272HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100273assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100274
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200275
2761.2.2. The request headers
277--------------------------
278
279The headers start at the second line. They are composed of a name at the
280beginning of the line, immediately followed by a colon (':'). Traditionally,
281an LWS is added after the colon but that's not required. Then come the values.
282Multiple identical headers may be folded into one single line, delimiting the
283values with commas, provided that their order is respected. This is commonly
284encountered in the "Cookie:" field. A header may span over multiple lines if
285the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
286define a total of 3 values for the "Accept:" header.
287
Davor Ocelice9ed2812017-12-25 17:49:28 +0100288Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200289their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100290"Connection:" header). In HTTP/2, header names are always sent in lower case,
Willy Tarreau253c2512020-07-07 15:55:23 +0200291as can be seen when running in debug mode. Internally, all header names are
292normalized to lower case so that HTTP/1.x and HTTP/2 use the exact same
293representation, and they are sent as-is on the other side. This explains why an
294HTTP/1.x request typed with camel case is delivered in lower case.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200295
296The end of the headers is indicated by the first empty line. People often say
297that it's a double line feed, which is not exact, even if a double line feed
298is one valid form of empty line.
299
300Fortunately, HAProxy takes care of all these complex combinations when indexing
301headers, checking values and counting them, so there is no reason to worry
302about the way they could be written, but it is important not to accuse an
303application of being buggy if it does unusual, valid things.
304
305Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000306 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200307 in the middle of headers by LWS in order to join multi-line headers. This
308 is necessary for proper analysis and helps less capable HTTP parsers to work
309 correctly and not to be fooled by such complex constructs.
310
311
3121.3. HTTP response
313------------------
314
315An HTTP response looks very much like an HTTP request. Both are called HTTP
316messages. Let's consider this HTTP response :
317
318 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100319 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200320 1 HTTP/1.1 200 OK
321 2 Content-length: 350
322 3 Content-Type: text/html
323
Willy Tarreau816b9792009-09-15 21:25:21 +0200324As a special case, HTTP supports so called "Informational responses" as status
325codes 1xx. These messages are special in that they don't convey any part of the
326response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100327continue to post its request for instance. In the case of a status 100 response
328the requested information will be carried by the next non-100 response message
329following the informational one. This implies that multiple responses may be
330sent to a single request, and that this only works when keep-alive is enabled
331(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
332correctly forward and skip them, and only process the next non-100 response. As
333such, these messages are neither logged nor transformed, unless explicitly
334state otherwise. Status 101 messages indicate that the protocol is changing
335over the same connection and that haproxy must switch to tunnel mode, just as
336if a CONNECT had occurred. Then the Upgrade header would contain additional
337information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200338
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200339
Davor Ocelice9ed2812017-12-25 17:49:28 +01003401.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200341------------------------
342
343Line 1 is the "response line". It is always composed of 3 fields :
344
345 - a version tag : HTTP/1.1
346 - a status code : 200
347 - a reason : OK
348
349The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100350 - 1xx = informational message to be skipped (e.g. 100, 101)
351 - 2xx = OK, content is following (e.g. 200, 206)
352 - 3xx = OK, no content following (e.g. 302, 304)
353 - 4xx = error caused by the client (e.g. 401, 403, 404)
354 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200355
Lukas Tribus23953682017-04-28 13:24:30 +0000356Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100357"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200358found there, but it's a common practice to respect the well-established
359messages. It can be composed of one or multiple words, such as "OK", "Found",
360or "Authentication Required".
361
Davor Ocelice9ed2812017-12-25 17:49:28 +0100362HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200363
364 Code When / reason
365 200 access to stats page, and when replying to monitoring requests
366 301 when performing a redirection, depending on the configured code
367 302 when performing a redirection, depending on the configured code
368 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100369 307 when performing a redirection, depending on the configured code
370 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200371 400 for an invalid or too large request
372 401 when an authentication is required to perform the action (when
373 accessing the stats page)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200374 403 when a request is forbidden by a "http-request deny" rule
Florian Tham9205fea2020-01-08 13:35:30 +0100375 404 when the requested resource could not be found
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200376 408 when the request timeout strikes before the request is complete
Florian Tham272e29b2020-01-08 10:19:05 +0100377 410 when the requested resource is no longer available and will not
378 be available again
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200379 500 when haproxy encounters an unrecoverable internal error, such as a
380 memory allocation failure, which should never happen
Christopher Faulete095f312020-12-07 11:22:24 +0100381 501 when haproxy is unable to satisfy a client request because of an
382 unsupported feature
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200383 502 when the server returns an empty, invalid or incomplete response, or
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200384 when an "http-response deny" rule blocks the response.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200385 503 when no server was available to handle the request, or in response to
386 monitoring requests which match the "monitor fail" condition
387 504 when the response timeout strikes before the server responds
388
389The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3904.2).
391
392
3931.3.2. The response headers
394---------------------------
395
396Response headers work exactly like request headers, and as such, HAProxy uses
397the same parsing function for both. Please refer to paragraph 1.2.2 for more
398details.
399
400
4012. Configuring HAProxy
402----------------------
403
4042.1. Configuration file format
405------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200406
407HAProxy's configuration process involves 3 major sources of parameters :
408
409 - the arguments from the command-line, which always take precedence
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100410 - the configuration file(s), whose format is described here
Thayne McCombscdbcca92021-01-07 21:24:41 -0700411 - the running process's environment, in case some environment variables are
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100412 explicitly referenced
Willy Tarreau6a06a402007-07-15 20:15:28 +0200413
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100414The configuration file follows a fairly simple hierarchical format which obey
415a few basic rules:
Willy Tarreau0ba27502007-12-24 16:55:16 +0100416
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100417 1. a configuration file is an ordered sequence of statements
418
419 2. a statement is a single non-empty line before any unprotected "#" (hash)
420
421 3. a line is a series of tokens or "words" delimited by unprotected spaces or
422 tab characters
423
424 4. the first word or sequence of words of a line is one of the keywords or
425 keyword sequences listed in this document
426
427 5. all other words are all arguments of the first one, some being well-known
428 keywords listed in this document, others being values, references to other
429 parts of the configuration, or expressions
430
431 6. certain keywords delimit a section inside which only a subset of keywords
432 are supported
433
434 7. a section ends at the end of a file or on a special keyword starting a new
435 section
436
437This is all that is needed to know to write a simple but reliable configuration
438generator, but this is not enough to reliably parse any configuration nor to
439figure how to deal with certain corner cases.
440
441First, there are a few consequences of the rules above. Rule 6 and 7 imply that
442the keywords used to define a new section are valid everywhere and cannot have
443a different meaning in a specific section. These keywords are always a single
444word (as opposed to a sequence of words), and traditionally the section that
445follows them is designated using the same name. For example when speaking about
446the "global section", it designates the section of configuration that follows
447the "global" keyword. This usage is used a lot in error messages to help locate
448the parts that need to be addressed.
449
450A number of sections create an internal object or configuration space, which
451requires to be distinguished from other ones. In this case they will take an
452extra word which will set the name of this particular section. For some of them
453the section name is mandatory. For example "frontend foo" will create a new
454section of type "frontend" named "foo". Usually a name is specific to its
455section and two sections of different types may use the same name, but this is
456not recommended as it tends to complexify configuration management.
457
458A direct consequence of rule 7 is that when multiple files are read at once,
459each of them must start with a new section, and the end of each file will end
460a section. A file cannot contain sub-sections nor end an existing section and
461start a new one.
462
463Rule 1 mentioned that ordering matters. Indeed, some keywords create directives
464that can be repeated multiple times to create ordered sequences of rules to be
465applied in a certain order. For example "tcp-request" can be used to alternate
466"accept" and "reject" rules on varying criteria. As such, a configuration file
467processor must always preserve a section's ordering when editing a file. The
468ordering of sections usually does not matter except for the global section
469which must be placed before other sections, but it may be repeated if needed.
470In addition, some automatic identifiers may automatically be assigned to some
471of the created objects (e.g. proxies), and by reordering sections, their
472identifiers will change. These ones appear in the statistics for example. As
473such, the configuration below will assign "foo" ID number 1 and "bar" ID number
4742, which will be swapped if the two sections are reversed:
475
476 listen foo
477 bind :80
478
479 listen bar
480 bind :81
481
482Another important point is that according to rules 2 and 3 above, empty lines,
483spaces, tabs, and comments following and unprotected "#" character are not part
484of the configuration as they are just used as delimiters. This implies that the
485following configurations are strictly equivalent:
486
487 global#this is the global section
488 daemon#daemonize
489 frontend foo
490 mode http # or tcp
491
492and:
493
494 global
495 daemon
496
497 # this is the public web frontend
498 frontend foo
499 mode http
500
501The common practice is to align to the left only the keyword that initiates a
502new section, and indent (i.e. prepend a tab character or a few spaces) all
503other keywords so that it's instantly visible that they belong to the same
504section (as done in the second example above). Placing comments before a new
505section helps the reader decide if it's the desired one. Leaving a blank line
506at the end of a section also visually helps spotting the end when editing it.
507
508Tabs are very convenient for indent but they do not copy-paste well. If spaces
509are used instead, it is recommended to avoid placing too many (2 to 4) so that
510editing in field doesn't become a burden with limited editors that do not
511support automatic indent.
512
513In the early days it used to be common to see arguments split at fixed tab
514positions because most keywords would not take more than two arguments. With
515modern versions featuring complex expressions this practice does not stand
516anymore, and is not recommended.
517
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200518
William Lallemandf9873ba2015-05-05 17:37:14 +02005192.2. Quoting and escaping
520-------------------------
521
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100522In modern configurations, some arguments require the use of some characters
523that were previously considered as pure delimiters. In order to make this
524possible, HAProxy supports character escaping by prepending a backslash ('\')
525in front of the character to be escaped, weak quoting within double quotes
526('"') and strong quoting within single quotes ("'").
William Lallemandf9873ba2015-05-05 17:37:14 +0200527
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100528This is pretty similar to what is done in a number of programming languages and
529very close to what is commonly encountered in Bourne shell. The principle is
530the following: while the configuration parser cuts the lines into words, it
531also takes care of quotes and backslashes to decide whether a character is a
532delimiter or is the raw representation of this character within the current
533word. The escape character is then removed, the quotes are removed, and the
534remaining word is used as-is as a keyword or argument for example.
William Lallemandf9873ba2015-05-05 17:37:14 +0200535
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100536If a backslash is needed in a word, it must either be escaped using itself
537(i.e. double backslash) or be strongly quoted.
538
539Escaping outside quotes is achieved by preceding a special character by a
540backslash ('\'):
William Lallemandf9873ba2015-05-05 17:37:14 +0200541
542 \ to mark a space and differentiate it from a delimiter
543 \# to mark a hash and differentiate it from a comment
544 \\ to use a backslash
545 \' to use a single quote and differentiate it from strong quoting
546 \" to use a double quote and differentiate it from weak quoting
547
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100548In addition, a few non-printable characters may be emitted using their usual
549C-language representation:
550
551 \n to insert a line feed (LF, character \x0a or ASCII 10 decimal)
552 \r to insert a carriage return (CR, character \x0d or ASCII 13 decimal)
553 \t to insert a tab (character \x09 or ASCII 9 decimal)
554 \xNN to insert character having ASCII code hex NN (e.g \x0a for LF).
555
556Weak quoting is achieved by surrounding double quotes ("") around the character
557or sequence of characters to protect. Weak quoting prevents the interpretation
558of:
William Lallemandf9873ba2015-05-05 17:37:14 +0200559
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100560 space or tab as a word separator
William Lallemandf9873ba2015-05-05 17:37:14 +0200561 ' single quote as a strong quoting delimiter
562 # hash as a comment start
563
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100564Weak quoting permits the interpretation of environment variables (which are not
565evaluated outside of quotes) by preceding them with a dollar sign ('$'). If a
566dollar character is needed inside double quotes, it must be escaped using a
567backslash.
William Lallemandb2f07452015-05-12 14:27:13 +0200568
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100569Strong quoting is achieved by surrounding single quotes ('') around the
570character or sequence of characters to protect. Inside single quotes, nothing
571is interpreted, it's the efficient way to quote regular expressions.
William Lallemandf9873ba2015-05-05 17:37:14 +0200572
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100573As a result, here is the matrix indicating how special characters can be
574entered in different contexts (unprintable characters are replaced with their
575name within angle brackets). Note that some characters that may only be
576represented escaped have no possible representation inside single quotes,
577hence the '-' there:
William Lallemandf9873ba2015-05-05 17:37:14 +0200578
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100579 Character | Unquoted | Weakly quoted | Strongly quoted
580 -----------+---------------+-----------------------------+-----------------
581 <TAB> | \<TAB>, \x09 | "<TAB>", "\<TAB>", "\x09" | '<TAB>'
582 <LF> | \n, \x0a | "\n", "\x0a" | -
583 <CR> | \r, \x0d | "\r", "\x0d" | -
584 <SPC> | \<SPC>, \x20 | "<SPC>", "\<SPC>", "\x20" | '<SPC>'
585 " | \", \x22 | "\"", "\x22" | '"'
586 # | \#, \x23 | "#", "\#", "\x23" | '#'
587 $ | $, \$, \x24 | "\$", "\x24" | '$'
588 ' | \', \x27 | "'", "\'", "\x27" | -
589 \ | \\, \x5c | "\\", "\x5c" | '\'
William Lallemandf9873ba2015-05-05 17:37:14 +0200590
591 Example:
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100592 # those are all strictly equivalent:
William Lallemandf9873ba2015-05-05 17:37:14 +0200593 log-format %{+Q}o\ %t\ %s\ %{-Q}r
594 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
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100599There is one particular case where a second level of quoting or escaping may be
600necessary. Some keywords take arguments within parenthesis, sometimes delimited
601by commas. These arguments are commonly integers or predefined words, but when
602they are arbitrary strings, it may be required to perform a separate level of
603escaping to disambiguate the characters that belong to the argument from the
604characters that are used to delimit the arguments themselves. A pretty common
605case is the "regsub" converter. It takes a regular expression in argument, and
606if a closing parenthesis is needed inside, this one will require to have its
607own quotes.
608
609The keyword argument parser is exactly the same as the top-level one regarding
610quotes, except that is will not make special cases of backslashes. But what is
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500611not always obvious is that the delimiters used inside must first be escaped or
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100612quoted so that they are not resolved at the top level.
613
614Let's take this example making use of the "regsub" converter which takes 3
615arguments, one regular expression, one replacement string and one set of flags:
616
617 # replace all occurrences of "foo" with "blah" in the path:
618 http-request set-path %[path,regsub(foo,blah,g)]
619
620Here no special quoting was necessary. But if now we want to replace either
621"foo" or "bar" with "blah", we'll need the regular expression "(foo|bar)". We
622cannot write:
623
624 http-request set-path %[path,regsub((foo|bar),blah,g)]
625
626because we would like the string to cut like this:
627
628 http-request set-path %[path,regsub((foo|bar),blah,g)]
629 |---------|----|-|
630 arg1 _/ / /
631 arg2 __________/ /
632 arg3 ______________/
633
634but actually what is passed is a string between the opening and closing
635parenthesis then garbage:
636
637 http-request set-path %[path,regsub((foo|bar),blah,g)]
638 |--------|--------|
639 arg1=(foo|bar _/ /
640 trailing garbage _________/
641
642The obvious solution here seems to be that the closing parenthesis needs to be
643quoted, but alone this will not work, because as mentioned above, quotes are
644processed by the top-level parser which will resolve them before processing
645this word:
646
647 http-request set-path %[path,regsub("(foo|bar)",blah,g)]
648 ------------ -------- ----------------------------------
649 word1 word2 word3=%[path,regsub((foo|bar),blah,g)]
650
651So we didn't change anything for the argument parser at the second level which
652still sees a truncated regular expression as the only argument, and garbage at
653the end of the string. By escaping the quotes they will be passed unmodified to
654the second level:
655
656 http-request set-path %[path,regsub(\"(foo|bar)\",blah,g)]
657 ------------ -------- ------------------------------------
658 word1 word2 word3=%[path,regsub("(foo|bar)",blah,g)]
659 |---------||----|-|
660 arg1=(foo|bar) _/ / /
661 arg2=blah ___________/ /
662 arg3=g _______________/
663
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500664Another approach consists in using single quotes outside the whole string and
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100665double quotes inside (so that the double quotes are not stripped again):
666
667 http-request set-path '%[path,regsub("(foo|bar)",blah,g)]'
668 ------------ -------- ----------------------------------
669 word1 word2 word3=%[path,regsub("(foo|bar)",blah,g)]
670 |---------||----|-|
671 arg1=(foo|bar) _/ / /
672 arg2 ___________/ /
673 arg3 _______________/
674
675When using regular expressions, it can happen that the dollar ('$') character
676appears in the expression or that a backslash ('\') is used in the replacement
677string. In this case these ones will also be processed inside the double quotes
678thus single quotes are preferred (or double escaping). Example:
679
680 http-request set-path '%[path,regsub("^/(here)(/|$)","my/\1",g)]'
681 ------------ -------- -----------------------------------------
682 word1 word2 word3=%[path,regsub("^/(here)(/|$)","my/\1",g)]
683 |-------------| |-----||-|
684 arg1=(here)(/|$) _/ / /
685 arg2=my/\1 ________________/ /
686 arg3 ______________________/
687
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +0500688Remember that backslahes are not escape characters within single quotes and
Willy Tarreau6f1129d2020-11-25 19:58:20 +0100689that the whole word3 above is already protected against them using the single
690quotes. Conversely, if double quotes had been used around the whole expression,
691single the dollar character and the backslashes would have been resolved at top
692level, breaking the argument contents at the second level.
693
694When in doubt, simply do not use quotes anywhere, and start to place single or
695double quotes around arguments that require a comma or a closing parenthesis,
696and think about escaping these quotes using a backslash of the string contains
697a dollar or a backslash. Again, this is pretty similar to what is used under
698a Bourne shell when double-escaping a command passed to "eval". For API writers
699the best is probably to place escaped quotes around each and every argument,
700regardless of their contents. Users will probably find that using single quotes
701around the whole expression and double quotes around each argument provides
702more readable configurations.
William Lallemandf9873ba2015-05-05 17:37:14 +0200703
704
William Lallemandb2f07452015-05-12 14:27:13 +02007052.3. Environment variables
706--------------------------
707
708HAProxy's configuration supports environment variables. Those variables are
709interpreted only within double quotes. Variables are expanded during the
710configuration parsing. Variable names must be preceded by a dollar ("$") and
711optionally enclosed with braces ("{}") similarly to what is done in Bourne
712shell. Variable names can contain alphanumerical characters or the character
Amaury Denoyellefa41cb62020-10-01 14:32:35 +0200713underscore ("_") but should not start with a digit. If the variable contains a
714list of several values separated by spaces, it can be expanded as individual
715arguments by enclosing the variable with braces and appending the suffix '[*]'
716before the closing brace.
William Lallemandb2f07452015-05-12 14:27:13 +0200717
718 Example:
719
720 bind "fd@${FD_APP1}"
721
722 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
723
724 user "$HAPROXY_USER"
725
William Lallemand4d03e432019-06-14 15:35:37 +0200726Some variables are defined by HAProxy, they can be used in the configuration
727file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200728
William Lallemand4d03e432019-06-14 15:35:37 +0200729* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
730 name of the local peer. (See "-L" in the management guide.)
731
732* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
733 separated by semicolons. Can be useful in the case you specified a
734 directory.
735
736* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
737
John Roeslerfb2fce12019-07-10 15:45:51 -0500738* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200739 processes, separated by semicolons.
740
John Roeslerfb2fce12019-07-10 15:45:51 -0500741* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200742 CLI, separated by semicolons.
743
744See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200745
7462.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200747----------------
748
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100749Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100750values are generally expressed in milliseconds (unless explicitly stated
751otherwise) but may be expressed in any other unit by suffixing the unit to the
752numeric value. It is important to consider this because it will not be repeated
753for every keyword. Supported units are :
754
755 - us : microseconds. 1 microsecond = 1/1000000 second
756 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
757 - s : seconds. 1s = 1000ms
758 - m : minutes. 1m = 60s = 60000ms
759 - h : hours. 1h = 60m = 3600s = 3600000ms
760 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
761
762
Lukas Tribusaa83a312017-03-21 09:25:09 +00007632.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200764-------------
765
766 # Simple configuration for an HTTP proxy listening on port 80 on all
767 # interfaces and forwarding requests to a single backend "servers" with a
768 # single server "server1" listening on 127.0.0.1:8000
769 global
770 daemon
771 maxconn 256
772
773 defaults
774 mode http
775 timeout connect 5000ms
776 timeout client 50000ms
777 timeout server 50000ms
778
779 frontend http-in
780 bind *:80
781 default_backend servers
782
783 backend servers
784 server server1 127.0.0.1:8000 maxconn 32
785
786
787 # The same configuration defined with a single listen block. Shorter but
788 # less expressive, especially in HTTP mode.
789 global
790 daemon
791 maxconn 256
792
793 defaults
794 mode http
795 timeout connect 5000ms
796 timeout client 50000ms
797 timeout server 50000ms
798
799 listen http-in
800 bind *:80
801 server server1 127.0.0.1:8000 maxconn 32
802
803
804Assuming haproxy is in $PATH, test these configurations in a shell with:
805
Willy Tarreauccb289d2010-12-11 20:19:38 +0100806 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200807
808
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008093. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200810--------------------
811
812Parameters in the "global" section are process-wide and often OS-specific. They
813are generally set once for all and do not need being changed once correct. Some
814of them have command-line equivalents.
815
816The following keywords are supported in the "global" section :
817
818 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200819 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200820 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200821 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200822 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200823 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200824 - description
825 - deviceatlas-json-file
826 - deviceatlas-log-level
827 - deviceatlas-separator
828 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900829 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200830 - gid
831 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100832 - hard-stop-after
Christopher Faulet98fbe952019-07-22 16:18:24 +0200833 - h1-case-adjust
834 - h1-case-adjust-file
Willy Tarreaud96f1122019-12-03 07:07:36 +0100835 - insecure-fork-wanted
Willy Tarreaua45a8b52019-12-06 16:31:45 +0100836 - insecure-setuid-wanted
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +0100837 - issuers-chain-path
Dragan Dosen13cd54c2020-06-18 18:24:05 +0200838 - localpeer
Willy Tarreau6a06a402007-07-15 20:15:28 +0200839 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200840 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100841 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200842 - lua-load
Thierry Fournier59f11be2020-11-29 00:37:41 +0100843 - lua-load-per-thread
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +0100844 - lua-prepend-path
William Lallemand27edc4b2019-05-07 17:49:33 +0200845 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200846 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200847 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200848 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200849 - pidfile
Willy Tarreau119e50e2020-05-22 13:53:29 +0200850 - pp2-never-send-local
Willy Tarreau1d549722016-02-16 12:41:57 +0100851 - presetenv
852 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200853 - uid
854 - ulimit-n
855 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200856 - set-dumpable
Willy Tarreau1d549722016-02-16 12:41:57 +0100857 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200858 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200859 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200860 - ssl-default-bind-ciphersuites
Jerome Magninb203ff62020-04-03 15:28:22 +0200861 - ssl-default-bind-curves
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200862 - ssl-default-bind-options
863 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200864 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200865 - ssl-default-server-options
866 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100867 - ssl-server-verify
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +0200868 - ssl-skip-self-issued-ca
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100869 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100870 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100871 - 51degrees-data-file
872 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200873 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200874 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200875 - wurfl-data-file
876 - wurfl-information-list
877 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200878 - wurfl-cache-size
William Dauchy0fec3ab2019-10-27 20:08:11 +0100879 - strict-limits
Willy Tarreaud72758d2010-01-12 10:42:19 +0100880
Willy Tarreau6a06a402007-07-15 20:15:28 +0200881 * Performance tuning
William Dauchy0a8824f2019-10-27 20:08:09 +0100882 - busy-polling
Willy Tarreau1746eec2014-04-25 10:46:47 +0200883 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200884 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200885 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100886 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100887 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100888 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200889 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200890 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200891 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200892 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200893 - noepoll
894 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000895 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200896 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100897 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300898 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000899 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100900 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200901 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200902 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200903 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000904 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000905 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200906 - tune.buffers.limit
907 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200908 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200909 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100910 - tune.comp.maxlevel
Willy Tarreaubc52bec2020-06-18 08:58:47 +0200911 - tune.fd.edge-triggered
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200912 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200913 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200914 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100915 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200916 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200917 - tune.http.maxhdr
Willy Tarreau76cc6992020-07-01 18:49:24 +0200918 - tune.idle-pool.shared
Willy Tarreau7e312732014-02-12 16:35:14 +0100919 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100920 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100921 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100922 - tune.lua.session-timeout
923 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200924 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100925 - tune.maxaccept
926 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200927 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200928 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200929 - tune.pipesize
Willy Tarreaua8e2d972020-07-01 18:27:16 +0200930 - tune.pool-high-fd-ratio
931 - tune.pool-low-fd-ratio
Willy Tarreaue803de22010-01-21 17:43:04 +0100932 - tune.rcvbuf.client
933 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100934 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200935 - tune.runqueue-depth
Willy Tarreaue7723bd2020-06-24 11:11:02 +0200936 - tune.sched.low-latency
Willy Tarreaue803de22010-01-21 17:43:04 +0100937 - tune.sndbuf.client
938 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100939 - tune.ssl.cachesize
William Lallemand7d42ef52020-07-06 11:41:30 +0200940 - tune.ssl.keylog
Willy Tarreaubfd59462013-02-21 07:46:09 +0100941 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200942 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100943 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200944 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200945 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100946 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200947 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100948 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200949 - tune.vars.reqres-max-size
950 - tune.vars.sess-max-size
951 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100952 - tune.zlib.memlevel
953 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100954
Willy Tarreau6a06a402007-07-15 20:15:28 +0200955 * Debugging
Willy Tarreau6a06a402007-07-15 20:15:28 +0200956 - quiet
Willy Tarreau3eb10b82020-04-15 16:42:39 +0200957 - zero-warning
Willy Tarreau6a06a402007-07-15 20:15:28 +0200958
959
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009603.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200961------------------------------------
962
Emeric Brunc8e8d122012-10-02 18:42:10 +0200963ca-base <dir>
964 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +0100965 relative path is used with "ca-file", "ca-verify-file" or "crl-file"
966 directives. Absolute locations specified in "ca-file", "ca-verify-file" and
967 "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200968
Willy Tarreau6a06a402007-07-15 20:15:28 +0200969chroot <jail dir>
970 Changes current directory to <jail dir> and performs a chroot() there before
971 dropping privileges. This increases the security level in case an unknown
972 vulnerability would be exploited, since it would make it very hard for the
973 attacker to exploit the system. This only works when the process is started
974 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100975 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100976
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100977cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
978 On Linux 2.6 and above, it is possible to bind a process or a thread to a
979 specific CPU set. This means that the process or the thread will never run on
980 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
981 sets. The first argument is a process set, eventually followed by a thread
982 set. These sets have the format
983
984 all | odd | even | number[-[number]]
985
986 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100987 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100988 ignored. It is possible to specify a range with two such number delimited by
989 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100990 "all", only odd numbers using "odd" or even numbers using "even", just like
991 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100992 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 +0100993 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100994 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100995 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100996 specified. Each "cpu-map" directive will replace the previous ones when they
997 overlap. A thread will be bound on the intersection of its mapping and the
998 one of the process on which it is attached. If the intersection is null, no
999 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +01001000
Christopher Fauletff4121f2017-11-22 16:38:49 +01001001 Ranges can be partially defined. The higher bound can be omitted. In such
1002 case, it is replaced by the corresponding maximum value, 32 or 64 depending
1003 on the machine's word size.
1004
Christopher Faulet26028f62017-11-22 15:01:51 +01001005 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001006 automatically bind a process or a thread to a CPU by incrementing
1007 process/thread and CPU sets. To be valid, both sets must have the same
1008 size. No matter the declaration order of the CPU sets, it will be bound from
1009 the lowest to the highest bound. Having a process and a thread range with the
1010 "auto:" prefix is not supported. Only one range is supported, the other one
1011 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +01001012
1013 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001014 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
1015
1016 cpu-map 1/all 0-3 # bind all threads of the first process on the
1017 # first 4 CPUs
1018
1019 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
1020 # or "cpu-map 1-32 0-31" depending on the machine's
1021 # word size.
1022
Christopher Faulet26028f62017-11-22 15:01:51 +01001023 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001024 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +01001025 cpu-map auto:1-4 0-3
1026 cpu-map auto:1-4 0-1 2-3
1027 cpu-map auto:1-4 3 2 1 0
1028
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001029 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
1030 # and so on.
1031 cpu-map auto:1/1-4 0-3
1032 cpu-map auto:1/1-4 0-1 2-3
1033 cpu-map auto:1/1-4 3 2 1 0
1034
Davor Ocelice9ed2812017-12-25 17:49:28 +01001035 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +01001036 cpu-map auto:all 0-63
1037 cpu-map auto:even 0-31
1038 cpu-map auto:odd 32-63
1039
1040 # invalid cpu-map because process and CPU sets have different sizes.
1041 cpu-map auto:1-4 0 # invalid
1042 cpu-map auto:1 0-3 # invalid
1043
Christopher Fauletcb6a9452017-11-22 16:50:41 +01001044 # invalid cpu-map because automatic binding is used with a process range
1045 # and a thread range.
1046 cpu-map auto:all/all 0 # invalid
1047 cpu-map auto:all/1-4 0 # invalid
1048 cpu-map auto:1-4/all 0 # invalid
1049
Emeric Brunc8e8d122012-10-02 18:42:10 +02001050crt-base <dir>
1051 Assigns a default directory to fetch SSL certificates from when a relative
William Dauchy238ea3b2020-01-11 13:09:12 +01001052 path is used with "crtfile" or "crt" directives. Absolute locations specified
1053 prevail and ignore "crt-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +02001054
Willy Tarreau6a06a402007-07-15 20:15:28 +02001055daemon
1056 Makes the process fork into background. This is the recommended mode of
1057 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +01001058 disabled by the command line "-db" argument. This option is ignored in
1059 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001060
David Carlier8167f302015-06-01 13:50:06 +02001061deviceatlas-json-file <path>
1062 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001063 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +02001064
1065deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001066 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +02001067 optional and set to 0 by default if not set.
1068
1069deviceatlas-separator <char>
1070 Sets the character separator for the API properties results. This directive
1071 is optional and set to | by default if not set.
1072
Cyril Bonté0306c4a2015-10-26 22:37:38 +01001073deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001074 Sets the client cookie's name used for the detection if the DeviceAtlas
1075 Client-side component was used during the request. This directive is optional
1076 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +01001077
Simon Horman98637e52014-06-20 12:30:16 +09001078external-check
Willy Tarreaud96f1122019-12-03 07:07:36 +01001079 Allows the use of an external agent to perform health checks. This is
1080 disabled by default as a security precaution, and even when enabled, checks
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001081 may still fail unless "insecure-fork-wanted" is enabled as well. If the
1082 program launched makes use of a setuid executable (it should really not),
1083 you may also need to set "insecure-setuid-wanted" in the global section.
1084 See "option external-check", and "insecure-fork-wanted", and
1085 "insecure-setuid-wanted".
Simon Horman98637e52014-06-20 12:30:16 +09001086
Willy Tarreau6a06a402007-07-15 20:15:28 +02001087gid <number>
Thayne McCombscdbcca92021-01-07 21:24:41 -07001088 Changes the process's group ID to <number>. It is recommended that the group
Willy Tarreau6a06a402007-07-15 20:15:28 +02001089 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1090 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +01001091 Note that if haproxy is started from a user having supplementary groups, it
1092 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001093 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +01001094
Willy Tarreau11770ce2019-12-03 08:29:22 +01001095group <group name>
1096 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
1097 See also "gid" and "user".
1098
Cyril Bonté203ec5a2017-03-23 22:44:13 +01001099hard-stop-after <time>
1100 Defines the maximum time allowed to perform a clean soft-stop.
1101
1102 Arguments :
1103 <time> is the maximum time (by default in milliseconds) for which the
1104 instance will remain alive when a soft-stop is received via the
1105 SIGUSR1 signal.
1106
1107 This may be used to ensure that the instance will quit even if connections
1108 remain opened during a soft-stop (for example with long timeouts for a proxy
1109 in tcp mode). It applies both in TCP and HTTP mode.
1110
1111 Example:
1112 global
1113 hard-stop-after 30s
1114
Christopher Faulet98fbe952019-07-22 16:18:24 +02001115h1-case-adjust <from> <to>
1116 Defines the case adjustment to apply, when enabled, to the header name
1117 <from>, to change it to <to> before sending it to HTTP/1 clients or
1118 servers. <from> must be in lower case, and <from> and <to> must not differ
1119 except for their case. It may be repeated if several header names need to be
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05001120 adjusted. Duplicate entries are not allowed. If a lot of header names have to
Christopher Faulet98fbe952019-07-22 16:18:24 +02001121 be adjusted, it might be more convenient to use "h1-case-adjust-file".
1122 Please note that no transformation will be applied unless "option
1123 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
1124 specified in a proxy.
1125
1126 There is no standard case for header names because, as stated in RFC7230,
1127 they are case-insensitive. So applications must handle them in a case-
1128 insensitive manner. But some bogus applications violate the standards and
1129 erroneously rely on the cases most commonly used by browsers. This problem
1130 becomes critical with HTTP/2 because all header names must be exchanged in
1131 lower case, and HAProxy follows the same convention. All header names are
1132 sent in lower case to clients and servers, regardless of the HTTP version.
1133
1134 Applications which fail to properly process requests or responses may require
1135 to temporarily use such workarounds to adjust header names sent to them for
1136 the time it takes the application to be fixed. Please note that an
1137 application which requires such workarounds might be vulnerable to content
1138 smuggling attacks and must absolutely be fixed.
1139
1140 Example:
1141 global
1142 h1-case-adjust content-length Content-Length
1143
1144 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
1145 "option h1-case-adjust-bogus-server".
1146
1147h1-case-adjust-file <hdrs-file>
1148 Defines a file containing a list of key/value pairs used to adjust the case
1149 of some header names before sending them to HTTP/1 clients or servers. The
1150 file <hdrs-file> must contain 2 header names per line. The first one must be
1151 in lower case and both must not differ except for their case. Lines which
1152 start with '#' are ignored, just like empty lines. Leading and trailing tabs
1153 and spaces are stripped. Duplicate entries are not allowed. Please note that
1154 no transformation will be applied unless "option h1-case-adjust-bogus-client"
1155 or "option h1-case-adjust-bogus-server" is specified in a proxy.
1156
1157 If this directive is repeated, only the last one will be processed. It is an
1158 alternative to the directive "h1-case-adjust" if a lot of header names need
1159 to be adjusted. Please read the risks associated with using this.
1160
1161 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
1162 "option h1-case-adjust-bogus-server".
1163
Willy Tarreaud96f1122019-12-03 07:07:36 +01001164insecure-fork-wanted
1165 By default haproxy tries hard to prevent any thread and process creation
1166 after it starts. Doing so is particularly important when using Lua files of
1167 uncertain origin, and when experimenting with development versions which may
1168 still contain bugs whose exploitability is uncertain. And generally speaking
1169 it's good hygiene to make sure that no unexpected background activity can be
1170 triggered by traffic. But this prevents external checks from working, and may
1171 break some very specific Lua scripts which actively rely on the ability to
1172 fork. This option is there to disable this protection. Note that it is a bad
1173 idea to disable it, as a vulnerability in a library or within haproxy itself
1174 will be easier to exploit once disabled. In addition, forking from Lua or
1175 anywhere else is not reliable as the forked process may randomly embed a lock
1176 set by another thread and never manage to finish an operation. As such it is
1177 highly recommended that this option is never used and that any workload
1178 requiring such a fork be reconsidered and moved to a safer solution (such as
1179 agents instead of external checks). This option supports the "no" prefix to
1180 disable it.
1181
Willy Tarreaua45a8b52019-12-06 16:31:45 +01001182insecure-setuid-wanted
1183 HAProxy doesn't need to call executables at run time (except when using
1184 external checks which are strongly recommended against), and is even expected
1185 to isolate itself into an empty chroot. As such, there basically is no valid
1186 reason to allow a setuid executable to be called without the user being fully
1187 aware of the risks. In a situation where haproxy would need to call external
1188 checks and/or disable chroot, exploiting a vulnerability in a library or in
1189 haproxy itself could lead to the execution of an external program. On Linux
1190 it is possible to lock the process so that any setuid bit present on such an
1191 executable is ignored. This significantly reduces the risk of privilege
1192 escalation in such a situation. This is what haproxy does by default. In case
1193 this causes a problem to an external check (for example one which would need
1194 the "ping" command), then it is possible to disable this protection by
1195 explicitly adding this directive in the global section. If enabled, it is
1196 possible to turn it back off by prefixing it with the "no" keyword.
1197
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +01001198issuers-chain-path <dir>
1199 Assigns a directory to load certificate chain for issuer completion. All
1200 files must be in PEM format. For certificates loaded with "crt" or "crt-list",
1201 if certificate chain is not included in PEM (also commonly known as
1202 intermediate certificate), haproxy will complete chain if the issuer of the
1203 certificate corresponds to the first certificate of the chain loaded with
1204 "issuers-chain-path".
1205 A "crt" file with PrivateKey+Certificate+IntermediateCA2+IntermediateCA1
1206 could be replaced with PrivateKey+Certificate. HAProxy will complete the
1207 chain if a file with IntermediateCA2+IntermediateCA1 is present in
1208 "issuers-chain-path" directory. All other certificates with the same issuer
1209 will share the chain in memory.
1210
Dragan Dosen13cd54c2020-06-18 18:24:05 +02001211localpeer <name>
1212 Sets the local instance's peer name. It will be ignored if the "-L"
1213 command line argument is specified or if used after "peers" section
1214 definitions. In such cases, a warning message will be emitted during
1215 the configuration parsing.
1216
1217 This option will also set the HAPROXY_LOCALPEER environment variable.
1218 See also "-L" in the management guide and "peers" section below.
1219
Jan Wagner3e678602020-12-17 22:22:32 +01001220log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001221 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +01001222 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +01001223 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +01001224 configured with "log global".
1225
1226 <address> can be one of:
1227
Willy Tarreau2769aa02007-12-27 18:26:09 +01001228 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +01001229 no port is specified, 514 is used by default (the standard syslog
1230 port).
1231
David du Colombier24bb5f52011-03-17 10:40:23 +01001232 - An IPv6 address followed by a colon and optionally a UDP port. If
1233 no port is specified, 514 is used by default (the standard syslog
1234 port).
1235
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001236 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +01001237 considerations for chroot (be sure the path is accessible inside
1238 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +01001239 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +01001240
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001241 - A file descriptor number in the form "fd@<number>", which may point
1242 to a pipe, terminal, or socket. In this case unbuffered logs are used
1243 and one writev() call per log is performed. This is a bit expensive
1244 but acceptable for most workloads. Messages sent this way will not be
1245 truncated but may be dropped, in which case the DroppedLogs counter
1246 will be incremented. The writev() call is atomic even on pipes for
1247 messages up to PIPE_BUF size, which POSIX recommends to be at least
1248 512 and which is 4096 bytes on most modern operating systems. Any
1249 larger message may be interleaved with messages from other processes.
1250 Exceptionally for debugging purposes the file descriptor may also be
1251 directed to a file, but doing so will significantly slow haproxy down
1252 as non-blocking calls will be ignored. Also there will be no way to
1253 purge nor rotate this file without restarting the process. Note that
1254 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +01001255 for use with a TCP syslog server. See also the "short" and "raw"
1256 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01001257
1258 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
1259 "fd@2", see above.
1260
Willy Tarreauc046d162019-08-30 15:24:59 +02001261 - A ring buffer in the form "ring@<name>", which will correspond to an
1262 in-memory ring buffer accessible over the CLI using the "show events"
1263 command, which will also list existing rings and their sizes. Such
1264 buffers are lost on reload or restart but when used as a complement
1265 this can help troubleshooting by having the logs instantly available.
1266
William Lallemandb2f07452015-05-12 14:27:13 +02001267 You may want to reference some environment variables in the address
1268 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001269
Willy Tarreau18324f52014-06-27 18:10:07 +02001270 <length> is an optional maximum line length. Log lines larger than this value
1271 will be truncated before being sent. The reason is that syslog
1272 servers act differently on log line length. All servers support the
1273 default value of 1024, but some servers simply drop larger lines
1274 while others do log them. If a server supports long lines, it may
1275 make sense to set this value here in order to avoid truncating long
1276 lines. Similarly, if a server drops long lines, it is preferable to
1277 truncate them before sending them. Accepted values are 80 to 65535
1278 inclusive. The default value of 1024 is generally fine for all
1279 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +01001280 JSON-formatted logs may require larger values. You may also need to
1281 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +02001282
Dragan Dosen7ad31542015-09-28 17:16:47 +02001283 <format> is the log format used when generating syslog messages. It may be
1284 one of the following :
1285
Emeric Brun0237c4e2020-11-27 16:24:34 +01001286 local Analog to rfc3164 syslog message format except that hostname
1287 field is stripped. This is the default.
1288 Note: option "log-send-hostname" switches the default to
1289 rfc3164.
1290
1291 rfc3164 The RFC3164 syslog message format.
Dragan Dosen7ad31542015-09-28 17:16:47 +02001292 (https://tools.ietf.org/html/rfc3164)
1293
1294 rfc5424 The RFC5424 syslog message format.
1295 (https://tools.ietf.org/html/rfc5424)
1296
Emeric Brun54648852020-07-06 15:54:06 +02001297 priority A message containing only a level plus syslog facility between
1298 angle brackets such as '<63>', followed by the text. The PID,
1299 date, time, process name and system name are omitted. This is
1300 designed to be used with a local log server.
1301
Willy Tarreaue8746a02018-11-12 08:45:00 +01001302 short A message containing only a level between angle brackets such as
1303 '<3>', followed by the text. The PID, date, time, process name
1304 and system name are omitted. This is designed to be used with a
1305 local log server. This format is compatible with what the systemd
1306 logger consumes.
1307
Emeric Brun54648852020-07-06 15:54:06 +02001308 timed A message containing only a level between angle brackets such as
1309 '<3>', followed by ISO date and by the text. The PID, process
1310 name and system name are omitted. This is designed to be
1311 used with a local log server.
1312
1313 iso A message containing only the ISO date, followed by the text.
1314 The PID, process name and system name are omitted. This is
1315 designed to be used with a local log server.
1316
Willy Tarreauc1b06452018-11-12 11:57:56 +01001317 raw A message containing only the text. The level, PID, date, time,
1318 process name and system name are omitted. This is designed to be
1319 used in containers or during development, where the severity only
1320 depends on the file descriptor used (stdout/stderr).
1321
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02001322 <ranges> A list of comma-separated ranges to identify the logs to sample.
1323 This is used to balance the load of the logs to send to the log
1324 server. The limits of the ranges cannot be null. They are numbered
1325 from 1. The size or period (in number of logs) of the sample must be
1326 set with <sample_size> parameter.
1327
1328 <sample_size>
1329 The size of the sample in number of logs to consider when balancing
1330 their logging loads. It is used to balance the load of the logs to
1331 send to the syslog server. This size must be greater or equal to the
1332 maximum of the high limits of the ranges.
1333 (see also <ranges> parameter).
1334
Robert Tsai81ae1952007-12-05 10:47:29 +01001335 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001336
Willy Tarreaue8746a02018-11-12 08:45:00 +01001337 kern user mail daemon auth syslog lpr news
1338 uucp cron auth2 ftp ntp audit alert cron2
1339 local0 local1 local2 local3 local4 local5 local6 local7
1340
Willy Tarreauc1b06452018-11-12 11:57:56 +01001341 Note that the facility is ignored for the "short" and "raw"
1342 formats, but still required as a positional field. It is
1343 recommended to use "daemon" in this case to make it clear that
1344 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001345
1346 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001347 all messages are sent. If a maximum level is specified, only messages with a
1348 severity at least as important as this level will be sent. An optional minimum
1349 level can be specified. If it is set, logs emitted with a more severe level
1350 than this one will be capped to this level. This is used to avoid sending
1351 "emerg" messages on all terminals on some default syslog configurations.
1352 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001353
Cyril Bontédc4d9032012-04-08 21:57:39 +02001354 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001355
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001356log-send-hostname [<string>]
1357 Sets the hostname field in the syslog header. If optional "string" parameter
1358 is set the header is set to the string contents, otherwise uses the hostname
1359 of the system. Generally used if one is not relaying logs through an
1360 intermediate syslog server or for simply customizing the hostname printed in
1361 the logs.
1362
Kevinm48936af2010-12-22 16:08:21 +00001363log-tag <string>
1364 Sets the tag field in the syslog header to this string. It defaults to the
1365 program name as launched from the command line, which usually is "haproxy".
1366 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001367 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001368
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001369lua-load <file>
Thierry Fournier59f11be2020-11-29 00:37:41 +01001370 This global directive loads and executes a Lua file in the shared context
1371 that is visible to all threads. Any variable set in such a context is visible
1372 from any thread. This is the easiest and recommended way to load Lua programs
1373 but it will not scale well if a lot of Lua calls are performed, as only one
1374 thread may be running on the global state at a time. A program loaded this
1375 way will always see 0 in the "core.thread" variable. This directive can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001376 used multiple times.
1377
Thierry Fournier59f11be2020-11-29 00:37:41 +01001378lua-load-per-thread <file>
1379 This global directive loads and executes a Lua file into each started thread.
1380 Any global variable has a thread-local visibility so that each thread could
1381 see a different value. As such it is strongly recommended not to use global
1382 variables in programs loaded this way. An independent copy is loaded and
1383 initialized for each thread, everything is done sequentially and in the
1384 thread's numeric order from 1 to nbthread. If some operations need to be
1385 performed only once, the program should check the "core.thread" variable to
1386 figure what thread is being initialized. Programs loaded this way will run
1387 concurrently on all threads and will be highly scalable. This is the
1388 recommended way to load simple functions that register sample-fetches,
1389 converters, actions or services once it is certain the program doesn't depend
1390 on global variables. For the sake of simplicity, the directive is available
1391 even if only one thread is used and even if threads are disabled (in which
1392 case it will be equivalent to lua-load). This directive can be used multiple
1393 times.
1394
Tim Duesterhusdd74b5f2020-01-12 13:55:40 +01001395lua-prepend-path <string> [<type>]
1396 Prepends the given string followed by a semicolon to Lua's package.<type>
1397 variable.
1398 <type> must either be "path" or "cpath". If <type> is not given it defaults
1399 to "path".
1400
1401 Lua's paths are semicolon delimited lists of patterns that specify how the
1402 `require` function attempts to find the source file of a library. Question
1403 marks (?) within a pattern will be replaced by module name. The path is
1404 evaluated left to right. This implies that paths that are prepended later
1405 will be checked earlier.
1406
1407 As an example by specifying the following path:
1408
1409 lua-prepend-path /usr/share/haproxy-lua/?/init.lua
1410 lua-prepend-path /usr/share/haproxy-lua/?.lua
1411
1412 When `require "example"` is being called Lua will first attempt to load the
1413 /usr/share/haproxy-lua/example.lua script, if that does not exist the
1414 /usr/share/haproxy-lua/example/init.lua will be attempted and the default
1415 paths if that does not exist either.
1416
1417 See https://www.lua.org/pil/8.1.html for the details within the Lua
1418 documentation.
1419
William Lallemand4cfede82017-11-24 22:02:34 +01001420master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001421 Master-worker mode. It is equivalent to the command line "-W" argument.
1422 This mode will launch a "master" which will monitor the "workers". Using
1423 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001424 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001425 or daemon mode. It is recommended to use this mode with multiprocess and
1426 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001427 By default, if a worker exits with a bad return code, in the case of a
1428 segfault for example, all workers will be killed, and the master will leave.
1429 It is convenient to combine this behavior with Restart=on-failure in a
1430 systemd unit file in order to relaunch the whole process. If you don't want
1431 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001432
William Lallemand4cfede82017-11-24 22:02:34 +01001433 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001434
William Lallemand27edc4b2019-05-07 17:49:33 +02001435mworker-max-reloads <number>
1436 In master-worker mode, this option limits the number of time a worker can
John Roeslerfb2fce12019-07-10 15:45:51 -05001437 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001438 number of reloads is greater than this number, the worker will receive a
1439 SIGTERM. This option helps to keep under control the number of workers.
1440 See also "show proc" in the Management Guide.
1441
Willy Tarreauf42d7942020-10-20 11:54:49 +02001442nbproc <number> (deprecated)
Willy Tarreau6a06a402007-07-15 20:15:28 +02001443 Creates <number> processes when going daemon. This requires the "daemon"
1444 mode. By default, only one process is created, which is the recommended mode
1445 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001446 process, it may be needed to fork multiple daemons. When set to a value
1447 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreauf42d7942020-10-20 11:54:49 +02001448 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. This directive is deprecated
1449 and scheduled for removal in 2.5. Please use "nbthread" instead. See also
1450 "daemon" and "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001451
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001452nbthread <number>
1453 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001454 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1455 "nbproc" historically used to be the only way to use multiple processors, it
1456 also involved a number of shortcomings related to the lack of synchronization
1457 between processes (health-checks, peers, stick-tables, stats, ...) which do
1458 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001459 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1460 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1461 when nbproc is not used, the default "nbthread" value is automatically set to
1462 the number of CPUs the process is bound to upon startup. This means that the
1463 thread count can easily be adjusted from the calling process using commands
1464 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1465 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001466
Willy Tarreau6a06a402007-07-15 20:15:28 +02001467pidfile <pidfile>
MIZUTA Takeshic32f3942020-08-26 13:46:19 +09001468 Writes PIDs of all daemons into file <pidfile> when daemon mode or writes PID
1469 of master process into file <pidfile> when master-worker mode. This option is
1470 equivalent to the "-p" command line argument. The file must be accessible to
1471 the user starting the process. See also "daemon" and "master-worker".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001472
Willy Tarreau119e50e2020-05-22 13:53:29 +02001473pp2-never-send-local
1474 A bug in the PROXY protocol v2 implementation was present in HAProxy up to
1475 version 2.1, causing it to emit a PROXY command instead of a LOCAL command
1476 for health checks. This is particularly minor but confuses some servers'
1477 logs. Sadly, the bug was discovered very late and revealed that some servers
1478 which possibly only tested their PROXY protocol implementation against
1479 HAProxy fail to properly handle the LOCAL command, and permanently remain in
1480 the "down" state when HAProxy checks them. When this happens, it is possible
1481 to enable this global option to revert to the older (bogus) behavior for the
1482 time it takes to contact the affected components' vendors and get them fixed.
1483 This option is disabled by default and acts on all servers having the
1484 "send-proxy-v2" statement.
1485
Willy Tarreau1d549722016-02-16 12:41:57 +01001486presetenv <name> <value>
1487 Sets environment variable <name> to value <value>. If the variable exists, it
1488 is NOT overwritten. The changes immediately take effect so that the next line
1489 in the configuration file sees the new value. See also "setenv", "resetenv",
1490 and "unsetenv".
1491
1492resetenv [<name> ...]
1493 Removes all environment variables except the ones specified in argument. It
1494 allows to use a clean controlled environment before setting new values with
1495 setenv or unsetenv. Please note that some internal functions may make use of
1496 some environment variables, such as time manipulation functions, but also
1497 OpenSSL or even external checks. This must be used with extreme care and only
1498 after complete validation. The changes immediately take effect so that the
1499 next line in the configuration file sees the new environment. See also
1500 "setenv", "presetenv", and "unsetenv".
1501
Christopher Fauletff4121f2017-11-22 16:38:49 +01001502stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001503 Limits the stats socket to a certain set of processes numbers. By default the
1504 stats socket is bound to all processes, causing a warning to be emitted when
1505 nbproc is greater than 1 because there is no way to select the target process
1506 when connecting. However, by using this setting, it becomes possible to pin
1507 the stats socket to a specific set of processes, typically the first one. The
1508 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001509 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001510 word size (32 or 64). Ranges can be partially defined. The higher bound can
1511 be omitted. In such case, it is replaced by the corresponding maximum
1512 value. A better option consists in using the "process" setting of the "stats
1513 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001514
Baptiste Assmann5626f482015-08-23 10:00:10 +02001515server-state-base <directory>
1516 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001517 file names which do not start with a '/'. See also "server-state-file",
1518 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001519
1520server-state-file <file>
1521 Specifies the path to the file containing state of servers. If the path starts
1522 with a slash ('/'), it is considered absolute, otherwise it is considered
1523 relative to the directory specified using "server-state-base" (if set) or to
1524 the current directory. Before reloading HAProxy, it is possible to save the
1525 servers' current state using the stats command "show servers state". The
1526 output of this command must be written in the file pointed by <file>. When
1527 starting up, before handling traffic, HAProxy will read, load and apply state
1528 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001529 configuration. See also "server-state-base" and "show servers state",
1530 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001531
Willy Tarreau1d549722016-02-16 12:41:57 +01001532setenv <name> <value>
1533 Sets environment variable <name> to value <value>. If the variable exists, it
1534 is overwritten. The changes immediately take effect so that the next line in
1535 the configuration file sees the new value. See also "presetenv", "resetenv",
1536 and "unsetenv".
1537
Willy Tarreau636848a2019-04-15 19:38:50 +02001538set-dumpable
1539 This option is better left disabled by default and enabled only upon a
William Dauchyec730982019-10-27 20:08:10 +01001540 developer's request. If it has been enabled, it may still be forcibly
1541 disabled by prefixing it with the "no" keyword. It has no impact on
1542 performance nor stability but will try hard to re-enable core dumps that were
1543 possibly disabled by file size limitations (ulimit -f), core size limitations
1544 (ulimit -c), or "dumpability" of a process after changing its UID/GID (such
1545 as /proc/sys/fs/suid_dumpable on Linux). Core dumps might still be limited by
1546 the current directory's permissions (check what directory the file is started
1547 from), the chroot directory's permission (it may be needed to temporarily
1548 disable the chroot directive or to move it to a dedicated writable location),
1549 or any other system-specific constraint. For example, some Linux flavours are
1550 notorious for replacing the default core file with a path to an executable
1551 not even installed on the system (check /proc/sys/kernel/core_pattern). Often,
1552 simply writing "core", "core.%p" or "/var/log/core/core.%p" addresses the
1553 issue. When trying to enable this option waiting for a rare issue to
1554 re-appear, it's often a good idea to first try to obtain such a dump by
1555 issuing, for example, "kill -11" to the haproxy process and verify that it
1556 leaves a core where expected when dying.
Willy Tarreau636848a2019-04-15 19:38:50 +02001557
Willy Tarreau610f04b2014-02-13 11:36:41 +01001558ssl-default-bind-ciphers <ciphers>
1559 This setting is only available when support for OpenSSL was built in. It sets
1560 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001561 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001562 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001563 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1564 information and recommendations see e.g.
1565 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1566 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1567 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1568 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001569
1570ssl-default-bind-ciphersuites <ciphersuites>
1571 This setting is only available when support for OpenSSL was built in and
1572 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1573 describing the list of cipher algorithms ("cipher suite") that are negotiated
1574 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1575 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001576 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1577 cipher configuration for TLSv1.2 and earlier, please check the
1578 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001579 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001580
Jerome Magninb203ff62020-04-03 15:28:22 +02001581ssl-default-bind-curves <curves>
1582 This setting is only available when support for OpenSSL was built in. It sets
1583 the default string describing the list of elliptic curves algorithms ("curve
1584 suite") that are negotiated during the SSL/TLS handshake with ECDHE. The format
1585 of the string is a colon-delimited list of curve name.
1586 Please check the "bind" keyword for more information.
1587
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001588ssl-default-bind-options [<option>]...
1589 This setting is only available when support for OpenSSL was built in. It sets
1590 default ssl-options to force on all "bind" lines. Please check the "bind"
1591 keyword to see available options.
1592
1593 Example:
1594 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001595 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001596
Willy Tarreau610f04b2014-02-13 11:36:41 +01001597ssl-default-server-ciphers <ciphers>
1598 This setting is only available when support for OpenSSL was built in. It
1599 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001600 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001601 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001602 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1603 information and recommendations see e.g.
1604 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1605 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1606 For TLSv1.3 cipher configuration, please check the
1607 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1608 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001609
1610ssl-default-server-ciphersuites <ciphersuites>
1611 This setting is only available when support for OpenSSL was built in and
1612 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1613 string describing the list of cipher algorithms that are negotiated during
1614 the TLSv1.3 handshake with the server, for all "server" lines which do not
1615 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001616 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1617 cipher configuration for TLSv1.2 and earlier, please check the
1618 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1619 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001620
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001621ssl-default-server-options [<option>]...
1622 This setting is only available when support for OpenSSL was built in. It sets
1623 default ssl-options to force on all "server" lines. Please check the "server"
1624 keyword to see available options.
1625
Remi Gacogne47783ef2015-05-29 15:53:22 +02001626ssl-dh-param-file <file>
1627 This setting is only available when support for OpenSSL was built in. It sets
1628 the default DH parameters that are used during the SSL/TLS handshake when
1629 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001630 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001631 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001632 are not specified either by using ssl-dh-param-file or by setting them
1633 directly in the certificate file, pre-generated DH parameters of the size
1634 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1635 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001636 Custom DH parameters may be generated by using the OpenSSL command
1637 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1638 parameters should not be considered secure anymore.
1639
William Lallemand8e8581e2020-10-20 17:36:46 +02001640ssl-load-extra-del-ext
1641 This setting allows to configure the way HAProxy does the lookup for the
1642 extra SSL files. By default HAProxy adds a new extension to the filename.
William Lallemand089c1382020-10-23 17:35:12 +02001643 (ex: with "foobar.crt" load "foobar.crt.key"). With this option enabled,
William Lallemand8e8581e2020-10-20 17:36:46 +02001644 HAProxy removes the extension before adding the new one (ex: with
William Lallemand089c1382020-10-23 17:35:12 +02001645 "foobar.crt" load "foobar.key").
1646
1647 Your crt file must have a ".crt" extension for this option to work.
William Lallemand8e8581e2020-10-20 17:36:46 +02001648
1649 This option is not compatible with bundle extensions (.ecdsa, .rsa. .dsa)
1650 and won't try to remove them.
1651
1652 This option is disabled by default. See also "ssl-load-extra-files".
1653
William Lallemand4c5adbf2020-02-24 14:23:22 +01001654ssl-load-extra-files <none|all|bundle|sctl|ocsp|issuer|key>*
William Lallemand3af48e72020-02-03 17:15:52 +01001655 This setting alters the way HAProxy will look for unspecified files during
Jerome Magnin587be9c2020-09-07 11:55:57 +02001656 the loading of the SSL certificates associated to "bind" lines. It does not
1657 apply to certificates used for client authentication on "server" lines.
William Lallemand3af48e72020-02-03 17:15:52 +01001658
1659 By default, HAProxy discovers automatically a lot of files not specified in
1660 the configuration, and you may want to disable this behavior if you want to
1661 optimize the startup time.
1662
1663 "none": Only load the files specified in the configuration. Don't try to load
1664 a certificate bundle if the file does not exist. In the case of a directory,
1665 it won't try to bundle the certificates if they have the same basename.
1666
1667 "all": This is the default behavior, it will try to load everything,
William Lallemand4c5adbf2020-02-24 14:23:22 +01001668 bundles, sctl, ocsp, issuer, key.
William Lallemand3af48e72020-02-03 17:15:52 +01001669
1670 "bundle": When a file specified in the configuration does not exist, HAProxy
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001671 will try to load a "cert bundle".
1672
1673 Starting from HAProxy 2.3, the bundles are not loaded in the same OpenSSL
1674 certificate store, instead it will loads each certificate in a separate
1675 store which is equivalent to declaring multiple "crt". OpenSSL 1.1.1 is
1676 required to achieve this. Which means that bundles are now used only for
1677 backward compatibility and are not mandatory anymore to do an hybrid RSA/ECC
1678 bind configuration..
1679
1680 To associate these PEM files into a "cert bundle" that is recognized by
1681 haproxy, they must be named in the following way: All PEM files that are to
1682 be bundled must have the same base name, with a suffix indicating the key
1683 type. Currently, three suffixes are supported: rsa, dsa and ecdsa. For
1684 example, if www.example.com has two PEM files, an RSA file and an ECDSA
1685 file, they must be named: "example.pem.rsa" and "example.pem.ecdsa". The
1686 first part of the filename is arbitrary; only the suffix matters. To load
1687 this bundle into haproxy, specify the base name only:
1688
1689 Example : bind :8443 ssl crt example.pem
1690
1691 Note that the suffix is not given to haproxy; this tells haproxy to look for
1692 a cert bundle.
1693
1694 HAProxy will load all PEM files in the bundle as if they were configured
1695 separately in several "crt".
1696
1697 The bundle loading does not have an impact anymore on the directory loading
1698 since files are loading separately.
1699
1700 On the CLI, bundles are seen as separate files, and the bundle extension is
1701 required to commit them.
1702
William Dauchy57dd6f12020-10-06 15:22:37 +02001703 OCSP files (.ocsp), issuer files (.issuer), Certificate Transparency (.sctl)
William Lallemandf9ff3ec2020-10-02 17:57:44 +02001704 as well as private keys (.key) are supported with multi-cert bundling.
William Lallemand3af48e72020-02-03 17:15:52 +01001705
1706 "sctl": Try to load "<basename>.sctl" for each crt keyword.
1707
1708 "ocsp": Try to load "<basename>.ocsp" for each crt keyword.
1709
1710 "issuer": Try to load "<basename>.issuer" if the issuer of the OCSP file is
1711 not provided in the PEM file.
1712
William Lallemand4c5adbf2020-02-24 14:23:22 +01001713 "key": If the private key was not provided by the PEM file, try to load a
1714 file "<basename>.key" containing a private key.
1715
William Lallemand3af48e72020-02-03 17:15:52 +01001716 The default behavior is "all".
1717
1718 Example:
1719 ssl-load-extra-files bundle sctl
1720 ssl-load-extra-files sctl ocsp issuer
1721 ssl-load-extra-files none
1722
1723 See also: "crt", section 5.1 about bind options.
1724
Emeric Brun850efd52014-01-29 12:24:34 +01001725ssl-server-verify [none|required]
1726 The default behavior for SSL verify on servers side. If specified to 'none',
1727 servers certificates are not verified. The default is 'required' except if
1728 forced using cmdline option '-dV'.
1729
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001730ssl-skip-self-issued-ca
Daniel Corbett67a82712020-07-06 23:01:19 -04001731 Self issued CA, aka x509 root CA, is the anchor for chain validation: as a
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001732 server is useless to send it, client must have it. Standard configuration
1733 need to not include such CA in PEM file. This option allows you to keep such
1734 CA in PEM file without sending it to the client. Use case is to provide
1735 issuer for ocsp without the need for '.issuer' file and be able to share it
1736 with 'issuers-chain-path'. This concerns all certificates without intermediate
1737 certificates. It's useless for BoringSSL, .issuer is ignored because ocsp
William Lallemand9a1d8392020-08-10 17:28:23 +02001738 bits does not need it. Requires at least OpenSSL 1.0.2.
Emmanuel Hocdetc3b7e742020-04-22 11:06:19 +02001739
Willy Tarreauabb175f2012-09-24 12:43:26 +02001740stats socket [<address:port>|<path>] [param*]
1741 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1742 Connections to this socket will return various statistics outputs and even
1743 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001744 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001745 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001746
Willy Tarreauabb175f2012-09-24 12:43:26 +02001747 All parameters supported by "bind" lines are supported, for instance to
1748 restrict access to some users or their access rights. Please consult
1749 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001750
1751stats timeout <timeout, in milliseconds>
1752 The default timeout on the stats socket is set to 10 seconds. It is possible
1753 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001754 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001755
1756stats maxconn <connections>
1757 By default, the stats socket is limited to 10 concurrent connections. It is
1758 possible to change this value with "stats maxconn".
1759
Willy Tarreau6a06a402007-07-15 20:15:28 +02001760uid <number>
Thayne McCombscdbcca92021-01-07 21:24:41 -07001761 Changes the process's user ID to <number>. It is recommended that the user ID
Willy Tarreau6a06a402007-07-15 20:15:28 +02001762 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1763 be started with superuser privileges in order to be able to switch to another
1764 one. See also "gid" and "user".
1765
1766ulimit-n <number>
1767 Sets the maximum number of per-process file-descriptors to <number>. By
1768 default, it is automatically computed, so it is recommended not to use this
1769 option.
1770
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001771unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1772 [ group <group> ] [ gid <gid> ]
1773
1774 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1775 This is mainly used to simplify declaration of those UNIX sockets and reduce
1776 the risk of errors, since those settings are most commonly required but are
1777 also process-specific. The <prefix> setting can be used to force all socket
1778 path to be relative to that directory. This might be needed to access another
1779 component's chroot. Note that those paths are resolved before haproxy chroots
1780 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1781 all have the same meaning as their homonyms used by the "bind" statement. If
1782 both are specified, the "bind" statement has priority, meaning that the
1783 "unix-bind" settings may be seen as process-wide default settings.
1784
Willy Tarreau1d549722016-02-16 12:41:57 +01001785unsetenv [<name> ...]
1786 Removes environment variables specified in arguments. This can be useful to
1787 hide some sensitive information that are occasionally inherited from the
1788 user's environment during some operations. Variables which did not exist are
1789 silently ignored so that after the operation, it is certain that none of
1790 these variables remain. The changes immediately take effect so that the next
1791 line in the configuration file will not see these variables. See also
1792 "setenv", "presetenv", and "resetenv".
1793
Willy Tarreau6a06a402007-07-15 20:15:28 +02001794user <user name>
1795 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1796 See also "uid" and "group".
1797
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001798node <name>
1799 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1800
1801 This statement is useful in HA configurations where two or more processes or
1802 servers share the same IP address. By setting a different node-name on all
1803 nodes, it becomes easy to immediately spot what server is handling the
1804 traffic.
1805
1806description <text>
1807 Add a text that describes the instance.
1808
1809 Please note that it is required to escape certain characters (# for example)
1810 and this text is inserted into a html page so you should avoid using
1811 "<" and ">" characters.
1812
Thomas Holmesdb04f192015-05-18 13:21:39 +0100181351degrees-data-file <file path>
1814 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001815 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001816
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001817 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001818 compiled with USE_51DEGREES.
1819
Ben Shillitof25e8e52016-12-02 14:25:37 +0000182051degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001821 A list of 51Degrees property names to be load from the dataset. A full list
1822 of names is available on the 51Degrees website:
1823 https://51degrees.com/resources/property-dictionary
1824
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001825 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001826 compiled with USE_51DEGREES.
1827
Dragan Dosen93b38d92015-06-29 16:43:25 +0200182851degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001829 A char that will be appended to every property value in a response header
1830 containing 51Degrees results. If not set that will be set as ','.
1831
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001832 Please note that this option is only available when haproxy has been
1833 compiled with USE_51DEGREES.
1834
183551degrees-cache-size <number>
1836 Sets the size of the 51Degrees converter cache to <number> entries. This
1837 is an LRU cache which reminds previous device detections and their results.
1838 By default, this cache is disabled.
1839
1840 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001841 compiled with USE_51DEGREES.
1842
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001843wurfl-data-file <file path>
1844 The path of the WURFL data file to provide device detection services. The
1845 file should be accessible by HAProxy with relevant permissions.
1846
1847 Please note that this option is only available when haproxy has been compiled
1848 with USE_WURFL=1.
1849
1850wurfl-information-list [<capability>]*
1851 A space-delimited list of WURFL capabilities, virtual capabilities, property
1852 names we plan to use in injected headers. A full list of capability and
1853 virtual capability names is available on the Scientiamobile website :
1854
1855 https://www.scientiamobile.com/wurflCapability
1856
1857 Valid WURFL properties are:
1858 - wurfl_id Contains the device ID of the matched device.
1859
1860 - wurfl_root_id Contains the device root ID of the matched
1861 device.
1862
1863 - wurfl_isdevroot Tells if the matched device is a root device.
1864 Possible values are "TRUE" or "FALSE".
1865
1866 - wurfl_useragent The original useragent coming with this
1867 particular web request.
1868
1869 - wurfl_api_version Contains a string representing the currently
1870 used Libwurfl API version.
1871
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001872 - wurfl_info A string containing information on the parsed
1873 wurfl.xml and its full path.
1874
1875 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1876 WURFL has been loaded successfully.
1877
1878 - wurfl_normalized_useragent The normalized useragent.
1879
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001880 Please note that this option is only available when haproxy has been compiled
1881 with USE_WURFL=1.
1882
1883wurfl-information-list-separator <char>
1884 A char that will be used to separate values in a response header containing
1885 WURFL results. If not set that a comma (',') will be used by default.
1886
1887 Please note that this option is only available when haproxy has been compiled
1888 with USE_WURFL=1.
1889
1890wurfl-patch-file [<file path>]
1891 A list of WURFL patch file paths. Note that patches are loaded during startup
1892 thus before the chroot.
1893
1894 Please note that this option is only available when haproxy has been compiled
1895 with USE_WURFL=1.
1896
paulborilebad132c2019-04-18 11:57:04 +02001897wurfl-cache-size <size>
1898 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1899 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001900 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001901 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001902
1903 Please note that this option is only available when haproxy has been compiled
1904 with USE_WURFL=1.
1905
William Dauchy0fec3ab2019-10-27 20:08:11 +01001906strict-limits
William Dauchya5194602020-03-28 19:29:58 +01001907 Makes process fail at startup when a setrlimit fails. Haproxy tries to set the
1908 best setrlimit according to what has been calculated. If it fails, it will
1909 emit a warning. This option is here to guarantee an explicit failure of
1910 haproxy when those limits fail. It is enabled by default. It may still be
1911 forcibly disabled by prefixing it with the "no" keyword.
William Dauchy0fec3ab2019-10-27 20:08:11 +01001912
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019133.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001914-----------------------
1915
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001916busy-polling
1917 In some situations, especially when dealing with low latency on processors
1918 supporting a variable frequency or when running inside virtual machines, each
1919 time the process waits for an I/O using the poller, the processor goes back
1920 to sleep or is offered to another VM for a long time, and it causes
1921 excessively high latencies. This option provides a solution preventing the
1922 processor from sleeping by always using a null timeout on the pollers. This
1923 results in a significant latency reduction (30 to 100 microseconds observed)
1924 at the expense of a risk to overheat the processor. It may even be used with
1925 threads, in which case improperly bound threads may heavily conflict,
1926 resulting in a worse performance and high values for the CPU stolen fields
1927 in "show info" output, indicating which threads are misconfigured. It is
1928 important not to let the process run on the same processor as the network
1929 interrupts when this option is used. It is also better to avoid using it on
1930 multiple CPU threads sharing the same core. This option is disabled by
1931 default. If it has been enabled, it may still be forcibly disabled by
1932 prefixing it with the "no" keyword. It is ignored by the "select" and
1933 "poll" pollers.
1934
William Dauchy3894d972019-12-28 15:36:02 +01001935 This option is automatically disabled on old processes in the context of
1936 seamless reload; it avoids too much cpu conflicts when multiple processes
1937 stay around for some time waiting for the end of their current connections.
1938
Willy Tarreau1746eec2014-04-25 10:46:47 +02001939max-spread-checks <delay in milliseconds>
1940 By default, haproxy tries to spread the start of health checks across the
1941 smallest health check interval of all the servers in a farm. The principle is
1942 to avoid hammering services running on the same server. But when using large
1943 check intervals (10 seconds or more), the last servers in the farm take some
1944 time before starting to be tested, which can be a problem. This parameter is
1945 used to enforce an upper bound on delay between the first and the last check,
1946 even if the servers' check intervals are larger. When servers run with
1947 shorter intervals, their intervals will be respected though.
1948
Willy Tarreau6a06a402007-07-15 20:15:28 +02001949maxconn <number>
1950 Sets the maximum per-process number of concurrent connections to <number>. It
1951 is equivalent to the command-line argument "-n". Proxies will stop accepting
1952 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001953 automatically adjusted according to this value. See also "ulimit-n". Note:
1954 the "select" poller cannot reliably use more than 1024 file descriptors on
1955 some platforms. If your platform only supports select and reports "select
1956 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001957 below 500 in general). If this value is not set, it will automatically be
1958 calculated based on the current file descriptors limit reported by the
1959 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1960 is enforced, based on the buffer size, memory allocated to compression, SSL
1961 cache size, and use or not of SSL and the associated maxsslconn (which can
1962 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001963
Willy Tarreau81c25d02011-09-07 15:17:21 +02001964maxconnrate <number>
1965 Sets the maximum per-process number of connections per second to <number>.
1966 Proxies will stop accepting connections when this limit is reached. It can be
1967 used to limit the global capacity regardless of each frontend capacity. It is
1968 important to note that this can only be used as a service protection measure,
1969 as there will not necessarily be a fair share between frontends when the
1970 limit is reached, so it's a good idea to also limit each frontend to some
1971 value close to its expected share. Also, lowering tune.maxaccept can improve
1972 fairness.
1973
William Lallemandd85f9172012-11-09 17:05:39 +01001974maxcomprate <number>
1975 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001976 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001977 level will be decreased during the session. If the maximum is reached at the
1978 beginning of a session, the session will not compress at all. If the maximum
1979 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001980 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001981 default value.
1982
William Lallemand072a2bf2012-11-20 17:01:01 +01001983maxcompcpuusage <number>
1984 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1985 for new requests or decreasing the compression level of current requests.
1986 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1987 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1988 case of multiple processes (nbproc > 1), each process manages its individual
1989 usage. A value of 100 disable the limit. The default value is 100. Setting
1990 a lower value will prevent the compression work from slowing the whole
1991 process down and from introducing high latencies.
1992
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001993maxpipes <number>
1994 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1995 are only used by kernel-based tcp splicing. Since a pipe contains two file
1996 descriptors, the "ulimit-n" value will be increased accordingly. The default
1997 value is maxconn/4, which seems to be more than enough for most heavy usages.
1998 The splice code dynamically allocates and releases pipes, and can fall back
1999 to standard copy, so setting this value too low may only impact performance.
2000
Willy Tarreau93e7c002013-10-07 18:51:07 +02002001maxsessrate <number>
2002 Sets the maximum per-process number of sessions per second to <number>.
2003 Proxies will stop accepting connections when this limit is reached. It can be
2004 used to limit the global capacity regardless of each frontend capacity. It is
2005 important to note that this can only be used as a service protection measure,
2006 as there will not necessarily be a fair share between frontends when the
2007 limit is reached, so it's a good idea to also limit each frontend to some
2008 value close to its expected share. Also, lowering tune.maxaccept can improve
2009 fairness.
2010
Willy Tarreau403edff2012-09-06 11:58:37 +02002011maxsslconn <number>
2012 Sets the maximum per-process number of concurrent SSL connections to
2013 <number>. By default there is no SSL-specific limit, which means that the
2014 global maxconn setting will apply to all connections. Setting this limit
2015 avoids having openssl use too much memory and crash when malloc returns NULL
2016 (since it unfortunately does not reliably check for such conditions). Note
2017 that the limit applies both to incoming and outgoing connections, so one
2018 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01002019 If this value is not set, but a memory limit is enforced, this value will be
2020 automatically computed based on the memory limit, maxconn, the buffer size,
2021 memory allocated to compression, SSL cache size, and use of SSL in either
2022 frontends, backends or both. If neither maxconn nor maxsslconn are specified
2023 when there is a memory limit, haproxy will automatically adjust these values
2024 so that 100% of the connections can be made over SSL with no risk, and will
2025 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02002026
Willy Tarreaue43d5322013-10-07 20:01:52 +02002027maxsslrate <number>
2028 Sets the maximum per-process number of SSL sessions per second to <number>.
2029 SSL listeners will stop accepting connections when this limit is reached. It
2030 can be used to limit the global SSL CPU usage regardless of each frontend
2031 capacity. It is important to note that this can only be used as a service
2032 protection measure, as there will not necessarily be a fair share between
2033 frontends when the limit is reached, so it's a good idea to also limit each
2034 frontend to some value close to its expected share. It is also important to
2035 note that the sessions are accounted before they enter the SSL stack and not
2036 after, which also protects the stack against bad handshakes. Also, lowering
2037 tune.maxaccept can improve fairness.
2038
William Lallemand9d5f5482012-11-07 16:12:57 +01002039maxzlibmem <number>
2040 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
2041 When the maximum amount is reached, future sessions will not compress as long
2042 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01002043 The default value is 0. The value is available in bytes on the UNIX socket
2044 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
2045 "ZlibMemUsage" in bytes.
2046
Willy Tarreau6a06a402007-07-15 20:15:28 +02002047noepoll
2048 Disables the use of the "epoll" event polling system on Linux. It is
2049 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01002050 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02002051
2052nokqueue
2053 Disables the use of the "kqueue" event polling system on BSD. It is
2054 equivalent to the command-line argument "-dk". The next polling system
2055 used will generally be "poll". See also "nopoll".
2056
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00002057noevports
2058 Disables the use of the event ports event polling system on SunOS systems
2059 derived from Solaris 10 and later. It is equivalent to the command-line
2060 argument "-dv". The next polling system used will generally be "poll". See
2061 also "nopoll".
2062
Willy Tarreau6a06a402007-07-15 20:15:28 +02002063nopoll
2064 Disables the use of the "poll" event polling system. It is equivalent to the
2065 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002066 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00002067 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
2068 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02002069
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002070nosplice
2071 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002072 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002073 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002074 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01002075 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
2076 be used. This option makes it easier to globally disable kernel splicing in
2077 case of doubt. See also "option splice-auto", "option splice-request" and
2078 "option splice-response".
2079
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002080nogetaddrinfo
2081 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
2082 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
2083
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00002084noreuseport
2085 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
2086 command line argument "-dR".
2087
Willy Tarreaud2d33482019-04-25 17:09:07 +02002088profiling.tasks { auto | on | off }
2089 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
2090 the profiling automatically turns on a thread when it starts to suffer from
2091 an average latency of 1000 microseconds or higher as reported in the
2092 "avg_loop_us" activity field, and automatically turns off when the latency
John Roeslerfb2fce12019-07-10 15:45:51 -05002093 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02002094 loops so it does not vary quickly and tends to significantly smooth short
2095 spikes). It may also spontaneously trigger from time to time on overloaded
2096 systems, containers, or virtual machines, or when the system swaps (which
2097 must absolutely never happen on a load balancer).
2098
2099 CPU profiling per task can be very convenient to report where the time is
2100 spent and which requests have what effect on which other request. Enabling
2101 it will typically affect the overall's performance by less than 1%, thus it
2102 is recommended to leave it to the default 'auto' value so that it only
2103 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01002104 supporting the clock_gettime(2) syscall with clock identifiers
2105 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
2106 be zero. This option may be changed at run time using "set profiling" on the
2107 CLI.
2108
Willy Tarreaufe255b72007-10-14 23:09:26 +02002109spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09002110 Sometimes it is desirable to avoid sending agent and health checks to
2111 servers at exact intervals, for instance when many logical servers are
2112 located on the same physical server. With the help of this parameter, it
2113 becomes possible to add some randomness in the check interval between 0
2114 and +/- 50%. A value between 2 and 5 seems to show good results. The
2115 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02002116
Davor Ocelice9ed2812017-12-25 17:49:28 +01002117ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00002118 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01002119 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00002120 multiple times, it will simply enable multiple crypto engines. Referencing an
2121 unsupported engine will prevent haproxy from starting. Note that many engines
2122 will lead to lower HTTPS performance than pure software with recent
2123 processors. The optional command "algo" sets the default algorithms an ENGINE
2124 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01002125 of "ALL" uses the engine for all cryptographic operations. If no list of
2126 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00002127 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
2128 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
2129 openssl configuration file uses:
2130 https://www.openssl.org/docs/man1.0.2/apps/config.html
2131
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00002132ssl-mode-async
2133 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02002134 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00002135 implementation supports a maximum of 32 engines. The Openssl ASYNC API
2136 doesn't support moving read/write buffers and is not compliant with
2137 haproxy's buffer management. So the asynchronous mode is disabled on
John Roeslerfb2fce12019-07-10 15:45:51 -05002138 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00002139 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00002140
Willy Tarreau33cb0652014-12-23 22:52:37 +01002141tune.buffers.limit <number>
2142 Sets a hard limit on the number of buffers which may be allocated per process.
2143 The default value is zero which means unlimited. The minimum non-zero value
2144 will always be greater than "tune.buffers.reserve" and should ideally always
2145 be about twice as large. Forcing this value can be particularly useful to
2146 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01002147 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01002148 another one to be released by another session. Since buffers are dynamically
2149 allocated and released, the waiting time is very short and not perceptible
2150 provided that limits remain reasonable. In fact sometimes reducing the limit
2151 may even increase performance by increasing the CPU cache's efficiency. Tests
2152 have shown good results on average HTTP traffic with a limit to 1/10 of the
2153 expected global maxconn setting, which also significantly reduces memory
2154 usage. The memory savings come from the fact that a number of connections
2155 will not allocate 2*tune.bufsize. It is best not to touch this value unless
2156 advised to do so by an haproxy core developer.
2157
Willy Tarreau1058ae72014-12-23 22:40:40 +01002158tune.buffers.reserve <number>
2159 Sets the number of buffers which are pre-allocated and reserved for use only
2160 during memory shortage conditions resulting in failed memory allocations. The
2161 minimum value is 2 and is also the default. There is no reason a user would
2162 want to change this value, it's mostly aimed at haproxy core developers.
2163
Willy Tarreau27a674e2009-08-17 07:23:33 +02002164tune.bufsize <number>
2165 Sets the buffer size to this size (in bytes). Lower values allow more
2166 sessions to coexist in the same amount of RAM, and higher values allow some
2167 applications with very large cookies to work. The default value is 16384 and
2168 can be changed at build time. It is strongly recommended not to change this
2169 from the default value, as very low values will break some services such as
2170 statistics, and values larger than default size will increase memory usage,
2171 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01002172 parameter should be decreased by the same factor as this one is increased. In
2173 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
2174 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04002175 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01002176 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
2177 value set using this parameter will automatically be rounded up to the next
2178 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02002179
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +01002180tune.chksize <number> (deprecated)
2181 This option is deprecated and ignored.
Willy Tarreau43961d52010-10-04 20:39:20 +02002182
William Lallemandf3747832012-11-09 12:33:10 +01002183tune.comp.maxlevel <number>
2184 Sets the maximum compression level. The compression level affects CPU
2185 usage during compression. This value affects CPU usage during compression.
2186 Each session using compression initializes the compression algorithm with
2187 this value. The default value is 1.
2188
Willy Tarreauc299e1e2019-02-27 11:35:12 +01002189tune.fail-alloc
2190 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
2191 allocation attempt fails. Must be between 0 (no failure) and 100 (no
2192 success). This is useful to debug and make sure memory failures are handled
2193 gracefully.
2194
Willy Tarreaubc52bec2020-06-18 08:58:47 +02002195tune.fd.edge-triggered { on | off } [ EXPERIMENTAL ]
2196 Enables ('on') or disables ('off') the edge-triggered polling mode for FDs
2197 that support it. This is currently only support with epoll. It may noticeably
2198 reduce the number of epoll_ctl() calls and slightly improve performance in
2199 certain scenarios. This is still experimental, it may result in frozen
2200 connections if bugs are still present, and is disabled by default.
2201
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02002202tune.h2.header-table-size <number>
2203 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
2204 cannot be larger than 65536 bytes. A larger value may help certain clients
2205 send more compact requests, depending on their capabilities. This amount of
2206 memory is consumed for each HTTP/2 connection. It is recommended not to
2207 change it.
2208
Willy Tarreaue6baec02017-07-27 11:45:11 +02002209tune.h2.initial-window-size <number>
2210 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01002211 can upload before waiting for an acknowledgment from haproxy. This setting
2212 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02002213 The default value is 65535, which roughly allows up to 5 Mbps of upload
2214 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
2215 over a 1-ms local network. It can make sense to increase this value to allow
2216 faster uploads, or to reduce it to increase fairness when dealing with many
2217 clients. It doesn't affect resource usage.
2218
Willy Tarreau5242ef82017-07-27 11:47:28 +02002219tune.h2.max-concurrent-streams <number>
2220 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
2221 number of outstanding requests on a single connection). The default value is
2222 100. A larger one may slightly improve page load time for complex sites when
2223 visited over high latency networks, but increases the amount of resources a
2224 single client may allocate. A value of zero disables the limit so a single
2225 client may create as many streams as allocatable by haproxy. It is highly
2226 recommended not to change this value.
2227
Willy Tarreaua24b35c2019-02-21 13:24:36 +01002228tune.h2.max-frame-size <number>
2229 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
2230 receive to its peers. The default value is the largest between 16384 and the
2231 buffer size (tune.bufsize). In any case, haproxy will not announce support
2232 for frame sizes larger than buffers. The main purpose of this setting is to
2233 allow to limit the maximum frame size setting when using large buffers. Too
2234 large frame sizes might have performance impact or cause some peers to
2235 misbehave. It is highly recommended not to change this value.
2236
Willy Tarreau193b8c62012-11-22 00:17:38 +01002237tune.http.cookielen <number>
2238 Sets the maximum length of captured cookies. This is the maximum value that
2239 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
2240 will automatically be truncated to this one. It is important not to set too
2241 high a value because all cookie captures still allocate this size whatever
2242 their configured value (they share a same pool). This value is per request
2243 per response, so the memory allocated is twice this value per connection.
2244 When not specified, the limit is set to 63 characters. It is recommended not
2245 to change this value.
2246
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002247tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002248 Sets the maximum length of request URI in logs. This prevents truncating long
2249 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002250 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002251 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02002252 configuration directives too.
2253 The default value is 1024.
2254
Willy Tarreauac1932d2011-10-24 19:14:41 +02002255tune.http.maxhdr <number>
2256 Sets the maximum number of headers in a request. When a request comes with a
2257 number of headers greater than this value (including the first line), it is
2258 rejected with a "400 Bad Request" status code. Similarly, too large responses
2259 are blocked with "502 Bad Gateway". The default value is 101, which is enough
2260 for all usages, considering that the widely deployed Apache server uses the
2261 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02002262 a buggy application to work by the time it gets fixed. The accepted range is
2263 1..32767. Keep in mind that each new header consumes 32bits of memory for
2264 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02002265
Willy Tarreau76cc6992020-07-01 18:49:24 +02002266tune.idle-pool.shared { on | off }
2267 Enables ('on') or disables ('off') sharing of idle connection pools between
2268 threads for a same server. The default is to share them between threads in
2269 order to minimize the number of persistent connections to a server, and to
2270 optimize the connection reuse rate. But to help with debugging or when
2271 suspecting a bug in HAProxy around connection reuse, it can be convenient to
2272 forcefully disable this idle pool sharing between multiple threads, and force
2273 this option to "off". The default is on.
2274
Willy Tarreau7e312732014-02-12 16:35:14 +01002275tune.idletimer <timeout>
2276 Sets the duration after which haproxy will consider that an empty buffer is
2277 probably associated with an idle stream. This is used to optimally adjust
2278 some packet sizes while forwarding large and small data alternatively. The
2279 decision to use splice() or to send large buffers in SSL is modulated by this
2280 parameter. The value is in milliseconds between 0 and 65535. A value of zero
2281 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01002282 which seems to correctly detect end user pauses (e.g. read a page before
John Roeslerfb2fce12019-07-10 15:45:51 -05002283 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01002284 tune.ssl.maxrecord below.
2285
Willy Tarreau7ac908b2019-02-27 12:02:18 +01002286tune.listener.multi-queue { on | off }
2287 Enables ('on') or disables ('off') the listener's multi-queue accept which
2288 spreads the incoming traffic to all threads a "bind" line is allowed to run
2289 on instead of taking them for itself. This provides a smoother traffic
2290 distribution and scales much better, especially in environments where threads
2291 may be unevenly loaded due to external activity (network interrupts colliding
2292 with one thread for example). This option is enabled by default, but it may
2293 be forcefully disabled for troubleshooting or for situations where it is
2294 estimated that the operating system already provides a good enough
2295 distribution and connections are extremely short-lived.
2296
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002297tune.lua.forced-yield <number>
2298 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01002299 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002300 HAProxy scheduler to process other tasks like accepting connections or
2301 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01002302 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002303 lowered. If the Lua code is quite long and its result is absolutely required
2304 to process the data, the <number> can be increased.
2305
Willy Tarreau32f61e22015-03-18 17:54:59 +01002306tune.lua.maxmem
2307 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
2308 default it is zero which means unlimited. It is important to set a limit to
2309 ensure that a bug in a script will not result in the system running out of
2310 memory.
2311
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002312tune.lua.session-timeout <timeout>
2313 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002314 preventing infinite loops or spending too much time in Lua. This timeout
2315 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002316 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01002317
2318tune.lua.task-timeout <timeout>
2319 Purpose is the same as "tune.lua.session-timeout", but this timeout is
2320 dedicated to the tasks. By default, this timeout isn't set because a task may
2321 remain alive during of the lifetime of HAProxy. For example, a task used to
2322 check servers.
2323
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002324tune.lua.service-timeout <timeout>
2325 This is the execution timeout for the Lua services. This is useful for
2326 preventing infinite loops or spending too much time in Lua. This timeout
2327 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002328 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02002329
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002330tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01002331 Sets the maximum number of consecutive connections a process may accept in a
2332 row before switching to other work. In single process mode, higher numbers
2333 give better performance at high connection rates. However in multi-process
2334 modes, keeping a bit of fairness between processes generally is better to
2335 increase performance. This value applies individually to each listener, so
2336 that the number of processes a listener is bound to is taken into account.
2337 This value defaults to 64. In multi-process mode, it is divided by twice
2338 the number of processes the listener is bound to. Setting this value to -1
2339 completely disables the limitation. It should normally not be needed to tweak
2340 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01002341
2342tune.maxpollevents <number>
2343 Sets the maximum amount of events that can be processed at once in a call to
2344 the polling system. The default value is adapted to the operating system. It
2345 has been noticed that reducing it below 200 tends to slightly decrease
2346 latency at the expense of network bandwidth, and increasing it above 200
2347 tends to trade latency for slightly increased bandwidth.
2348
Willy Tarreau27a674e2009-08-17 07:23:33 +02002349tune.maxrewrite <number>
2350 Sets the reserved buffer space to this size in bytes. The reserved space is
2351 used for header rewriting or appending. The first reads on sockets will never
2352 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
2353 bufsize, though that does not make much sense since there are rarely large
2354 numbers of headers to add. Setting it too high prevents processing of large
2355 requests or responses. Setting it too low prevents addition of new headers
2356 to already large requests or to POST requests. It is generally wise to set it
2357 to about 1024. It is automatically readjusted to half of bufsize if it is
2358 larger than that. This means you don't have to worry about it when changing
2359 bufsize.
2360
Willy Tarreauf3045d22015-04-29 16:24:50 +02002361tune.pattern.cache-size <number>
2362 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
2363 cache which reminds previous lookups and their results. It is used by ACLs
2364 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
2365 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
2366 strings. It applies to pattern expressions which means that it will be able
2367 to memorize the result of a lookup among all the patterns specified on a
2368 configuration line (including all those loaded from files). It automatically
2369 invalidates entries which are updated using HTTP actions or on the CLI. The
2370 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau403bfbb2019-10-23 06:59:31 +02002371 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
2372 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02002373 risk of collision in this cache, which is in the order of the size of the
2374 cache divided by 2^64. Typically, at 10000 requests per second with the
2375 default cache size of 10000 entries, there's 1% chance that a brute force
2376 attack could cause a single collision after 60 years, or 0.1% after 6 years.
2377 This is considered much lower than the risk of a memory corruption caused by
2378 aging components. If this is not acceptable, the cache can be disabled by
2379 setting this parameter to 0.
2380
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02002381tune.pipesize <number>
2382 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
2383 are the default size for the system. But sometimes when using TCP splicing,
2384 it can improve performance to increase pipe sizes, especially if it is
2385 suspected that pipes are not filled and that many calls to splice() are
2386 performed. This has an impact on the kernel's memory footprint, so this must
2387 not be changed if impacts are not understood.
2388
Olivier Houchard88698d92019-04-16 19:07:22 +02002389tune.pool-high-fd-ratio <number>
2390 This setting sets the max number of file descriptors (in percentage) used by
2391 haproxy globally against the maximum number of file descriptors haproxy can
2392 use before we start killing idle connections when we can't reuse a connection
2393 and we have to create a new one. The default is 25 (one quarter of the file
2394 descriptor will mean that roughly half of the maximum front connections can
2395 keep an idle connection behind, anything beyond this probably doesn't make
John Roeslerfb2fce12019-07-10 15:45:51 -05002396 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02002397
Willy Tarreau83ca3052020-07-01 18:30:16 +02002398tune.pool-low-fd-ratio <number>
2399 This setting sets the max number of file descriptors (in percentage) used by
2400 haproxy globally against the maximum number of file descriptors haproxy can
2401 use before we stop putting connection into the idle pool for reuse. The
2402 default is 20.
2403
Willy Tarreaue803de22010-01-21 17:43:04 +01002404tune.rcvbuf.client <number>
2405tune.rcvbuf.server <number>
2406 Forces the kernel socket receive buffer size on the client or the server side
2407 to the specified value in bytes. This value applies to all TCP/HTTP frontends
2408 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002409 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002410 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002411 order to save kernel memory by preventing it from buffering too large amounts
2412 of received data. Lower values will significantly increase CPU usage though.
2413
Willy Tarreaub22fc302015-12-14 12:04:35 +01002414tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01002415 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01002416 socket buffers. One of them is that a read returns more than <recv_enough>
2417 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
2418 may be changed by this setting to better deal with workloads involving lots
2419 of short messages such as telnet or SSH sessions.
2420
Olivier Houchard1599b802018-05-24 18:59:04 +02002421tune.runqueue-depth <number>
John Roeslerfb2fce12019-07-10 15:45:51 -05002422 Sets the maximum amount of task that can be processed at once when running
Olivier Houchard1599b802018-05-24 18:59:04 +02002423 tasks. The default value is 200. Increasing it may incur latency when
Willy Tarreaue7723bd2020-06-24 11:11:02 +02002424 dealing with I/Os, making it too small can incur extra overhead. When
2425 experimenting with much larger values, it may be useful to also enable
2426 tune.sched.low-latency to limit the maximum latency to the lowest possible.
2427
2428tune.sched.low-latency { on | off }
2429 Enables ('on') or disables ('off') the low-latency task scheduler. By default
2430 haproxy processes tasks from several classes one class at a time as this is
2431 the most efficient. But when running with large values of tune.runqueue-depth
2432 this can have a measurable effect on request or connection latency. When this
2433 low-latency setting is enabled, tasks of lower priority classes will always
2434 be executed before other ones if they exist. This will permit to lower the
2435 maximum latency experienced by new requests or connections in the middle of
2436 massive traffic, at the expense of a higher impact on this large traffic.
2437 For regular usage it is better to leave this off. The default value is off.
Olivier Houchard1599b802018-05-24 18:59:04 +02002438
Willy Tarreaue803de22010-01-21 17:43:04 +01002439tune.sndbuf.client <number>
2440tune.sndbuf.server <number>
2441 Forces the kernel socket send buffer size on the client or the server side to
2442 the specified value in bytes. This value applies to all TCP/HTTP frontends
2443 and backends. It should normally never be set, and the default size (0) lets
John Roeslerfb2fce12019-07-10 15:45:51 -05002444 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01002445 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01002446 order to save kernel memory by preventing it from buffering too large amounts
2447 of received data. Lower values will significantly increase CPU usage though.
2448 Another use case is to prevent write timeouts with extremely slow clients due
2449 to the kernel waiting for a large part of the buffer to be read before
2450 notifying haproxy again.
2451
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002452tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01002453 Sets the size of the global SSL session cache, in a number of blocks. A block
2454 is large enough to contain an encoded session without peer certificate.
2455 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002456 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01002457 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01002458 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01002459 and reassigned. Higher values reduce the occurrence of such a purge, hence
2460 the number of CPU-intensive SSL handshakes by ensuring that all users keep
2461 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01002462 and are shared between all processes if "nbproc" is greater than 1. Setting
2463 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01002464
Emeric Brun8dc60392014-05-09 13:52:00 +02002465tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02002466 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02002467 should normally not be used since it will force many renegotiations due to
2468 clients hitting a random process. But it may be required on some operating
2469 systems where none of the SSL cache synchronization method may be used. In
2470 this case, adding a first layer of hash-based load balancing before the SSL
2471 layer might limit the impact of the lack of session sharing.
2472
William Lallemand7d42ef52020-07-06 11:41:30 +02002473tune.ssl.keylog { on | off }
2474 This option activates the logging of the TLS keys. It should be used with
2475 care as it will consume more memory per SSL session and could decrease
2476 performances. This is disabled by default.
2477
2478 These sample fetches should be used to generate the SSLKEYLOGFILE that is
2479 required to decipher traffic with wireshark.
2480
2481 https://developer.mozilla.org/en-US/docs/Mozilla/Projects/NSS/Key_Log_Format
2482
2483 The SSLKEYLOG is a series of lines which are formatted this way:
2484
2485 <Label> <space> <ClientRandom> <space> <Secret>
2486
2487 The ClientRandom is provided by the %[ssl_fc_client_random,hex] sample
2488 fetch, the secret and the Label could be find in the array below. You need
2489 to generate a SSLKEYLOGFILE with all the labels in this array.
2490
2491 The following sample fetches are hexadecimal strings and does not need to be
2492 converted.
2493
2494 SSLKEYLOGFILE Label | Sample fetches for the Secrets
2495 --------------------------------|-----------------------------------------
2496 CLIENT_EARLY_TRAFFIC_SECRET | %[ssl_fc_client_early_traffic_secret]
2497 CLIENT_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_client_handshake_traffic_secret]
2498 SERVER_HANDSHAKE_TRAFFIC_SECRET | %[ssl_fc_server_handshake_traffic_secret]
2499 CLIENT_TRAFFIC_SECRET_0 | %[ssl_fc_client_traffic_secret_0]
2500 SERVER_TRAFFIC_SECRET_0 | %[ssl_fc_server_traffic_secret_0]
William Lallemandd742b6c2020-07-07 10:14:56 +02002501 EXPORTER_SECRET | %[ssl_fc_exporter_secret]
2502 EARLY_EXPORTER_SECRET | %[ssl_fc_early_exporter_secret]
William Lallemand7d42ef52020-07-06 11:41:30 +02002503
2504 This is only available with OpenSSL 1.1.1, and useful with TLS1.3 session.
2505
2506 If you want to generate the content of a SSLKEYLOGFILE with TLS < 1.3, you
2507 only need this line:
2508
2509 "CLIENT_RANDOM %[ssl_fc_client_random,hex] %[ssl_fc_session_key,hex]"
2510
Emeric Brun4f65bff2012-11-16 15:11:00 +01002511tune.ssl.lifetime <timeout>
2512 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002513 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01002514 does not guarantee that sessions will last that long, because if the cache is
2515 full, the longest idle sessions will be purged despite their configured
2516 lifetime. The real usefulness of this setting is to prevent sessions from
2517 being used for too long.
2518
Willy Tarreaubfd59462013-02-21 07:46:09 +01002519tune.ssl.maxrecord <number>
2520 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
2521 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
2522 data only once it has received a full record. With large records, it means
2523 that clients might have to download up to 16kB of data before starting to
2524 process them. Limiting the value can improve page load times on browsers
2525 located over high latency or low bandwidth networks. It is suggested to find
2526 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
2527 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
2528 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
2529 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01002530 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01002531 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01002532
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002533tune.ssl.default-dh-param <number>
2534 Sets the maximum size of the Diffie-Hellman parameters used for generating
2535 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
2536 final size will try to match the size of the server's RSA (or DSA) key (e.g,
2537 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
Willy Tarreau3ba77d22020-05-08 09:31:18 +02002538 this maximum value. Default value if 2048. Only 1024 or higher values are
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002539 allowed. Higher values will increase the CPU load, and values greater than
2540 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02002541 used if static Diffie-Hellman parameters are supplied either directly
2542 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02002543
Christopher Faulet31af49d2015-06-09 17:29:50 +02002544tune.ssl.ssl-ctx-cache-size <number>
2545 Sets the size of the cache used to store generated certificates to <number>
2546 entries. This is a LRU cache. Because generating a SSL certificate
2547 dynamically is expensive, they are cached. The default cache size is set to
2548 1000 entries.
2549
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01002550tune.ssl.capture-cipherlist-size <number>
2551 Sets the maximum size of the buffer used for capturing client-hello cipher
2552 list. If the value is 0 (default value) the capture is disabled, otherwise
2553 a buffer is allocated for each SSL/TLS connection.
2554
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002555tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002556tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002557tune.vars.reqres-max-size <size>
2558tune.vars.sess-max-size <size>
2559tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01002560 These five tunes help to manage the maximum amount of memory used by the
2561 variables system. "global" limits the overall amount of memory available for
2562 all scopes. "proc" limits the memory for the process scope, "sess" limits the
2563 memory for the session scope, "txn" for the transaction scope, and "reqres"
2564 limits the memory for each request or response processing.
2565 Memory accounting is hierarchical, meaning more coarse grained limits include
2566 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
2567 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002568
Daniel Schneller0b547052016-03-21 20:46:57 +01002569 For example, when "tune.vars.sess-max-size" is limited to 100,
2570 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
2571 100 either. If we create a variable "txn.var" that contains 100 bytes,
2572 all available space is consumed.
2573 Notice that exceeding the limits at runtime will not result in an error
2574 message, but values might be cut off or corrupted. So make sure to accurately
2575 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02002576
William Lallemanda509e4c2012-11-07 16:54:34 +01002577tune.zlib.memlevel <number>
2578 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002579 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002580 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002581 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002582 between 1 and 9. The default value is 8.
2583
2584tune.zlib.windowsize <number>
2585 Sets the window size (the size of the history buffer) as a parameter of the
2586 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002587 in better compression at the expense of memory usage. Can be a value between
2588 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002589
Willy Tarreauc57f0e22009-05-10 13:12:33 +020025903.3. Debugging
2591--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002592
Willy Tarreau6a06a402007-07-15 20:15:28 +02002593quiet
2594 Do not display any message during startup. It is equivalent to the command-
2595 line argument "-q".
2596
Willy Tarreau3eb10b82020-04-15 16:42:39 +02002597zero-warning
2598 When this option is set, haproxy will refuse to start if any warning was
2599 emitted while processing the configuration. It is highly recommended to set
2600 this option on configurations that are not changed often, as it helps detect
2601 subtle mistakes and keep the configuration clean and forward-compatible. Note
2602 that "haproxy -c" will also report errors in such a case. This option is
2603 equivalent to command line argument "-dW".
2604
Emeric Brunf099e792010-09-27 12:05:28 +02002605
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010026063.4. Userlists
2607--------------
2608It is possible to control access to frontend/backend/listen sections or to
2609http stats by allowing only authenticated and authorized users. To do this,
2610it is required to create at least one userlist and to define users.
2611
2612userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002613 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002614 used to store authentication & authorization data for independent customers.
2615
2616group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002617 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002618 attach users to this group by using a comma separated list of names
2619 proceeded by "users" keyword.
2620
Cyril Bontéf0c60612010-02-06 14:44:47 +01002621user <username> [password|insecure-password <password>]
2622 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002623 Adds user <username> to the current userlist. Both secure (encrypted) and
2624 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002625 evaluated using the crypt(3) function, so depending on the system's
2626 capabilities, different algorithms are supported. For example, modern Glibc
2627 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2628 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002629
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002630 Attention: Be aware that using encrypted passwords might cause significantly
2631 increased CPU usage, depending on the number of requests, and the algorithm
2632 used. For any of the hashed variants, the password for each request must
2633 be processed through the chosen algorithm, before it can be compared to the
2634 value specified in the config file. Most current algorithms are deliberately
2635 designed to be expensive to compute to achieve resistance against brute
2636 force attacks. They do not simply salt/hash the clear text password once,
2637 but thousands of times. This can quickly become a major factor in haproxy's
2638 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002639
2640 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002641 userlist L1
2642 group G1 users tiger,scott
2643 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002644
Cyril Bontéf0c60612010-02-06 14:44:47 +01002645 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2646 user scott insecure-password elgato
2647 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002648
Cyril Bontéf0c60612010-02-06 14:44:47 +01002649 userlist L2
2650 group G1
2651 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002652
Cyril Bontéf0c60612010-02-06 14:44:47 +01002653 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2654 user scott insecure-password elgato groups G1,G2
2655 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002656
2657 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002658
Emeric Brunf099e792010-09-27 12:05:28 +02002659
26603.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002661----------
Emeric Brun94900952015-06-11 18:25:54 +02002662It is possible to propagate entries of any data-types in stick-tables between
2663several haproxy instances over TCP connections in a multi-master fashion. Each
2664instance pushes its local updates and insertions to remote peers. The pushed
2665values overwrite remote ones without aggregation. Interrupted exchanges are
2666automatically detected and recovered from the last known point.
2667In addition, during a soft restart, the old process connects to the new one
2668using such a TCP connection to push all its entries before the new process
2669tries to connect to other peers. That ensures very fast replication during a
2670reload, it typically takes a fraction of a second even for large tables.
2671Note that Server IDs are used to identify servers remotely, so it is important
2672that configurations look similar or at least that the same IDs are forced on
2673each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002674
2675peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002676 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002677 which is referenced by one or more stick-tables.
2678
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002679bind [<address>]:<port_range> [, ...] [param*]
2680 Defines the binding parameters of the local peer of this "peers" section.
2681 Such lines are not supported with "peer" line in the same "peers" section.
2682
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002683disabled
2684 Disables a peers section. It disables both listening and any synchronization
2685 related to this section. This is provided to disable synchronization of stick
2686 tables without having to comment out all "peers" references.
2687
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002688default-bind [param*]
2689 Defines the binding parameters for the local peer, excepted its address.
2690
2691default-server [param*]
2692 Change default options for a server in a "peers" section.
2693
2694 Arguments:
2695 <param*> is a list of parameters for this server. The "default-server"
2696 keyword accepts an important number of options and has a complete
2697 section dedicated to it. Please refer to section 5 for more
2698 details.
2699
2700
2701 See also: "server" and section 5 about server options
2702
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002703enable
2704 This re-enables a disabled peers section which was previously disabled.
2705
Jan Wagner3e678602020-12-17 22:22:32 +01002706log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailleb6f759b2019-11-05 09:57:45 +01002707 <facility> [<level> [<minlevel>]]
2708 "peers" sections support the same "log" keyword as for the proxies to
2709 log information about the "peers" listener. See "log" option for proxies for
2710 more details.
2711
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002712peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002713 Defines a peer inside a peers section.
2714 If <peername> is set to the local peer name (by default hostname, or forced
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002715 using "-L" command line option or "localpeer" global configuration setting),
2716 haproxy will listen for incoming remote peer connection on <ip>:<port>.
2717 Otherwise, <ip>:<port> defines where to connect to in order to join the
2718 remote peer, and <peername> is used at the protocol level to identify and
2719 validate the remote peer on the server side.
Emeric Brunf099e792010-09-27 12:05:28 +02002720
2721 During a soft restart, local peer <ip>:<port> is used by the old instance to
2722 connect the new one and initiate a complete replication (teaching process).
2723
2724 It is strongly recommended to have the exact same peers declaration on all
Dragan Dosen13cd54c2020-06-18 18:24:05 +02002725 peers and to only rely on the "-L" command line argument or the "localpeer"
2726 global configuration setting to change the local peer name. This makes it
2727 easier to maintain coherent configuration files across all peers.
Emeric Brunf099e792010-09-27 12:05:28 +02002728
William Lallemandb2f07452015-05-12 14:27:13 +02002729 You may want to reference some environment variables in the address
2730 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002731
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002732 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2733 "server" keyword explanation below).
2734
2735server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002736 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002737 with a support for all "server" parameters found in 5.2 paragraph.
2738 If the underlying peer is local, <ip>:<port> parameters must not be present.
2739 These parameters must be provided on a "bind" line (see "bind" keyword
2740 of this "peers" section).
2741 Some of these parameters are irrelevant for "peers" sections.
2742
2743
Cyril Bontédc4d9032012-04-08 21:57:39 +02002744 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002745 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002746 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002747 peer haproxy1 192.168.0.1:1024
2748 peer haproxy2 192.168.0.2:1024
2749 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002750
2751 backend mybackend
2752 mode tcp
2753 balance roundrobin
2754 stick-table type ip size 20k peers mypeers
2755 stick on src
2756
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002757 server srv1 192.168.0.30:80
2758 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002759
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002760 Example:
2761 peers mypeers
2762 bind 127.0.0.11:10001 ssl crt mycerts/pem
2763 default-server ssl verify none
2764 server hostA 127.0.0.10:10000
2765 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002766
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002767
2768table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2769 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2770
2771 Configure a stickiness table for the current section. This line is parsed
2772 exactly the same way as the "stick-table" keyword in others section, except
John Roeslerfb2fce12019-07-10 15:45:51 -05002773 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002774 mandatory first parameter to designate the stick-table. Contrary to others
2775 sections, there may be several "table" lines in "peers" sections (see also
2776 "stick-table" keyword).
2777
2778 Also be aware of the fact that "peers" sections have their own stick-table
2779 namespaces to avoid collisions between stick-table names identical in
2780 different "peers" section. This is internally handled prepending the "peers"
2781 sections names to the name of the stick-tables followed by a '/' character.
2782 If somewhere else in the configuration file you have to refer to such
2783 stick-tables declared in "peers" sections you must use the prefixed version
2784 of the stick-table name as follows:
2785
2786 peers mypeers
2787 peer A ...
2788 peer B ...
2789 table t1 ...
2790
2791 frontend fe1
2792 tcp-request content track-sc0 src table mypeers/t1
2793
2794 This is also this prefixed version of the stick-table names which must be
2795 used to refer to stick-tables through the CLI.
2796
2797 About "peers" protocol, as only "peers" belonging to the same section may
2798 communicate with each others, there is no need to do such a distinction.
2799 Several "peers" sections may declare stick-tables with the same name.
2800 This is shorter version of the stick-table name which is sent over the network.
2801 There is only a '/' character as prefix to avoid stick-table name collisions between
2802 stick-tables declared as backends and stick-table declared in "peers" sections
2803 as follows in this weird but supported configuration:
2804
2805 peers mypeers
2806 peer A ...
2807 peer B ...
2808 table t1 type string size 10m store gpc0
2809
2810 backend t1
2811 stick-table type string size 10m store gpc0 peers mypeers
2812
Daniel Corbett67a82712020-07-06 23:01:19 -04002813 Here "t1" table declared in "mypeers" section has "mypeers/t1" as global name.
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002814 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2815 level the former table is named "/t1", the latter is again named "t1".
2816
Simon Horman51a1cf62015-02-03 13:00:44 +090028173.6. Mailers
2818------------
2819It is possible to send email alerts when the state of servers changes.
2820If configured email alerts are sent to each mailer that is configured
2821in a mailers section. Email is sent to mailers using SMTP.
2822
Pieter Baauw386a1272015-08-16 15:26:24 +02002823mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002824 Creates a new mailer list with the name <mailersect>. It is an
2825 independent section which is referenced by one or more proxies.
2826
2827mailer <mailername> <ip>:<port>
2828 Defines a mailer inside a mailers section.
2829
2830 Example:
2831 mailers mymailers
2832 mailer smtp1 192.168.0.1:587
2833 mailer smtp2 192.168.0.2:587
2834
2835 backend mybackend
2836 mode tcp
2837 balance roundrobin
2838
2839 email-alert mailers mymailers
2840 email-alert from test1@horms.org
2841 email-alert to test2@horms.org
2842
2843 server srv1 192.168.0.30:80
2844 server srv2 192.168.0.31:80
2845
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002846timeout mail <time>
2847 Defines the time available for a mail/connection to be made and send to
2848 the mail-server. If not defined the default value is 10 seconds. To allow
2849 for at least two SYN-ACK packets to be send during initial TCP handshake it
2850 is advised to keep this value above 4 seconds.
2851
2852 Example:
2853 mailers mymailers
2854 timeout mail 20s
2855 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002856
William Lallemandc9515522019-06-12 16:32:11 +020028573.7. Programs
2858-------------
2859In master-worker mode, it is possible to launch external binaries with the
2860master, these processes are called programs. These programs are launched and
2861managed the same way as the workers.
2862
2863During a reload of HAProxy, those processes are dealing with the same
2864sequence as a worker:
2865
2866 - the master is re-executed
2867 - the master sends a SIGUSR1 signal to the program
2868 - if "option start-on-reload" is not disabled, the master launches a new
2869 instance of the program
2870
2871During a stop, or restart, a SIGTERM is sent to the programs.
2872
2873program <name>
2874 This is a new program section, this section will create an instance <name>
2875 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2876 the management guide).
2877
2878command <command> [arguments*]
2879 Define the command to start with optional arguments. The command is looked
2880 up in the current PATH if it does not include an absolute path. This is a
2881 mandatory option of the program section. Arguments containing spaces must
2882 be enclosed in quotes or double quotes or be prefixed by a backslash.
2883
Andrew Heberle97236962019-07-12 11:50:26 +08002884user <user name>
2885 Changes the executed command user ID to the <user name> from /etc/passwd.
2886 See also "group".
2887
2888group <group name>
2889 Changes the executed command group ID to the <group name> from /etc/group.
2890 See also "user".
2891
William Lallemandc9515522019-06-12 16:32:11 +02002892option start-on-reload
2893no option start-on-reload
2894 Start (or not) a new instance of the program upon a reload of the master.
2895 The default is to start a new instance. This option may only be used in a
2896 program section.
2897
2898
Christopher Faulet76edc0f2020-01-13 15:52:01 +010028993.8. HTTP-errors
2900----------------
2901
2902It is possible to globally declare several groups of HTTP errors, to be
2903imported afterwards in any proxy section. Same group may be referenced at
2904several places and can be fully or partially imported.
2905
2906http-errors <name>
2907 Create a new http-errors group with the name <name>. It is an independent
2908 section that may be referenced by one or more proxies using its name.
2909
2910errorfile <code> <file>
2911 Associate a file contents to an HTTP error code
2912
2913 Arguments :
2914 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02002915 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01002916 425, 429, 500, 501, 502, 503, and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01002917
2918 <file> designates a file containing the full HTTP response. It is
2919 recommended to follow the common practice of appending ".http" to
2920 the filename so that people do not confuse the response with HTML
2921 error pages, and to use absolute paths, since files are read
2922 before any chroot is performed.
2923
2924 Please referrers to "errorfile" keyword in section 4 for details.
2925
2926 Example:
2927 http-errors website-1
2928 errorfile 400 /etc/haproxy/errorfiles/site1/400.http
2929 errorfile 404 /etc/haproxy/errorfiles/site1/404.http
2930 errorfile 408 /dev/null # work around Chrome pre-connect bug
2931
2932 http-errors website-2
2933 errorfile 400 /etc/haproxy/errorfiles/site2/400.http
2934 errorfile 404 /etc/haproxy/errorfiles/site2/404.http
2935 errorfile 408 /dev/null # work around Chrome pre-connect bug
2936
Emeric Brun99c453d2020-05-25 15:01:04 +020029373.9. Rings
2938----------
2939
2940It is possible to globally declare ring-buffers, to be used as target for log
2941servers or traces.
2942
2943ring <ringname>
2944 Creates a new ring-buffer with name <ringname>.
2945
2946description <text>
Daniel Corbett67a82712020-07-06 23:01:19 -04002947 The description is an optional description string of the ring. It will
Emeric Brun99c453d2020-05-25 15:01:04 +02002948 appear on CLI. By default, <name> is reused to fill this field.
2949
2950format <format>
2951 Format used to store events into the ring buffer.
2952
2953 Arguments:
2954 <format> is the log format used when generating syslog messages. It may be
2955 one of the following :
2956
2957 iso A message containing only the ISO date, followed by the text.
2958 The PID, process name and system name are omitted. This is
2959 designed to be used with a local log server.
2960
Emeric Brun0237c4e2020-11-27 16:24:34 +01002961 local Analog to rfc3164 syslog message format except that hostname
2962 field is stripped. This is the default.
2963 Note: option "log-send-hostname" switches the default to
2964 rfc3164.
2965
Emeric Brun99c453d2020-05-25 15:01:04 +02002966 raw A message containing only the text. The level, PID, date, time,
2967 process name and system name are omitted. This is designed to be
2968 used in containers or during development, where the severity
2969 only depends on the file descriptor used (stdout/stderr). This
2970 is the default.
2971
Emeric Brun0237c4e2020-11-27 16:24:34 +01002972 rfc3164 The RFC3164 syslog message format.
Emeric Brun99c453d2020-05-25 15:01:04 +02002973 (https://tools.ietf.org/html/rfc3164)
2974
2975 rfc5424 The RFC5424 syslog message format.
2976 (https://tools.ietf.org/html/rfc5424)
2977
2978 short A message containing only a level between angle brackets such as
2979 '<3>', followed by the text. The PID, date, time, process name
2980 and system name are omitted. This is designed to be used with a
2981 local log server. This format is compatible with what the systemd
2982 logger consumes.
2983
Emeric Brun54648852020-07-06 15:54:06 +02002984 priority A message containing only a level plus syslog facility between angle
2985 brackets such as '<63>', followed by the text. The PID, date, time,
2986 process name and system name are omitted. This is designed to be used
2987 with a local log server.
2988
Emeric Brun99c453d2020-05-25 15:01:04 +02002989 timed A message containing only a level between angle brackets such as
2990 '<3>', followed by ISO date and by the text. The PID, process
2991 name and system name are omitted. This is designed to be
2992 used with a local log server.
2993
2994maxlen <length>
2995 The maximum length of an event message stored into the ring,
2996 including formatted header. If an event message is longer than
2997 <length>, it will be truncated to this length.
2998
Emeric Brun494c5052020-05-28 11:13:15 +02002999server <name> <address> [param*]
3000 Used to configure a syslog tcp server to forward messages from ring buffer.
3001 This supports for all "server" parameters found in 5.2 paragraph. Some of
3002 these parameters are irrelevant for "ring" sections. Important point: there
3003 is little reason to add more than one server to a ring, because all servers
3004 will receive the exact same copy of the ring contents, and as such the ring
3005 will progress at the speed of the slowest server. If one server does not
3006 respond, it will prevent old messages from being purged and may block new
3007 messages from being inserted into the ring. The proper way to send messages
3008 to multiple servers is to use one distinct ring per log server, not to
Emeric Brun97556472020-05-30 01:42:45 +02003009 attach multiple servers to the same ring. Note that specific server directive
3010 "log-proto" is used to set the protocol used to send messages.
Emeric Brun494c5052020-05-28 11:13:15 +02003011
Emeric Brun99c453d2020-05-25 15:01:04 +02003012size <size>
3013 This is the optional size in bytes for the ring-buffer. Default value is
3014 set to BUFSIZE.
3015
Emeric Brun494c5052020-05-28 11:13:15 +02003016timeout connect <timeout>
3017 Set the maximum time to wait for a connection attempt to a server to succeed.
3018
3019 Arguments :
3020 <timeout> is the timeout value specified in milliseconds by default, but
3021 can be in any other unit if the number is suffixed by the unit,
3022 as explained at the top of this document.
3023
3024timeout server <timeout>
3025 Set the maximum time for pending data staying into output buffer.
3026
3027 Arguments :
3028 <timeout> is the timeout value specified in milliseconds by default, but
3029 can be in any other unit if the number is suffixed by the unit,
3030 as explained at the top of this document.
3031
Emeric Brun99c453d2020-05-25 15:01:04 +02003032 Example:
3033 global
3034 log ring@myring local7
3035
3036 ring myring
3037 description "My local buffer"
3038 format rfc3164
3039 maxlen 1200
3040 size 32764
Emeric Brun494c5052020-05-28 11:13:15 +02003041 timeout connect 5s
3042 timeout server 10s
Emeric Brun97556472020-05-30 01:42:45 +02003043 server mysyslogsrv 127.0.0.1:6514 log-proto octet-count
Emeric Brun99c453d2020-05-25 15:01:04 +02003044
Emeric Brun12941c82020-07-07 14:19:42 +020030453.10. Log forwarding
3046-------------------
3047
3048It is possible to declare one or multiple log forwarding section,
3049haproxy will forward all received log messages to a log servers list.
3050
3051log-forward <name>
3052 Creates a new log forwarder proxy identified as <name>.
3053
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003054backlog <conns>
3055 Give hints to the system about the approximate listen backlog desired size
3056 on connections accept.
3057
3058bind <addr> [param*]
3059 Used to configure a stream log listener to receive messages to forward.
Emeric Brunda46c1c2020-10-08 08:39:02 +02003060 This supports the "bind" parameters found in 5.1 paragraph including
3061 those about ssl but some statements such as "alpn" may be irrelevant for
3062 syslog protocol over TCP.
3063 Those listeners support both "Octet Counting" and "Non-Transparent-Framing"
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003064 modes as defined in rfc-6587.
3065
Willy Tarreau76aaa7f2020-09-16 15:07:22 +02003066dgram-bind <addr> [param*]
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003067 Used to configure a datagram log listener to receive messages to forward.
3068 Addresses must be in IPv4 or IPv6 form,followed by a port. This supports
3069 for some of the "bind" parameters found in 5.1 paragraph among which
3070 "interface", "namespace" or "transparent", the other ones being
Willy Tarreau26ff5da2020-09-16 15:22:19 +02003071 silently ignored as irrelevant for UDP/syslog case.
Emeric Brun12941c82020-07-07 14:19:42 +02003072
3073log global
Jan Wagner3e678602020-12-17 22:22:32 +01003074log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Emeric Brun12941c82020-07-07 14:19:42 +02003075 <facility> [<level> [<minlevel>]]
3076 Used to configure target log servers. See more details on proxies
3077 documentation.
3078 If no format specified, haproxy tries to keep the incoming log format.
3079 Configured facility is ignored, except if incoming message does not
3080 present a facility but one is mandatory on the outgoing format.
3081 If there is no timestamp available in the input format, but the field
3082 exists in output format, haproxy will use the local date.
3083
3084 Example:
3085 global
3086 log stderr format iso local7
3087
3088 ring myring
3089 description "My local buffer"
3090 format rfc5424
3091 maxlen 1200
3092 size 32764
3093 timeout connect 5s
3094 timeout server 10s
3095 # syslog tcp server
3096 server mysyslogsrv 127.0.0.1:514 log-proto octet-count
3097
3098 log-forward sylog-loadb
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003099 dgram-bind 127.0.0.1:1514
3100 bind 127.0.0.1:1514
Emeric Brun12941c82020-07-07 14:19:42 +02003101 # all messages on stderr
3102 log global
3103 # all messages on local tcp syslog server
3104 log ring@myring local0
3105 # load balance messages on 4 udp syslog servers
3106 log 127.0.0.1:10001 sample 1:4 local0
3107 log 127.0.0.1:10002 sample 2:4 local0
3108 log 127.0.0.1:10003 sample 3:4 local0
3109 log 127.0.0.1:10004 sample 4:4 local0
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003110
Emeric Bruncbb7bf72020-10-05 14:39:35 +02003111maxconn <conns>
3112 Fix the maximum number of concurrent connections on a log forwarder.
3113 10 is the default.
3114
3115timeout client <timeout>
3116 Set the maximum inactivity time on the client side.
3117
Willy Tarreauc57f0e22009-05-10 13:12:33 +020031184. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02003119----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003120
Willy Tarreau6a06a402007-07-15 20:15:28 +02003121Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02003122 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02003123 - frontend <name>
3124 - backend <name>
3125 - listen <name>
3126
3127A "defaults" section sets default parameters for all other sections following
3128its declaration. Those default parameters are reset by the next "defaults"
3129section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01003130section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003131
3132A "frontend" section describes a set of listening sockets accepting client
3133connections.
3134
3135A "backend" section describes a set of servers to which the proxy will connect
3136to forward incoming connections.
3137
3138A "listen" section defines a complete proxy with its frontend and backend
3139parts combined in one section. It is generally useful for TCP-only traffic.
3140
Willy Tarreau0ba27502007-12-24 16:55:16 +01003141All proxy names must be formed from upper and lower case letters, digits,
3142'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
3143case-sensitive, which means that "www" and "WWW" are two different proxies.
3144
3145Historically, all proxy names could overlap, it just caused troubles in the
3146logs. Since the introduction of content switching, it is mandatory that two
3147proxies with overlapping capabilities (frontend/backend) have different names.
3148However, it is still permitted that a frontend and a backend share the same
3149name, as this configuration seems to be commonly encountered.
3150
3151Right now, two major proxy modes are supported : "tcp", also known as layer 4,
3152and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003153bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003154protocol, and can interact with it by allowing, blocking, switching, adding,
3155modifying, or removing arbitrary contents in requests or responses, based on
3156arbitrary criteria.
3157
Willy Tarreau70dffda2014-01-30 03:07:23 +01003158In HTTP mode, the processing applied to requests and responses flowing over
3159a connection depends in the combination of the frontend's HTTP options and
Julien Pivotto21ad3152019-12-10 13:11:17 +01003160the backend's. HAProxy supports 3 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01003161
3162 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
3163 requests and responses are processed, and connections remain open but idle
3164 between responses and new requests.
3165
Willy Tarreau70dffda2014-01-30 03:07:23 +01003166 - SCL: server close ("option http-server-close") : the server-facing
3167 connection is closed after the end of the response is received, but the
3168 client-facing connection remains open.
3169
Christopher Faulet315b39c2018-09-21 16:26:19 +02003170 - CLO: close ("option httpclose"): the connection is closed after the end of
3171 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003172
3173The effective mode that will be applied to a connection passing through a
3174frontend and a backend can be determined by both proxy modes according to the
3175following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02003176weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01003177
Christopher Faulet315b39c2018-09-21 16:26:19 +02003178 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01003179
Christopher Faulet315b39c2018-09-21 16:26:19 +02003180 | KAL | SCL | CLO
3181 ----+-----+-----+----
3182 KAL | KAL | SCL | CLO
3183 ----+-----+-----+----
Christopher Faulet315b39c2018-09-21 16:26:19 +02003184 mode SCL | SCL | SCL | CLO
3185 ----+-----+-----+----
3186 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01003187
Willy Tarreau0ba27502007-12-24 16:55:16 +01003188
Willy Tarreau70dffda2014-01-30 03:07:23 +01003189
Willy Tarreauc57f0e22009-05-10 13:12:33 +020031904.1. Proxy keywords matrix
3191--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003192
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003193The following list of keywords is supported. Most of them may only be used in a
3194limited set of section types. Some of them are marked as "deprecated" because
3195they are inherited from an old syntax which may be confusing or functionally
3196limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01003197marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003198option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02003199and must be disabled for a specific instance. Such options may also be prefixed
3200with "default" in order to restore default settings regardless of what has been
3201specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003202
Willy Tarreau6a06a402007-07-15 20:15:28 +02003203
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003204 keyword defaults frontend listen backend
3205------------------------------------+----------+----------+---------+---------
3206acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003207backlog X X X -
3208balance X - X X
3209bind - X X -
3210bind-process X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003211capture cookie - X X -
3212capture request header - X X -
3213capture response header - X X -
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003214clitcpka-cnt X X X -
3215clitcpka-idle X X X -
3216clitcpka-intvl X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02003217compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003218cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003219declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003220default-server X - X X
3221default_backend X X X -
3222description - X X X
3223disabled X X X X
3224dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09003225email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09003226email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09003227email-alert mailers X X X X
3228email-alert myhostname X X X X
3229email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003230enabled X X X X
3231errorfile X X X X
Christopher Faulet76edc0f2020-01-13 15:52:01 +01003232errorfiles X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003233errorloc X X X X
3234errorloc302 X X X X
3235-- keyword -------------------------- defaults - frontend - listen -- backend -
3236errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003237force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003238filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003239fullconn X - X X
3240grace X X X X
3241hash-type X - X X
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01003242http-after-response - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003243http-check comment X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003244http-check connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003245http-check disable-on-404 X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003246http-check expect X - X X
Peter Gervai8912ae62020-06-11 18:26:36 +02003247http-check send X - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02003248http-check send-state X - X X
Christopher Faulete5870d82020-04-15 11:32:03 +02003249http-check set-var X - X X
3250http-check unset-var X - X X
Christopher Faulet3b967c12020-05-15 15:47:44 +02003251http-error X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003252http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02003253http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02003254http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02003255http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003256id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003257ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003258load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02003259log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01003260log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02003261log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01003262log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02003263max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003264maxconn X X X -
3265mode X X X X
3266monitor fail - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003267monitor-uri X X X -
3268option abortonclose (*) X - X X
3269option accept-invalid-http-request (*) X X X -
3270option accept-invalid-http-response (*) X - X X
3271option allbackups (*) X - X X
3272option checkcache (*) X - X X
3273option clitcpka (*) X X X -
3274option contstats (*) X X X -
Christopher Faulet89aed322020-06-02 17:33:56 +02003275option disable-h2-upgrade (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003276option dontlog-normal (*) X X X -
3277option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003278-- keyword -------------------------- defaults - frontend - listen -- backend -
3279option forwardfor X X X X
Christopher Faulet98fbe952019-07-22 16:18:24 +02003280option h1-case-adjust-bogus-client (*) X X X -
3281option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02003282option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02003283option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01003284option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02003285option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02003286option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003287option http-server-close (*) X X X X
3288option http-use-proxy-header (*) X X X -
3289option httpchk X - X X
3290option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01003291option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003292option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003293option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02003294option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003295option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003296option log-health-checks (*) X - X X
3297option log-separate-errors (*) X X X -
3298option logasap (*) X X X -
3299option mysql-check X - X X
3300option nolinger (*) X X X X
3301option originalto X X X X
3302option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02003303option pgsql-check X - X X
3304option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003305option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02003306option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003307option smtpchk X - X X
3308option socket-stats (*) X X X -
3309option splice-auto (*) X X X X
3310option splice-request (*) X X X X
3311option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01003312option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003313option srvtcpka (*) X - X X
3314option ssl-hello-chk X - X X
3315-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01003316option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003317option tcp-smart-accept (*) X X X -
3318option tcp-smart-connect (*) X - X X
3319option tcpka X X X X
3320option tcplog X X X X
3321option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09003322external-check command X - X X
3323external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003324persist rdp-cookie X - X X
3325rate-limit sessions X X X -
3326redirect - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003327-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003328retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02003329retry-on X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003330server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02003331server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02003332server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003333source X - X X
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003334srvtcpka-cnt X - X X
3335srvtcpka-idle X - X X
3336srvtcpka-intvl X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02003337stats admin - X X X
3338stats auth X X X X
3339stats enable X X X X
3340stats hide-version X X X X
3341stats http-request - X X X
3342stats realm X X X X
3343stats refresh X X X X
3344stats scope X X X X
3345stats show-desc X X X X
3346stats show-legends X X X X
3347stats show-node X X X X
3348stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003349-- keyword -------------------------- defaults - frontend - listen -- backend -
3350stick match - - X X
3351stick on - - X X
3352stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02003353stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01003354stick-table - X X X
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02003355tcp-check comment X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003356tcp-check connect X - X X
3357tcp-check expect X - X X
3358tcp-check send X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003359tcp-check send-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003360tcp-check send-binary X - X X
Christopher Fauletb50b3e62020-05-05 18:43:43 +02003361tcp-check send-binary-lf X - X X
Christopher Faulet404f9192020-04-09 23:13:54 +02003362tcp-check set-var X - X X
3363tcp-check unset-var X - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02003364tcp-request connection - X X -
3365tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02003366tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02003367tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02003368tcp-response content - - X X
3369tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003370timeout check X - X X
3371timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02003372timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003373timeout connect X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003374timeout http-keep-alive X X X X
3375timeout http-request X X X X
3376timeout queue X - X X
3377timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02003378timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003379timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02003380timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003381transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01003382unique-id-format X X X -
3383unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003384use_backend - X X -
Christopher Fauletb30b3102019-09-12 23:03:09 +02003385use-fcgi-app - - X X
Willy Tarreau4a5cade2012-04-05 21:09:48 +02003386use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01003387------------------------------------+----------+----------+---------+---------
3388 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02003389
Willy Tarreau0ba27502007-12-24 16:55:16 +01003390
Willy Tarreauc57f0e22009-05-10 13:12:33 +020033914.2. Alphabetically sorted keywords reference
3392---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01003393
3394This section provides a description of each keyword and its usage.
3395
3396
3397acl <aclname> <criterion> [flags] [operator] <value> ...
3398 Declare or complete an access list.
3399 May be used in sections : defaults | frontend | listen | backend
3400 no | yes | yes | yes
3401 Example:
3402 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
3403 acl invalid_src src_port 0:1023
3404 acl local_dst hdr(host) -i localhost
3405
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003406 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003407
3408
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003409backlog <conns>
3410 Give hints to the system about the approximate listen backlog desired size
3411 May be used in sections : defaults | frontend | listen | backend
3412 yes | yes | yes | no
3413 Arguments :
3414 <conns> is the number of pending connections. Depending on the operating
3415 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02003416 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01003417
3418 In order to protect against SYN flood attacks, one solution is to increase
3419 the system's SYN backlog size. Depending on the system, sometimes it is just
3420 tunable via a system parameter, sometimes it is not adjustable at all, and
3421 sometimes the system relies on hints given by the application at the time of
3422 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
3423 to the listen() syscall. On systems which can make use of this value, it can
3424 sometimes be useful to be able to specify a different value, hence this
3425 backlog parameter.
3426
3427 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
3428 used as a hint and the system accepts up to the smallest greater power of
3429 two, and never more than some limits (usually 32768).
3430
3431 See also : "maxconn" and the target operating system's tuning guide.
3432
3433
Willy Tarreau0ba27502007-12-24 16:55:16 +01003434balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02003435balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003436 Define the load balancing algorithm to be used in a backend.
3437 May be used in sections : defaults | frontend | listen | backend
3438 yes | no | yes | yes
3439 Arguments :
3440 <algorithm> is the algorithm used to select a server when doing load
3441 balancing. This only applies when no persistence information
3442 is available, or when a connection is redispatched to another
3443 server. <algorithm> may be one of the following :
3444
3445 roundrobin Each server is used in turns, according to their weights.
3446 This is the smoothest and fairest algorithm when the server's
3447 processing time remains equally distributed. This algorithm
3448 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02003449 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08003450 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02003451 large farms, when a server becomes up after having been down
3452 for a very short time, it may sometimes take a few hundreds
3453 requests for it to be re-integrated into the farm and start
3454 receiving traffic. This is normal, though very rare. It is
3455 indicated here in case you would have the chance to observe
3456 it, so that you don't worry.
3457
3458 static-rr Each server is used in turns, according to their weights.
3459 This algorithm is as similar to roundrobin except that it is
3460 static, which means that changing a server's weight on the
3461 fly will have no effect. On the other hand, it has no design
3462 limitation on the number of servers, and when a server goes
3463 up, it is always immediately reintroduced into the farm, once
3464 the full map is recomputed. It also uses slightly less CPU to
3465 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003466
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003467 leastconn The server with the lowest number of connections receives the
3468 connection. Round-robin is performed within groups of servers
3469 of the same load to ensure that all servers will be used. Use
3470 of this algorithm is recommended where very long sessions are
3471 expected, such as LDAP, SQL, TSE, etc... but is not very well
3472 suited for protocols using short sessions such as HTTP. This
3473 algorithm is dynamic, which means that server weights may be
Willy Tarreau8c855f62020-10-22 17:41:45 +02003474 adjusted on the fly for slow starts for instance. It will
3475 also consider the number of queued connections in addition to
3476 the established ones in order to minimize queuing.
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01003477
Willy Tarreauf09c6602012-02-13 17:12:08 +01003478 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003479 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01003480 identifier to the highest (see server parameter "id"), which
3481 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02003482 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01003483 not make sense to use this algorithm without setting maxconn.
3484 The purpose of this algorithm is to always use the smallest
3485 number of servers so that extra servers can be powered off
3486 during non-intensive hours. This algorithm ignores the server
3487 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02003488 or IMAP than HTTP, though it can be useful there too. In
3489 order to use this algorithm efficiently, it is recommended
3490 that a cloud controller regularly checks server usage to turn
3491 them off when unused, and regularly checks backend queue to
3492 turn new servers on when the queue inflates. Alternatively,
3493 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01003494
Willy Tarreau0ba27502007-12-24 16:55:16 +01003495 source The source IP address is hashed and divided by the total
3496 weight of the running servers to designate which server will
3497 receive the request. This ensures that the same client IP
3498 address will always reach the same server as long as no
3499 server goes down or up. If the hash result changes due to the
3500 number of running servers changing, many clients will be
3501 directed to a different server. This algorithm is generally
3502 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003503 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01003504 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003505 static by default, which means that changing a server's
3506 weight on the fly will have no effect, but this can be
3507 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003508
Oskar Stolc8dc41842012-05-19 10:19:54 +01003509 uri This algorithm hashes either the left part of the URI (before
3510 the question mark) or the whole URI (if the "whole" parameter
3511 is present) and divides the hash value by the total weight of
3512 the running servers. The result designates which server will
3513 receive the request. This ensures that the same URI will
3514 always be directed to the same server as long as no server
3515 goes up or down. This is used with proxy caches and
3516 anti-virus proxies in order to maximize the cache hit rate.
3517 Note that this algorithm may only be used in an HTTP backend.
3518 This algorithm is static by default, which means that
3519 changing a server's weight on the fly will have no effect,
3520 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003521
Oskar Stolc8dc41842012-05-19 10:19:54 +01003522 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003523 "depth", both followed by a positive integer number. These
3524 options may be helpful when it is needed to balance servers
3525 based on the beginning of the URI only. The "len" parameter
3526 indicates that the algorithm should only consider that many
3527 characters at the beginning of the URI to compute the hash.
3528 Note that having "len" set to 1 rarely makes sense since most
3529 URIs start with a leading "/".
3530
3531 The "depth" parameter indicates the maximum directory depth
3532 to be used to compute the hash. One level is counted for each
3533 slash in the request. If both parameters are specified, the
3534 evaluation stops when either is reached.
3535
Willy Tarreau57a37412020-09-23 08:56:29 +02003536 A "path-only" parameter indicates that the hashing key starts
3537 at the first '/' of the path. This can be used to ignore the
3538 authority part of absolute URIs, and to make sure that HTTP/1
3539 and HTTP/2 URIs will provide the same hash.
3540
Willy Tarreau0ba27502007-12-24 16:55:16 +01003541 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003542 the query string of each HTTP GET request.
3543
3544 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02003545 request entity will be searched for the parameter argument,
3546 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02003547 ('?') in the URL. The message body will only start to be
3548 analyzed once either the advertised amount of data has been
3549 received or the request buffer is full. In the unlikely event
3550 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02003551 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02003552 be randomly balanced if at all. This keyword used to support
3553 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003554
3555 If the parameter is found followed by an equal sign ('=') and
3556 a value, then the value is hashed and divided by the total
3557 weight of the running servers. The result designates which
3558 server will receive the request.
3559
3560 This is used to track user identifiers in requests and ensure
3561 that a same user ID will always be sent to the same server as
3562 long as no server goes up or down. If no value is found or if
3563 the parameter is not found, then a round robin algorithm is
3564 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003565 backend. This algorithm is static by default, which means
3566 that changing a server's weight on the fly will have no
3567 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003568
Cyril Bontédc4d9032012-04-08 21:57:39 +02003569 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
3570 request. Just as with the equivalent ACL 'hdr()' function,
3571 the header name in parenthesis is not case sensitive. If the
3572 header is absent or if it does not contain any value, the
3573 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01003574
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003575 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01003576 reducing the hash algorithm to the main domain part with some
3577 specific headers such as 'Host'. For instance, in the Host
3578 value "haproxy.1wt.eu", only "1wt" will be considered.
3579
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003580 This algorithm is static by default, which means that
3581 changing a server's weight on the fly will have no effect,
3582 but this can be changed using "hash-type".
3583
Willy Tarreau21c741a2019-01-14 18:14:27 +01003584 random
3585 random(<draws>)
3586 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02003587 hashing function. This means that the servers' weights are
3588 respected, dynamic weight changes immediately take effect, as
3589 well as new server additions. Random load balancing can be
3590 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01003591 or removed as it may avoid the hammering effect that could
3592 result from roundrobin or leastconn in this situation. The
3593 hash-balance-factor directive can be used to further improve
3594 fairness of the load balancing, especially in situations
3595 where servers show highly variable response times. When an
3596 argument <draws> is present, it must be an integer value one
3597 or greater, indicating the number of draws before selecting
3598 the least loaded of these servers. It was indeed demonstrated
3599 that picking the least loaded of two servers is enough to
3600 significantly improve the fairness of the algorithm, by
3601 always avoiding to pick the most loaded server within a farm
3602 and getting rid of any bias that could be induced by the
3603 unfair distribution of the consistent list. Higher values N
3604 will take away N-1 of the highest loaded servers at the
3605 expense of performance. With very high values, the algorithm
3606 will converge towards the leastconn's result but much slower.
3607 The default value is 2, which generally shows very good
3608 distribution and performance. This algorithm is also known as
3609 the Power of Two Random Choices and is described here :
3610 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02003611
Emeric Brun736aa232009-06-30 17:56:00 +02003612 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02003613 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02003614 The RDP cookie <name> (or "mstshash" if omitted) will be
3615 looked up and hashed for each incoming TCP request. Just as
3616 with the equivalent ACL 'req_rdp_cookie()' function, the name
3617 is not case-sensitive. This mechanism is useful as a degraded
3618 persistence mode, as it makes it possible to always send the
3619 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003620 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02003621 used instead.
3622
3623 Note that for this to work, the frontend must ensure that an
3624 RDP cookie is already present in the request buffer. For this
3625 you must use 'tcp-request content accept' rule combined with
3626 a 'req_rdp_cookie_cnt' ACL.
3627
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003628 This algorithm is static by default, which means that
3629 changing a server's weight on the fly will have no effect,
3630 but this can be changed using "hash-type".
3631
Cyril Bontédc4d9032012-04-08 21:57:39 +02003632 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09003633
Willy Tarreau0ba27502007-12-24 16:55:16 +01003634 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02003635 algorithms. Right now, only "url_param" and "uri" support an
3636 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003637
Willy Tarreau3cd9af22009-03-15 14:06:41 +01003638 The load balancing algorithm of a backend is set to roundrobin when no other
3639 algorithm, mode nor option have been set. The algorithm may only be set once
3640 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003641
Lukas Tribus80512b12018-10-27 20:07:40 +02003642 With authentication schemes that require the same connection like NTLM, URI
John Roeslerfb2fce12019-07-10 15:45:51 -05003643 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02003644 to be routed to different backend servers, breaking the invalid assumptions
3645 NTLM relies on.
3646
Willy Tarreau0ba27502007-12-24 16:55:16 +01003647 Examples :
3648 balance roundrobin
3649 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003650 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01003651 balance hdr(User-Agent)
3652 balance hdr(host)
3653 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003654
3655 Note: the following caveats and limitations on using the "check_post"
3656 extension with "url_param" must be considered :
3657
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003658 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003659 to determine if the parameters will be found in the body or entity which
3660 may contain binary data. Therefore another method may be required to
3661 restrict consideration of POST requests that have no URL parameters in
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02003662 the body. (see acl http_end)
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003663
3664 - using a <max_wait> value larger than the request buffer size does not
3665 make sense and is useless. The buffer size is set at build time, and
3666 defaults to 16 kB.
3667
3668 - Content-Encoding is not supported, the parameter search will probably
3669 fail; and load balancing will fall back to Round Robin.
3670
3671 - Expect: 100-continue is not supported, load balancing will fall back to
3672 Round Robin.
3673
Lukas Tribus23953682017-04-28 13:24:30 +00003674 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003675 If the entire parameter value is not present in the first chunk, the
3676 selection of server is undefined (actually, defined by how little
3677 actually appeared in the first chunk).
3678
3679 - This feature does not support generation of a 100, 411 or 501 response.
3680
3681 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003682 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02003683 white space or control characters are found, indicating the end of what
3684 might be a URL parameter list. This is probably not a concern with SGML
3685 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003686
Willy Tarreau294d0f02015-08-10 19:40:12 +02003687 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003688
3689
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003690bind [<address>]:<port_range> [, ...] [param*]
3691bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003692 Define one or several listening addresses and/or ports in a frontend.
3693 May be used in sections : defaults | frontend | listen | backend
3694 no | yes | yes | no
3695 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003696 <address> is optional and can be a host name, an IPv4 address, an IPv6
3697 address, or '*'. It designates the address the frontend will
3698 listen on. If unset, all IPv4 addresses of the system will be
3699 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01003700 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01003701 Optionally, an address family prefix may be used before the
3702 address to force the family regardless of the address format,
3703 which can be useful to specify a path to a unix socket with
3704 no slash ('/'). Currently supported prefixes are :
3705 - 'ipv4@' -> address is always IPv4
3706 - 'ipv6@' -> address is always IPv6
Emeric Brun3835c0d2020-07-07 09:46:09 +02003707 - 'udp@' -> address is resolved as IPv4 or IPv6 and
Emeric Brun12941c82020-07-07 14:19:42 +02003708 protocol UDP is used. Currently those listeners are
3709 supported only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003710 - 'udp4@' -> address is always IPv4 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003711 is used. Currently those listeners are supported
3712 only in log-forward sections.
Emeric Brun3835c0d2020-07-07 09:46:09 +02003713 - 'udp6@' -> address is always IPv6 and protocol UDP
Emeric Brun12941c82020-07-07 14:19:42 +02003714 is used. Currently those listeners are supported
3715 only in log-forward sections.
Willy Tarreau24709282013-03-10 21:32:12 +01003716 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02003717 - 'abns@' -> address is in abstract namespace (Linux only).
3718 Note: since abstract sockets are not "rebindable", they
3719 do not cope well with multi-process mode during
3720 soft-restart, so it is better to avoid them if
3721 nbproc is greater than 1. The effect is that if the
3722 new process fails to start, only one of the old ones
3723 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01003724 - 'fd@<n>' -> use file descriptor <n> inherited from the
3725 parent. The fd must be bound and may or may not already
3726 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02003727 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
3728 connected unix socket or of a socketpair. The bind waits
3729 to receive a FD over the unix socket and uses it as if it
3730 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02003731 You may want to reference some environment variables in the
3732 address parameter, see section 2.3 about environment
3733 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01003734
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003735 <port_range> is either a unique TCP port, or a port range for which the
3736 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003737 above. The port is mandatory for TCP listeners. Note that in
3738 the case of an IPv6 address, the port is always the number
3739 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003740 - a numerical port (ex: '80')
3741 - a dash-delimited ports range explicitly stating the lower
3742 and upper bounds (ex: '2000-2100') which are included in
3743 the range.
3744
3745 Particular care must be taken against port ranges, because
3746 every <address:port> couple consumes one socket (= a file
3747 descriptor), so it's easy to consume lots of descriptors
3748 with a simple range, and to run out of sockets. Also, each
3749 <address:port> couple must be used only once among all
3750 instances running on a same system. Please note that binding
3751 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04003752 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01003753 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003754
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003755 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01003756 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003757 receive UNIX connections on the socket located at this place.
3758 The path must begin with a slash and by default is absolute.
3759 It can be relative to the prefix defined by "unix-bind" in
3760 the global section. Note that the total length of the prefix
3761 followed by the socket path cannot exceed some system limits
3762 for UNIX sockets, which commonly are set to 107 characters.
3763
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003764 <param*> is a list of parameters common to all sockets declared on the
3765 same line. These numerous parameters depend on OS and build
3766 options and have a complete section dedicated to them. Please
3767 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003768
Willy Tarreau0ba27502007-12-24 16:55:16 +01003769 It is possible to specify a list of address:port combinations delimited by
3770 commas. The frontend will then listen on all of these addresses. There is no
3771 fixed limit to the number of addresses and ports which can be listened on in
3772 a frontend, as well as there is no limit to the number of "bind" statements
3773 in a frontend.
3774
3775 Example :
3776 listen http_proxy
3777 bind :80,:443
3778 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003779 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01003780
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003781 listen http_https_proxy
3782 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02003783 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02003784
Willy Tarreau24709282013-03-10 21:32:12 +01003785 listen http_https_proxy_explicit
3786 bind ipv6@:80
3787 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
3788 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
3789
Willy Tarreaudad36a32013-03-11 01:20:04 +01003790 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02003791 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01003792
Willy Tarreau55dcaf62015-09-27 15:03:15 +02003793 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
3794 sun_path length is used for the address length. Some other programs
3795 such as socat use the string length only by default. Pass the option
3796 ",unix-tightsocklen=0" to any abstract socket definition in socat to
3797 make it compatible with HAProxy's.
3798
Willy Tarreauceb24bc2010-11-09 12:46:41 +01003799 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02003800 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003801
3802
Christopher Fauletff4121f2017-11-22 16:38:49 +01003803bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003804 Limit visibility of an instance to a certain set of processes numbers.
3805 May be used in sections : defaults | frontend | listen | backend
3806 yes | yes | yes | yes
3807 Arguments :
3808 all All process will see this instance. This is the default. It
3809 may be used to override a default value.
3810
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003811 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003812 option may be combined with other numbers.
3813
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003814 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003815 option may be combined with other numbers. Do not use it
3816 with less than 2 processes otherwise some instances might be
3817 missing from all processes.
3818
Christopher Fauletff4121f2017-11-22 16:38:49 +01003819 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003820 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01003821 the machine's word size. Ranges can be partially defined. The
3822 higher bound can be omitted. In such case, it is replaced by
3823 the corresponding maximum value. If a proxy is bound to
3824 process numbers greater than the configured global.nbproc, it
3825 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003826 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003827
3828 This keyword limits binding of certain instances to certain processes. This
3829 is useful in order not to have too many processes listening to the same
3830 ports. For instance, on a dual-core machine, it might make sense to set
3831 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3832 and 'even' instances.
3833
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003834 At the moment, it is not possible to reference more than 32 or 64 processes
3835 using this keyword, but this should be more than enough for most setups.
3836 Please note that 'all' really means all processes regardless of the machine's
3837 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003838
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003839 Each "bind" line may further be limited to a subset of the proxy's processes,
3840 please consult the "process" bind keyword in section 5.1.
3841
Willy Tarreaub369a042014-09-16 13:21:03 +02003842 When a frontend has no explicit "bind-process" line, it tries to bind to all
3843 the processes referenced by its "bind" lines. That means that frontends can
3844 easily adapt to their listeners' processes.
3845
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003846 If some backends are referenced by frontends bound to other processes, the
3847 backend automatically inherits the frontend's processes.
3848
3849 Example :
3850 listen app_ip1
3851 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003852 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003853
3854 listen app_ip2
3855 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003856 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003857
3858 listen management
3859 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003860 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003861
Willy Tarreau110ecc12012-11-15 17:50:01 +01003862 listen management
3863 bind 10.0.0.4:80
3864 bind-process 1-4
3865
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003866 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003867
3868
Willy Tarreau0ba27502007-12-24 16:55:16 +01003869capture cookie <name> len <length>
3870 Capture and log a cookie in the request and in the response.
3871 May be used in sections : defaults | frontend | listen | backend
3872 no | yes | yes | no
3873 Arguments :
3874 <name> is the beginning of the name of the cookie to capture. In order
3875 to match the exact name, simply suffix the name with an equal
3876 sign ('='). The full name will appear in the logs, which is
3877 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003878 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003879
3880 <length> is the maximum number of characters to report in the logs, which
3881 include the cookie name, the equal sign and the value, all in the
3882 standard "name=value" form. The string will be truncated on the
3883 right if it exceeds <length>.
3884
3885 Only the first cookie is captured. Both the "cookie" request headers and the
3886 "set-cookie" response headers are monitored. This is particularly useful to
3887 check for application bugs causing session crossing or stealing between
3888 users, because generally the user's cookies can only change on a login page.
3889
3890 When the cookie was not presented by the client, the associated log column
3891 will report "-". When a request does not cause a cookie to be assigned by the
3892 server, a "-" is reported in the response column.
3893
3894 The capture is performed in the frontend only because it is necessary that
3895 the log format does not change for a given frontend depending on the
3896 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003897 "capture cookie" statement in a frontend. The maximum capture length is set
3898 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3899 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003900
3901 Example:
3902 capture cookie ASPSESSION len 32
3903
3904 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003905 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003906
3907
3908capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003909 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003910 May be used in sections : defaults | frontend | listen | backend
3911 no | yes | yes | no
3912 Arguments :
3913 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003914 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003915 appear in the requests, with the first letter of each word in
3916 upper case. The header name will not appear in the logs, only the
3917 value is reported, but the position in the logs is respected.
3918
3919 <length> is the maximum number of characters to extract from the value and
3920 report in the logs. The string will be truncated on the right if
3921 it exceeds <length>.
3922
Willy Tarreau4460d032012-11-21 23:37:37 +01003923 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003924 value will be added to the logs between braces ('{}'). If multiple headers
3925 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003926 in the same order they were declared in the configuration. Non-existent
3927 headers will be logged just as an empty string. Common uses for request
3928 header captures include the "Host" field in virtual hosting environments, the
3929 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003930 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003931 environments to find where the request came from.
3932
3933 Note that when capturing headers such as "User-agent", some spaces may be
3934 logged, making the log analysis more difficult. Thus be careful about what
3935 you log if you know your log parser is not smart enough to rely on the
3936 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003937
Willy Tarreau0900abb2012-11-22 00:21:46 +01003938 There is no limit to the number of captured request headers nor to their
3939 length, though it is wise to keep them low to limit memory usage per session.
3940 In order to keep log format consistent for a same frontend, header captures
3941 can only be declared in a frontend. It is not possible to specify a capture
3942 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003943
3944 Example:
3945 capture request header Host len 15
3946 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003947 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003948
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003949 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003950 about logging.
3951
3952
3953capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003954 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003955 May be used in sections : defaults | frontend | listen | backend
3956 no | yes | yes | no
3957 Arguments :
3958 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003959 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003960 appear in the response, with the first letter of each word in
3961 upper case. The header name will not appear in the logs, only the
3962 value is reported, but the position in the logs is respected.
3963
3964 <length> is the maximum number of characters to extract from the value and
3965 report in the logs. The string will be truncated on the right if
3966 it exceeds <length>.
3967
Willy Tarreau4460d032012-11-21 23:37:37 +01003968 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003969 result will be added to the logs between braces ('{}') after the captured
3970 request headers. If multiple headers are captured, they will be delimited by
3971 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003972 the configuration. Non-existent headers will be logged just as an empty
3973 string. Common uses for response header captures include the "Content-length"
3974 header which indicates how many bytes are expected to be returned, the
3975 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003976
Willy Tarreau0900abb2012-11-22 00:21:46 +01003977 There is no limit to the number of captured response headers nor to their
3978 length, though it is wise to keep them low to limit memory usage per session.
3979 In order to keep log format consistent for a same frontend, header captures
3980 can only be declared in a frontend. It is not possible to specify a capture
3981 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003982
3983 Example:
3984 capture response header Content-length len 9
3985 capture response header Location len 15
3986
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003987 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003988 about logging.
3989
3990
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09003991clitcpka-cnt <count>
3992 Sets the maximum number of keepalive probes TCP should send before dropping
3993 the connection on the client side.
3994 May be used in sections : defaults | frontend | listen | backend
3995 yes | yes | yes | no
3996 Arguments :
3997 <count> is the maximum number of keepalive probes.
3998
3999 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
4000 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004001 The availability of this setting depends on the operating system. It is
4002 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004003
4004 See also : "option clitcpka", "clitcpka-idle", "clitcpka-intvl".
4005
4006
4007clitcpka-idle <timeout>
4008 Sets the time the connection needs to remain idle before TCP starts sending
4009 keepalive probes, if enabled the sending of TCP keepalive packets on the
4010 client side.
4011 May be used in sections : defaults | frontend | listen | backend
4012 yes | yes | yes | no
4013 Arguments :
4014 <timeout> is the time the connection needs to remain idle before TCP starts
4015 sending keepalive probes. It is specified in seconds by default,
4016 but can be in any other unit if the number is suffixed by the
4017 unit, as explained at the top of this document.
4018
4019 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
4020 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004021 The availability of this setting depends on the operating system. It is
4022 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004023
4024 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-intvl".
4025
4026
4027clitcpka-intvl <timeout>
4028 Sets the time between individual keepalive probes on the client side.
4029 May be used in sections : defaults | frontend | listen | backend
4030 yes | yes | yes | no
4031 Arguments :
4032 <timeout> is the time between individual keepalive probes. It is specified
4033 in seconds by default, but can be in any other unit if the number
4034 is suffixed by the unit, as explained at the top of this
4035 document.
4036
4037 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
4038 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02004039 The availability of this setting depends on the operating system. It is
4040 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09004041
4042 See also : "option clitcpka", "clitcpka-cnt", "clitcpka-idle".
4043
4044
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004045compression algo <algorithm> ...
4046compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02004047compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02004048 Enable HTTP compression.
4049 May be used in sections : defaults | frontend | listen | backend
4050 yes | yes | yes | yes
4051 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004052 algo is followed by the list of supported compression algorithms.
4053 type is followed by the list of MIME types that will be compressed.
4054 offload makes haproxy work as a compression offloader only (see notes).
4055
4056 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01004057 identity this is mostly for debugging, and it was useful for developing
4058 the compression feature. Identity does not apply any change on
4059 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004060
Willy Tarreauc91840a2015-03-28 17:00:39 +01004061 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01004062 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01004063
4064 deflate same as "gzip", but with deflate algorithm and zlib format.
4065 Note that this algorithm has ambiguous support on many
4066 browsers and no support at all from recent ones. It is
4067 strongly recommended not to use it for anything else than
4068 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01004069 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004070
Willy Tarreauc91840a2015-03-28 17:00:39 +01004071 raw-deflate same as "deflate" without the zlib wrapper, and used as an
4072 alternative when the browser wants "deflate". All major
4073 browsers understand it and despite violating the standards,
4074 it is known to work better than "deflate", at least on MSIE
4075 and some versions of Safari. Do not use it in conjunction
4076 with "deflate", use either one or the other since both react
4077 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01004078 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004079
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04004080 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01004081 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004082 If backend servers support HTTP compression, these directives
4083 will be no-op: haproxy will see the compressed response and will not
4084 compress again. If backend servers do not support HTTP compression and
4085 there is Accept-Encoding header in request, haproxy will compress the
4086 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02004087
4088 The "offload" setting makes haproxy remove the Accept-Encoding header to
4089 prevent backend servers from compressing responses. It is strongly
4090 recommended not to do this because this means that all the compression work
4091 will be done on the single point where haproxy is located. However in some
4092 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04004093 with broken HTTP compression implementation which can't be turned off.
4094 In that case haproxy can be used to prevent that gateway from emitting
4095 invalid payloads. In this case, simply removing the header in the
4096 configuration does not work because it applies before the header is parsed,
4097 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02004098 then be used for such scenarios. Note: for now, the "offload" setting is
4099 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02004100
William Lallemand05097442012-11-20 12:14:28 +01004101 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01004102 * the request does not advertise a supported compression algorithm in the
4103 "Accept-Encoding" header
4104 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01004105 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01004106 * response contain neither a "Content-Length" header nor a
4107 "Transfer-Encoding" whose last value is "chunked"
4108 * response contains a "Content-Type" header whose first value starts with
4109 "multipart"
4110 * the response contains the "no-transform" value in the "Cache-control"
4111 header
4112 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
4113 and later
4114 * The response contains a "Content-Encoding" header, indicating that the
4115 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01004116 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01004117
Tim Duesterhusb229f012019-01-29 16:38:56 +01004118 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01004119
William Lallemand82fe75c2012-10-23 10:25:10 +02004120 Examples :
4121 compression algo gzip
4122 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01004123
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004124
Willy Tarreau55165fe2009-05-10 12:02:55 +02004125cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02004126 [ postonly ] [ preserve ] [ httponly ] [ secure ]
4127 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Christopher Faulet2f533902020-01-21 11:06:48 +01004128 [ dynamic ] [ attr <value> ]*
Willy Tarreau0ba27502007-12-24 16:55:16 +01004129 Enable cookie-based persistence in a backend.
4130 May be used in sections : defaults | frontend | listen | backend
4131 yes | no | yes | yes
4132 Arguments :
4133 <name> is the name of the cookie which will be monitored, modified or
4134 inserted in order to bring persistence. This cookie is sent to
4135 the client via a "Set-Cookie" header in the response, and is
4136 brought back by the client in a "Cookie" header in all requests.
4137 Special care should be taken to choose a name which does not
4138 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01004139 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004140 HTTP/HTTPS), care should be taken to use different cookie names
4141 between all backends if persistence between them is not desired.
4142
4143 rewrite This keyword indicates that the cookie will be provided by the
4144 server and that haproxy will have to modify its value to set the
4145 server's identifier in it. This mode is handy when the management
4146 of complex combinations of "Set-cookie" and "Cache-control"
4147 headers is left to the application. The application can then
4148 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004149 cookie. Since all responses should be monitored, this mode
4150 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01004151 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004152 start with this mode for new deployments. This keyword is
4153 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004154
4155 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02004156 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004157
Willy Tarreaua79094d2010-08-31 22:54:15 +02004158 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004159 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02004160 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01004161 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004162 existing configurations running in the "rewrite" mode. The cookie
4163 will only be a session cookie and will not be stored on the
4164 client's disk. By default, unless the "indirect" option is added,
4165 the server will see the cookies emitted by the client. Due to
4166 caching effects, it is generally wise to add the "nocache" or
4167 "postonly" keywords (see below). The "insert" keyword is not
4168 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004169
4170 prefix This keyword indicates that instead of relying on a dedicated
4171 cookie for the persistence, an existing one will be completed.
4172 This may be needed in some specific environments where the client
4173 does not support more than one single cookie and the application
4174 already needs it. In this case, whenever the server sets a cookie
4175 named <name>, it will be prefixed with the server's identifier
4176 and a delimiter. The prefix will be removed from all client
4177 requests so that the server still finds the cookie it emitted.
4178 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01004179 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02004180 not compatible with "rewrite" and "insert". Note: it is highly
4181 recommended not to use "indirect" with "prefix", otherwise server
4182 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004183
Willy Tarreaua79094d2010-08-31 22:54:15 +02004184 indirect When this option is specified, no cookie will be emitted to a
4185 client which already has a valid one for the server which has
4186 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004187 it will be removed, unless the "preserve" option is also set. In
4188 "insert" mode, this will additionally remove cookies from the
4189 requests transmitted to the server, making the persistence
4190 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02004191 Note: it is highly recommended not to use "indirect" with
4192 "prefix", otherwise server cookie updates would not be sent to
4193 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004194
4195 nocache This option is recommended in conjunction with the insert mode
4196 when there is a cache between the client and HAProxy, as it
4197 ensures that a cacheable response will be tagged non-cacheable if
4198 a cookie needs to be inserted. This is important because if all
4199 persistence cookies are added on a cacheable home page for
4200 instance, then all customers will then fetch the page from an
4201 outer cache and will all share the same persistence cookie,
4202 leading to one server receiving much more traffic than others.
4203 See also the "insert" and "postonly" options.
4204
4205 postonly This option ensures that cookie insertion will only be performed
4206 on responses to POST requests. It is an alternative to the
4207 "nocache" option, because POST responses are not cacheable, so
4208 this ensures that the persistence cookie will never get cached.
4209 Since most sites do not need any sort of persistence before the
4210 first POST which generally is a login request, this is a very
4211 efficient method to optimize caching without risking to find a
4212 persistence cookie in the cache.
4213 See also the "insert" and "nocache" options.
4214
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004215 preserve This option may only be used with "insert" and/or "indirect". It
4216 allows the server to emit the persistence cookie itself. In this
4217 case, if a cookie is found in the response, haproxy will leave it
4218 untouched. This is useful in order to end persistence after a
4219 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01004220 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02004221 the past. By combining this mechanism with the "disable-on-404"
4222 check option, it is possible to perform a completely graceful
4223 shutdown because users will definitely leave the server after
4224 they logout.
4225
Willy Tarreau4992dd22012-05-31 21:02:17 +02004226 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
4227 when a cookie is inserted. This attribute is used so that a
4228 user agent doesn't share the cookie with non-HTTP components.
4229 Please check RFC6265 for more information on this attribute.
4230
4231 secure This option tells haproxy to add a "Secure" cookie attribute when
4232 a cookie is inserted. This attribute is used so that a user agent
4233 never emits this cookie over non-secure channels, which means
4234 that a cookie learned with this flag will be presented only over
4235 SSL/TLS connections. Please check RFC6265 for more information on
4236 this attribute.
4237
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02004238 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004239 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01004240 name. If the domain begins with a dot, the browser is allowed to
4241 use it for any host ending with that name. It is also possible to
4242 specify several domain names by invoking this option multiple
4243 times. Some browsers might have small limits on the number of
4244 domains, so be careful when doing that. For the record, sending
4245 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02004246
Willy Tarreau996a92c2010-10-13 19:30:47 +02004247 maxidle This option allows inserted cookies to be ignored after some idle
4248 time. It only works with insert-mode cookies. When a cookie is
4249 sent to the client, the date this cookie was emitted is sent too.
4250 Upon further presentations of this cookie, if the date is older
4251 than the delay indicated by the parameter (in seconds), it will
4252 be ignored. Otherwise, it will be refreshed if needed when the
4253 response is sent to the client. This is particularly useful to
4254 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004255 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02004256 this option is set and a cookie has no date, it is always
4257 accepted, but gets refreshed in the response. This maintains the
4258 ability for admins to access their sites. Cookies that have a
4259 date in the future further than 24 hours are ignored. Doing so
4260 lets admins fix timezone issues without risking kicking users off
4261 the site.
4262
4263 maxlife This option allows inserted cookies to be ignored after some life
4264 time, whether they're in use or not. It only works with insert
4265 mode cookies. When a cookie is first sent to the client, the date
4266 this cookie was emitted is sent too. Upon further presentations
4267 of this cookie, if the date is older than the delay indicated by
4268 the parameter (in seconds), it will be ignored. If the cookie in
4269 the request has no date, it is accepted and a date will be set.
4270 Cookies that have a date in the future further than 24 hours are
4271 ignored. Doing so lets admins fix timezone issues without risking
4272 kicking users off the site. Contrary to maxidle, this value is
4273 not refreshed, only the first visit date counts. Both maxidle and
4274 maxlife may be used at the time. This is particularly useful to
4275 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01004276 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02004277 is stronger than the maxidle method in that it forces a
4278 redispatch after some absolute delay.
4279
Olivier Houchard4e694042017-03-14 20:01:29 +01004280 dynamic Activate dynamic cookies. When used, a session cookie is
4281 dynamically created for each server, based on the IP and port
4282 of the server, and a secret key, specified in the
4283 "dynamic-cookie-key" backend directive.
4284 The cookie will be regenerated each time the IP address change,
4285 and is only generated for IPv4/IPv6.
4286
Christopher Faulet2f533902020-01-21 11:06:48 +01004287 attr This option tells haproxy to add an extra attribute when a
4288 cookie is inserted. The attribute value can contain any
4289 characters except control ones or ";". This option may be
4290 repeated.
4291
Willy Tarreau0ba27502007-12-24 16:55:16 +01004292 There can be only one persistence cookie per HTTP backend, and it can be
4293 declared in a defaults section. The value of the cookie will be the value
4294 indicated after the "cookie" keyword in a "server" statement. If no cookie
4295 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02004296
Willy Tarreau0ba27502007-12-24 16:55:16 +01004297 Examples :
4298 cookie JSESSIONID prefix
4299 cookie SRV insert indirect nocache
4300 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02004301 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01004302
Willy Tarreau294d0f02015-08-10 19:40:12 +02004303 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01004304
Willy Tarreau983e01e2010-01-11 18:42:06 +01004305
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004306declare capture [ request | response ] len <length>
4307 Declares a capture slot.
4308 May be used in sections : defaults | frontend | listen | backend
4309 no | yes | yes | no
4310 Arguments:
4311 <length> is the length allowed for the capture.
4312
4313 This declaration is only available in the frontend or listen section, but the
4314 reserved slot can be used in the backends. The "request" keyword allocates a
4315 capture slot for use in the request, and "response" allocates a capture slot
4316 for use in the response.
4317
4318 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02004319 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02004320 "http-request capture" and "http-response capture".
4321
4322
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004323default-server [param*]
4324 Change default options for a server in a backend
4325 May be used in sections : defaults | frontend | listen | backend
4326 yes | no | yes | yes
4327 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01004328 <param*> is a list of parameters for this server. The "default-server"
4329 keyword accepts an important number of options and has a complete
4330 section dedicated to it. Please refer to section 5 for more
4331 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004332
Willy Tarreau983e01e2010-01-11 18:42:06 +01004333 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01004334 default-server inter 1000 weight 13
4335
4336 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01004337
Willy Tarreau983e01e2010-01-11 18:42:06 +01004338
Willy Tarreau0ba27502007-12-24 16:55:16 +01004339default_backend <backend>
4340 Specify the backend to use when no "use_backend" rule has been matched.
4341 May be used in sections : defaults | frontend | listen | backend
4342 yes | yes | yes | no
4343 Arguments :
4344 <backend> is the name of the backend to use.
4345
4346 When doing content-switching between frontend and backends using the
4347 "use_backend" keyword, it is often useful to indicate which backend will be
4348 used when no rule has matched. It generally is the dynamic backend which
4349 will catch all undetermined requests.
4350
Willy Tarreau0ba27502007-12-24 16:55:16 +01004351 Example :
4352
4353 use_backend dynamic if url_dyn
4354 use_backend static if url_css url_img extension_img
4355 default_backend dynamic
4356
Willy Tarreau98d04852015-05-26 12:18:29 +02004357 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004358
Willy Tarreau0ba27502007-12-24 16:55:16 +01004359
Baptiste Assmann27f51342013-10-09 06:51:49 +02004360description <string>
4361 Describe a listen, frontend or backend.
4362 May be used in sections : defaults | frontend | listen | backend
4363 no | yes | yes | yes
4364 Arguments : string
4365
4366 Allows to add a sentence to describe the related object in the HAProxy HTML
4367 stats page. The description will be printed on the right of the object name
4368 it describes.
4369 No need to backslash spaces in the <string> arguments.
4370
4371
Willy Tarreau0ba27502007-12-24 16:55:16 +01004372disabled
4373 Disable a proxy, frontend or backend.
4374 May be used in sections : defaults | frontend | listen | backend
4375 yes | yes | yes | yes
4376 Arguments : none
4377
4378 The "disabled" keyword is used to disable an instance, mainly in order to
4379 liberate a listening port or to temporarily disable a service. The instance
4380 will still be created and its configuration will be checked, but it will be
4381 created in the "stopped" state and will appear as such in the statistics. It
4382 will not receive any traffic nor will it send any health-checks or logs. It
4383 is possible to disable many instances at once by adding the "disabled"
4384 keyword in a "defaults" section.
4385
4386 See also : "enabled"
4387
4388
Willy Tarreau5ce94572010-06-07 14:35:41 +02004389dispatch <address>:<port>
4390 Set a default server address
4391 May be used in sections : defaults | frontend | listen | backend
4392 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02004393 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02004394
4395 <address> is the IPv4 address of the default server. Alternatively, a
4396 resolvable hostname is supported, but this name will be resolved
4397 during start-up.
4398
4399 <ports> is a mandatory port specification. All connections will be sent
4400 to this port, and it is not permitted to use port offsets as is
4401 possible with normal servers.
4402
Willy Tarreau787aed52011-04-15 06:45:37 +02004403 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02004404 server can take the connection. In the past it was used to forward non
4405 persistent connections to an auxiliary load balancer. Due to its simple
4406 syntax, it has also been used for simple TCP relays. It is recommended not to
4407 use it for more clarity, and to use the "server" directive instead.
4408
4409 See also : "server"
4410
Olivier Houchard4e694042017-03-14 20:01:29 +01004411
4412dynamic-cookie-key <string>
4413 Set the dynamic cookie secret key for a backend.
4414 May be used in sections : defaults | frontend | listen | backend
4415 yes | no | yes | yes
4416 Arguments : The secret key to be used.
4417
4418 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01004419 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01004420 specified on the "server" line), using a hash of the IP address of the
4421 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01004422 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01004423 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02004424
Willy Tarreau0ba27502007-12-24 16:55:16 +01004425enabled
4426 Enable a proxy, frontend or backend.
4427 May be used in sections : defaults | frontend | listen | backend
4428 yes | yes | yes | yes
4429 Arguments : none
4430
4431 The "enabled" keyword is used to explicitly enable an instance, when the
4432 defaults has been set to "disabled". This is very rarely used.
4433
4434 See also : "disabled"
4435
4436
4437errorfile <code> <file>
4438 Return a file contents instead of errors generated by HAProxy
4439 May be used in sections : defaults | frontend | listen | backend
4440 yes | yes | yes | yes
4441 Arguments :
4442 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004443 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004444 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004445
4446 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004447 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01004448 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01004449 error pages, and to use absolute paths, since files are read
4450 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01004451
4452 It is important to understand that this keyword is not meant to rewrite
4453 errors returned by the server, but errors detected and returned by HAProxy.
4454 This is why the list of supported errors is limited to a small set.
4455
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004456 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4457
Christopher Faulet70170672020-05-18 17:42:48 +02004458 The files are parsed when HAProxy starts and must be valid according to the
4459 HTTP specification. They should not exceed the configured buffer size
4460 (BUFSIZE), which generally is 16 kB, otherwise an internal error will be
4461 returned. It is also wise not to put any reference to local contents
4462 (e.g. images) in order to avoid loops between the client and HAProxy when all
4463 servers are down, causing an error to be returned instead of an
4464 image. Finally, The response cannot exceed (tune.bufsize - tune.maxrewrite)
4465 so that "http-after-response" rules still have room to operate (see
4466 "tune.maxrewrite").
Willy Tarreau59140a22009-02-22 12:02:30 +01004467
Willy Tarreau0ba27502007-12-24 16:55:16 +01004468 The files are read at the same time as the configuration and kept in memory.
4469 For this reason, the errors continue to be returned even when the process is
4470 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01004471 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01004472 403 status code and interrogating a blocked URL.
4473
Christopher Faulet3b967c12020-05-15 15:47:44 +02004474 See also : "http-error", "errorloc", "errorloc302", "errorloc303"
Willy Tarreau0ba27502007-12-24 16:55:16 +01004475
Willy Tarreau59140a22009-02-22 12:02:30 +01004476 Example :
4477 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01004478 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01004479 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
4480 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
4481
Willy Tarreau2769aa02007-12-27 18:26:09 +01004482
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004483errorfiles <name> [<code> ...]
4484 Import, fully or partially, the error files defined in the <name> http-errors
4485 section.
4486 May be used in sections : defaults | frontend | listen | backend
4487 yes | yes | yes | yes
4488 Arguments :
4489 <name> is the name of an existing http-errors section.
4490
4491 <code> is a HTTP status code. Several status code may be listed.
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004492 Currently, HAProxy is capable of generating codes 200, 400, 401,
Christopher Faulete095f312020-12-07 11:22:24 +01004493 403, 404, 405, 407, 408, 410, 413, 425, 429, 500, 501, 502, 503,
4494 and 504.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004495
4496 Errors defined in the http-errors section with the name <name> are imported
4497 in the current proxy. If no status code is specified, all error files of the
4498 http-errors section are imported. Otherwise, only error files associated to
4499 the listed status code are imported. Those error files override the already
4500 defined custom errors for the proxy. And they may be overridden by following
Daniel Corbett67a82712020-07-06 23:01:19 -04004501 ones. Functionally, it is exactly the same as declaring all error files by
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004502 hand using "errorfile" directives.
4503
Christopher Faulet3b967c12020-05-15 15:47:44 +02004504 See also : "http-error", "errorfile", "errorloc", "errorloc302" ,
4505 "errorloc303" and section 3.8 about http-errors.
Christopher Faulet76edc0f2020-01-13 15:52:01 +01004506
4507 Example :
4508 errorfiles generic
4509 errorfiles site-1 403 404
4510
4511
Willy Tarreau2769aa02007-12-27 18:26:09 +01004512errorloc <code> <url>
4513errorloc302 <code> <url>
4514 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4515 May be used in sections : defaults | frontend | listen | backend
4516 yes | yes | yes | yes
4517 Arguments :
4518 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004519 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004520 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004521
4522 <url> it is the exact contents of the "Location" header. It may contain
4523 either a relative URI to an error page hosted on the same site,
4524 or an absolute URI designating an error page on another site.
4525 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004526 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004527
4528 It is important to understand that this keyword is not meant to rewrite
4529 errors returned by the server, but errors detected and returned by HAProxy.
4530 This is why the list of supported errors is limited to a small set.
4531
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004532 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4533
Willy Tarreau2769aa02007-12-27 18:26:09 +01004534 Note that both keyword return the HTTP 302 status code, which tells the
4535 client to fetch the designated URL using the same HTTP method. This can be
4536 quite problematic in case of non-GET methods such as POST, because the URL
4537 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01004538 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01004539 status code, indicating to the client that the URL must be fetched with a GET
4540 request.
4541
Christopher Faulet3b967c12020-05-15 15:47:44 +02004542 See also : "http-error", "errorfile", "errorloc303"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004543
4544
4545errorloc303 <code> <url>
4546 Return an HTTP redirection to a URL instead of errors generated by HAProxy
4547 May be used in sections : defaults | frontend | listen | backend
4548 yes | yes | yes | yes
4549 Arguments :
4550 <code> is the HTTP status code. Currently, HAProxy is capable of
Christopher Faulet612f2ea2020-05-27 09:57:28 +02004551 generating codes 200, 400, 401, 403, 404, 405, 407, 408, 410,
Christopher Faulete095f312020-12-07 11:22:24 +01004552 413, 425, 429, 500, 501, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004553
4554 <url> it is the exact contents of the "Location" header. It may contain
4555 either a relative URI to an error page hosted on the same site,
4556 or an absolute URI designating an error page on another site.
4557 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01004558 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01004559
4560 It is important to understand that this keyword is not meant to rewrite
4561 errors returned by the server, but errors detected and returned by HAProxy.
4562 This is why the list of supported errors is limited to a small set.
4563
Willy Tarreauae94d4d2011-05-11 16:28:49 +02004564 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
4565
Willy Tarreau2769aa02007-12-27 18:26:09 +01004566 Note that both keyword return the HTTP 303 status code, which tells the
4567 client to fetch the designated URL using the same HTTP GET method. This
4568 solves the usual problems associated with "errorloc" and the 302 code. It is
4569 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01004570 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004571
Christopher Faulet3b967c12020-05-15 15:47:44 +02004572 See also : "http-error", "errorfile", "errorloc", "errorloc302"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004573
4574
Simon Horman51a1cf62015-02-03 13:00:44 +09004575email-alert from <emailaddr>
4576 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01004577 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09004578 May be used in sections: defaults | frontend | listen | backend
4579 yes | yes | yes | yes
4580
4581 Arguments :
4582
4583 <emailaddr> is the from email address to use when sending email alerts
4584
4585 Also requires "email-alert mailers" and "email-alert to" to be set
4586 and if so sending email alerts is enabled for the proxy.
4587
Simon Horman64e34162015-02-06 11:11:57 +09004588 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02004589 "email-alert myhostname", "email-alert to", section 3.6 about
4590 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09004591
4592
4593email-alert level <level>
4594 Declare the maximum log level of messages for which email alerts will be
4595 sent. This acts as a filter on the sending of email alerts.
4596 May be used in sections: defaults | frontend | listen | backend
4597 yes | yes | yes | yes
4598
4599 Arguments :
4600
4601 <level> One of the 8 syslog levels:
4602 emerg alert crit err warning notice info debug
4603 The above syslog levels are ordered from lowest to highest.
4604
4605 By default level is alert
4606
4607 Also requires "email-alert from", "email-alert mailers" and
4608 "email-alert to" to be set and if so sending email alerts is enabled
4609 for the proxy.
4610
Simon Horman1421e212015-04-30 13:10:35 +09004611 Alerts are sent when :
4612
4613 * An un-paused server is marked as down and <level> is alert or lower
4614 * A paused server is marked as down and <level> is notice or lower
4615 * A server is marked as up or enters the drain state and <level>
4616 is notice or lower
4617 * "option log-health-checks" is enabled, <level> is info or lower,
4618 and a health check status update occurs
4619
Simon Horman64e34162015-02-06 11:11:57 +09004620 See also : "email-alert from", "email-alert mailers",
4621 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09004622 section 3.6 about mailers.
4623
4624
4625email-alert mailers <mailersect>
4626 Declare the mailers to be used when sending email alerts
4627 May be used in sections: defaults | frontend | listen | backend
4628 yes | yes | yes | yes
4629
4630 Arguments :
4631
4632 <mailersect> is the name of the mailers section to send email alerts.
4633
4634 Also requires "email-alert from" and "email-alert to" to be set
4635 and if so sending email alerts is enabled for the proxy.
4636
Simon Horman64e34162015-02-06 11:11:57 +09004637 See also : "email-alert from", "email-alert level", "email-alert myhostname",
4638 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004639
4640
4641email-alert myhostname <hostname>
4642 Declare the to hostname address to be used when communicating with
4643 mailers.
4644 May be used in sections: defaults | frontend | listen | backend
4645 yes | yes | yes | yes
4646
4647 Arguments :
4648
Baptiste Assmann738bad92015-12-21 15:27:53 +01004649 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09004650
4651 By default the systems hostname is used.
4652
4653 Also requires "email-alert from", "email-alert mailers" and
4654 "email-alert to" to be set and if so sending email alerts is enabled
4655 for the proxy.
4656
Simon Horman64e34162015-02-06 11:11:57 +09004657 See also : "email-alert from", "email-alert level", "email-alert mailers",
4658 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09004659
4660
4661email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01004662 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09004663 header of email alerts. This is the address that email alerts are sent to.
4664 May be used in sections: defaults | frontend | listen | backend
4665 yes | yes | yes | yes
4666
4667 Arguments :
4668
4669 <emailaddr> is the to email address to use when sending email alerts
4670
4671 Also requires "email-alert mailers" and "email-alert to" to be set
4672 and if so sending email alerts is enabled for the proxy.
4673
Simon Horman64e34162015-02-06 11:11:57 +09004674 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09004675 "email-alert myhostname", section 3.6 about mailers.
4676
4677
Willy Tarreau4de91492010-01-22 19:10:05 +01004678force-persist { if | unless } <condition>
4679 Declare a condition to force persistence on down servers
4680 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01004681 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01004682
4683 By default, requests are not dispatched to down servers. It is possible to
4684 force this using "option persist", but it is unconditional and redispatches
4685 to a valid server if "option redispatch" is set. That leaves with very little
4686 possibilities to force some requests to reach a server which is artificially
4687 marked down for maintenance operations.
4688
4689 The "force-persist" statement allows one to declare various ACL-based
4690 conditions which, when met, will cause a request to ignore the down status of
4691 a server and still try to connect to it. That makes it possible to start a
4692 server, still replying an error to the health checks, and run a specially
4693 configured browser to test the service. Among the handy methods, one could
4694 use a specific source IP address, or a specific cookie. The cookie also has
4695 the advantage that it can easily be added/removed on the browser from a test
4696 page. Once the service is validated, it is then possible to open the service
4697 to the world by returning a valid response to health checks.
4698
4699 The forced persistence is enabled when an "if" condition is met, or unless an
4700 "unless" condition is met. The final redispatch is always disabled when this
4701 is used.
4702
Cyril Bonté0d4bf012010-04-25 23:21:46 +02004703 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02004704 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01004705
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004706
4707filter <name> [param*]
4708 Add the filter <name> in the filter list attached to the proxy.
4709 May be used in sections : defaults | frontend | listen | backend
4710 no | yes | yes | yes
4711 Arguments :
4712 <name> is the name of the filter. Officially supported filters are
4713 referenced in section 9.
4714
Tim Düsterhus4896c442016-11-29 02:15:19 +01004715 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004716 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01004717 filter. Please refer to the documentation of the corresponding
4718 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02004719
4720 Multiple occurrences of the filter line can be used for the same proxy. The
4721 same filter can be referenced many times if needed.
4722
4723 Example:
4724 listen
4725 bind *:80
4726
4727 filter trace name BEFORE-HTTP-COMP
4728 filter compression
4729 filter trace name AFTER-HTTP-COMP
4730
4731 compression algo gzip
4732 compression offload
4733
4734 server srv1 192.168.0.1:80
4735
4736 See also : section 9.
4737
Willy Tarreau4de91492010-01-22 19:10:05 +01004738
Willy Tarreau2769aa02007-12-27 18:26:09 +01004739fullconn <conns>
4740 Specify at what backend load the servers will reach their maxconn
4741 May be used in sections : defaults | frontend | listen | backend
4742 yes | no | yes | yes
4743 Arguments :
4744 <conns> is the number of connections on the backend which will make the
4745 servers use the maximal number of connections.
4746
Willy Tarreau198a7442008-01-17 12:05:32 +01004747 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01004748 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01004749 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01004750 load. The server will then always accept at least <minconn> connections,
4751 never more than <maxconn>, and the limit will be on the ramp between both
4752 values when the backend has less than <conns> concurrent connections. This
4753 makes it possible to limit the load on the servers during normal loads, but
4754 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004755 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01004756
Willy Tarreaufbb78422011-06-05 15:38:35 +02004757 Since it's hard to get this value right, haproxy automatically sets it to
4758 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01004759 backend (based on "use_backend" and "default_backend" rules). That way it's
4760 safe to leave it unset. However, "use_backend" involving dynamic names are
4761 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02004762
Willy Tarreau2769aa02007-12-27 18:26:09 +01004763 Example :
4764 # The servers will accept between 100 and 1000 concurrent connections each
4765 # and the maximum of 1000 will be reached when the backend reaches 10000
4766 # connections.
4767 backend dynamic
4768 fullconn 10000
4769 server srv1 dyn1:80 minconn 100 maxconn 1000
4770 server srv2 dyn2:80 minconn 100 maxconn 1000
4771
4772 See also : "maxconn", "server"
4773
4774
Willy Tarreauab0a5192020-10-09 19:07:01 +02004775grace <time> (deprecated)
Willy Tarreau2769aa02007-12-27 18:26:09 +01004776 Maintain a proxy operational for some time after a soft stop
4777 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01004778 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01004779 Arguments :
4780 <time> is the time (by default in milliseconds) for which the instance
4781 will remain operational with the frontend sockets still listening
4782 when a soft-stop is received via the SIGUSR1 signal.
4783
4784 This may be used to ensure that the services disappear in a certain order.
4785 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01004786 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01004787 needed by the equipment to detect the failure.
4788
4789 Note that currently, there is very little benefit in using this parameter,
4790 and it may in fact complicate the soft-reconfiguration process more than
4791 simplify it.
4792
Willy Tarreau0ba27502007-12-24 16:55:16 +01004793
Andrew Rodland17be45e2016-10-25 17:04:12 -04004794hash-balance-factor <factor>
4795 Specify the balancing factor for bounded-load consistent hashing
4796 May be used in sections : defaults | frontend | listen | backend
4797 yes | no | no | yes
4798 Arguments :
4799 <factor> is the control for the maximum number of concurrent requests to
4800 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01004801 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04004802
4803 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
4804 enables an algorithm that prevents any one server from getting too many
4805 requests at once, even if some hash buckets receive many more requests than
4806 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
4807 <factor> is a percentage greater than 100. For example, if <factor> is 150,
4808 then no server will be allowed to have a load more than 1.5 times the average.
4809 If server weights are used, they will be respected.
4810
4811 If the first-choice server is disqualified, the algorithm will choose another
4812 server based on the request hash, until a server with additional capacity is
4813 found. A higher <factor> allows more imbalance between the servers, while a
4814 lower <factor> means that more servers will be checked on average, affecting
4815 performance. Reasonable values are from 125 to 200.
4816
Willy Tarreau760e81d2018-05-03 07:20:40 +02004817 This setting is also used by "balance random" which internally relies on the
4818 consistent hashing mechanism.
4819
Andrew Rodland17be45e2016-10-25 17:04:12 -04004820 See also : "balance" and "hash-type".
4821
4822
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004823hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004824 Specify a method to use for mapping hashes to servers
4825 May be used in sections : defaults | frontend | listen | backend
4826 yes | no | yes | yes
4827 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04004828 <method> is the method used to select a server from the hash computed by
4829 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004830
Bhaskar98634f02013-10-29 23:30:51 -04004831 map-based the hash table is a static array containing all alive servers.
4832 The hashes will be very smooth, will consider weights, but
4833 will be static in that weight changes while a server is up
4834 will be ignored. This means that there will be no slow start.
4835 Also, since a server is selected by its position in the array,
4836 most mappings are changed when the server count changes. This
4837 means that when a server goes up or down, or when a server is
4838 added to a farm, most connections will be redistributed to
4839 different servers. This can be inconvenient with caches for
4840 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01004841
Bhaskar98634f02013-10-29 23:30:51 -04004842 consistent the hash table is a tree filled with many occurrences of each
4843 server. The hash key is looked up in the tree and the closest
4844 server is chosen. This hash is dynamic, it supports changing
4845 weights while the servers are up, so it is compatible with the
4846 slow start feature. It has the advantage that when a server
4847 goes up or down, only its associations are moved. When a
4848 server is added to the farm, only a few part of the mappings
4849 are redistributed, making it an ideal method for caches.
4850 However, due to its principle, the distribution will never be
4851 very smooth and it may sometimes be necessary to adjust a
4852 server's weight or its ID to get a more balanced distribution.
4853 In order to get the same distribution on multiple load
4854 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004855 same IDs. Note: consistent hash uses sdbm and avalanche if no
4856 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04004857
4858 <function> is the hash function to be used :
4859
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004860 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04004861 reimplementation of ndbm) database library. It was found to do
4862 well in scrambling bits, causing better distribution of the keys
4863 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004864 function with good distribution, unless the total server weight
4865 is a multiple of 64, in which case applying the avalanche
4866 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04004867
4868 djb2 this function was first proposed by Dan Bernstein many years ago
4869 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004870 function provides a better distribution than sdbm. It generally
4871 works well with text-based inputs though it can perform extremely
4872 poorly with numeric-only input or when the total server weight is
4873 a multiple of 33, unless the avalanche modifier is also used.
4874
Willy Tarreaua0f42712013-11-14 14:30:35 +01004875 wt6 this function was designed for haproxy while testing other
4876 functions in the past. It is not as smooth as the other ones, but
4877 is much less sensible to the input data set or to the number of
4878 servers. It can make sense as an alternative to sdbm+avalanche or
4879 djb2+avalanche for consistent hashing or when hashing on numeric
4880 data such as a source IP address or a visitor identifier in a URL
4881 parameter.
4882
Willy Tarreau324f07f2015-01-20 19:44:50 +01004883 crc32 this is the most common CRC32 implementation as used in Ethernet,
4884 gzip, PNG, etc. It is slower than the other ones but may provide
4885 a better distribution or less predictable results especially when
4886 used on strings.
4887
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004888 <modifier> indicates an optional method applied after hashing the key :
4889
4890 avalanche This directive indicates that the result from the hash
4891 function above should not be used in its raw form but that
4892 a 4-byte full avalanche hash must be applied first. The
4893 purpose of this step is to mix the resulting bits from the
4894 previous hash in order to avoid any undesired effect when
4895 the input contains some limited values or when the number of
4896 servers is a multiple of one of the hash's components (64
4897 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
4898 result less predictable, but it's also not as smooth as when
4899 using the original function. Some testing might be needed
4900 with some workloads. This hash is one of the many proposed
4901 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004902
Bhaskar98634f02013-10-29 23:30:51 -04004903 The default hash type is "map-based" and is recommended for most usages. The
4904 default function is "sdbm", the selection of a function should be based on
4905 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004906
Andrew Rodland17be45e2016-10-25 17:04:12 -04004907 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004908
4909
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004910http-after-response <action> <options...> [ { if | unless } <condition> ]
4911 Access control for all Layer 7 responses (server, applet/service and internal
4912 ones).
4913
4914 May be used in sections: defaults | frontend | listen | backend
4915 no | yes | yes | yes
4916
4917 The http-after-response statement defines a set of rules which apply to layer
4918 7 processing. The rules are evaluated in their declaration order when they
4919 are met in a frontend, listen or backend section. Any rule may optionally be
4920 followed by an ACL-based condition, in which case it will only be evaluated
4921 if the condition is true. Since these rules apply on responses, the backend
4922 rules are applied first, followed by the frontend's rules.
4923
4924 Unlike http-response rules, these ones are applied on all responses, the
4925 server ones but also to all responses generated by HAProxy. These rules are
4926 evaluated at the end of the responses analysis, before the data forwarding.
4927
4928 The first keyword is the rule's action. The supported actions are described
4929 below.
4930
4931 There is no limit to the number of http-after-response statements per
4932 instance.
4933
Christopher Fauletd5ac6de2020-12-02 08:40:14 +01004934 Note: Errors emitted in early stage of the request parsing are handled by the
4935 multiplexer at a lower level, before any http analysis. Thus no
4936 http-after-response ruleset is evaluated on these errors.
4937
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004938 Example:
4939 http-after-response set-header Strict-Transport-Security "max-age=31536000"
4940 http-after-response set-header Cache-Control "no-store,no-cache,private"
4941 http-after-response set-header Pragma "no-cache"
4942
4943http-after-response add-header <name> <fmt> [ { if | unless } <condition> ]
4944
4945 This appends an HTTP header field whose name is specified in <name> and whose
4946 value is defined by <fmt> which follows the log-format rules (see Custom Log
4947 Format in section 8.2.4). This may be used to send a cookie to a client for
4948 example, or to pass some internal information.
4949 This rule is not final, so it is possible to add other similar rules.
4950 Note that header addition is performed immediately, so one rule might reuse
4951 the resulting header from a previous rule.
4952
4953http-after-response allow [ { if | unless } <condition> ]
4954
4955 This stops the evaluation of the rules and lets the response pass the check.
4956 No further "http-after-response" rules are evaluated.
4957
Maciej Zdebebdd4c52020-11-20 13:58:48 +00004958http-after-response del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004959
Maciej Zdebebdd4c52020-11-20 13:58:48 +00004960 This removes all HTTP header fields whose name is specified in <name>. <meth>
4961 is the matching method, applied on the header name. Supported matching methods
4962 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
4963 (substring match) and "reg" (regex match). If not specified, exact matching
4964 method is used.
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01004965
4966http-after-response replace-header <name> <regex-match> <replace-fmt>
4967 [ { if | unless } <condition> ]
4968
4969 This works like "http-response replace-header".
4970
4971 Example:
4972 http-after-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
4973
4974 # applied to:
4975 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
4976
4977 # outputs:
4978 Set-Cookie: C=1;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT
4979
4980 # assuming the backend IP is 192.168.1.20.
4981
4982http-after-response replace-value <name> <regex-match> <replace-fmt>
4983 [ { if | unless } <condition> ]
4984
4985 This works like "http-response replace-value".
4986
4987 Example:
4988 http-after-response replace-value Cache-control ^public$ private
4989
4990 # applied to:
4991 Cache-Control: max-age=3600, public
4992
4993 # outputs:
4994 Cache-Control: max-age=3600, private
4995
4996http-after-response set-header <name> <fmt> [ { if | unless } <condition> ]
4997
4998 This does the same as "add-header" except that the header name is first
4999 removed if it existed. This is useful when passing security information to
5000 the server, where the header must not be manipulated by external users.
5001
5002http-after-response set-status <status> [reason <str>]
5003 [ { if | unless } <condition> ]
5004
5005 This replaces the response status code with <status> which must be an integer
5006 between 100 and 999. Optionally, a custom reason text can be provided defined
5007 by <str>, or the default reason for the specified code will be used as a
5008 fallback.
5009
5010 Example:
5011 # return "431 Request Header Fields Too Large"
5012 http-response set-status 431
5013 # return "503 Slow Down", custom reason
5014 http-response set-status 503 reason "Slow Down"
5015
5016http-after-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5017
5018 This is used to set the contents of a variable. The variable is declared
5019 inline.
5020
5021 Arguments:
5022 <var-name> The name of the variable starts with an indication about its
5023 scope. The scopes allowed are:
5024 "proc" : the variable is shared with the whole process
5025 "sess" : the variable is shared with the whole session
5026 "txn" : the variable is shared with the transaction
5027 (request and response)
5028 "req" : the variable is shared only during request
5029 processing
5030 "res" : the variable is shared only during response
5031 processing
5032 This prefix is followed by a name. The separator is a '.'.
5033 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5034 and '_'.
5035
5036 <expr> Is a standard HAProxy expression formed by a sample-fetch
5037 followed by some converters.
5038
5039 Example:
5040 http-after-response set-var(sess.last_redir) res.hdr(location)
5041
5042http-after-response strict-mode { on | off }
5043
5044 This enables or disables the strict rewriting mode for following rules. It
5045 does not affect rules declared before it and it is only applicable on rules
5046 performing a rewrite on the responses. When the strict mode is enabled, any
5047 rewrite failure triggers an internal error. Otherwise, such errors are
5048 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005049 rewrites optional while others must be performed to continue the response
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005050 processing.
5051
5052 By default, the strict rewriting mode is enabled. Its value is also reset
5053 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04005054 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005055 rules evaluation.
5056
5057http-after-response unset-var(<var-name>) [ { if | unless } <condition> ]
5058
5059 This is used to unset a variable. See "http-after-response set-var" for
5060 details about <var-name>.
5061
5062 Example:
5063 http-after-response unset-var(sess.last_redir)
5064
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005065
5066http-check comment <string>
5067 Defines a comment for the following the http-check rule, reported in logs if
5068 it fails.
5069 May be used in sections : defaults | frontend | listen | backend
5070 yes | no | yes | yes
5071
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005072 Arguments :
5073 <string> is the comment message to add in logs if the following http-check
5074 rule fails.
5075
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005076 It only works for connect, send and expect rules. It is useful to make
5077 user-friendly error reporting.
5078
Daniel Corbett67a82712020-07-06 23:01:19 -04005079 See also : "option httpchk", "http-check connect", "http-check send" and
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005080 "http-check expect".
5081
5082
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005083http-check connect [default] [port <expr>] [addr <ip>] [send-proxy]
5084 [via-socks4] [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +02005085 [proto <name>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02005086 Opens a new connection to perform an HTTP health check
5087 May be used in sections : defaults | frontend | listen | backend
5088 yes | no | yes | yes
5089
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005090 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005091 comment <msg> defines a message to report if the rule evaluation fails.
5092
Christopher Faulete5870d82020-04-15 11:32:03 +02005093 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -04005094 checks. The server options are used only if not redefined.
Christopher Faulete5870d82020-04-15 11:32:03 +02005095
5096 port <expr> if not set, check port or server port is used.
5097 It tells HAProxy where to open the connection to.
5098 <port> must be a valid TCP port source integer, from 1 to
5099 65535 or an sample-fetch expression.
5100
5101 addr <ip> defines the IP address to do the health check.
5102
5103 send-proxy send a PROXY protocol string
5104
5105 via-socks4 enables outgoing health checks using upstream socks4 proxy.
5106
5107 ssl opens a ciphered connection
5108
5109 sni <sni> specifies the SNI to use to do health checks over SSL.
5110
5111 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
5112 list consists in a comma-delimited list of protocol names,
5113 for instance: "h2,http/1.1". If it is not set, the server ALPN
5114 is used.
5115
Christopher Fauletedc6ed92020-04-23 16:27:59 +02005116 proto <name> forces the multiplexer's protocol to use for this connection.
5117 It must be an HTTP mux protocol and it must be usable on the
5118 backend side. The list of available protocols is reported in
5119 haproxy -vv.
5120
Christopher Faulete5870d82020-04-15 11:32:03 +02005121 linger cleanly close the connection instead of using a single RST.
5122
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005123 Just like tcp-check health checks, it is possible to configure the connection
5124 to use to perform HTTP health check. This directive should also be used to
5125 describe a scenario involving several request/response exchanges, possibly on
5126 different ports or with different servers.
5127
5128 When there are no TCP port configured on the server line neither server port
5129 directive, then the first step of the http-check sequence must be to specify
5130 the port with a "http-check connect".
5131
5132 In an http-check ruleset a 'connect' is required, it is also mandatory to start
5133 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
5134 do.
5135
5136 When a connect must start the ruleset, if may still be preceded by set-var,
5137 unset-var or comment rules.
5138
5139 Examples :
Christopher Faulete5870d82020-04-15 11:32:03 +02005140 # check HTTP and HTTPs services on a server.
5141 # first open port 80 thanks to server line port directive, then
5142 # tcp-check opens port 443, ciphered and run a request on it:
5143 option httpchk
5144
5145 http-check connect
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02005146 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005147 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02005148 http-check connect port 443 ssl sni haproxy.1wt.eu
Christopher Fauleta5c14ef2020-04-29 14:19:13 +02005149 http-check send meth GET uri / ver HTTP/1.1 hdr host haproxy.1wt.eu
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005150 http-check expect status 200-399
Christopher Faulete5870d82020-04-15 11:32:03 +02005151
5152 server www 10.0.0.1 check port 80
5153
5154 See also : "option httpchk", "http-check send", "http-check expect"
Christopher Faulet6d0c3df2020-01-22 09:26:35 +01005155
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005156
Willy Tarreau0ba27502007-12-24 16:55:16 +01005157http-check disable-on-404
5158 Enable a maintenance mode upon HTTP/404 response to health-checks
5159 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01005160 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01005161 Arguments : none
5162
5163 When this option is set, a server which returns an HTTP code 404 will be
5164 excluded from further load-balancing, but will still receive persistent
5165 connections. This provides a very convenient method for Web administrators
5166 to perform a graceful shutdown of their servers. It is also important to note
5167 that a server which is detected as failed while it was in this mode will not
5168 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
5169 will immediately be reinserted into the farm. The status on the stats page
5170 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01005171 option only works in conjunction with the "httpchk" option. If this option
5172 is used with "http-check expect", then it has precedence over it so that 404
Christopher Fauletfa8b89a2020-11-20 18:54:13 +01005173 responses will still be considered as soft-stop. Note also that a stopped
5174 server will stay stopped even if it replies 404s. This option is only
5175 evaluated for running servers.
Willy Tarreaubd741542010-03-16 18:46:54 +01005176
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005177 See also : "option httpchk" and "http-check expect".
Willy Tarreaubd741542010-03-16 18:46:54 +01005178
5179
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005180http-check expect [min-recv <int>] [comment <msg>]
Christopher Faulete5870d82020-04-15 11:32:03 +02005181 [ok-status <st>] [error-status <st>] [tout-status <st>]
5182 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
5183 [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005184 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01005185 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02005186 yes | no | yes | yes
Christopher Faulete5870d82020-04-15 11:32:03 +02005187
Willy Tarreaubd741542010-03-16 18:46:54 +01005188 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005189 comment <msg> defines a message to report if the rule evaluation fails.
5190
Christopher Faulete5870d82020-04-15 11:32:03 +02005191 min-recv is optional and can define the minimum amount of data required to
5192 evaluate the current expect rule. If the number of received bytes
5193 is under this limit, the check will wait for more data. This
5194 option can be used to resolve some ambiguous matching rules or to
5195 avoid executing costly regex matches on content known to be still
5196 incomplete. If an exact string is used, the minimum between the
5197 string length and this parameter is used. This parameter is
5198 ignored if it is set to -1. If the expect rule does not match,
5199 the check will wait for more data. If set to 0, the evaluation
5200 result is always conclusive.
5201
5202 ok-status <st> is optional and can be used to set the check status if
5203 the expect rule is successfully evaluated and if it is
5204 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005205 "L6OK" and "L4OK" are supported :
5206 - L7OK : check passed on layer 7
Christopher Faulet83662b52020-11-20 17:47:47 +01005207 - L7OKC : check conditionally passed on layer 7, set
5208 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005209 - L6OK : check passed on layer 6
5210 - L4OK : check passed on layer 4
5211 By default "L7OK" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02005212
5213 error-status <st> is optional and can be used to set the check status if
5214 an error occurred during the expect rule evaluation.
Christopher Faulet83662b52020-11-20 17:47:47 +01005215 "L7OKC", "L7RSP", "L7STS", "L6RSP" and "L4CON" are
5216 supported :
5217 - L7OKC : check conditionally passed on layer 7, set
5218 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005219 - L7RSP : layer 7 invalid response - protocol error
5220 - L7STS : layer 7 response error, for example HTTP 5xx
5221 - L6RSP : layer 6 invalid response - protocol error
5222 - L4CON : layer 1-4 connection problem
5223 By default "L7RSP" is used.
Christopher Faulete5870d82020-04-15 11:32:03 +02005224
5225 tout-status <st> is optional and can be used to set the check status if
5226 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +02005227 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
5228 - L7TOUT : layer 7 (HTTP/SMTP) timeout
5229 - L6TOUT : layer 6 (SSL) timeout
5230 - L4TOUT : layer 1-4 timeout
Christopher Faulete5870d82020-04-15 11:32:03 +02005231 By default "L7TOUT" is used.
5232
5233 on-success <fmt> is optional and can be used to customize the
5234 informational message reported in logs if the expect
5235 rule is successfully evaluated and if it is the last rule
5236 in the tcp-check ruleset. <fmt> is a log-format string.
5237
5238 on-error <fmt> is optional and can be used to customize the
5239 informational message reported in logs if an error
5240 occurred during the expect rule evaluation. <fmt> is a
5241 log-format string.
5242
Willy Tarreaubd741542010-03-16 18:46:54 +01005243 <match> is a keyword indicating how to look for a specific pattern in the
Christopher Fauletb5594262020-05-05 20:23:13 +02005244 response. The keyword may be one of "status", "rstatus", "hdr",
5245 "fhdr", "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01005246 exclamation mark ("!") to negate the match. Spaces are allowed
5247 between the exclamation mark and the keyword. See below for more
5248 details on the supported keywords.
5249
Christopher Faulet39708192020-05-05 10:47:36 +02005250 <pattern> is the pattern to look for. It may be a string, a regular
5251 expression or a more complex pattern with several arguments. If
5252 the string pattern contains spaces, they must be escaped with the
5253 usual backslash ('\').
Willy Tarreaubd741542010-03-16 18:46:54 +01005254
5255 By default, "option httpchk" considers that response statuses 2xx and 3xx
5256 are valid, and that others are invalid. When "http-check expect" is used,
5257 it defines what is considered valid or invalid. Only one "http-check"
5258 statement is supported in a backend. If a server fails to respond or times
5259 out, the check obviously fails. The available matches are :
5260
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005261 status <codes> : test the status codes found parsing <codes> string. it
5262 must be a comma-separated list of status codes or range
5263 codes. A health check response will be considered as
5264 valid if the response's status code matches any status
5265 code or is inside any range of the list. If the "status"
5266 keyword is prefixed with "!", then the response will be
5267 considered invalid if the status code matches.
Willy Tarreaubd741542010-03-16 18:46:54 +01005268
5269 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005270 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005271 response's status code matches the expression. If the
5272 "rstatus" keyword is prefixed with "!", then the response
5273 will be considered invalid if the status code matches.
5274 This is mostly used to check for multiple codes.
5275
Christopher Fauletb5594262020-05-05 20:23:13 +02005276 hdr { name | name-lf } [ -m <meth> ] <name>
5277 [ { value | value-lf } [ -m <meth> ] <value> :
Christopher Faulet39708192020-05-05 10:47:36 +02005278 test the specified header pattern on the HTTP response
5279 headers. The name pattern is mandatory but the value
5280 pattern is optional. If not specified, only the header
5281 presence is verified. <meth> is the matching method,
5282 applied on the header name or the header value. Supported
5283 matching methods are "str" (exact match), "beg" (prefix
5284 match), "end" (suffix match), "sub" (substring match) or
5285 "reg" (regex match). If not specified, exact matching
Christopher Fauletb5594262020-05-05 20:23:13 +02005286 method is used. If the "name-lf" parameter is used,
5287 <name> is evaluated as a log-format string. If "value-lf"
5288 parameter is used, <value> is evaluated as a log-format
5289 string. These parameters cannot be used with the regex
5290 matching method. Finally, the header value is considered
5291 as comma-separated list. Note that matchings are case
5292 insensitive on the header names.
5293
5294 fhdr { name | name-lf } [ -m <meth> ] <name>
5295 [ { value | value-lf } [ -m <meth> ] <value> :
5296 test the specified full header pattern on the HTTP
5297 response headers. It does exactly the same than "hdr"
5298 keyword, except the full header value is tested, commas
5299 are not considered as delimiters.
Christopher Faulet39708192020-05-05 10:47:36 +02005300
Willy Tarreaubd741542010-03-16 18:46:54 +01005301 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005302 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005303 response's body contains this exact string. If the
5304 "string" keyword is prefixed with "!", then the response
5305 will be considered invalid if the body contains this
5306 string. This can be used to look for a mandatory word at
5307 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005308 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01005309 trace).
5310
5311 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04005312 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01005313 response's body matches this expression. If the "rstring"
5314 keyword is prefixed with "!", then the response will be
5315 considered invalid if the body matches the expression.
5316 This can be used to look for a mandatory word at the end
5317 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01005318 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01005319
Christopher Fauletaaab0832020-05-05 15:54:22 +02005320 string-lf <fmt> : test a log-format string match in the HTTP response body.
5321 A health check response will be considered valid if the
5322 response's body contains the string resulting of the
5323 evaluation of <fmt>, which follows the log-format rules.
5324 If prefixed with "!", then the response will be
5325 considered invalid if the body contains the string.
5326
Willy Tarreaubd741542010-03-16 18:46:54 +01005327 It is important to note that the responses will be limited to a certain size
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +01005328 defined by the global "tune.bufsize" option, which defaults to 16384 bytes.
Willy Tarreaubd741542010-03-16 18:46:54 +01005329 Thus, too large responses may not contain the mandatory pattern when using
5330 "string" or "rstring". If a large response is absolutely required, it is
5331 possible to change the default max size by setting the global variable.
5332 However, it is worth keeping in mind that parsing very large responses can
5333 waste some CPU cycles, especially when regular expressions are used, and that
5334 it is always better to focus the checks on smaller resources.
5335
Christopher Faulete5870d82020-04-15 11:32:03 +02005336 In an http-check ruleset, the last expect rule may be implicit. If no expect
5337 rule is specified after the last "http-check send", an implicit expect rule
5338 is defined to match on 2xx or 3xx status codes. It means this rule is also
5339 defined if there is no "http-check" rule at all, when only "option httpchk"
5340 is set.
Cyril Bonté32602d22015-01-30 00:07:07 +01005341
Willy Tarreaubd741542010-03-16 18:46:54 +01005342 Last, if "http-check expect" is combined with "http-check disable-on-404",
5343 then this last one has precedence when the server responds with 404.
5344
5345 Examples :
5346 # only accept status 200 as valid
Christopher Faulet8021a5f2020-04-24 13:53:12 +02005347 http-check expect status 200,201,300-310
Willy Tarreaubd741542010-03-16 18:46:54 +01005348
Christopher Faulet39708192020-05-05 10:47:36 +02005349 # be sure a sessid coookie is set
5350 http-check expect header name "set-cookie" value -m beg "sessid="
5351
Willy Tarreaubd741542010-03-16 18:46:54 +01005352 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005353 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01005354
5355 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01005356 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01005357
5358 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005359 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01005360
Christopher Faulete5870d82020-04-15 11:32:03 +02005361 See also : "option httpchk", "http-check connect", "http-check disable-on-404"
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005362 and "http-check send".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005363
5364
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005365http-check send [meth <method>] [{ uri <uri> | uri-lf <fmt> }>] [ver <version>]
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005366 [hdr <name> <fmt>]* [{ body <string> | body-lf <fmt> }]
5367 [comment <msg>]
Christopher Faulet8acb1282020-04-09 08:44:06 +02005368 Add a possible list of headers and/or a body to the request sent during HTTP
5369 health checks.
5370 May be used in sections : defaults | frontend | listen | backend
5371 yes | no | yes | yes
5372 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +02005373 comment <msg> defines a message to report if the rule evaluation fails.
5374
Christopher Faulete5870d82020-04-15 11:32:03 +02005375 meth <method> is the optional HTTP method used with the requests. When not
5376 set, the "OPTIONS" method is used, as it generally requires
5377 low server processing and is easy to filter out from the
5378 logs. Any method may be used, though it is not recommended
5379 to invent non-standard ones.
5380
Christopher Faulet7c95f5f2020-05-06 15:06:34 +02005381 uri <uri> is optional and set the URI referenced in the HTTP requests
5382 to the string <uri>. It defaults to "/" which is accessible
5383 by default on almost any server, but may be changed to any
5384 other URI. Query strings are permitted.
5385
5386 uri-lf <fmt> is optional and set the URI referenced in the HTTP requests
5387 using the log-format string <fmt>. It defaults to "/" which
5388 is accessible by default on almost any server, but may be
5389 changed to any other URI. Query strings are permitted.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005390
Christopher Faulet907701b2020-04-28 09:37:00 +02005391 ver <version> is the optional HTTP version string. It defaults to
Christopher Faulete5870d82020-04-15 11:32:03 +02005392 "HTTP/1.0" but some servers might behave incorrectly in HTTP
Daniel Corbett67a82712020-07-06 23:01:19 -04005393 1.0, so turning it to HTTP/1.1 may sometimes help. Note that
Christopher Faulete5870d82020-04-15 11:32:03 +02005394 the Host field is mandatory in HTTP/1.1, use "hdr" argument
5395 to add it.
5396
5397 hdr <name> <fmt> adds the HTTP header field whose name is specified in
5398 <name> and whose value is defined by <fmt>, which follows
5399 to the log-format rules.
5400
5401 body <string> add the body defined by <string> to the request sent during
5402 HTTP health checks. If defined, the "Content-Length" header
5403 is thus automatically added to the request.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005404
Christopher Faulet574e7bd2020-05-06 15:38:58 +02005405 body-lf <fmt> add the body defined by the log-format string <fmt> to the
5406 request sent during HTTP health checks. If defined, the
5407 "Content-Length" header is thus automatically added to the
5408 request.
5409
Christopher Faulet8acb1282020-04-09 08:44:06 +02005410 In addition to the request line defined by the "option httpchk" directive,
5411 this one is the valid way to add some headers and optionally a body to the
5412 request sent during HTTP health checks. If a body is defined, the associate
Christopher Faulet9df910c2020-04-29 14:20:47 +02005413 "Content-Length" header is automatically added. Thus, this header or
5414 "Transfer-encoding" header should not be present in the request provided by
5415 "http-check send". If so, it will be ignored. The old trick consisting to add
5416 headers after the version string on the "option httpchk" line is now
Amaury Denoyelle6d975f02020-12-22 14:08:52 +01005417 deprecated.
Christopher Faulet8acb1282020-04-09 08:44:06 +02005418
Christopher Faulete5870d82020-04-15 11:32:03 +02005419 Also "http-check send" doesn't support HTTP keep-alive. Keep in mind that it
Amaury Denoyelle6d975f02020-12-22 14:08:52 +01005420 will automatically append a "Connection: close" header, unless a Connection
5421 header has already already been configured via a hdr entry.
Christopher Faulet9df910c2020-04-29 14:20:47 +02005422
5423 Note that the Host header and the request authority, when both defined, are
5424 automatically synchronized. It means when the HTTP request is sent, when a
5425 Host is inserted in the request, the request authority is accordingly
5426 updated. Thus, don't be surprised if the Host header value overwrites the
5427 configured request authority.
5428
5429 Note also for now, no Host header is automatically added in HTTP/1.1 or above
5430 requests. You should add it explicitly.
Christopher Faulete5870d82020-04-15 11:32:03 +02005431
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005432 See also : "option httpchk", "http-check send-state" and "http-check expect".
Christopher Faulet8acb1282020-04-09 08:44:06 +02005433
5434
Willy Tarreauef781042010-01-27 11:53:01 +01005435http-check send-state
5436 Enable emission of a state header with HTTP health checks
5437 May be used in sections : defaults | frontend | listen | backend
5438 yes | no | yes | yes
5439 Arguments : none
5440
5441 When this option is set, haproxy will systematically send a special header
5442 "X-Haproxy-Server-State" with a list of parameters indicating to each server
5443 how they are seen by haproxy. This can be used for instance when a server is
5444 manipulated without access to haproxy and the operator needs to know whether
5445 haproxy still sees it up or not, or if the server is the last one in a farm.
5446
5447 The header is composed of fields delimited by semi-colons, the first of which
5448 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
5449 checks on the total number before transition, just as appears in the stats
5450 interface. Next headers are in the form "<variable>=<value>", indicating in
5451 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08005452 - a variable "address", containing the address of the backend server.
5453 This corresponds to the <address> field in the server declaration. For
5454 unix domain sockets, it will read "unix".
5455
5456 - a variable "port", containing the port of the backend server. This
5457 corresponds to the <port> field in the server declaration. For unix
5458 domain sockets, it will read "unix".
5459
Willy Tarreauef781042010-01-27 11:53:01 +01005460 - a variable "name", containing the name of the backend followed by a slash
5461 ("/") then the name of the server. This can be used when a server is
5462 checked in multiple backends.
5463
5464 - a variable "node" containing the name of the haproxy node, as set in the
5465 global "node" variable, otherwise the system's hostname if unspecified.
5466
5467 - a variable "weight" indicating the weight of the server, a slash ("/")
5468 and the total weight of the farm (just counting usable servers). This
5469 helps to know if other servers are available to handle the load when this
5470 one fails.
5471
5472 - a variable "scur" indicating the current number of concurrent connections
5473 on the server, followed by a slash ("/") then the total number of
5474 connections on all servers of the same backend.
5475
5476 - a variable "qcur" indicating the current number of requests in the
5477 server's queue.
5478
5479 Example of a header received by the application server :
5480 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
5481 scur=13/22; qcur=0
5482
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005483 See also : "option httpchk", "http-check disable-on-404" and
5484 "http-check send".
Willy Tarreauef781042010-01-27 11:53:01 +01005485
Christopher Faulete5870d82020-04-15 11:32:03 +02005486
5487http-check set-var(<var-name>) <expr>
Christopher Faulete5870d82020-04-15 11:32:03 +02005488 This operation sets the content of a variable. The variable is declared inline.
Christopher Faulete5870d82020-04-15 11:32:03 +02005489 May be used in sections: defaults | frontend | listen | backend
5490 yes | no | yes | yes
5491
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005492 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005493 <var-name> The name of the variable starts with an indication about its
5494 scope. The scopes allowed for http-check are:
5495 "proc" : the variable is shared with the whole process.
5496 "sess" : the variable is shared with the tcp-check session.
5497 "check": the variable is declared for the lifetime of the tcp-check.
5498 This prefix is followed by a name. The separator is a '.'.
5499 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5500 and '-'.
5501
5502 <expr> Is a sample-fetch expression potentially followed by converters.
5503
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005504 Examples :
5505 http-check set-var(check.port) int(1234)
Christopher Faulete5870d82020-04-15 11:32:03 +02005506
5507
5508http-check unset-var(<var-name>)
Christopher Faulete5870d82020-04-15 11:32:03 +02005509 Free a reference to a variable within its scope.
Christopher Faulete5870d82020-04-15 11:32:03 +02005510 May be used in sections: defaults | frontend | listen | backend
5511 yes | no | yes | yes
5512
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005513 Arguments :
Christopher Faulete5870d82020-04-15 11:32:03 +02005514 <var-name> The name of the variable starts with an indication about its
5515 scope. The scopes allowed for http-check are:
5516 "proc" : the variable is shared with the whole process.
5517 "sess" : the variable is shared with the tcp-check session.
5518 "check": the variable is declared for the lifetime of the tcp-check.
5519 This prefix is followed by a name. The separator is a '.'.
5520 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
5521 and '-'.
5522
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02005523 Examples :
5524 http-check unset-var(check.port)
Christopher Faulete5870d82020-04-15 11:32:03 +02005525
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005526
Christopher Faulet3b967c12020-05-15 15:47:44 +02005527http-error status <code> [content-type <type>]
5528 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5529 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5530 [ hdr <name> <fmt> ]*
5531 Defines a custom error message to use instead of errors generated by HAProxy.
5532 May be used in sections : defaults | frontend | listen | backend
5533 yes | yes | yes | yes
5534 Arguments :
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005535 status <code> is the HTTP status code. It must be specified.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005536 Currently, HAProxy is capable of generating codes
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02005537 200, 400, 401, 403, 404, 405, 407, 408, 410, 413, 425,
Christopher Faulete095f312020-12-07 11:22:24 +01005538 429, 500, 501, 502, 503, and 504.
Christopher Faulet3b967c12020-05-15 15:47:44 +02005539
5540 content-type <type> is the response content type, for instance
5541 "text/plain". This parameter is ignored and should be
5542 omitted when an errorfile is configured or when the
5543 payload is empty. Otherwise, it must be defined.
5544
5545 default-errorfiles Reset the previously defined error message for current
5546 proxy for the status <code>. If used on a backend, the
5547 frontend error message is used, if defined. If used on
5548 a frontend, the default error message is used.
5549
5550 errorfile <file> designates a file containing the full HTTP response.
5551 It is recommended to follow the common practice of
5552 appending ".http" to the filename so that people do
5553 not confuse the response with HTML error pages, and to
5554 use absolute paths, since files are read before any
5555 chroot is performed.
5556
5557 errorfiles <name> designates the http-errors section to use to import
5558 the error message with the status code <code>. If no
5559 such message is found, the proxy's error messages are
5560 considered.
5561
5562 file <file> specifies the file to use as response payload. If the
5563 file is not empty, its content-type must be set as
5564 argument to "content-type", otherwise, any
5565 "content-type" argument is ignored. <file> is
5566 considered as a raw string.
5567
5568 string <str> specifies the raw string to use as response payload.
5569 The content-type must always be set as argument to
5570 "content-type".
5571
5572 lf-file <file> specifies the file to use as response payload. If the
5573 file is not empty, its content-type must be set as
5574 argument to "content-type", otherwise, any
5575 "content-type" argument is ignored. <file> is
5576 evaluated as a log-format string.
5577
5578 lf-string <str> specifies the log-format string to use as response
5579 payload. The content-type must always be set as
5580 argument to "content-type".
5581
5582 hdr <name> <fmt> adds to the response the HTTP header field whose name
5583 is specified in <name> and whose value is defined by
5584 <fmt>, which follows to the log-format rules.
5585 This parameter is ignored if an errorfile is used.
5586
5587 This directive may be used instead of "errorfile", to define a custom error
5588 message. As "errorfile" directive, it is used for errors detected and
5589 returned by HAProxy. If an errorfile is defined, it is parsed when HAProxy
5590 starts and must be valid according to the HTTP standards. The generated
5591 response must not exceed the configured buffer size (BUFFSIZE), otherwise an
5592 internal error will be returned. Finally, if you consider to use some
5593 http-after-response rules to rewrite these errors, the reserved buffer space
5594 should be available (see "tune.maxrewrite").
5595
5596 The files are read at the same time as the configuration and kept in memory.
5597 For this reason, the errors continue to be returned even when the process is
5598 chrooted, and no file change is considered while the process is running.
5599
Christopher Fauletd5ac6de2020-12-02 08:40:14 +01005600 Note: 400/408/500 errors emitted in early stage of the request parsing are
5601 handled by the multiplexer at a lower level. No custom formatting is
5602 supported at this level. Thus only static error messages, defined with
5603 "errorfile" directive, are supported. However, this limitation only
5604 exists during the request headers parsing or between two transactions.
5605
Christopher Faulet3b967c12020-05-15 15:47:44 +02005606 See also : "errorfile", "errorfiles", "errorloc", "errorloc302",
5607 "errorloc303" and section 3.8 about http-errors.
5608
5609
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005610http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005611 Access control for Layer 7 requests
5612
5613 May be used in sections: defaults | frontend | listen | backend
5614 no | yes | yes | yes
5615
Willy Tarreau20b0de52012-12-24 15:45:22 +01005616 The http-request statement defines a set of rules which apply to layer 7
5617 processing. The rules are evaluated in their declaration order when they are
5618 met in a frontend, listen or backend section. Any rule may optionally be
5619 followed by an ACL-based condition, in which case it will only be evaluated
5620 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01005621
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005622 The first keyword is the rule's action. The supported actions are described
5623 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005624
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005625 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01005626
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005627 Example:
5628 acl nagios src 192.168.129.3
5629 acl local_net src 192.168.0.0/16
5630 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01005631
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005632 http-request allow if nagios
5633 http-request allow if local_net auth_ok
5634 http-request auth realm Gimme if local_net auth_ok
5635 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01005636
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005637 Example:
5638 acl key req.hdr(X-Add-Acl-Key) -m found
5639 acl add path /addacl
5640 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01005641
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005642 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01005643
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005644 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
5645 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02005646
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005647 Example:
5648 acl value req.hdr(X-Value) -m found
5649 acl setmap path /setmap
5650 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005651
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005652 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06005653
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005654 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
5655 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06005656
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005657 See also : "stats http-request", section 3.4 about userlists and section 7
5658 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06005659
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005660http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005661
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005662 This is used to add a new entry into an ACL. The ACL must be loaded from a
5663 file (even a dummy empty file). The file name of the ACL to be updated is
5664 passed between parentheses. It takes one argument: <key fmt>, which follows
5665 log-format rules, to collect content of the new entry. It performs a lookup
5666 in the ACL before insertion, to avoid duplicated (or more) values. This
5667 lookup is done by a linear search and can be expensive with large lists!
5668 It is the equivalent of the "add acl" command from the stats socket, but can
5669 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06005670
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005671http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005672
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005673 This appends an HTTP header field whose name is specified in <name> and
5674 whose value is defined by <fmt> which follows the log-format rules (see
5675 Custom Log Format in section 8.2.4). This is particularly useful to pass
5676 connection-specific information to the server (e.g. the client's SSL
5677 certificate), or to combine several headers into one. This rule is not
5678 final, so it is possible to add other similar rules. Note that header
5679 addition is performed immediately, so one rule might reuse the resulting
5680 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06005681
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005682http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005683
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005684 This stops the evaluation of the rules and lets the request pass the check.
5685 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06005686
Sasha Pachev218f0642014-06-16 12:05:59 -06005687
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005688http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005689
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005690 This stops the evaluation of the rules and immediately responds with an
5691 HTTP 401 or 407 error code to invite the user to present a valid user name
5692 and password. No further "http-request" rules are evaluated. An optional
5693 "realm" parameter is supported, it sets the authentication realm that is
5694 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06005695
Christopher Faulet612f2ea2020-05-27 09:57:28 +02005696 The corresponding proxy's error message is used. It may be customized using
5697 an "errorfile" or an "http-error" directive. For 401 responses, all
5698 occurrences of the WWW-Authenticate header are removed and replaced by a new
5699 one with a basic authentication challenge for realm "<realm>". For 407
5700 responses, the same is done on the Proxy-Authenticate header. If the error
5701 message must not be altered, consider to use "http-request return" rule
5702 instead.
5703
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005704 Example:
5705 acl auth_ok http_auth_group(L1) G1
5706 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06005707
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02005708http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005709
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02005710 See section 6.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005711
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005712http-request capture <sample> [ len <length> | id <id> ]
5713 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005714
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005715 This captures sample expression <sample> from the request buffer, and
5716 converts it to a string of at most <len> characters. The resulting string is
5717 stored into the next request "capture" slot, so it will possibly appear next
5718 to some captured HTTP headers. It will then automatically appear in the logs,
5719 and it will be possible to extract it using sample fetch rules to feed it
5720 into headers or anything. The length should be limited given that this size
5721 will be allocated for each capture during the whole session life.
5722 Please check section 7.3 (Fetching samples) and "capture request header" for
5723 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005724
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005725 If the keyword "id" is used instead of "len", the action tries to store the
5726 captured string in a previously declared capture slot. This is useful to run
5727 captures in backends. The slot id can be declared by a previous directive
Baptiste Assmann19a69b32020-01-16 14:34:22 +01005728 "http-request capture" or with the "declare capture" keyword.
5729
5730 When using this action in a backend, double check that the relevant
5731 frontend(s) have the required capture slots otherwise, this rule will be
5732 ignored at run time. This can't be detected at configuration parsing time
5733 due to HAProxy's ability to dynamically resolve backend name at runtime.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005734
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005735http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005736
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005737 This is used to delete an entry from an ACL. The ACL must be loaded from a
5738 file (even a dummy empty file). The file name of the ACL to be updated is
5739 passed between parentheses. It takes one argument: <key fmt>, which follows
5740 log-format rules, to collect content of the entry to delete.
5741 It is the equivalent of the "del acl" command from the stats socket, but can
5742 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01005743
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005744http-request del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02005745
Maciej Zdebebdd4c52020-11-20 13:58:48 +00005746 This removes all HTTP header fields whose name is specified in <name>. <meth>
5747 is the matching method, applied on the header name. Supported matching methods
5748 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
5749 (substring match) and "reg" (regex match). If not specified, exact matching
5750 method is used.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02005751
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005752http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02005753
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005754 This is used to delete an entry from a MAP. The MAP must be loaded from a
5755 file (even a dummy empty file). The file name of the MAP to be updated is
5756 passed between parentheses. It takes one argument: <key fmt>, which follows
5757 log-format rules, to collect content of the entry to delete.
5758 It takes one argument: "file name" It is the equivalent of the "del map"
5759 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02005760
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005761http-request deny [deny_status <status>] [ { if | unless } <condition> ]
5762http-request deny [ { status | deny_status } <code>] [content-type <type>]
5763 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5764 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
5765 [ hdr <name> <fmt> ]*
5766 [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04005767
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005768 This stops the evaluation of the rules and immediately rejects the request.
5769 By default an HTTP 403 error is returned. But the response may be customized
5770 using same syntax than "http-request return" rules. Thus, see "http-request
Ilya Shipitsin11057a32020-06-21 21:18:27 +05005771 return" for details. For compatibility purpose, when no argument is defined,
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005772 or only "deny_status", the argument "default-errorfiles" is implied. It means
5773 "http-request deny [deny_status <status>]" is an alias of
5774 "http-request deny [status <status>] default-errorfiles".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005775 No further "http-request" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02005776 See also "http-request return".
Patrick Hemmer268a7072018-05-11 12:52:31 -04005777
Olivier Houchard602bf7d2019-05-10 13:59:15 +02005778http-request disable-l7-retry [ { if | unless } <condition> ]
5779 This disables any attempt to retry the request if it fails for any other
5780 reason than a connection failure. This can be useful for example to make
5781 sure POST requests aren't retried on failure.
5782
Baptiste Assmann333939c2019-01-21 08:34:50 +01005783http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
5784
5785 This action performs a DNS resolution of the output of <expr> and stores
5786 the result in the variable <var>. It uses the DNS resolvers section
5787 pointed by <resolvers>.
5788 It is possible to choose a resolution preference using the optional
5789 arguments 'ipv4' or 'ipv6'.
5790 When performing the DNS resolution, the client side connection is on
5791 pause waiting till the end of the resolution.
5792 If an IP address can be found, it is stored into <var>. If any kind of
5793 error occurs, then <var> is not set.
5794 One can use this action to discover a server IP address at run time and
5795 based on information found in the request (IE a Host header).
5796 If this action is used to find the server's IP address (using the
5797 "set-dst" action), then the server IP address in the backend must be set
5798 to 0.0.0.0.
5799
5800 Example:
5801 resolvers mydns
5802 nameserver local 127.0.0.53:53
5803 nameserver google 8.8.8.8:53
5804 timeout retry 1s
5805 hold valid 10s
5806 hold nx 3s
5807 hold other 3s
5808 hold obsolete 0s
5809 accepted_payload_size 8192
5810
5811 frontend fe
5812 bind 10.42.0.1:80
5813 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
5814 http-request capture var(txn.myip) len 40
5815
5816 # return 503 when the variable is not set,
5817 # which mean DNS resolution error
5818 use_backend b_503 unless { var(txn.myip) -m found }
5819
5820 default_backend be
5821
5822 backend b_503
5823 # dummy backend used to return 503.
5824 # one can use the errorfile directive to send a nice
5825 # 503 error page to end users
5826
5827 backend be
5828 # rule to prevent HAProxy from reconnecting to services
5829 # on the local network (forged DNS name used to scan the network)
5830 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
5831 http-request set-dst var(txn.myip)
5832 server clear 0.0.0.0:0
5833
5834 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
5835 be used to scan the network or worst won't loop over itself...
5836
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01005837http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
5838
5839 This is used to build an HTTP 103 Early Hints response prior to any other one.
5840 This appends an HTTP header field to this response whose name is specified in
5841 <name> and whose value is defined by <fmt> which follows the log-format rules
5842 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01005843 to the client some Link headers to preload resources required to render the
5844 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01005845
5846 See RFC 8297 for more information.
5847
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005848http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005849
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005850 This performs an HTTP redirection based on a redirect rule. This is exactly
5851 the same as the "redirect" statement except that it inserts a redirect rule
5852 which can be processed in the middle of other "http-request" rules and that
5853 these rules use the "log-format" strings. See the "redirect" keyword for the
5854 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005855
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005856http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005857
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005858 This stops the evaluation of the rules and immediately closes the connection
5859 without sending any response. It acts similarly to the
5860 "tcp-request content reject" rules. It can be useful to force an immediate
5861 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005862
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005863http-request replace-header <name> <match-regex> <replace-fmt>
5864 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02005865
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005866 This matches the value of all occurrences of header field <name> against
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005867 <match-regex>. Matching is performed case-sensitively. Matching values are
5868 completely replaced by <replace-fmt>. Format characters are allowed in
5869 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
5870 Standard back-references using the backslash ('\') followed by a number are
5871 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02005872
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005873 This action acts on whole header lines, regardless of the number of values
5874 they may contain. Thus it is well-suited to process headers naturally
5875 containing commas in their value, such as If-Modified-Since. Headers that
5876 contain a comma-separated list of values, such as Accept, should be processed
5877 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01005878
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005879 Example:
5880 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
5881
5882 # applied to:
5883 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
5884
5885 # outputs:
5886 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
5887
5888 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02005889
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005890 http-request replace-header User-Agent curl foo
5891
5892 # applied to:
5893 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02005894
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005895 # outputs:
5896 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02005897
Willy Tarreau262c3f12019-12-17 06:52:51 +01005898http-request replace-path <match-regex> <replace-fmt>
5899 [ { if | unless } <condition> ]
5900
5901 This works like "replace-header" except that it works on the request's path
5902 component instead of a header. The path component starts at the first '/'
Christopher Faulet82c83322020-09-02 14:16:59 +02005903 after an optional scheme+authority and ends before the question mark. Thus,
5904 the replacement does not modify the scheme, the authority and the
5905 query-string.
Willy Tarreau262c3f12019-12-17 06:52:51 +01005906
5907 It is worth noting that regular expressions may be more expensive to evaluate
5908 than certain ACLs, so rare replacements may benefit from a condition to avoid
5909 performing the evaluation at all if it does not match.
5910
5911 Example:
5912 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
5913 http-request replace-path (.*) /foo\1
5914
Willy Tarreau262c3f12019-12-17 06:52:51 +01005915 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
5916 http-request replace-path /foo/(.*) /\1
5917 # or more efficient if only some requests match :
5918 http-request replace-path /foo/(.*) /\1 if { url_beg /foo/ }
5919
Christopher Faulet312294f2020-09-02 17:17:44 +02005920http-request replace-pathq <match-regex> <replace-fmt>
5921 [ { if | unless } <condition> ]
5922
5923 This does the same as "http-request replace-path" except that the path
5924 contains the query-string if any is present. Thus, the path and the
5925 query-string are replaced.
5926
5927 Example:
5928 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
5929 http-request replace-pathq ([^?]*)(\?(.*))? \1/foo\2
5930
Willy Tarreau33810222019-06-12 17:44:02 +02005931http-request replace-uri <match-regex> <replace-fmt>
5932 [ { if | unless } <condition> ]
5933
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005934 This works like "replace-header" except that it works on the request's URI part
5935 instead of a header. The URI part may contain an optional scheme, authority or
5936 query string. These are considered to be part of the value that is matched
5937 against.
5938
5939 It is worth noting that regular expressions may be more expensive to evaluate
5940 than certain ACLs, so rare replacements may benefit from a condition to avoid
5941 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02005942
Willy Tarreau62b59132019-12-17 06:51:20 +01005943 IMPORTANT NOTE: historically in HTTP/1.x, the vast majority of requests sent
5944 by browsers use the "origin form", which differs from the "absolute form" in
5945 that they do not contain a scheme nor authority in the URI portion. Mostly
5946 only requests sent to proxies, those forged by hand and some emitted by
5947 certain applications use the absolute form. As such, "replace-uri" usually
5948 works fine most of the time in HTTP/1.x with rules starting with a "/". But
5949 with HTTP/2, clients are encouraged to send absolute URIs only, which look
5950 like the ones HTTP/1 clients use to talk to proxies. Such partial replace-uri
5951 rules may then fail in HTTP/2 when they work in HTTP/1. Either the rules need
Willy Tarreau262c3f12019-12-17 06:52:51 +01005952 to be adapted to optionally match a scheme and authority, or replace-path
5953 should be used.
Willy Tarreau33810222019-06-12 17:44:02 +02005954
Willy Tarreau62b59132019-12-17 06:51:20 +01005955 Example:
5956 # rewrite all "http" absolute requests to "https":
5957 http-request replace-uri ^http://(.*) https://\1
Willy Tarreau33810222019-06-12 17:44:02 +02005958
Willy Tarreau62b59132019-12-17 06:51:20 +01005959 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
5960 http-request replace-uri ([^/:]*://[^/]*)?(.*) \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02005961
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005962http-request replace-value <name> <match-regex> <replace-fmt>
5963 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02005964
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005965 This works like "replace-header" except that it matches the regex against
5966 every comma-delimited value of the header field <name> instead of the
5967 entire header. This is suited for all headers which are allowed to carry
5968 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02005969
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005970 Example:
5971 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02005972
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005973 # applied to:
5974 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02005975
Tim Duesterhus2252beb2019-10-29 00:05:13 +01005976 # outputs:
5977 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01005978
Christopher Faulet24231ab2020-01-24 17:44:23 +01005979http-request return [status <code>] [content-type <type>]
5980 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
5981 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01005982 [ hdr <name> <fmt> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01005983 [ { if | unless } <condition> ]
5984
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005985 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01005986 default status code used for the response is 200. It can be optionally
5987 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04005988 specified as an argument to "content-type". Finally the response itself may
Sébastien Grossab877122020-10-08 10:06:03 +02005989 be defined. It can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05005990 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01005991 are followed to create the response :
5992
5993 * If neither the errorfile nor the payload to use is defined, a dummy
5994 response is returned. Only the "status" argument is considered. It can be
5995 any code in the range [200, 599]. The "content-type" argument, if any, is
5996 ignored.
5997
5998 * If "default-errorfiles" argument is set, the proxy's errorfiles are
5999 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04006000 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Christopher Faulete095f312020-12-07 11:22:24 +01006001 425, 429, 500, 501, 502, 503, and 504). The "content-type" argument, if
6002 any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006003
6004 * If a specific errorfile is defined, with an "errorfile" argument, the
6005 corresponding file, containing a full HTTP response, is returned. Only the
6006 "status" argument is considered. It must be one of the status code handled
Christopher Faulete095f312020-12-07 11:22:24 +01006007 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 501,
6008 502, 503, and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006009
6010 * If an http-errors section is defined, with an "errorfiles" argument, the
6011 corresponding file in the specified http-errors section, containing a full
6012 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04006013 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Christopher Faulete095f312020-12-07 11:22:24 +01006014 408, 410, 413, 425, 429, 500, 501, 502, 503, and 504). The "content-type"
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006015 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006016
6017 * If a "file" or a "lf-file" argument is specified, the file's content is
6018 used as the response payload. If the file is not empty, its content-type
6019 must be set as argument to "content-type". Otherwise, any "content-type"
6020 argument is ignored. With a "lf-file" argument, the file's content is
6021 evaluated as a log-format string. With a "file" argument, it is considered
6022 as a raw content.
6023
6024 * If a "string" or "lf-string" argument is specified, the defined string is
6025 used as the response payload. The content-type must always be set as
6026 argument to "content-type". With a "lf-string" argument, the string is
6027 evaluated as a log-format string. With a "string" argument, it is
6028 considered as a raw string.
6029
Sébastien Grossab877122020-10-08 10:06:03 +02006030 When the response is not based on an errorfile, it is possible to append HTTP
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006031 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
6032 arguments are ignored. For each one, the header name is specified in <name>
6033 and its value is defined by <fmt> which follows the log-format rules.
6034
Christopher Faulet24231ab2020-01-24 17:44:23 +01006035 Note that the generated response must be smaller than a buffer. And to avoid
6036 any warning, when an errorfile or a raw file is loaded, the buffer space
Sébastien Grossab877122020-10-08 10:06:03 +02006037 reserved for the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006038
6039 No further "http-request" rules are evaluated.
6040
6041 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006042 http-request return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006043 if { path /ping }
6044
6045 http-request return content-type image/x-icon file /var/www/favicon.ico \
6046 if { path /favicon.ico }
6047
6048 http-request return status 403 content-type text/plain \
6049 lf-string "Access denied. IP %[src] is blacklisted." \
6050 if { src -f /etc/haproxy/blacklist.lst }
6051
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006052http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6053http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006054
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006055 This actions increments the GPC0 or GPC1 counter according with the sticky
6056 counter designated by <sc-id>. If an error occurs, this action silently fails
6057 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006058
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006059http-request sc-set-gpt0(<sc-id>) { <int> | <expr> }
6060 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006061
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006062 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6063 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
6064 boolean. If an error occurs, this action silently fails and the actions
6065 evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006066
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006067http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006068
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006069 This is used to set the destination IP address to the value of specified
6070 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
6071 but provides the correct IP in a HTTP header; or you want to mask the IP for
6072 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
6073 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01006074
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006075 Arguments:
6076 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6077 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01006078
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006079 Example:
6080 http-request set-dst hdr(x-dst)
6081 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01006082
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006083 When possible, set-dst preserves the original destination port as long as the
6084 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006085
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006086http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006087
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006088 This is used to set the destination port address to the value of specified
6089 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
6090 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006091
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006092 Arguments:
6093 <expr> Is a standard HAProxy expression formed by a sample-fetch
6094 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006095
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006096 Example:
6097 http-request set-dst-port hdr(x-port)
6098 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02006099
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006100 When possible, set-dst-port preserves the original destination address as
6101 long as the address family supports a port, otherwise it forces the
6102 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02006103
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006104http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02006105
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006106 This does the same as "http-request add-header" except that the header name
6107 is first removed if it existed. This is useful when passing security
6108 information to the server, where the header must not be manipulated by
6109 external users. Note that the new value is computed before the removal so it
6110 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02006111
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006112 Example:
6113 http-request set-header X-Haproxy-Current-Date %T
6114 http-request set-header X-SSL %[ssl_fc]
6115 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
6116 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
6117 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
6118 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
6119 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
6120 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
6121 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02006122
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006123http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02006124
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006125 This is used to change the log level of the current request when a certain
6126 condition is met. Valid levels are the 8 syslog levels (see the "log"
6127 keyword) plus the special level "silent" which disables logging for this
6128 request. This rule is not final so the last matching rule wins. This rule
6129 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006130
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006131http-request set-map(<file-name>) <key fmt> <value fmt>
6132 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006133
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006134 This is used to add a new entry into a MAP. The MAP must be loaded from a
6135 file (even a dummy empty file). The file name of the MAP to be updated is
6136 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
6137 log-format rules, used to collect MAP key, and <value fmt>, which follows
6138 log-format rules, used to collect content for the new entry.
6139 It performs a lookup in the MAP before insertion, to avoid duplicated (or
6140 more) values. This lookup is done by a linear search and can be expensive
6141 with large lists! It is the equivalent of the "set map" command from the
6142 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006143
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006144http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006145
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006146 This is used to set the Netfilter MARK on all packets sent to the client to
6147 the value passed in <mark> on platforms which support it. This value is an
6148 unsigned 32 bit value which can be matched by netfilter and by the routing
6149 table. It can be expressed both in decimal or hexadecimal format (prefixed by
6150 "0x"). This can be useful to force certain packets to take a different route
6151 (for example a cheaper network path for bulk downloads). This works on Linux
6152 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02006153
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006154http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006155
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006156 This rewrites the request method with the result of the evaluation of format
6157 string <fmt>. There should be very few valid reasons for having to do so as
6158 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006159
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006160http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02006161
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006162 This sets the "nice" factor of the current request being processed. It only
6163 has effect against the other requests being processed at the same time.
6164 The default value is 0, unless altered by the "nice" setting on the "bind"
6165 line. The accepted range is -1024..1024. The higher the value, the nicest
6166 the request will be. Lower values will make the request more important than
6167 other ones. This can be useful to improve the speed of some requests, or
6168 lower the priority of non-important requests. Using this setting without
6169 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02006170
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006171http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02006172
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006173 This rewrites the request path with the result of the evaluation of format
6174 string <fmt>. The query string, if any, is left intact. If a scheme and
6175 authority is found before the path, they are left intact as well. If the
6176 request doesn't have a path ("*"), this one is replaced with the format.
6177 This can be used to prepend a directory component in front of a path for
6178 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02006179
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006180 Example :
6181 # prepend the host name before the path
6182 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006183
Christopher Faulet312294f2020-09-02 17:17:44 +02006184http-request set-pathq <fmt> [ { if | unless } <condition> ]
6185
6186 This does the same as "http-request set-path" except that the query-string is
6187 also rewritten. It may be used to remove the query-string, including the
6188 question mark (it is not possible using "http-request set-query").
6189
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006190http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02006191
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006192 This is used to set the queue priority class of the current request.
6193 The value must be a sample expression which converts to an integer in the
6194 range -2047..2047. Results outside this range will be truncated.
6195 The priority class determines the order in which queued requests are
6196 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006197
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006198http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006199
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006200 This is used to set the queue priority timestamp offset of the current
6201 request. The value must be a sample expression which converts to an integer
6202 in the range -524287..524287. Results outside this range will be truncated.
6203 When a request is queued, it is ordered first by the priority class, then by
6204 the current timestamp adjusted by the given offset in milliseconds. Lower
6205 values have higher priority.
6206 Note that the resulting timestamp is is only tracked with enough precision
6207 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
6208 adjusted timestamp exceeds this value, it will be misidentified as highest
6209 priority. Thus it is important to set "timeout queue" to a value, where when
6210 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006211
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006212http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01006213
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006214 This rewrites the request's query string which appears after the first
6215 question mark ("?") with the result of the evaluation of format string <fmt>.
6216 The part prior to the question mark is left intact. If the request doesn't
6217 contain a question mark and the new value is not empty, then one is added at
6218 the end of the URI, followed by the new value. If a question mark was
6219 present, it will never be removed even if the value is empty. This can be
6220 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08006221
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006222 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006223
6224 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006225 # replace "%3D" with "=" in the query string
6226 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006227
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006228http-request set-src <expr> [ { if | unless } <condition> ]
6229 This is used to set the source IP address to the value of specified
6230 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
6231 provides the correct IP in a HTTP header; or you want to mask source IP for
Olivier Doucet56e31202020-04-21 09:32:56 +02006232 privacy. All subsequent calls to "src" fetch will return this value
6233 (see example).
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006234
6235 Arguments :
6236 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6237 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006238
Olivier Doucet56e31202020-04-21 09:32:56 +02006239 See also "option forwardfor".
6240
Cyril Bonté78caf842010-03-10 22:41:43 +01006241 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006242 http-request set-src hdr(x-forwarded-for)
6243 http-request set-src src,ipmask(24)
6244
Olivier Doucet56e31202020-04-21 09:32:56 +02006245 # After the masking this will track connections
6246 # based on the IP address with the last byte zeroed out.
6247 http-request track-sc0 src
6248
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006249 When possible, set-src preserves the original source port as long as the
6250 address family allows it, otherwise the source port is set to 0.
6251
6252http-request set-src-port <expr> [ { if | unless } <condition> ]
6253
6254 This is used to set the source port address to the value of specified
6255 expression.
6256
6257 Arguments:
6258 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
6259 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01006260
Willy Tarreau20b0de52012-12-24 15:45:22 +01006261 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006262 http-request set-src-port hdr(x-port)
6263 http-request set-src-port int(4000)
6264
6265 When possible, set-src-port preserves the original source address as long as
6266 the address family supports a port, otherwise it forces the source address to
6267 IPv4 "0.0.0.0" before rewriting the port.
6268
Amaury Denoyelle8d228232020-12-10 13:43:54 +01006269http-request set-timeout server|tunnel { <timeout> | <expr> }
6270 [ { if | unless } <condition> ]
6271
6272 This action overrides the specified "server" or "tunnel" timeout for the
6273 current stream only. The timeout can be specified in millisecond or with any
6274 other unit if the number is suffixed by the unit as explained at the top of
6275 this document. It is also possible to write an expression which must returns
6276 a number interpreted as a timeout in millisecond.
6277
6278 Note that the server/tunnel timeouts are only relevant on the backend side
6279 and thus this rule is only available for the proxies with backend
6280 capabilities. Also the timeout value must be non-null to obtain the expected
6281 results.
6282
6283 Example:
6284 http-request set-timeout server 5s
6285 http-request set-timeout hdr(host),map_int(host.lst)
6286
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006287http-request set-tos <tos> [ { if | unless } <condition> ]
6288
6289 This is used to set the TOS or DSCP field value of packets sent to the client
6290 to the value passed in <tos> on platforms which support this. This value
6291 represents the whole 8 bits of the IP TOS field, and can be expressed both in
6292 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
6293 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
6294 be used to adjust some routing behavior on border routers based on some
6295 information from the request.
6296
6297 See RFC 2474, 2597, 3260 and 4594 for more information.
6298
6299http-request set-uri <fmt> [ { if | unless } <condition> ]
6300
6301 This rewrites the request URI with the result of the evaluation of format
6302 string <fmt>. The scheme, authority, path and query string are all replaced
6303 at once. This can be used to rewrite hosts in front of proxies, or to
6304 perform complex modifications to the URI such as moving parts between the
6305 path and the query string.
6306 See also "http-request set-path" and "http-request set-query".
6307
6308http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
6309
6310 This is used to set the contents of a variable. The variable is declared
6311 inline.
6312
6313 Arguments:
6314 <var-name> The name of the variable starts with an indication about its
6315 scope. The scopes allowed are:
6316 "proc" : the variable is shared with the whole process
6317 "sess" : the variable is shared with the whole session
6318 "txn" : the variable is shared with the transaction
6319 (request and response)
6320 "req" : the variable is shared only during request
6321 processing
6322 "res" : the variable is shared only during response
6323 processing
6324 This prefix is followed by a name. The separator is a '.'.
6325 The name may only contain characters 'a-z', 'A-Z', '0-9'
6326 and '_'.
6327
6328 <expr> Is a standard HAProxy expression formed by a sample-fetch
6329 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01006330
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006331 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006332 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006333
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006334http-request send-spoe-group <engine-name> <group-name>
6335 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006336
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006337 This action is used to trigger sending of a group of SPOE messages. To do so,
6338 the SPOE engine used to send messages must be defined, as well as the SPOE
6339 group to send. Of course, the SPOE engine must refer to an existing SPOE
6340 filter. If not engine name is provided on the SPOE filter line, the SPOE
6341 agent name must be used.
6342
6343 Arguments:
6344 <engine-name> The SPOE engine name.
6345
6346 <group-name> The SPOE group name as specified in the engine
6347 configuration.
6348
6349http-request silent-drop [ { if | unless } <condition> ]
6350
6351 This stops the evaluation of the rules and makes the client-facing connection
6352 suddenly disappear using a system-dependent way that tries to prevent the
6353 client from being notified. The effect it then that the client still sees an
6354 established connection while there's none on HAProxy. The purpose is to
6355 achieve a comparable effect to "tarpit" except that it doesn't use any local
6356 resource at all on the machine running HAProxy. It can resist much higher
6357 loads than "tarpit", and slow down stronger attackers. It is important to
6358 understand the impact of using this mechanism. All stateful equipment placed
6359 between the client and HAProxy (firewalls, proxies, load balancers) will also
6360 keep the established connection for a long time and may suffer from this
6361 action.
6362 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6363 option is used to block the emission of a TCP reset. On other systems, the
6364 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6365 router, though it's still delivered to local networks. Do not use it unless
6366 you fully understand how it works.
6367
Christopher Faulet46f95542019-12-20 10:07:22 +01006368http-request strict-mode { on | off }
6369
6370 This enables or disables the strict rewriting mode for following rules. It
6371 does not affect rules declared before it and it is only applicable on rules
6372 performing a rewrite on the requests. When the strict mode is enabled, any
6373 rewrite failure triggers an internal error. Otherwise, such errors are
6374 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006375 rewrites optional while others must be performed to continue the request
Christopher Faulet46f95542019-12-20 10:07:22 +01006376 processing.
6377
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006378 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006379 when a ruleset evaluation ends. So, for instance, if you change the mode on
6380 the frontend, the default mode is restored when HAProxy starts the backend
6381 rules evaluation.
6382
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006383http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
6384http-request tarpit [ { status | deny_status } <code>] [content-type <type>]
6385 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6386 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6387 [ hdr <name> <fmt> ]*
6388 [ { if | unless } <condition> ]
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006389
6390 This stops the evaluation of the rules and immediately blocks the request
6391 without responding for a delay specified by "timeout tarpit" or
6392 "timeout connect" if the former is not set. After that delay, if the client
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006393 is still connected, a response is returned so that the client does not
6394 suspect it has been tarpitted. Logs will report the flags "PT". The goal of
6395 the tarpit rule is to slow down robots during an attack when they're limited
6396 on the number of concurrent requests. It can be very efficient against very
6397 dumb robots, and will significantly reduce the load on firewalls compared to
6398 a "deny" rule. But when facing "correctly" developed robots, it can make
6399 things worse by forcing haproxy and the front firewall to support insane
6400 number of concurrent connections. By default an HTTP error 500 is returned.
6401 But the response may be customized using same syntax than
6402 "http-request return" rules. Thus, see "http-request return" for details.
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006403 For compatibility purpose, when no argument is defined, or only "deny_status",
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006404 the argument "default-errorfiles" is implied. It means
6405 "http-request tarpit [deny_status <status>]" is an alias of
6406 "http-request tarpit [status <status>] default-errorfiles".
6407 No further "http-request" rules are evaluated.
6408 See also "http-request return" and "http-request silent-drop".
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006409
6410http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6411http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6412http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
6413
6414 This enables tracking of sticky counters from current request. These rules do
6415 not stop evaluation and do not change default action. The number of counters
6416 that may be simultaneously tracked by the same connection is set in
6417 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
Matteo Contrini1857b8c2020-10-16 17:35:54 +02006418 so the track-sc number is between 0 and (MAX_SESS_STKCTR-1). The first
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006419 "track-sc0" rule executed enables tracking of the counters of the specified
6420 table as the first set. The first "track-sc1" rule executed enables tracking
6421 of the counters of the specified table as the second set. The first
6422 "track-sc2" rule executed enables tracking of the counters of the specified
6423 table as the third set. It is a recommended practice to use the first set of
6424 counters for the per-frontend counters and the second set for the per-backend
6425 ones. But this is just a guideline, all may be used everywhere.
6426
6427 Arguments :
6428 <key> is mandatory, and is a sample expression rule as described in
6429 section 7.3. It describes what elements of the incoming request or
6430 connection will be analyzed, extracted, combined, and used to
6431 select which table entry to update the counters.
6432
6433 <table> is an optional table to be used instead of the default one, which
6434 is the stick-table declared in the current proxy. All the counters
6435 for the matches and updates for the key will then be performed in
6436 that table until the session ends.
6437
6438 Once a "track-sc*" rule is executed, the key is looked up in the table and if
6439 it is not found, an entry is allocated for it. Then a pointer to that entry
6440 is kept during all the session's life, and this entry's counters are updated
6441 as often as possible, every time the session's counters are updated, and also
6442 systematically when the session ends. Counters are only updated for events
6443 that happen after the tracking has been started. As an exception, connection
6444 counters and request counters are systematically updated so that they reflect
6445 useful information.
6446
6447 If the entry tracks concurrent connection counters, one connection is counted
6448 for as long as the entry is tracked, and the entry will not expire during
6449 that time. Tracking counters also provides a performance advantage over just
6450 checking the keys, because only one table lookup is performed for all ACL
6451 checks that make use of it.
6452
6453http-request unset-var(<var-name>) [ { if | unless } <condition> ]
6454
6455 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006456
6457 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006458 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006459
Christopher Faulet579d83b2019-11-22 15:34:17 +01006460http-request use-service <service-name> [ { if | unless } <condition> ]
6461
6462 This directive executes the configured HTTP service to reply to the request
6463 and stops the evaluation of the rules. An HTTP service may choose to reply by
6464 sending any valid HTTP response or it may immediately close the connection
6465 without sending any response. Outside natives services, for instance the
6466 Prometheus exporter, it is possible to write your own services in Lua. No
6467 further "http-request" rules are evaluated.
6468
6469 Arguments :
6470 <service-name> is mandatory. It is the service to call
6471
6472 Example:
6473 http-request use-service prometheus-exporter if { path /metrics }
6474
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006475http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006476
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02006477 This will delay the processing of the request until the SSL handshake
6478 happened. This is mostly useful to delay processing early data until we're
6479 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006480
Willy Tarreauef781042010-01-27 11:53:01 +01006481
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006482http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006483 Access control for Layer 7 responses
6484
6485 May be used in sections: defaults | frontend | listen | backend
6486 no | yes | yes | yes
6487
6488 The http-response statement defines a set of rules which apply to layer 7
6489 processing. The rules are evaluated in their declaration order when they are
6490 met in a frontend, listen or backend section. Any rule may optionally be
6491 followed by an ACL-based condition, in which case it will only be evaluated
6492 if the condition is true. Since these rules apply on responses, the backend
6493 rules are applied first, followed by the frontend's rules.
6494
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006495 The first keyword is the rule's action. The supported actions are described
6496 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006497
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006498 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006499
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006500 Example:
6501 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02006502
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006503 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06006504
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006505 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
6506 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06006507
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006508 Example:
6509 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06006510
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006511 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06006512
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006513 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
6514 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06006515
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006516 See also : "http-request", section 3.4 about userlists and section 7 about
6517 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06006518
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006519http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006520
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006521 This is used to add a new entry into an ACL. The ACL must be loaded from a
6522 file (even a dummy empty file). The file name of the ACL to be updated is
6523 passed between parentheses. It takes one argument: <key fmt>, which follows
6524 log-format rules, to collect content of the new entry. It performs a lookup
6525 in the ACL before insertion, to avoid duplicated (or more) values.
6526 This lookup is done by a linear search and can be expensive with large lists!
6527 It is the equivalent of the "add acl" command from the stats socket, but can
6528 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06006529
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006530http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006531
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006532 This appends an HTTP header field whose name is specified in <name> and whose
6533 value is defined by <fmt> which follows the log-format rules (see Custom Log
6534 Format in section 8.2.4). This may be used to send a cookie to a client for
6535 example, or to pass some internal information.
6536 This rule is not final, so it is possible to add other similar rules.
6537 Note that header addition is performed immediately, so one rule might reuse
6538 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06006539
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006540http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006541
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006542 This stops the evaluation of the rules and lets the response pass the check.
6543 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06006544
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02006545http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006546
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02006547 See section 6.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06006548
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006549http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06006550
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006551 This captures sample expression <sample> from the response buffer, and
6552 converts it to a string. The resulting string is stored into the next request
6553 "capture" slot, so it will possibly appear next to some captured HTTP
6554 headers. It will then automatically appear in the logs, and it will be
6555 possible to extract it using sample fetch rules to feed it into headers or
6556 anything. Please check section 7.3 (Fetching samples) and
6557 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006558
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006559 The keyword "id" is the id of the capture slot which is used for storing the
6560 string. The capture slot must be defined in an associated frontend.
6561 This is useful to run captures in backends. The slot id can be declared by a
6562 previous directive "http-response capture" or with the "declare capture"
6563 keyword.
Baptiste Assmann19a69b32020-01-16 14:34:22 +01006564
6565 When using this action in a backend, double check that the relevant
6566 frontend(s) have the required capture slots otherwise, this rule will be
6567 ignored at run time. This can't be detected at configuration parsing time
6568 due to HAProxy's ability to dynamically resolve backend name at runtime.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006569
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006570http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02006571
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006572 This is used to delete an entry from an ACL. The ACL must be loaded from a
6573 file (even a dummy empty file). The file name of the ACL to be updated is
6574 passed between parentheses. It takes one argument: <key fmt>, which follows
6575 log-format rules, to collect content of the entry to delete.
6576 It is the equivalent of the "del acl" command from the stats socket, but can
6577 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02006578
Maciej Zdebebdd4c52020-11-20 13:58:48 +00006579http-response del-header <name> [ -m <meth> ] [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02006580
Maciej Zdebebdd4c52020-11-20 13:58:48 +00006581 This removes all HTTP header fields whose name is specified in <name>. <meth>
6582 is the matching method, applied on the header name. Supported matching methods
6583 are "str" (exact match), "beg" (prefix match), "end" (suffix match), "sub"
6584 (substring match) and "reg" (regex match). If not specified, exact matching
6585 method is used.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02006586
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006587http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02006588
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006589 This is used to delete an entry from a MAP. The MAP must be loaded from a
6590 file (even a dummy empty file). The file name of the MAP to be updated is
6591 passed between parentheses. It takes one argument: <key fmt>, which follows
6592 log-format rules, to collect content of the entry to delete.
6593 It takes one argument: "file name" It is the equivalent of the "del map"
6594 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006595
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006596http-response deny [deny_status <status>] [ { if | unless } <condition> ]
6597http-response deny [ { status | deny_status } <code>] [content-type <type>]
6598 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6599 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
6600 [ hdr <name> <fmt> ]*
6601 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006602
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006603 This stops the evaluation of the rules and immediately rejects the response.
6604 By default an HTTP 502 error is returned. But the response may be customized
6605 using same syntax than "http-response return" rules. Thus, see
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006606 "http-response return" for details. For compatibility purpose, when no
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006607 argument is defined, or only "deny_status", the argument "default-errorfiles"
6608 is implied. It means "http-response deny [deny_status <status>]" is an alias
6609 of "http-response deny [status <status>] default-errorfiles".
Christopher Faulet040c8cd2020-01-13 16:43:45 +01006610 No further "http-response" rules are evaluated.
Christopher Faulet5cb513a2020-05-13 17:56:56 +02006611 See also "http-response return".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006612
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006613http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006614
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006615 This performs an HTTP redirection based on a redirect rule.
6616 This supports a format string similarly to "http-request redirect" rules,
6617 with the exception that only the "location" type of redirect is possible on
6618 the response. See the "redirect" keyword for the rule's syntax. When a
6619 redirect rule is applied during a response, connections to the server are
6620 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006621
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006622http-response replace-header <name> <regex-match> <replace-fmt>
6623 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02006624
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006625 This works like "http-request replace-header" except that it works on the
6626 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01006627
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006628 Example:
6629 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02006630
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006631 # applied to:
6632 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006633
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006634 # outputs:
6635 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 +02006636
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006637 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006638
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006639http-response replace-value <name> <regex-match> <replace-fmt>
6640 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006641
Tim Duesterhus6bd909b2020-01-17 15:53:18 +01006642 This works like "http-request replace-value" except that it works on the
Tim Duesterhus2252beb2019-10-29 00:05:13 +01006643 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02006644
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006645 Example:
6646 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006647
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006648 # applied to:
6649 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01006650
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006651 # outputs:
6652 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01006653
Christopher Faulet24231ab2020-01-24 17:44:23 +01006654http-response return [status <code>] [content-type <type>]
6655 [ { default-errorfiles | errorfile <file> | errorfiles <name> |
6656 file <file> | lf-file <file> | string <str> | lf-string <fmt> } ]
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006657 [ hdr <name> <value> ]*
Christopher Faulet24231ab2020-01-24 17:44:23 +01006658 [ { if | unless } <condition> ]
6659
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006660 This stops the evaluation of the rules and immediately returns a response. The
Christopher Faulet24231ab2020-01-24 17:44:23 +01006661 default status code used for the response is 200. It can be optionally
6662 specified as an arguments to "status". The response content-type may also be
Daniel Corbett67a82712020-07-06 23:01:19 -04006663 specified as an argument to "content-type". Finally the response itself may
Christopher Faulet24231ab2020-01-24 17:44:23 +01006664 be defined. If can be a full HTTP response specifying the errorfile to use,
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006665 or the response payload specifying the file or the string to use. These rules
Christopher Faulet24231ab2020-01-24 17:44:23 +01006666 are followed to create the response :
6667
6668 * If neither the errorfile nor the payload to use is defined, a dummy
6669 response is returned. Only the "status" argument is considered. It can be
6670 any code in the range [200, 599]. The "content-type" argument, if any, is
6671 ignored.
6672
6673 * If "default-errorfiles" argument is set, the proxy's errorfiles are
6674 considered. If the "status" argument is defined, it must be one of the
Daniel Corbett67a82712020-07-06 23:01:19 -04006675 status code handled by haproxy (200, 400, 403, 404, 405, 408, 410, 413,
Christopher Faulete095f312020-12-07 11:22:24 +01006676 425, 429, 500, 501, 502, 503, and 504). The "content-type" argument, if
6677 any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006678
6679 * If a specific errorfile is defined, with an "errorfile" argument, the
6680 corresponding file, containing a full HTTP response, is returned. Only the
6681 "status" argument is considered. It must be one of the status code handled
Christopher Faulete095f312020-12-07 11:22:24 +01006682 by haproxy (200, 400, 403, 404, 405, 408, 410, 413, 425, 429, 500, 501,
6683 502, 503, and 504). The "content-type" argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006684
6685 * If an http-errors section is defined, with an "errorfiles" argument, the
6686 corresponding file in the specified http-errors section, containing a full
6687 HTTP response, is returned. Only the "status" argument is considered. It
Daniel Corbett67a82712020-07-06 23:01:19 -04006688 must be one of the status code handled by haproxy (200, 400, 403, 404, 405,
Christopher Faulete095f312020-12-07 11:22:24 +01006689 408, 410, 413, 425, 429, 500, 501, 502, 503, and 504). The "content-type"
Anthonin Bonnefoy85048f82020-06-22 09:17:01 +02006690 argument, if any, is ignored.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006691
6692 * If a "file" or a "lf-file" argument is specified, the file's content is
6693 used as the response payload. If the file is not empty, its content-type
6694 must be set as argument to "content-type". Otherwise, any "content-type"
6695 argument is ignored. With a "lf-file" argument, the file's content is
6696 evaluated as a log-format string. With a "file" argument, it is considered
6697 as a raw content.
6698
6699 * If a "string" or "lf-string" argument is specified, the defined string is
6700 used as the response payload. The content-type must always be set as
6701 argument to "content-type". With a "lf-string" argument, the string is
6702 evaluated as a log-format string. With a "string" argument, it is
6703 considered as a raw string.
6704
Christopher Faulet4a2c1422020-01-31 17:36:01 +01006705 When the response is not based an errorfile, it is possible to appends HTTP
6706 header fields to the response using "hdr" arguments. Otherwise, all "hdr"
6707 arguments are ignored. For each one, the header name is specified in <name>
6708 and its value is defined by <fmt> which follows the log-format rules.
6709
Christopher Faulet24231ab2020-01-24 17:44:23 +01006710 Note that the generated response must be smaller than a buffer. And to avoid
6711 any warning, when an errorfile or a raw file is loaded, the buffer space
Ilya Shipitsin11057a32020-06-21 21:18:27 +05006712 reserved to the headers rewriting should also be free.
Christopher Faulet24231ab2020-01-24 17:44:23 +01006713
6714 No further "http-response" rules are evaluated.
6715
6716 Example:
Daniel Corbett67a82712020-07-06 23:01:19 -04006717 http-response return errorfile /etc/haproxy/errorfiles/200.http \
Christopher Faulet24231ab2020-01-24 17:44:23 +01006718 if { status eq 404 }
6719
6720 http-response return content-type text/plain \
6721 string "This is the end !" \
6722 if { status eq 500 }
6723
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006724http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
6725http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08006726
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006727 This action increments the GPC0 or GPC1 counter according with the sticky
6728 counter designated by <sc-id>. If an error occurs, this action silently fails
6729 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02006730
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006731http-response sc-set-gpt0(<sc-id>) { <int> | <expr> }
6732 [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02006733
Cédric Dufour0d7712d2019-11-06 18:38:53 +01006734 This action sets the 32-bit unsigned GPT0 tag according to the sticky counter
6735 designated by <sc-id> and the value of <int>/<expr>. The expected result is a
6736 boolean. If an error occurs, this action silently fails and the actions
6737 evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01006738
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006739http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02006740
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006741 This action is used to trigger sending of a group of SPOE messages. To do so,
6742 the SPOE engine used to send messages must be defined, as well as the SPOE
6743 group to send. Of course, the SPOE engine must refer to an existing SPOE
6744 filter. If not engine name is provided on the SPOE filter line, the SPOE
6745 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006746
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006747 Arguments:
6748 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006749
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006750 <group-name> The SPOE group name as specified in the engine
6751 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02006752
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006753http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006754
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006755 This does the same as "add-header" except that the header name is first
6756 removed if it existed. This is useful when passing security information to
6757 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006758
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006759http-response set-log-level <level> [ { if | unless } <condition> ]
6760
6761 This is used to change the log level of the current request when a certain
6762 condition is met. Valid levels are the 8 syslog levels (see the "log"
6763 keyword) plus the special level "silent" which disables logging for this
6764 request. This rule is not final so the last matching rule wins. This rule can
6765 be useful to disable health checks coming from another equipment.
6766
6767http-response set-map(<file-name>) <key fmt> <value fmt>
6768
6769 This is used to add a new entry into a MAP. The MAP must be loaded from a
6770 file (even a dummy empty file). The file name of the MAP to be updated is
6771 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
6772 log-format rules, used to collect MAP key, and <value fmt>, which follows
6773 log-format rules, used to collect content for the new entry. It performs a
6774 lookup in the MAP before insertion, to avoid duplicated (or more) values.
6775 This lookup is done by a linear search and can be expensive with large lists!
6776 It is the equivalent of the "set map" command from the stats socket, but can
6777 be triggered by an HTTP response.
6778
6779http-response set-mark <mark> [ { if | unless } <condition> ]
6780
6781 This is used to set the Netfilter MARK on all packets sent to the client to
6782 the value passed in <mark> on platforms which support it. This value is an
6783 unsigned 32 bit value which can be matched by netfilter and by the routing
6784 table. It can be expressed both in decimal or hexadecimal format (prefixed
6785 by "0x"). This can be useful to force certain packets to take a different
6786 route (for example a cheaper network path for bulk downloads). This works on
6787 Linux kernels 2.6.32 and above and requires admin privileges.
6788
6789http-response set-nice <nice> [ { if | unless } <condition> ]
6790
6791 This sets the "nice" factor of the current request being processed.
6792 It only has effect against the other requests being processed at the same
6793 time. The default value is 0, unless altered by the "nice" setting on the
6794 "bind" line. The accepted range is -1024..1024. The higher the value, the
6795 nicest the request will be. Lower values will make the request more important
6796 than other ones. This can be useful to improve the speed of some requests, or
6797 lower the priority of non-important requests. Using this setting without
6798 prior experimentation can cause some major slowdown.
6799
6800http-response set-status <status> [reason <str>]
6801 [ { if | unless } <condition> ]
6802
6803 This replaces the response status code with <status> which must be an integer
6804 between 100 and 999. Optionally, a custom reason text can be provided defined
6805 by <str>, or the default reason for the specified code will be used as a
6806 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08006807
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006808 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006809 # return "431 Request Header Fields Too Large"
6810 http-response set-status 431
6811 # return "503 Slow Down", custom reason
6812 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006813
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006814http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006815
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006816 This is used to set the TOS or DSCP field value of packets sent to the client
6817 to the value passed in <tos> on platforms which support this.
6818 This value represents the whole 8 bits of the IP TOS field, and can be
6819 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
6820 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
6821 always 0. This can be used to adjust some routing behavior on border routers
6822 based on some information from the request.
6823
6824 See RFC 2474, 2597, 3260 and 4594 for more information.
6825
6826http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
6827
6828 This is used to set the contents of a variable. The variable is declared
6829 inline.
6830
6831 Arguments:
6832 <var-name> The name of the variable starts with an indication about its
6833 scope. The scopes allowed are:
6834 "proc" : the variable is shared with the whole process
6835 "sess" : the variable is shared with the whole session
6836 "txn" : the variable is shared with the transaction
6837 (request and response)
6838 "req" : the variable is shared only during request
6839 processing
6840 "res" : the variable is shared only during response
6841 processing
6842 This prefix is followed by a name. The separator is a '.'.
6843 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
6844 and '_'.
6845
6846 <expr> Is a standard HAProxy expression formed by a sample-fetch
6847 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006848
6849 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006850 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006851
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006852http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006853
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006854 This stops the evaluation of the rules and makes the client-facing connection
6855 suddenly disappear using a system-dependent way that tries to prevent the
6856 client from being notified. The effect it then that the client still sees an
6857 established connection while there's none on HAProxy. The purpose is to
6858 achieve a comparable effect to "tarpit" except that it doesn't use any local
6859 resource at all on the machine running HAProxy. It can resist much higher
6860 loads than "tarpit", and slow down stronger attackers. It is important to
6861 understand the impact of using this mechanism. All stateful equipment placed
6862 between the client and HAProxy (firewalls, proxies, load balancers) will also
6863 keep the established connection for a long time and may suffer from this
6864 action.
6865 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
6866 option is used to block the emission of a TCP reset. On other systems, the
6867 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
6868 router, though it's still delivered to local networks. Do not use it unless
6869 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02006870
Christopher Faulet46f95542019-12-20 10:07:22 +01006871http-response strict-mode { on | off }
6872
6873 This enables or disables the strict rewriting mode for following rules. It
6874 does not affect rules declared before it and it is only applicable on rules
6875 performing a rewrite on the responses. When the strict mode is enabled, any
6876 rewrite failure triggers an internal error. Otherwise, such errors are
6877 silently ignored. The purpose of the strict rewriting mode is to make some
Ilya Shipitsin8525fd92020-02-29 12:34:59 +05006878 rewrites optional while others must be performed to continue the response
Christopher Faulet46f95542019-12-20 10:07:22 +01006879 processing.
6880
Christopher Faulet1aea50e2020-01-17 16:03:53 +01006881 By default, the strict rewriting mode is enabled. Its value is also reset
Christopher Faulet46f95542019-12-20 10:07:22 +01006882 when a ruleset evaluation ends. So, for instance, if you change the mode on
Daniel Corbett67a82712020-07-06 23:01:19 -04006883 the backend, the default mode is restored when HAProxy starts the frontend
Christopher Faulet46f95542019-12-20 10:07:22 +01006884 rules evaluation.
6885
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006886http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
6887http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
6888http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02006889
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02006890 This enables tracking of sticky counters from current response. Please refer
6891 to "http-request track-sc" for a complete description. The only difference
6892 from "http-request track-sc" is the <key> sample expression can only make use
6893 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
6894 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
6895 supported, haproxy will fail and warn while parsing the config.
6896
6897http-response unset-var(<var-name>) [ { if | unless } <condition> ]
6898
6899 This is used to unset a variable. See "http-response set-var" for details
6900 about <var-name>.
6901
6902 Example:
6903 http-response unset-var(sess.last_redir)
6904
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02006905
Willy Tarreau30631952015-08-06 15:05:24 +02006906http-reuse { never | safe | aggressive | always }
6907 Declare how idle HTTP connections may be shared between requests
6908
6909 May be used in sections: defaults | frontend | listen | backend
6910 yes | no | yes | yes
6911
6912 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01006913 which is considered safe for reuse is moved back to the server's idle
6914 connections pool so that any other request can make use of it. This is the
6915 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02006916
6917 The argument indicates the desired connection reuse strategy :
6918
Olivier Houchard86006a52018-12-14 19:37:49 +01006919 - "never" : idle connections are never shared between sessions. This mode
6920 may be enforced to cancel a different strategy inherited from
6921 a defaults section or for troubleshooting. For example, if an
6922 old bogus application considers that multiple requests over
6923 the same connection come from the same client and it is not
6924 possible to fix the application, it may be desirable to
6925 disable connection sharing in a single backend. An example of
6926 such an application could be an old haproxy using cookie
6927 insertion in tunnel mode and not checking any request past the
6928 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02006929
Olivier Houchard86006a52018-12-14 19:37:49 +01006930 - "safe" : this is the default and the recommended strategy. The first
6931 request of a session is always sent over its own connection,
6932 and only subsequent requests may be dispatched over other
6933 existing connections. This ensures that in case the server
6934 closes the connection when the request is being sent, the
6935 browser can decide to silently retry it. Since it is exactly
6936 equivalent to regular keep-alive, there should be no side
Amaury Denoyelle27179652020-10-14 18:17:12 +02006937 effects. There is also a special handling for the connections
6938 using protocols subject to Head-of-line blocking (backend with
6939 h2 or fcgi). In this case, when at least one stream is
6940 processed, the used connection is reserved to handle streams
6941 of the same session. When no more streams are processed, the
6942 connection is released and can be reused.
Willy Tarreau30631952015-08-06 15:05:24 +02006943
6944 - "aggressive" : this mode may be useful in webservices environments where
6945 all servers are not necessarily known and where it would be
6946 appreciable to deliver most first requests over existing
6947 connections. In this case, first requests are only delivered
6948 over existing connections that have been reused at least once,
6949 proving that the server correctly supports connection reuse.
6950 It should only be used when it's sure that the client can
6951 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02006952 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02006953 downsides of rare connection failures.
6954
6955 - "always" : this mode is only recommended when the path to the server is
6956 known for never breaking existing connections quickly after
6957 releasing them. It allows the first request of a session to be
6958 sent to an existing connection. This can provide a significant
6959 performance increase over the "safe" strategy when the backend
6960 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01006961 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02006962 sharing. It is recommended that the "http-keep-alive" timeout
6963 remains low in this mode so that no dead connections remain
6964 usable. In most cases, this will lead to the same performance
6965 gains as "aggressive" but with more risks. It should only be
6966 used when it improves the situation over "aggressive".
6967
6968 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006969 connection properties and compatibility. Specifically :
6970 - connections made with "usesrc" followed by a client-dependent value
6971 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02006972
Amaury Denoyelle7239c242020-10-15 16:41:09 +02006973 - connections sent to a server with a variable value as TLS SNI extension
6974 are marked private and are never shared. This is not the case if the SNI
6975 is guaranteed to be a constant, as for example using a literal string;
Willy Tarreau30631952015-08-06 15:05:24 +02006976
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02006977 - connections with certain bogus authentication schemes (relying on the
6978 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02006979
Lukas Tribuse8adfeb2019-11-06 11:50:25 +01006980 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02006981
6982 Note: connection reuse improves the accuracy of the "server maxconn" setting,
6983 because almost no new connection will be established while idle connections
6984 remain available. This is particularly true with the "always" strategy.
6985
6986 See also : "option http-keep-alive", "server maxconn"
6987
6988
Mark Lamourinec2247f02012-01-04 13:02:01 -05006989http-send-name-header [<header>]
6990 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05006991 May be used in sections: defaults | frontend | listen | backend
6992 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05006993 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05006994 <header> The header string to use to send the server name
6995
Willy Tarreau81bef7e2019-10-07 14:58:02 +02006996 The "http-send-name-header" statement causes the header field named <header>
6997 to be set to the name of the target server at the moment the request is about
6998 to be sent on the wire. Any existing occurrences of this header are removed.
6999 Upon retries and redispatches, the header field is updated to always reflect
7000 the server being attempted to connect to. Given that this header is modified
7001 very late in the connection setup, it may have unexpected effects on already
7002 modified headers. For example using it with transport-level header such as
7003 connection, content-length, transfer-encoding and so on will likely result in
7004 invalid requests being sent to the server. Additionally it has been reported
7005 that this directive is currently being used as a way to overwrite the Host
7006 header field in outgoing requests; while this trick has been known to work
7007 as a side effect of the feature for some time, it is not officially supported
7008 and might possibly not work anymore in a future version depending on the
7009 technical difficulties this feature induces. A long-term solution instead
7010 consists in fixing the application which required this trick so that it binds
7011 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05007012
7013 See also : "server"
7014
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01007015id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02007016 Set a persistent ID to a proxy.
7017 May be used in sections : defaults | frontend | listen | backend
7018 no | yes | yes | yes
7019 Arguments : none
7020
7021 Set a persistent ID for the proxy. This ID must be unique and positive.
7022 An unused ID will automatically be assigned if unset. The first assigned
7023 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01007024
7025
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007026ignore-persist { if | unless } <condition>
7027 Declare a condition to ignore persistence
7028 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01007029 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007030
7031 By default, when cookie persistence is enabled, every requests containing
7032 the cookie are unconditionally persistent (assuming the target server is up
7033 and running).
7034
7035 The "ignore-persist" statement allows one to declare various ACL-based
7036 conditions which, when met, will cause a request to ignore persistence.
7037 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007038 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007039 persistence for a specific User-Agent (for example, some web crawler bots).
7040
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007041 The persistence is ignored when an "if" condition is met, or unless an
7042 "unless" condition is met.
7043
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03007044 Example:
7045 acl url_static path_beg /static /images /img /css
7046 acl url_static path_end .gif .png .jpg .css .js
7047 ignore-persist if url_static
7048
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007049 See also : "force-persist", "cookie", and section 7 about ACL usage.
7050
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007051load-server-state-from-file { global | local | none }
7052 Allow seamless reload of HAProxy
7053 May be used in sections: defaults | frontend | listen | backend
7054 yes | no | yes | yes
7055
7056 This directive points HAProxy to a file where server state from previous
7057 running process has been saved. That way, when starting up, before handling
7058 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01007059 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007060 to tell haproxy which file to use. For now, only 2 arguments to either prevent
7061 loading state or load states from a file containing all backends and servers.
7062 The state file can be generated by running the command "show servers state"
7063 over the stats socket and redirect output.
7064
Davor Ocelice9ed2812017-12-25 17:49:28 +01007065 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007066 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02007067 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007068
7069 Arguments:
7070 global load the content of the file pointed by the global directive
7071 named "server-state-file".
7072
7073 local load the content of the file pointed by the directive
7074 "server-state-file-name" if set. If not set, then the backend
7075 name is used as a file name.
7076
7077 none don't load any stat for this backend
7078
7079 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01007080 - server's IP address is preserved across reloads by default, but the
7081 order can be changed thanks to the server's "init-addr" setting. This
7082 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007083 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01007084 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007085
7086 - server's weight is applied from previous running process unless it has
7087 has changed between previous and new configuration files.
7088
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007089 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007090
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007091 global
7092 stats socket /tmp/socket
7093 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007094
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007095 defaults
7096 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007097
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007098 backend bk
7099 server s1 127.0.0.1:22 check weight 11
7100 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007101
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007102
7103 Then one can run :
7104
7105 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
7106
7107 Content of the file /tmp/server_state would be like this:
7108
7109 1
7110 # <field names skipped for the doc example>
7111 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
7112 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
7113
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007114 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007115
7116 global
7117 stats socket /tmp/socket
7118 server-state-base /etc/haproxy/states
7119
7120 defaults
7121 load-server-state-from-file local
7122
7123 backend bk
7124 server s1 127.0.0.1:22 check weight 11
7125 server s2 127.0.0.1:22 check weight 12
7126
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007127
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02007128 Then one can run :
7129
7130 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
7131
7132 Content of the file /etc/haproxy/states/bk would be like this:
7133
7134 1
7135 # <field names skipped for the doc example>
7136 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
7137 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
7138
7139 See also: "server-state-file", "server-state-file-name", and
7140 "show servers state"
7141
Cyril Bonté0d4bf012010-04-25 23:21:46 +02007142
Willy Tarreau2769aa02007-12-27 18:26:09 +01007143log global
Jan Wagner3e678602020-12-17 22:22:32 +01007144log <address> [len <length>] [format <format>] [sample <ranges>:<sample_size>]
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02007145 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02007146no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01007147 Enable per-instance logging of events and traffic.
7148 May be used in sections : defaults | frontend | listen | backend
7149 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02007150
7151 Prefix :
7152 no should be used when the logger list must be flushed. For example,
7153 if you don't want to inherit from the default logger list. This
7154 prefix does not allow arguments.
7155
Willy Tarreau2769aa02007-12-27 18:26:09 +01007156 Arguments :
7157 global should be used when the instance's logging parameters are the
7158 same as the global ones. This is the most common usage. "global"
7159 replaces <address>, <facility> and <level> with those of the log
7160 entries found in the "global" section. Only one "log global"
7161 statement may be used per instance, and this form takes no other
7162 parameter.
7163
7164 <address> indicates where to send the logs. It takes the same format as
7165 for the "global" section's logs, and can be one of :
7166
7167 - An IPv4 address optionally followed by a colon (':') and a UDP
7168 port. If no port is specified, 514 is used by default (the
7169 standard syslog port).
7170
David du Colombier24bb5f52011-03-17 10:40:23 +01007171 - An IPv6 address followed by a colon (':') and optionally a UDP
7172 port. If no port is specified, 514 is used by default (the
7173 standard syslog port).
7174
Willy Tarreau2769aa02007-12-27 18:26:09 +01007175 - A filesystem path to a UNIX domain socket, keeping in mind
7176 considerations for chroot (be sure the path is accessible
7177 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007178 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007179
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007180 - A file descriptor number in the form "fd@<number>", which may
7181 point to a pipe, terminal, or socket. In this case unbuffered
7182 logs are used and one writev() call per log is performed. This
7183 is a bit expensive but acceptable for most workloads. Messages
7184 sent this way will not be truncated but may be dropped, in
7185 which case the DroppedLogs counter will be incremented. The
7186 writev() call is atomic even on pipes for messages up to
7187 PIPE_BUF size, which POSIX recommends to be at least 512 and
7188 which is 4096 bytes on most modern operating systems. Any
7189 larger message may be interleaved with messages from other
7190 processes. Exceptionally for debugging purposes the file
7191 descriptor may also be directed to a file, but doing so will
7192 significantly slow haproxy down as non-blocking calls will be
7193 ignored. Also there will be no way to purge nor rotate this
7194 file without restarting the process. Note that the configured
7195 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01007196 with a TCP syslog server. See also the "short" and "raw"
7197 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007198
7199 - "stdout" / "stderr", which are respectively aliases for "fd@1"
7200 and "fd@2", see above.
7201
Willy Tarreauc046d162019-08-30 15:24:59 +02007202 - A ring buffer in the form "ring@<name>", which will correspond
7203 to an in-memory ring buffer accessible over the CLI using the
7204 "show events" command, which will also list existing rings and
7205 their sizes. Such buffers are lost on reload or restart but
7206 when used as a complement this can help troubleshooting by
7207 having the logs instantly available.
7208
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01007209 You may want to reference some environment variables in the
7210 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01007211
Willy Tarreau18324f52014-06-27 18:10:07 +02007212 <length> is an optional maximum line length. Log lines larger than this
7213 value will be truncated before being sent. The reason is that
7214 syslog servers act differently on log line length. All servers
7215 support the default value of 1024, but some servers simply drop
7216 larger lines while others do log them. If a server supports long
7217 lines, it may make sense to set this value here in order to avoid
7218 truncating long lines. Similarly, if a server drops long lines,
7219 it is preferable to truncate them before sending them. Accepted
7220 values are 80 to 65535 inclusive. The default value of 1024 is
7221 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01007222 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02007223
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02007224 <ranges> A list of comma-separated ranges to identify the logs to sample.
7225 This is used to balance the load of the logs to send to the log
7226 server. The limits of the ranges cannot be null. They are numbered
7227 from 1. The size or period (in number of logs) of the sample must
7228 be set with <sample_size> parameter.
7229
7230 <sample_size>
7231 The size of the sample in number of logs to consider when balancing
7232 their logging loads. It is used to balance the load of the logs to
7233 send to the syslog server. This size must be greater or equal to the
7234 maximum of the high limits of the ranges.
7235 (see also <ranges> parameter).
7236
Willy Tarreauadb345d2018-11-12 07:56:13 +01007237 <format> is the log format used when generating syslog messages. It may be
7238 one of the following :
7239
Emeric Brun0237c4e2020-11-27 16:24:34 +01007240 local Analog to rfc3164 syslog message format except that hostname
7241 field is stripped. This is the default.
7242 Note: option "log-send-hostname" switches the default to
7243 rfc3164.
7244
7245 rfc3164 The RFC3164 syslog message format.
Willy Tarreauadb345d2018-11-12 07:56:13 +01007246 (https://tools.ietf.org/html/rfc3164)
7247
7248 rfc5424 The RFC5424 syslog message format.
7249 (https://tools.ietf.org/html/rfc5424)
7250
Emeric Brun54648852020-07-06 15:54:06 +02007251 priority A message containing only a level plus syslog facility between
7252 angle brackets such as '<63>', followed by the text. The PID,
7253 date, time, process name and system name are omitted. This is
7254 designed to be used with a local log server.
7255
Willy Tarreaue8746a02018-11-12 08:45:00 +01007256 short A message containing only a level between angle brackets such as
7257 '<3>', followed by the text. The PID, date, time, process name
7258 and system name are omitted. This is designed to be used with a
7259 local log server. This format is compatible with what the
7260 systemd logger consumes.
7261
Emeric Brun54648852020-07-06 15:54:06 +02007262 timed A message containing only a level between angle brackets such as
7263 '<3>', followed by ISO date and by the text. The PID, process
7264 name and system name are omitted. This is designed to be
7265 used with a local log server.
7266
7267 iso A message containing only the ISO date, followed by the text.
7268 The PID, process name and system name are omitted. This is
7269 designed to be used with a local log server.
7270
Willy Tarreauc1b06452018-11-12 11:57:56 +01007271 raw A message containing only the text. The level, PID, date, time,
7272 process name and system name are omitted. This is designed to
7273 be used in containers or during development, where the severity
7274 only depends on the file descriptor used (stdout/stderr).
7275
Willy Tarreau2769aa02007-12-27 18:26:09 +01007276 <facility> must be one of the 24 standard syslog facilities :
7277
Willy Tarreaue8746a02018-11-12 08:45:00 +01007278 kern user mail daemon auth syslog lpr news
7279 uucp cron auth2 ftp ntp audit alert cron2
7280 local0 local1 local2 local3 local4 local5 local6 local7
7281
Willy Tarreauc1b06452018-11-12 11:57:56 +01007282 Note that the facility is ignored for the "short" and "raw"
7283 formats, but still required as a positional field. It is
7284 recommended to use "daemon" in this case to make it clear that
7285 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007286
7287 <level> is optional and can be specified to filter outgoing messages. By
7288 default, all messages are sent. If a level is specified, only
7289 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007290 will be sent. An optional minimum level can be specified. If it
7291 is set, logs emitted with a more severe level than this one will
7292 be capped to this level. This is used to avoid sending "emerg"
7293 messages on all terminals on some default syslog configurations.
7294 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007295
7296 emerg alert crit err warning notice info debug
7297
William Lallemand0f99e342011-10-12 17:50:54 +02007298 It is important to keep in mind that it is the frontend which decides what to
7299 log from a connection, and that in case of content switching, the log entries
7300 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01007301
7302 However, backend log declaration define how and where servers status changes
7303 will be logged. Level "notice" will be used to indicate a server going up,
7304 "warning" will be used for termination signals and definitive service
7305 termination, and "alert" will be used for when a server goes down.
7306
7307 Note : According to RFC3164, messages are truncated to 1024 bytes before
7308 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007309
7310 Example :
7311 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01007312 log stdout format short daemon # send log to systemd
7313 log stdout format raw daemon # send everything to stdout
7314 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02007315 log 127.0.0.1:514 local0 notice # only send important events
7316 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02007317 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01007318
Willy Tarreau2769aa02007-12-27 18:26:09 +01007319
William Lallemand48940402012-01-30 16:47:22 +01007320log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007321 Specifies the log format string to use for traffic logs
7322 May be used in sections: defaults | frontend | listen | backend
7323 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01007324
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01007325 This directive specifies the log format string that will be used for all logs
7326 resulting from traffic passing through the frontend using this line. If the
7327 directive is used in a defaults section, all subsequent frontends will use
7328 the same log format. Please see section 8.2.4 which covers the log format
7329 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01007330
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007331 "log-format" directive overrides previous "option tcplog", "log-format" and
7332 "option httplog" directives.
7333
Dragan Dosen7ad31542015-09-28 17:16:47 +02007334log-format-sd <string>
7335 Specifies the RFC5424 structured-data log format string
7336 May be used in sections: defaults | frontend | listen | backend
7337 yes | yes | yes | no
7338
7339 This directive specifies the RFC5424 structured-data log format string that
7340 will be used for all logs resulting from traffic passing through the frontend
7341 using this line. If the directive is used in a defaults section, all
7342 subsequent frontends will use the same log format. Please see section 8.2.4
7343 which covers the log format string in depth.
7344
7345 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
7346 about the RFC5424 structured-data part.
7347
7348 Note : This log format string will be used only for loggers that have set
7349 log format to "rfc5424".
7350
7351 Example :
7352 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
7353
7354
Willy Tarreau094af4e2015-01-07 15:03:42 +01007355log-tag <string>
7356 Specifies the log tag to use for all outgoing logs
7357 May be used in sections: defaults | frontend | listen | backend
7358 yes | yes | yes | yes
7359
7360 Sets the tag field in the syslog header to this string. It defaults to the
7361 log-tag set in the global section, otherwise the program name as launched
7362 from the command line, which usually is "haproxy". Sometimes it can be useful
7363 to differentiate between multiple processes running on the same host, or to
7364 differentiate customer instances running in the same process. In the backend,
7365 logs about servers up/down will use this tag. As a hint, it can be convenient
7366 to set a log-tag related to a hosted customer in a defaults section then put
7367 all the frontends and backends for that customer, then start another customer
7368 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007369
Willy Tarreauc35362a2014-04-25 13:58:37 +02007370max-keep-alive-queue <value>
7371 Set the maximum server queue size for maintaining keep-alive connections
7372 May be used in sections: defaults | frontend | listen | backend
7373 yes | no | yes | yes
7374
7375 HTTP keep-alive tries to reuse the same server connection whenever possible,
7376 but sometimes it can be counter-productive, for example if a server has a lot
7377 of connections while other ones are idle. This is especially true for static
7378 servers.
7379
7380 The purpose of this setting is to set a threshold on the number of queued
7381 connections at which haproxy stops trying to reuse the same server and prefers
7382 to find another one. The default value, -1, means there is no limit. A value
7383 of zero means that keep-alive requests will never be queued. For very close
7384 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007385 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02007386 use a value of 10 or less). For remote servers suffering from a high latency,
7387 higher values might be needed to cover for the latency and/or the cost of
7388 picking a different server.
7389
7390 Note that this has no impact on responses which are maintained to the same
7391 server consecutively to a 401 response. They will still go to the same server
7392 even if they have to be queued.
7393
7394 See also : "option http-server-close", "option prefer-last-server", server
7395 "maxconn" and cookie persistence.
7396
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01007397max-session-srv-conns <nb>
7398 Set the maximum number of outgoing connections we can keep idling for a given
7399 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
7400 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02007401
Willy Tarreau2769aa02007-12-27 18:26:09 +01007402maxconn <conns>
7403 Fix the maximum number of concurrent connections on a frontend
7404 May be used in sections : defaults | frontend | listen | backend
7405 yes | yes | yes | no
7406 Arguments :
7407 <conns> is the maximum number of concurrent connections the frontend will
7408 accept to serve. Excess connections will be queued by the system
7409 in the socket's listen queue and will be served once a connection
7410 closes.
7411
7412 If the system supports it, it can be useful on big sites to raise this limit
7413 very high so that haproxy manages connection queues, instead of leaving the
7414 clients with unanswered connection attempts. This value should not exceed the
7415 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01007416 of tune.bufsize (16kB by default) each, as well as some other data resulting
7417 in about 33 kB of RAM being consumed per established connection. That means
7418 that a medium system equipped with 1GB of RAM can withstand around
7419 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007420
7421 Also, when <conns> is set to large values, it is possible that the servers
7422 are not sized to accept such loads, and for this reason it is generally wise
7423 to assign them some reasonable connection limits.
7424
Willy Tarreauc8d5b952019-02-27 17:25:52 +01007425 When this value is set to zero, which is the default, the global "maxconn"
7426 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02007427
Willy Tarreau2769aa02007-12-27 18:26:09 +01007428 See also : "server", global section's "maxconn", "fullconn"
7429
7430
Willy Tarreau77e0dae2020-10-14 15:44:27 +02007431mode { tcp|http }
Willy Tarreau2769aa02007-12-27 18:26:09 +01007432 Set the running mode or protocol of the instance
7433 May be used in sections : defaults | frontend | listen | backend
7434 yes | yes | yes | yes
7435 Arguments :
7436 tcp The instance will work in pure TCP mode. A full-duplex connection
7437 will be established between clients and servers, and no layer 7
7438 examination will be performed. This is the default mode. It
7439 should be used for SSL, SSH, SMTP, ...
7440
7441 http The instance will work in HTTP mode. The client request will be
7442 analyzed in depth before connecting to any server. Any request
7443 which is not RFC-compliant will be rejected. Layer 7 filtering,
7444 processing and switching will be possible. This is the mode which
7445 brings HAProxy most of its value.
7446
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007447 When doing content switching, it is mandatory that the frontend and the
7448 backend are in the same mode (generally HTTP), otherwise the configuration
7449 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01007450
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007451 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01007452 defaults http_instances
7453 mode http
7454
Willy Tarreau0ba27502007-12-24 16:55:16 +01007455
Cyril Bontéf0c60612010-02-06 14:44:47 +01007456monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01007457 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007458 May be used in sections : defaults | frontend | listen | backend
7459 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01007460 Arguments :
7461 if <cond> the monitor request will fail if the condition is satisfied,
7462 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007463 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01007464 are met, for instance a low number of servers both in a
7465 backend and its backup.
7466
7467 unless <cond> the monitor request will succeed only if the condition is
7468 satisfied, and will fail otherwise. Such a condition may be
7469 based on a test on the presence of a minimum number of active
7470 servers in a list of backends.
7471
7472 This statement adds a condition which can force the response to a monitor
7473 request to report a failure. By default, when an external component queries
7474 the URI dedicated to monitoring, a 200 response is returned. When one of the
7475 conditions above is met, haproxy will return 503 instead of 200. This is
7476 very useful to report a site failure to an external component which may base
7477 routing advertisements between multiple sites on the availability reported by
7478 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02007479 criterion. Note that "monitor fail" only works in HTTP mode. Both status
7480 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007481
7482 Example:
7483 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01007484 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01007485 acl site_dead nbsrv(dynamic) lt 2
7486 acl site_dead nbsrv(static) lt 2
7487 monitor-uri /site_alive
7488 monitor fail if site_dead
7489
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007490 See also : "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007491
7492
Willy Tarreau2769aa02007-12-27 18:26:09 +01007493monitor-uri <uri>
7494 Intercept a URI used by external components' monitor requests
7495 May be used in sections : defaults | frontend | listen | backend
7496 yes | yes | yes | no
7497 Arguments :
7498 <uri> is the exact URI which we want to intercept to return HAProxy's
7499 health status instead of forwarding the request.
7500
7501 When an HTTP request referencing <uri> will be received on a frontend,
7502 HAProxy will not forward it nor log it, but instead will return either
7503 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
7504 conditions defined with "monitor fail". This is normally enough for any
7505 front-end HTTP probe to detect that the service is UP and running without
7506 forwarding the request to a backend server. Note that the HTTP method, the
7507 version and all headers are ignored, but the request must at least be valid
7508 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
7509
Willy Tarreau721d8e02017-12-01 18:25:08 +01007510 Monitor requests are processed very early, just after the request is parsed
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007511 and even before any "http-request". The only rulesets applied before are the
7512 tcp-request ones. They cannot be logged either, and it is the intended
7513 purpose. They are only used to report HAProxy's health to an upper component,
7514 nothing more. However, it is possible to add any number of conditions using
7515 "monitor fail" and ACLs so that the result can be adjusted to whatever check
7516 can be imagined (most often the number of available servers in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01007517
Christopher Faulet6072beb2020-02-18 15:34:58 +01007518 Note: if <uri> starts by a slash ('/'), the matching is performed against the
7519 request's path instead of the request's uri. It is a workaround to let
7520 the HTTP/2 requests match the monitor-uri. Indeed, in HTTP/2, clients
7521 are encouraged to send absolute URIs only.
7522
Willy Tarreau2769aa02007-12-27 18:26:09 +01007523 Example :
7524 # Use /haproxy_test to report haproxy's status
7525 frontend www
7526 mode http
7527 monitor-uri /haproxy_test
7528
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007529 See also : "monitor fail"
Willy Tarreau2769aa02007-12-27 18:26:09 +01007530
Willy Tarreau0ba27502007-12-24 16:55:16 +01007531
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007532option abortonclose
7533no option abortonclose
7534 Enable or disable early dropping of aborted requests pending in queues.
7535 May be used in sections : defaults | frontend | listen | backend
7536 yes | no | yes | yes
7537 Arguments : none
7538
7539 In presence of very high loads, the servers will take some time to respond.
7540 The per-instance connection queue will inflate, and the response time will
7541 increase respective to the size of the queue times the average per-session
7542 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01007543 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007544 the queue, and slowing down other users, and the servers as well, because the
7545 request will eventually be served, then aborted at the first error
7546 encountered while delivering the response.
7547
7548 As there is no way to distinguish between a full STOP and a simple output
7549 close on the client side, HTTP agents should be conservative and consider
7550 that the client might only have closed its output channel while waiting for
7551 the response. However, this introduces risks of congestion when lots of users
7552 do the same, and is completely useless nowadays because probably no client at
7553 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01007554 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007555 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01007556 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007557 of being the single component to break rare but valid traffic is extremely
7558 low, which adds to the temptation to be able to abort a session early while
7559 still not served and not pollute the servers.
7560
Davor Ocelice9ed2812017-12-25 17:49:28 +01007561 In HAProxy, the user can choose the desired behavior using the option
7562 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007563 compliant and aborted requests will be served. But when the option is
7564 specified, a session with an incoming channel closed will be aborted while
7565 it is still possible, either pending in the queue for a connection slot, or
7566 during the connection establishment if the server has not yet acknowledged
7567 the connection request. This considerably reduces the queue size and the load
7568 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01007569 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007570
7571 If this option has been enabled in a "defaults" section, it can be disabled
7572 in a specific instance by prepending the "no" keyword before it.
7573
7574 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
7575
7576
Willy Tarreau4076a152009-04-02 15:18:36 +02007577option accept-invalid-http-request
7578no option accept-invalid-http-request
7579 Enable or disable relaxing of HTTP request parsing
7580 May be used in sections : defaults | frontend | listen | backend
7581 yes | yes | yes | no
7582 Arguments : none
7583
Willy Tarreau91852eb2015-05-01 13:26:00 +02007584 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007585 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007586 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007587 forbidden characters are essentially used to build attacks exploiting server
7588 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7589 server will emit invalid header names for whatever reason (configuration,
7590 implementation) and the issue will not be immediately fixed. In such a case,
7591 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01007592 even if that does not make sense, by specifying this option. Similarly, the
7593 list of characters allowed to appear in a URI is well defined by RFC3986, and
7594 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
7595 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01007596 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02007597 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02007598 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
7599 to pass through (no version specified) and multiple digits for both the major
7600 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007601
7602 This option should never be enabled by default as it hides application bugs
7603 and open security breaches. It should only be deployed after a problem has
7604 been confirmed.
7605
7606 When this option is enabled, erroneous header names will still be accepted in
7607 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01007608 analysis using the "show errors" request on the UNIX stats socket. Similarly,
7609 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02007610 also helps confirming that the issue has been solved.
7611
7612 If this option has been enabled in a "defaults" section, it can be disabled
7613 in a specific instance by prepending the "no" keyword before it.
7614
7615 See also : "option accept-invalid-http-response" and "show errors" on the
7616 stats socket.
7617
7618
7619option accept-invalid-http-response
7620no option accept-invalid-http-response
7621 Enable or disable relaxing of HTTP response parsing
7622 May be used in sections : defaults | frontend | listen | backend
7623 yes | no | yes | yes
7624 Arguments : none
7625
Willy Tarreau91852eb2015-05-01 13:26:00 +02007626 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02007627 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01007628 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02007629 forbidden characters are essentially used to build attacks exploiting server
7630 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
7631 server will emit invalid header names for whatever reason (configuration,
7632 implementation) and the issue will not be immediately fixed. In such a case,
7633 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02007634 even if that does not make sense, by specifying this option. This option also
7635 relaxes the test on the HTTP version format, it allows multiple digits for
7636 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02007637
7638 This option should never be enabled by default as it hides application bugs
7639 and open security breaches. It should only be deployed after a problem has
7640 been confirmed.
7641
7642 When this option is enabled, erroneous header names will still be accepted in
7643 responses, but the complete response will be captured in order to permit
7644 later analysis using the "show errors" request on the UNIX stats socket.
7645 Doing this also helps confirming that the issue has been solved.
7646
7647 If this option has been enabled in a "defaults" section, it can be disabled
7648 in a specific instance by prepending the "no" keyword before it.
7649
7650 See also : "option accept-invalid-http-request" and "show errors" on the
7651 stats socket.
7652
7653
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007654option allbackups
7655no option allbackups
7656 Use either all backup servers at a time or only the first one
7657 May be used in sections : defaults | frontend | listen | backend
7658 yes | no | yes | yes
7659 Arguments : none
7660
7661 By default, the first operational backup server gets all traffic when normal
7662 servers are all down. Sometimes, it may be preferred to use multiple backups
7663 at once, because one will not be enough. When "option allbackups" is enabled,
7664 the load balancing will be performed among all backup servers when all normal
7665 ones are unavailable. The same load balancing algorithm will be used and the
7666 servers' weights will be respected. Thus, there will not be any priority
7667 order between the backup servers anymore.
7668
7669 This option is mostly used with static server farms dedicated to return a
7670 "sorry" page when an application is completely offline.
7671
7672 If this option has been enabled in a "defaults" section, it can be disabled
7673 in a specific instance by prepending the "no" keyword before it.
7674
7675
7676option checkcache
7677no option checkcache
Godbach7056a352013-12-11 20:01:07 +08007678 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007679 May be used in sections : defaults | frontend | listen | backend
7680 yes | no | yes | yes
7681 Arguments : none
7682
7683 Some high-level frameworks set application cookies everywhere and do not
7684 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007685 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007686 high risk of session crossing or stealing between users traversing the same
7687 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007688 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007689
7690 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007691 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01007692 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007693 response to check if there's a risk of caching a cookie on a client-side
7694 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01007695 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007696 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01007697 - all those with a return code other than 200, 203, 204, 206, 300, 301,
7698 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01007699 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01007700 - all those that result from a request using a method other than GET, HEAD,
7701 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01007702 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007703 - those with a 'Pragma: no-cache' header
7704 - those with a 'Cache-control: private' header
7705 - those with a 'Cache-control: no-store' header
7706 - those with a 'Cache-control: max-age=0' header
7707 - those with a 'Cache-control: s-maxage=0' header
7708 - those with a 'Cache-control: no-cache' header
7709 - those with a 'Cache-control: no-cache="set-cookie"' header
7710 - those with a 'Cache-control: no-cache="set-cookie,' header
7711 (allowing other fields after set-cookie)
7712
7713 If a response doesn't respect these requirements, then it will be blocked
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02007714 just as if it was from an "http-response deny" rule, with an "HTTP 502 bad
7715 gateway". The session state shows "PH--" meaning that the proxy blocked the
7716 response during headers processing. Additionally, an alert will be sent in
7717 the logs so that admins are informed that there's something to be fixed.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007718
7719 Due to the high impact on the application, the application should be tested
7720 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007721 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01007722 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007723
7724 If this option has been enabled in a "defaults" section, it can be disabled
7725 in a specific instance by prepending the "no" keyword before it.
7726
7727
7728option clitcpka
7729no option clitcpka
7730 Enable or disable the sending of TCP keepalive packets on the client side
7731 May be used in sections : defaults | frontend | listen | backend
7732 yes | yes | yes | no
7733 Arguments : none
7734
7735 When there is a firewall or any session-aware component between a client and
7736 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007737 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007738 components decides to expire a session which has remained idle for too long.
7739
7740 Enabling socket-level TCP keep-alives makes the system regularly send packets
7741 to the other end of the connection, leaving it active. The delay between
7742 keep-alive probes is controlled by the system only and depends both on the
7743 operating system and its tuning parameters.
7744
7745 It is important to understand that keep-alive packets are neither emitted nor
7746 received at the application level. It is only the network stacks which sees
7747 them. For this reason, even if one side of the proxy already uses keep-alives
7748 to maintain its connection alive, those keep-alive packets will not be
7749 forwarded to the other side of the proxy.
7750
7751 Please note that this has nothing to do with HTTP keep-alive.
7752
7753 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
7754 client side of a connection, which should help when session expirations are
7755 noticed between HAProxy and a client.
7756
7757 If this option has been enabled in a "defaults" section, it can be disabled
7758 in a specific instance by prepending the "no" keyword before it.
7759
7760 See also : "option srvtcpka", "option tcpka"
7761
7762
Willy Tarreau0ba27502007-12-24 16:55:16 +01007763option contstats
7764 Enable continuous traffic statistics updates
7765 May be used in sections : defaults | frontend | listen | backend
7766 yes | yes | yes | no
7767 Arguments : none
7768
7769 By default, counters used for statistics calculation are incremented
7770 only when a session finishes. It works quite well when serving small
7771 objects, but with big ones (for example large images or archives) or
7772 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01007773 a hedgehog. With this option enabled counters get incremented frequently
7774 along the session, typically every 5 seconds, which is often enough to
7775 produce clean graphs. Recounting touches a hotpath directly so it is not
7776 not enabled by default, as it can cause a lot of wakeups for very large
7777 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007778
Christopher Faulet89aed322020-06-02 17:33:56 +02007779option disable-h2-upgrade
7780no option disable-h2-upgrade
7781 Enable or disable the implicit HTTP/2 upgrade from an HTTP/1.x client
7782 connection.
7783 May be used in sections : defaults | frontend | listen | backend
7784 yes | yes | yes | no
7785 Arguments : none
7786
7787 By default, HAProxy is able to implicitly upgrade an HTTP/1.x client
7788 connection to an HTTP/2 connection if the first request it receives from a
7789 given HTTP connection matches the HTTP/2 connection preface (i.e. the string
7790 "PRI * HTTP/2.0\r\n\r\nSM\r\n\r\n"). This way, it is possible to support
7791 HTTP/1.x and HTTP/2 clients on a non-SSL connections. This option must be used to
7792 disable the implicit upgrade. Note this implicit upgrade is only supported
7793 for HTTP proxies, thus this option too. Note also it is possible to force the
7794 HTTP/2 on clear connections by specifying "proto h2" on the bind line.
7795
7796 If this option has been enabled in a "defaults" section, it can be disabled
7797 in a specific instance by prepending the "no" keyword before it.
Willy Tarreau0ba27502007-12-24 16:55:16 +01007798
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007799option dontlog-normal
7800no option dontlog-normal
7801 Enable or disable logging of normal, successful connections
7802 May be used in sections : defaults | frontend | listen | backend
7803 yes | yes | yes | no
7804 Arguments : none
7805
7806 There are large sites dealing with several thousand connections per second
7807 and for which logging is a major pain. Some of them are even forced to turn
7808 logs off and cannot debug production issues. Setting this option ensures that
7809 normal connections, those which experience no error, no timeout, no retry nor
7810 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
7811 mode, the response status code is checked and return codes 5xx will still be
7812 logged.
7813
7814 It is strongly discouraged to use this option as most of the time, the key to
7815 complex issues is in the normal logs which will not be logged here. If you
7816 need to separate logs, see the "log-separate-errors" option instead.
7817
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007818 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007819 logging.
7820
7821
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007822option dontlognull
7823no option dontlognull
7824 Enable or disable logging of null connections
7825 May be used in sections : defaults | frontend | listen | backend
7826 yes | yes | yes | no
7827 Arguments : none
7828
7829 In certain environments, there are components which will regularly connect to
7830 various systems to ensure that they are still alive. It can be the case from
7831 another load balancer as well as from monitoring systems. By default, even a
7832 simple port probe or scan will produce a log. If those connections pollute
7833 the logs too much, it is possible to enable option "dontlognull" to indicate
7834 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02007835 which typically corresponds to those probes. Note that errors will still be
7836 returned to the client and accounted for in the stats. If this is not what is
7837 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007838
7839 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01007840 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007841 would not be logged.
7842
7843 If this option has been enabled in a "defaults" section, it can be disabled
7844 in a specific instance by prepending the "no" keyword before it.
7845
Willy Tarreau9e9919d2020-10-14 15:55:23 +02007846 See also : "log", "http-ignore-probes", "monitor-uri", and
Willy Tarreau0f228a02015-05-01 15:37:53 +02007847 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007848
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007849
Willy Tarreau87cf5142011-08-19 22:57:24 +02007850option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01007851 Enable insertion of the X-Forwarded-For header to requests sent to servers
7852 May be used in sections : defaults | frontend | listen | backend
7853 yes | yes | yes | yes
7854 Arguments :
7855 <network> is an optional argument used to disable this option for sources
7856 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02007857 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01007858 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007859
7860 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
7861 their client address. This is sometimes annoying when the client's IP address
7862 is expected in server logs. To solve this problem, the well-known HTTP header
7863 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
7864 This header contains a value representing the client's IP address. Since this
7865 header is always appended at the end of the existing header list, the server
7866 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02007867 the server's manual to find how to enable use of this standard header. Note
7868 that only the last occurrence of the header must be used, since it is really
7869 possible that the client has already brought one.
7870
Willy Tarreaud72758d2010-01-12 10:42:19 +01007871 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02007872 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01007873 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01007874 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007875 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02007876 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01007877
7878 Sometimes, a same HAProxy instance may be shared between a direct client
7879 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
7880 used to decrypt HTTPS traffic). It is possible to disable the addition of the
7881 header for a known source address or network by adding the "except" keyword
7882 followed by the network address. In this case, any source IP matching the
7883 network will not cause an addition of this header. Most common uses are with
7884 private networks or 127.0.0.1.
7885
Willy Tarreau87cf5142011-08-19 22:57:24 +02007886 Alternatively, the keyword "if-none" states that the header will only be
7887 added if it is not present. This should only be used in perfectly trusted
7888 environment, as this might cause a security issue if headers reaching haproxy
7889 are under the control of the end-user.
7890
Willy Tarreauc27debf2008-01-06 08:57:02 +01007891 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02007892 least one of them uses it, the header will be added. Note that the backend's
7893 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02007894 both are defined. In the case of the "if-none" argument, if at least one of
7895 the frontend or the backend does not specify it, it wants the addition to be
7896 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007897
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02007898 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01007899 # Public HTTP address also used by stunnel on the same machine
7900 frontend www
7901 mode http
7902 option forwardfor except 127.0.0.1 # stunnel already adds the header
7903
Ross Westaf72a1d2008-08-03 10:51:45 +02007904 # Those servers want the IP Address in X-Client
7905 backend www
7906 mode http
7907 option forwardfor header X-Client
7908
Willy Tarreau87cf5142011-08-19 22:57:24 +02007909 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02007910 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01007911
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02007912
Christopher Faulet98fbe952019-07-22 16:18:24 +02007913option h1-case-adjust-bogus-client
7914no option h1-case-adjust-bogus-client
7915 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
7916 May be used in sections : defaults | frontend | listen | backend
7917 yes | yes | yes | no
7918 Arguments : none
7919
7920 There is no standard case for header names because, as stated in RFC7230,
7921 they are case-insensitive. So applications must handle them in a case-
7922 insensitive manner. But some bogus applications violate the standards and
7923 erroneously rely on the cases most commonly used by browsers. This problem
7924 becomes critical with HTTP/2 because all header names must be exchanged in
7925 lower case, and HAProxy follows the same convention. All header names are
7926 sent in lower case to clients and servers, regardless of the HTTP version.
7927
7928 When HAProxy receives an HTTP/1 response, its header names are converted to
7929 lower case and manipulated and sent this way to the clients. If a client is
7930 known to violate the HTTP standards and to fail to process a response coming
7931 from HAProxy, it is possible to transform the lower case header names to a
7932 different format when the response is formatted and sent to the client, by
7933 enabling this option and specifying the list of headers to be reformatted
7934 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
7935 must only be a temporary workaround for the time it takes the client to be
7936 fixed, because clients which require such workarounds might be vulnerable to
7937 content smuggling attacks and must absolutely be fixed.
7938
7939 Please note that this option will not affect standards-compliant clients.
7940
7941 If this option has been enabled in a "defaults" section, it can be disabled
7942 in a specific instance by prepending the "no" keyword before it.
7943
7944 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
7945 "h1-case-adjust-file".
7946
7947
7948option h1-case-adjust-bogus-server
7949no option h1-case-adjust-bogus-server
7950 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
7951 May be used in sections : defaults | frontend | listen | backend
7952 yes | no | yes | yes
7953 Arguments : none
7954
7955 There is no standard case for header names because, as stated in RFC7230,
7956 they are case-insensitive. So applications must handle them in a case-
7957 insensitive manner. But some bogus applications violate the standards and
7958 erroneously rely on the cases most commonly used by browsers. This problem
7959 becomes critical with HTTP/2 because all header names must be exchanged in
7960 lower case, and HAProxy follows the same convention. All header names are
7961 sent in lower case to clients and servers, regardless of the HTTP version.
7962
7963 When HAProxy receives an HTTP/1 request, its header names are converted to
7964 lower case and manipulated and sent this way to the servers. If a server is
7965 known to violate the HTTP standards and to fail to process a request coming
7966 from HAProxy, it is possible to transform the lower case header names to a
7967 different format when the request is formatted and sent to the server, by
7968 enabling this option and specifying the list of headers to be reformatted
7969 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
7970 must only be a temporary workaround for the time it takes the server to be
7971 fixed, because servers which require such workarounds might be vulnerable to
7972 content smuggling attacks and must absolutely be fixed.
7973
7974 Please note that this option will not affect standards-compliant servers.
7975
7976 If this option has been enabled in a "defaults" section, it can be disabled
7977 in a specific instance by prepending the "no" keyword before it.
7978
7979 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
7980 "h1-case-adjust-file".
7981
7982
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02007983option http-buffer-request
7984no option http-buffer-request
7985 Enable or disable waiting for whole HTTP request body before proceeding
7986 May be used in sections : defaults | frontend | listen | backend
7987 yes | yes | yes | yes
7988 Arguments : none
7989
7990 It is sometimes desirable to wait for the body of an HTTP request before
7991 taking a decision. This is what is being done by "balance url_param" for
7992 example. The first use case is to buffer requests from slow clients before
7993 connecting to the server. Another use case consists in taking the routing
7994 decision based on the request body's contents. This option placed in a
7995 frontend or backend forces the HTTP processing to wait until either the whole
Christopher Faulet6db8a2e2019-11-19 16:27:25 +01007996 body is received or the request buffer is full. It can have undesired side
7997 effects with some applications abusing HTTP by expecting unbuffered
7998 transmissions between the frontend and the backend, so this should definitely
7999 not be used by default.
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008000
Baptiste Assmanneccdf432015-10-28 13:49:01 +01008001 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008002
8003
Willy Tarreau0f228a02015-05-01 15:37:53 +02008004option http-ignore-probes
8005no option http-ignore-probes
8006 Enable or disable logging of null connections and request timeouts
8007 May be used in sections : defaults | frontend | listen | backend
8008 yes | yes | yes | no
8009 Arguments : none
8010
8011 Recently some browsers started to implement a "pre-connect" feature
8012 consisting in speculatively connecting to some recently visited web sites
8013 just in case the user would like to visit them. This results in many
8014 connections being established to web sites, which end up in 408 Request
8015 Timeout if the timeout strikes first, or 400 Bad Request when the browser
8016 decides to close them first. These ones pollute the log and feed the error
8017 counters. There was already "option dontlognull" but it's insufficient in
8018 this case. Instead, this option does the following things :
8019 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01008020 was received over a connection before it was closed;
8021 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02008022 - prevent any error counter from being incremented
8023
8024 That way the empty connection is silently ignored. Note that it is better
8025 not to use this unless it is clear that it is needed, because it will hide
8026 real problems. The most common reason for not receiving a request and seeing
8027 a 408 is due to an MTU inconsistency between the client and an intermediary
8028 element such as a VPN, which blocks too large packets. These issues are
8029 generally seen with POST requests as well as GET with large cookies. The logs
8030 are often the only way to detect them.
8031
8032 If this option has been enabled in a "defaults" section, it can be disabled
8033 in a specific instance by prepending the "no" keyword before it.
8034
8035 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
8036
8037
Willy Tarreau16bfb022010-01-16 19:48:41 +01008038option http-keep-alive
8039no option http-keep-alive
8040 Enable or disable HTTP keep-alive from client to server
8041 May be used in sections : defaults | frontend | listen | backend
8042 yes | yes | yes | yes
8043 Arguments : none
8044
Willy Tarreau70dffda2014-01-30 03:07:23 +01008045 By default HAProxy operates in keep-alive mode with regards to persistent
8046 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02008047 leaves the connection idle on both sides between the end of a response and
8048 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008049 as "option http-server-close" or "option httpclose". This option allows to
8050 set back the keep-alive mode, which can be useful when another mode was used
8051 in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01008052
8053 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
8054 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01008055 network) and the fastest session reuse on the server side at the expense
8056 of maintaining idle connections to the servers. In general, it is possible
8057 with this option to achieve approximately twice the request rate that the
8058 "http-server-close" option achieves on small objects. There are mainly two
8059 situations where this option may be useful :
8060
8061 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008062 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01008063
8064 - when the cost of establishing the connection to the server is significant
8065 compared to the cost of retrieving the associated object from the server.
8066
8067 This last case can happen when the server is a fast static server of cache.
8068 In this case, the server will need to be properly tuned to support high enough
8069 connection counts because connections will last until the client sends another
8070 request.
8071
8072 If the client request has to go to another backend or another server due to
8073 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01008074 immediately be closed and a new one re-opened. Option "prefer-last-server" is
8075 available to try optimize server selection so that if the server currently
8076 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01008077
Willy Tarreau16bfb022010-01-16 19:48:41 +01008078 At the moment, logs will not indicate whether requests came from the same
8079 session or not. The accept date reported in the logs corresponds to the end
8080 of the previous request, and the request time corresponds to the time spent
8081 waiting for a new request. The keep-alive request time is still bound to the
8082 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
8083 not set.
8084
Christopher Faulet159e6672019-07-16 15:09:52 +02008085 This option disables and replaces any previous "option httpclose" or "option
8086 http-server-close". When backend and frontend options differ, all of these 4
8087 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01008088
Christopher Faulet315b39c2018-09-21 16:26:19 +02008089 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01008090 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01008091 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01008092
8093
Willy Tarreau96e31212011-05-30 18:10:30 +02008094option http-no-delay
8095no option http-no-delay
8096 Instruct the system to favor low interactive delays over performance in HTTP
8097 May be used in sections : defaults | frontend | listen | backend
8098 yes | yes | yes | yes
8099 Arguments : none
8100
8101 In HTTP, each payload is unidirectional and has no notion of interactivity.
8102 Any agent is expected to queue data somewhat for a reasonably low delay.
8103 There are some very rare server-to-server applications that abuse the HTTP
8104 protocol and expect the payload phase to be highly interactive, with many
8105 interleaved data chunks in both directions within a single request. This is
8106 absolutely not supported by the HTTP specification and will not work across
8107 most proxies or servers. When such applications attempt to do this through
8108 haproxy, it works but they will experience high delays due to the network
8109 optimizations which favor performance by instructing the system to wait for
8110 enough data to be available in order to only send full packets. Typical
8111 delays are around 200 ms per round trip. Note that this only happens with
8112 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
8113 affected.
8114
8115 When "option http-no-delay" is present in either the frontend or the backend
8116 used by a connection, all such optimizations will be disabled in order to
8117 make the exchanges as fast as possible. Of course this offers no guarantee on
8118 the functionality, as it may break at any other place. But if it works via
8119 HAProxy, it will work as fast as possible. This option should never be used
8120 by default, and should never be used at all unless such a buggy application
8121 is discovered. The impact of using this option is an increase of bandwidth
8122 usage and CPU usage, which may significantly lower performance in high
8123 latency environments.
8124
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02008125 See also : "option http-buffer-request"
8126
Willy Tarreau96e31212011-05-30 18:10:30 +02008127
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008128option http-pretend-keepalive
8129no option http-pretend-keepalive
8130 Define whether haproxy will announce keepalive to the server or not
8131 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02008132 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008133 Arguments : none
8134
Christopher Faulet315b39c2018-09-21 16:26:19 +02008135 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008136 adds a "Connection: close" header to the request forwarded to the server.
8137 Unfortunately, when some servers see this header, they automatically refrain
8138 from using the chunked encoding for responses of unknown length, while this
8139 is totally unrelated. The immediate effect is that this prevents haproxy from
8140 maintaining the client connection alive. A second effect is that a client or
8141 a cache could receive an incomplete response without being aware of it, and
8142 consider the response complete.
8143
8144 By setting "option http-pretend-keepalive", haproxy will make the server
8145 believe it will keep the connection alive. The server will then not fall back
8146 to the abnormal undesired above. When haproxy gets the whole response, it
8147 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02008148 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008149 connection is correctly closed on the server side.
8150
8151 It is recommended not to enable this option by default, because most servers
8152 will more efficiently close the connection themselves after the last packet,
8153 and release its buffers slightly earlier. Also, the added packet on the
8154 network could slightly reduce the overall peak performance. However it is
8155 worth noting that when this option is enabled, haproxy will have slightly
8156 less work to do. So if haproxy is the bottleneck on the whole architecture,
8157 enabling this option might save a few CPU cycles.
8158
Christopher Faulet98db9762018-09-21 10:25:19 +02008159 This option may be set in backend and listen sections. Using it in a frontend
8160 section will be ignored and a warning will be reported during startup. It is
8161 a backend related option, so there is no real reason to set it on a
8162 frontend. This option may be combined with "option httpclose", which will
8163 cause keepalive to be announced to the server and close to be announced to
8164 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008165
8166 If this option has been enabled in a "defaults" section, it can be disabled
8167 in a specific instance by prepending the "no" keyword before it.
8168
Christopher Faulet315b39c2018-09-21 16:26:19 +02008169 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01008170 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02008171
Willy Tarreauc27debf2008-01-06 08:57:02 +01008172
Willy Tarreaub608feb2010-01-02 22:47:18 +01008173option http-server-close
8174no option http-server-close
8175 Enable or disable HTTP connection closing on the server side
8176 May be used in sections : defaults | frontend | listen | backend
8177 yes | yes | yes | yes
8178 Arguments : none
8179
Willy Tarreau70dffda2014-01-30 03:07:23 +01008180 By default HAProxy operates in keep-alive mode with regards to persistent
8181 connections: for each connection it processes each request and response, and
8182 leaves the connection idle on both sides between the end of a response and
8183 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008184 as "option http-server-close" or "option httpclose". Setting "option
8185 http-server-close" enables HTTP connection-close mode on the server side
8186 while keeping the ability to support HTTP keep-alive and pipelining on the
8187 client side. This provides the lowest latency on the client side (slow
8188 network) and the fastest session reuse on the server side to save server
8189 resources, similarly to "option httpclose". It also permits non-keepalive
8190 capable servers to be served in keep-alive mode to the clients if they
8191 conform to the requirements of RFC7230. Please note that some servers do not
8192 always conform to those requirements when they see "Connection: close" in the
8193 request. The effect will be that keep-alive will never be used. A workaround
8194 consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01008195
8196 At the moment, logs will not indicate whether requests came from the same
8197 session or not. The accept date reported in the logs corresponds to the end
8198 of the previous request, and the request time corresponds to the time spent
8199 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01008200 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
8201 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01008202
8203 This option may be set both in a frontend and in a backend. It is enabled if
8204 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008205 It disables and replaces any previous "option httpclose" or "option
8206 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8207 combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01008208
8209 If this option has been enabled in a "defaults" section, it can be disabled
8210 in a specific instance by prepending the "no" keyword before it.
8211
Christopher Faulet315b39c2018-09-21 16:26:19 +02008212 See also : "option httpclose", "option http-pretend-keepalive",
8213 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01008214
Willy Tarreau88d349d2010-01-25 12:15:43 +01008215option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01008216no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01008217 Make use of non-standard Proxy-Connection header instead of Connection
8218 May be used in sections : defaults | frontend | listen | backend
8219 yes | yes | yes | no
8220 Arguments : none
8221
Lukas Tribus23953682017-04-28 13:24:30 +00008222 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01008223 Connection header to indicate their wish of persistent or non-persistent
8224 connections, both browsers and proxies ignore this header for proxied
8225 connections and make use of the undocumented, non-standard Proxy-Connection
8226 header instead. The issue begins when trying to put a load balancer between
8227 browsers and such proxies, because there will be a difference between what
8228 haproxy understands and what the client and the proxy agree on.
8229
8230 By setting this option in a frontend, haproxy can automatically switch to use
8231 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008232 defined here as one where the URI begins with neither a '/' nor a '*'. This
8233 is incompatible with the HTTP tunnel mode. Note that this option can only be
8234 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01008235
Willy Tarreau844a7e72010-01-31 21:46:18 +01008236 Also, when this option is set, a request which requires authentication will
8237 automatically switch to use proxy authentication headers if it is itself a
8238 proxied request. That makes it possible to check or enforce authentication in
8239 front of an existing proxy.
8240
Willy Tarreau88d349d2010-01-25 12:15:43 +01008241 This option should normally never be used, except in front of a proxy.
8242
Christopher Faulet315b39c2018-09-21 16:26:19 +02008243 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01008244
Willy Tarreaud63335a2010-02-26 12:56:52 +01008245option httpchk
8246option httpchk <uri>
8247option httpchk <method> <uri>
8248option httpchk <method> <uri> <version>
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008249 Enables HTTP protocol to check on the servers health
Willy Tarreaud63335a2010-02-26 12:56:52 +01008250 May be used in sections : defaults | frontend | listen | backend
8251 yes | no | yes | yes
8252 Arguments :
8253 <method> is the optional HTTP method used with the requests. When not set,
8254 the "OPTIONS" method is used, as it generally requires low server
8255 processing and is easy to filter out from the logs. Any method
8256 may be used, though it is not recommended to invent non-standard
8257 ones.
8258
8259 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
8260 which is accessible by default on almost any server, but may be
8261 changed to any other URI. Query strings are permitted.
8262
8263 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
8264 but some servers might behave incorrectly in HTTP 1.0, so turning
8265 it to HTTP/1.1 may sometimes help. Note that the Host field is
Christopher Faulet8acb1282020-04-09 08:44:06 +02008266 mandatory in HTTP/1.1, use "http-check send" directive to add it.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008267
8268 By default, server health checks only consist in trying to establish a TCP
8269 connection. When "option httpchk" is specified, a complete HTTP request is
8270 sent once the TCP connection is established, and responses 2xx and 3xx are
8271 considered valid, while all other ones indicate a server failure, including
8272 the lack of any response.
8273
Christopher Faulete5870d82020-04-15 11:32:03 +02008274 Combined with "http-check" directives, it is possible to customize the
8275 request sent during the HTTP health checks or the matching rules on the
8276 response. It is also possible to configure a send/expect sequence, just like
8277 with the directive "tcp-check" for TCP health checks.
8278
8279 The server configuration is used by default to open connections to perform
8280 HTTP health checks. By it is also possible to overwrite server parameters
8281 using "http-check connect" rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008282
Christopher Faulete5870d82020-04-15 11:32:03 +02008283 "httpchk" option does not necessarily require an HTTP backend, it also works
8284 with plain TCP backends. This is particularly useful to check simple scripts
Christopher Faulet14cd3162020-04-16 14:50:06 +02008285 bound to some dedicated ports using the inetd daemon. However, it will always
Daniel Corbett67a82712020-07-06 23:01:19 -04008286 internally relies on an HTX multiplexer. Thus, it means the request
Christopher Faulet14cd3162020-04-16 14:50:06 +02008287 formatting and the response parsing will be strict.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008288
Christopher Faulet8acb1282020-04-09 08:44:06 +02008289 Note : For a while, there was no way to add headers or body in the request
8290 used for HTTP health checks. So a workaround was to hide it at the end
8291 of the version string with a "\r\n" after the version. It is now
8292 deprecated. The directive "http-check send" must be used instead.
8293
Willy Tarreaud63335a2010-02-26 12:56:52 +01008294 Examples :
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02008295 # Relay HTTPS traffic to Apache instance and check service availability
8296 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
8297 backend https_relay
8298 mode tcp
8299 option httpchk OPTIONS * HTTP/1.1
8300 http-check send hdr Host www
8301 server apache1 192.168.1.1:443 check port 80
Willy Tarreaud63335a2010-02-26 12:56:52 +01008302
Simon Hormanafc47ee2013-11-25 10:46:35 +09008303 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
8304 "option pgsql-check", "http-check" and the "check", "port" and
8305 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008306
8307
Willy Tarreauc27debf2008-01-06 08:57:02 +01008308option httpclose
8309no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02008310 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01008311 May be used in sections : defaults | frontend | listen | backend
8312 yes | yes | yes | yes
8313 Arguments : none
8314
Willy Tarreau70dffda2014-01-30 03:07:23 +01008315 By default HAProxy operates in keep-alive mode with regards to persistent
8316 connections: for each connection it processes each request and response, and
8317 leaves the connection idle on both sides between the end of a response and
8318 the start of a new request. This mode may be changed by several options such
Christopher Faulet159e6672019-07-16 15:09:52 +02008319 as "option http-server-close" or "option httpclose".
Willy Tarreau70dffda2014-01-30 03:07:23 +01008320
Christopher Faulet315b39c2018-09-21 16:26:19 +02008321 If "option httpclose" is set, HAProxy will close connections with the server
8322 and the client as soon as the request and the response are received. It will
John Roeslerfb2fce12019-07-10 15:45:51 -05008323 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02008324 and will add one if missing. Any "Connection" header different from "close"
8325 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008326
Christopher Faulet315b39c2018-09-21 16:26:19 +02008327 This option may also be combined with "option http-pretend-keepalive", which
8328 will disable sending of the "Connection: close" header, but will still cause
8329 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008330
8331 This option may be set both in a frontend and in a backend. It is enabled if
8332 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet159e6672019-07-16 15:09:52 +02008333 It disables and replaces any previous "option http-server-close" or "option
8334 http-keep-alive". Please check section 4 ("Proxies") to see how this option
8335 combines with others when frontend and backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008336
8337 If this option has been enabled in a "defaults" section, it can be disabled
8338 in a specific instance by prepending the "no" keyword before it.
8339
Christopher Faulet315b39c2018-09-21 16:26:19 +02008340 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01008341
8342
Emeric Brun3a058f32009-06-30 18:26:00 +02008343option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01008344 Enable logging of HTTP request, session state and timers
8345 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01008346 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02008347 Arguments :
8348 clf if the "clf" argument is added, then the output format will be
8349 the CLF format instead of HAProxy's default HTTP format. You can
8350 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01008351 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02008352 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008353
8354 By default, the log output format is very poor, as it only contains the
8355 source and destination addresses, and the instance name. By specifying
8356 "option httplog", each log line turns into a much richer format including,
8357 but not limited to, the HTTP request, the connection timers, the session
8358 status, the connections numbers, the captured headers and cookies, the
8359 frontend, backend and server name, and of course the source address and
8360 ports.
8361
PiBa-NLbd556bf2014-12-11 21:31:54 +01008362 Specifying only "option httplog" will automatically clear the 'clf' mode
8363 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02008364
Guillaume de Lafond29f45602017-03-31 19:52:15 +02008365 "option httplog" overrides any previous "log-format" directive.
8366
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008367 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008368
Willy Tarreau55165fe2009-05-10 12:02:55 +02008369
8370option http_proxy
8371no option http_proxy
8372 Enable or disable plain HTTP proxy mode
8373 May be used in sections : defaults | frontend | listen | backend
8374 yes | yes | yes | yes
8375 Arguments : none
8376
8377 It sometimes happens that people need a pure HTTP proxy which understands
8378 basic proxy requests without caching nor any fancy feature. In this case,
8379 it may be worth setting up an HAProxy instance with the "option http_proxy"
8380 set. In this mode, no server is declared, and the connection is forwarded to
8381 the IP address and port found in the URL after the "http://" scheme.
8382
8383 No host address resolution is performed, so this only works when pure IP
8384 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01008385 it will probably be used only by experts who know they need exactly it. This
8386 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02008387
8388 If this option has been enabled in a "defaults" section, it can be disabled
8389 in a specific instance by prepending the "no" keyword before it.
8390
8391 Example :
8392 # this backend understands HTTP proxy requests and forwards them directly.
8393 backend direct_forward
8394 option httpclose
8395 option http_proxy
8396
8397 See also : "option httpclose"
8398
Willy Tarreau211ad242009-10-03 21:45:07 +02008399
Jamie Gloudon801a0a32012-08-25 00:18:33 -04008400option independent-streams
8401no option independent-streams
8402 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008403 May be used in sections : defaults | frontend | listen | backend
8404 yes | yes | yes | yes
8405 Arguments : none
8406
8407 By default, when data is sent over a socket, both the write timeout and the
8408 read timeout for that socket are refreshed, because we consider that there is
8409 activity on that socket, and we have no other means of guessing if we should
8410 receive data or not.
8411
Davor Ocelice9ed2812017-12-25 17:49:28 +01008412 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008413 exists a situation where it is desirable to disable it, and only refresh the
8414 read timeout if there are incoming data. This happens on sessions with large
8415 timeouts and low amounts of exchanged data such as telnet session. If the
8416 server suddenly disappears, the output data accumulates in the system's
8417 socket buffers, both timeouts are correctly refreshed, and there is no way
8418 to know the server does not receive them, so we don't timeout. However, when
8419 the underlying protocol always echoes sent data, it would be enough by itself
8420 to detect the issue using the read timeout. Note that this problem does not
8421 happen with more verbose protocols because data won't accumulate long in the
8422 socket buffers.
8423
8424 When this option is set on the frontend, it will disable read timeout updates
8425 on data sent to the client. There probably is little use of this case. When
8426 the option is set on the backend, it will disable read timeout updates on
8427 data sent to the server. Doing so will typically break large HTTP posts from
8428 slow lines, so use it with caution.
8429
Willy Tarreauce887fd2012-05-12 12:50:00 +02008430 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02008431
8432
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02008433option ldap-check
8434 Use LDAPv3 health checks for server testing
8435 May be used in sections : defaults | frontend | listen | backend
8436 yes | no | yes | yes
8437 Arguments : none
8438
8439 It is possible to test that the server correctly talks LDAPv3 instead of just
8440 testing that it accepts the TCP connection. When this option is set, an
8441 LDAPv3 anonymous simple bind message is sent to the server, and the response
8442 is analyzed to find an LDAPv3 bind response message.
8443
8444 The server is considered valid only when the LDAP response contains success
8445 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
8446
8447 Logging of bind requests is server dependent see your documentation how to
8448 configure it.
8449
8450 Example :
8451 option ldap-check
8452
8453 See also : "option httpchk"
8454
8455
Simon Horman98637e52014-06-20 12:30:16 +09008456option external-check
8457 Use external processes for server health checks
8458 May be used in sections : defaults | frontend | listen | backend
8459 yes | no | yes | yes
8460
8461 It is possible to test the health of a server using an external command.
8462 This is achieved by running the executable set using "external-check
8463 command".
8464
8465 Requires the "external-check" global to be set.
8466
8467 See also : "external-check", "external-check command", "external-check path"
8468
8469
Willy Tarreau211ad242009-10-03 21:45:07 +02008470option log-health-checks
8471no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02008472 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02008473 May be used in sections : defaults | frontend | listen | backend
8474 yes | no | yes | yes
8475 Arguments : none
8476
Willy Tarreaubef1b322014-05-13 21:01:39 +02008477 By default, failed health check are logged if server is UP and successful
8478 health checks are logged if server is DOWN, so the amount of additional
8479 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02008480
Willy Tarreaubef1b322014-05-13 21:01:39 +02008481 When this option is enabled, any change of the health check status or to
8482 the server's health will be logged, so that it becomes possible to know
8483 that a server was failing occasional checks before crashing, or exactly when
8484 it failed to respond a valid HTTP status, then when the port started to
8485 reject connections, then when the server stopped responding at all.
8486
Davor Ocelice9ed2812017-12-25 17:49:28 +01008487 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02008488 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02008489
Willy Tarreaubef1b322014-05-13 21:01:39 +02008490 See also: "option httpchk", "option ldap-check", "option mysql-check",
8491 "option pgsql-check", "option redis-check", "option smtpchk",
8492 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02008493
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008494
8495option log-separate-errors
8496no option log-separate-errors
8497 Change log level for non-completely successful connections
8498 May be used in sections : defaults | frontend | listen | backend
8499 yes | yes | yes | no
8500 Arguments : none
8501
8502 Sometimes looking for errors in logs is not easy. This option makes haproxy
8503 raise the level of logs containing potentially interesting information such
8504 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
8505 level changes from "info" to "err". This makes it possible to log them
8506 separately to a different file with most syslog daemons. Be careful not to
8507 remove them from the original file, otherwise you would lose ordering which
8508 provides very important information.
8509
8510 Using this option, large sites dealing with several thousand connections per
8511 second may log normal traffic to a rotating buffer and only archive smaller
8512 error logs.
8513
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008514 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02008515 logging.
8516
Willy Tarreauc27debf2008-01-06 08:57:02 +01008517
8518option logasap
8519no option logasap
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008520 Enable or disable early logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008521 May be used in sections : defaults | frontend | listen | backend
8522 yes | yes | yes | no
8523 Arguments : none
8524
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008525 By default, logs are emitted when all the log format variables and sample
8526 fetches used in the definition of the log-format string return a value, or
8527 when the session is terminated. This allows the built in log-format strings
8528 to account for the transfer time, or the number of bytes in log messages.
8529
8530 When handling long lived connections such as large file transfers or RDP,
8531 it may take a while for the request or connection to appear in the logs.
8532 Using "option logasap", the log message is created as soon as the server
8533 connection is established in mode tcp, or as soon as the server sends the
8534 complete headers in mode http. Missing information in the logs will be the
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008535 total number of bytes which will only indicate the amount of data transferred
Jerome Magnin95fb57b2020-04-23 19:01:17 +02008536 before the message was created and the total time which will not take the
8537 remainder of the connection life or transfer time into account. For the case
8538 of HTTP, it is good practice to capture the Content-Length response header
8539 so that the logs at least indicate how many bytes are expected to be
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +05008540 transferred.
Willy Tarreauc27debf2008-01-06 08:57:02 +01008541
Willy Tarreaucc6c8912009-02-22 10:53:55 +01008542 Examples :
8543 listen http_proxy 0.0.0.0:80
8544 mode http
8545 option httplog
8546 option logasap
8547 log 192.168.2.200 local3
8548
8549 >>> Feb 6 12:14:14 localhost \
8550 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
8551 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
8552 "GET /image.iso HTTP/1.0"
8553
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008554 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01008555 logging.
8556
8557
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008558option mysql-check [ user <username> [ { post-41 | pre-41 } ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008559 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008560 May be used in sections : defaults | frontend | listen | backend
8561 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008562 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02008563 <username> This is the username which will be used when connecting to MySQL
8564 server.
Christopher Faulet62f79fe2020-05-18 18:13:03 +02008565 post-41 Send post v4.1 client compatible checks (the default)
8566 pre-41 Send pre v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008567
8568 If you specify a username, the check consists of sending two MySQL packet,
8569 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01008570 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008571 Error packet. It is a basic but useful test which does not produce error nor
8572 aborted connect on the server. However, it requires adding an authorization
8573 in the MySQL table, like this :
8574
8575 USE mysql;
8576 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
8577 FLUSH PRIVILEGES;
8578
8579 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008580 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02008581 Error packet, we don't send anything in this mode. It was reported that it
8582 can generate lockout if check is too frequent and/or if there is not enough
8583 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
8584 value as if a connection is established successfully within fewer than MySQL
8585 "max_connect_errors" attempts after a previous connection was interrupted,
8586 the error count for the host is cleared to zero. If HAProxy's server get
8587 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
8588
8589 Remember that this does not check database presence nor database consistency.
8590 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008591
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02008592 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008593
8594 Most often, an incoming MySQL server needs to see the client's IP address for
8595 various purposes, including IP privilege matching and connection logging.
8596 When possible, it is often wise to masquerade the client's IP address when
8597 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02008598 which requires the transparent proxy feature to be compiled in, and the MySQL
8599 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01008600
8601 See also: "option httpchk"
8602
8603
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008604option nolinger
8605no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008606 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008607 May be used in sections: defaults | frontend | listen | backend
8608 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008609 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008610
Davor Ocelice9ed2812017-12-25 17:49:28 +01008611 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008612 physically disconnected), the session timeouts triggers and the session is
8613 closed. But it will remain in FIN_WAIT1 state for some time in the system,
8614 using some resources and possibly limiting the ability to establish newer
8615 connections.
8616
8617 When this happens, it is possible to activate "option nolinger" which forces
8618 the system to immediately remove any socket's pending data on close. Thus,
Willy Tarreau4a321032020-10-16 04:55:19 +02008619 a TCP RST is emitted, any pending data are truncated, and the session is
8620 instantly purged from the system's tables. The generally visible effect for
8621 a client is that responses are truncated if the close happens with a last
8622 block of data (e.g. on a redirect or error response). On the server side,
8623 it may help release the source ports immediately when forwarding a client
8624 aborts in tunnels. In both cases, TCP resets are emitted and given that
8625 the session is instantly destroyed, there will be no retransmit. On a lossy
8626 network this can increase problems, especially when there is a firewall on
8627 the lossy side, because the firewall might see and process the reset (hence
8628 purge its session) and block any further traffic for this session,, including
8629 retransmits from the other side. So if the other side doesn't receive it,
8630 it will never receive any RST again, and the firewall might log many blocked
8631 packets.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008632
Willy Tarreau4a321032020-10-16 04:55:19 +02008633 For all these reasons, it is strongly recommended NOT to use this option,
8634 unless absolutely needed as a last resort. In most situations, using the
8635 "client-fin" or "server-fin" timeouts achieves similar results with a more
8636 reliable behavior. On Linux it's also possible to use the "tcp-ut" bind or
8637 server setting.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008638
8639 This option may be used both on frontends and backends, depending on the side
8640 where it is required. Use it on the frontend for clients, and on the backend
Willy Tarreau4a321032020-10-16 04:55:19 +02008641 for servers. While this option is technically supported in "defaults"
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +05008642 sections, it must really not be used there as it risks to accidentally
Willy Tarreau4a321032020-10-16 04:55:19 +02008643 propagate to sections that must no use it and to cause problems there.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008644
8645 If this option has been enabled in a "defaults" section, it can be disabled
8646 in a specific instance by prepending the "no" keyword before it.
8647
Willy Tarreau4a321032020-10-16 04:55:19 +02008648 See also: "timeout client-fin", "timeout server-fin", "tcp-ut" bind or server
8649 keywords.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008650
Willy Tarreau55165fe2009-05-10 12:02:55 +02008651option originalto [ except <network> ] [ header <name> ]
8652 Enable insertion of the X-Original-To header to requests sent to servers
8653 May be used in sections : defaults | frontend | listen | backend
8654 yes | yes | yes | yes
8655 Arguments :
8656 <network> is an optional argument used to disable this option for sources
8657 matching <network>
8658 <name> an optional argument to specify a different "X-Original-To"
8659 header name.
8660
8661 Since HAProxy can work in transparent mode, every request from a client can
8662 be redirected to the proxy and HAProxy itself can proxy every request to a
8663 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
8664 be lost. This is annoying when you want access rules based on destination ip
8665 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
8666 added by HAProxy to all requests sent to the server. This header contains a
8667 value representing the original destination IP address. Since this must be
8668 configured to always use the last occurrence of this header only. Note that
8669 only the last occurrence of the header must be used, since it is really
8670 possible that the client has already brought one.
8671
8672 The keyword "header" may be used to supply a different header name to replace
8673 the default "X-Original-To". This can be useful where you might already
8674 have a "X-Original-To" header from a different application, and you need
8675 preserve it. Also if your backend server doesn't use the "X-Original-To"
8676 header and requires different one.
8677
8678 Sometimes, a same HAProxy instance may be shared between a direct client
8679 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
8680 used to decrypt HTTPS traffic). It is possible to disable the addition of the
8681 header for a known source address or network by adding the "except" keyword
8682 followed by the network address. In this case, any source IP matching the
8683 network will not cause an addition of this header. Most common uses are with
8684 private networks or 127.0.0.1.
8685
8686 This option may be specified either in the frontend or in the backend. If at
8687 least one of them uses it, the header will be added. Note that the backend's
8688 setting of the header subargument takes precedence over the frontend's if
8689 both are defined.
8690
Willy Tarreau55165fe2009-05-10 12:02:55 +02008691 Examples :
8692 # Original Destination address
8693 frontend www
8694 mode http
8695 option originalto except 127.0.0.1
8696
8697 # Those servers want the IP Address in X-Client-Dst
8698 backend www
8699 mode http
8700 option originalto header X-Client-Dst
8701
Christopher Faulet315b39c2018-09-21 16:26:19 +02008702 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02008703
8704
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008705option persist
8706no option persist
8707 Enable or disable forced persistence on down servers
8708 May be used in sections: defaults | frontend | listen | backend
8709 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008710 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008711
8712 When an HTTP request reaches a backend with a cookie which references a dead
8713 server, by default it is redispatched to another server. It is possible to
8714 force the request to be sent to the dead server first using "option persist"
8715 if absolutely needed. A common use case is when servers are under extreme
8716 load and spend their time flapping. In this case, the users would still be
8717 directed to the server they opened the session on, in the hope they would be
8718 correctly served. It is recommended to use "option redispatch" in conjunction
8719 with this option so that in the event it would not be possible to connect to
8720 the server at all (server definitely dead), the client would finally be
8721 redirected to another valid server.
8722
8723 If this option has been enabled in a "defaults" section, it can be disabled
8724 in a specific instance by prepending the "no" keyword before it.
8725
Willy Tarreau4de91492010-01-22 19:10:05 +01008726 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008727
8728
Willy Tarreau0c122822013-12-15 18:49:01 +01008729option pgsql-check [ user <username> ]
8730 Use PostgreSQL health checks for server testing
8731 May be used in sections : defaults | frontend | listen | backend
8732 yes | no | yes | yes
8733 Arguments :
8734 <username> This is the username which will be used when connecting to
8735 PostgreSQL server.
8736
8737 The check sends a PostgreSQL StartupMessage and waits for either
8738 Authentication request or ErrorResponse message. It is a basic but useful
8739 test which does not produce error nor aborted connect on the server.
8740 This check is identical with the "mysql-check".
8741
8742 See also: "option httpchk"
8743
8744
Willy Tarreau9420b122013-12-15 18:58:25 +01008745option prefer-last-server
8746no option prefer-last-server
8747 Allow multiple load balanced requests to remain on the same server
8748 May be used in sections: defaults | frontend | listen | backend
8749 yes | no | yes | yes
8750 Arguments : none
8751
8752 When the load balancing algorithm in use is not deterministic, and a previous
8753 request was sent to a server to which haproxy still holds a connection, it is
8754 sometimes desirable that subsequent requests on a same session go to the same
8755 server as much as possible. Note that this is different from persistence, as
8756 we only indicate a preference which haproxy tries to apply without any form
8757 of warranty. The real use is for keep-alive connections sent to servers. When
8758 this option is used, haproxy will try to reuse the same connection that is
8759 attached to the server instead of rebalancing to another server, causing a
8760 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01008761 not make much sense to use this in combination with hashing algorithms. Note,
8762 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02008763 to a proxy which sends a 407 (authentication required), when the load
8764 balancing algorithm is not deterministic. This is mandatory for use with the
8765 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01008766 troubleshooting some faulty applications. Option prefer-last-server might be
8767 desirable in these environments as well, to avoid redistributing the traffic
8768 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01008769
8770 If this option has been enabled in a "defaults" section, it can be disabled
8771 in a specific instance by prepending the "no" keyword before it.
8772
8773 See also: "option http-keep-alive"
8774
8775
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008776option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07008777option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008778no option redispatch
8779 Enable or disable session redistribution in case of connection failure
8780 May be used in sections: defaults | frontend | listen | backend
8781 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07008782 Arguments :
8783 <interval> The optional integer value that controls how often redispatches
8784 occur when retrying connections. Positive value P indicates a
8785 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01008786 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07008787 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008788 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07008789 positive value of 1 would indicate a redispatch on every retry,
8790 and a positive value of 3 would indicate a redispatch on every
8791 third retry. You can disable redispatches with a value of 0.
8792
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008793
8794 In HTTP mode, if a server designated by a cookie is down, clients may
8795 definitely stick to it because they cannot flush the cookie, so they will not
8796 be able to access the service anymore.
8797
Willy Tarreau59884a62019-01-02 14:48:31 +01008798 Specifying "option redispatch" will allow the proxy to break cookie or
8799 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008800
Olivier Carrère6e6f59b2020-04-15 11:30:18 +02008801 Active servers are selected from a subset of the list of available
8802 servers. Active servers that are not down or in maintenance (i.e., whose
8803 health is not checked or that have been checked as "up"), are selected in the
8804 following order:
8805
8806 1. Any active, non-backup server, if any, or,
8807
8808 2. If the "allbackups" option is not set, the first backup server in the
8809 list, or
8810
8811 3. If the "allbackups" option is set, any backup server.
8812
8813 When a retry occurs, HAProxy tries to select another server than the last
8814 one. The new server is selected from the current list of servers.
8815
8816 Sometimes, if the list is updated between retries (e.g., if numerous retries
8817 occur and last longer than the time needed to check that a server is down,
8818 remove it from the list and fall back on the list of backup servers),
8819 connections may be redirected to a backup server, though.
8820
Joseph Lynch726ab712015-05-11 23:25:34 -07008821 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008822 connection failures. Of course, it requires having "retries" set to a nonzero
8823 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01008824
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008825 If this option has been enabled in a "defaults" section, it can be disabled
8826 in a specific instance by prepending the "no" keyword before it.
8827
Christopher Faulet87f1f3d2019-07-18 14:51:20 +02008828 See also : "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008829
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008830
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02008831option redis-check
8832 Use redis health checks for server testing
8833 May be used in sections : defaults | frontend | listen | backend
8834 yes | no | yes | yes
8835 Arguments : none
8836
8837 It is possible to test that the server correctly talks REDIS protocol instead
8838 of just testing that it accepts the TCP connection. When this option is set,
8839 a PING redis command is sent to the server, and the response is analyzed to
8840 find the "+PONG" response message.
8841
8842 Example :
8843 option redis-check
8844
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03008845 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02008846
8847
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008848option smtpchk
8849option smtpchk <hello> <domain>
8850 Use SMTP health checks for server testing
8851 May be used in sections : defaults | frontend | listen | backend
8852 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01008853 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008854 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02008855 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008856 values will be turned into the default command ("HELO").
8857
8858 <domain> is the domain name to present to the server. It may only be
8859 specified (and is mandatory) if the hello command has been
8860 specified. By default, "localhost" is used.
8861
8862 When "option smtpchk" is set, the health checks will consist in TCP
8863 connections followed by an SMTP command. By default, this command is
8864 "HELO localhost". The server's return code is analyzed and only return codes
8865 starting with a "2" will be considered as valid. All other responses,
8866 including a lack of response will constitute an error and will indicate a
8867 dead server.
8868
8869 This test is meant to be used with SMTP servers or relays. Depending on the
8870 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01008871 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008872 25 is often easier than adjusting the configuration.
8873
8874 Most often, an incoming SMTP server needs to see the client's IP address for
8875 various purposes, including spam filtering, anti-spoofing and logging. When
8876 possible, it is often wise to masquerade the client's IP address when
8877 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02008878 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01008879
8880 Example :
8881 option smtpchk HELO mydomain.org
8882
8883 See also : "option httpchk", "source"
8884
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008885
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02008886option socket-stats
8887no option socket-stats
8888
8889 Enable or disable collecting & providing separate statistics for each socket.
8890 May be used in sections : defaults | frontend | listen | backend
8891 yes | yes | yes | no
8892
8893 Arguments : none
8894
8895
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008896option splice-auto
8897no option splice-auto
8898 Enable or disable automatic kernel acceleration on sockets in both directions
8899 May be used in sections : defaults | frontend | listen | backend
8900 yes | yes | yes | yes
8901 Arguments : none
8902
8903 When this option is enabled either on a frontend or on a backend, haproxy
8904 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01008905 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008906 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008907 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008908 are not much aggressive in order to limit excessive use of splicing. This
8909 option requires splicing to be enabled at compile time, and may be globally
8910 disabled with the global option "nosplice". Since splice uses pipes, using it
8911 requires that there are enough spare pipes.
8912
8913 Important note: kernel-based TCP splicing is a Linux-specific feature which
8914 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
8915 transfer data between sockets without copying these data to user-space, thus
8916 providing noticeable performance gains and CPU cycles savings. Since many
8917 early implementations are buggy, corrupt data and/or are inefficient, this
8918 feature is not enabled by default, and it should be used with extreme care.
8919 While it is not possible to detect the correctness of an implementation,
8920 2.6.29 is the first version offering a properly working implementation. In
8921 case of doubt, splicing may be globally disabled using the global "nosplice"
8922 keyword.
8923
8924 Example :
8925 option splice-auto
8926
8927 If this option has been enabled in a "defaults" section, it can be disabled
8928 in a specific instance by prepending the "no" keyword before it.
8929
8930 See also : "option splice-request", "option splice-response", and global
8931 options "nosplice" and "maxpipes"
8932
8933
8934option splice-request
8935no option splice-request
8936 Enable or disable automatic kernel acceleration on sockets for requests
8937 May be used in sections : defaults | frontend | listen | backend
8938 yes | yes | yes | yes
8939 Arguments : none
8940
8941 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008942 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008943 the client to the server. It might still use the recv/send scheme if there
8944 are no spare pipes left. This option requires splicing to be enabled at
8945 compile time, and may be globally disabled with the global option "nosplice".
8946 Since splice uses pipes, using it requires that there are enough spare pipes.
8947
8948 Important note: see "option splice-auto" for usage limitations.
8949
8950 Example :
8951 option splice-request
8952
8953 If this option has been enabled in a "defaults" section, it can be disabled
8954 in a specific instance by prepending the "no" keyword before it.
8955
8956 See also : "option splice-auto", "option splice-response", and global options
8957 "nosplice" and "maxpipes"
8958
8959
8960option splice-response
8961no option splice-response
8962 Enable or disable automatic kernel acceleration on sockets for responses
8963 May be used in sections : defaults | frontend | listen | backend
8964 yes | yes | yes | yes
8965 Arguments : none
8966
8967 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008968 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01008969 the server to the client. It might still use the recv/send scheme if there
8970 are no spare pipes left. This option requires splicing to be enabled at
8971 compile time, and may be globally disabled with the global option "nosplice".
8972 Since splice uses pipes, using it requires that there are enough spare pipes.
8973
8974 Important note: see "option splice-auto" for usage limitations.
8975
8976 Example :
8977 option splice-response
8978
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 splice-auto", "option splice-request", and global options
8983 "nosplice" and "maxpipes"
8984
8985
Christopher Fauletba7bc162016-11-07 21:07:38 +01008986option spop-check
8987 Use SPOP health checks for server testing
8988 May be used in sections : defaults | frontend | listen | backend
8989 no | no | no | yes
8990 Arguments : none
8991
8992 It is possible to test that the server correctly talks SPOP protocol instead
8993 of just testing that it accepts the TCP connection. When this option is set,
8994 a HELLO handshake is performed between HAProxy and the server, and the
8995 response is analyzed to check no error is reported.
8996
8997 Example :
8998 option spop-check
8999
9000 See also : "option httpchk"
9001
9002
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009003option srvtcpka
9004no option srvtcpka
9005 Enable or disable the sending of TCP keepalive packets on the server side
9006 May be used in sections : defaults | frontend | listen | backend
9007 yes | no | yes | yes
9008 Arguments : none
9009
9010 When there is a firewall or any session-aware component between a client and
9011 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01009012 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009013 components decides to expire a session which has remained idle for too long.
9014
9015 Enabling socket-level TCP keep-alives makes the system regularly send packets
9016 to the other end of the connection, leaving it active. The delay between
9017 keep-alive probes is controlled by the system only and depends both on the
9018 operating system and its tuning parameters.
9019
9020 It is important to understand that keep-alive packets are neither emitted nor
9021 received at the application level. It is only the network stacks which sees
9022 them. For this reason, even if one side of the proxy already uses keep-alives
9023 to maintain its connection alive, those keep-alive packets will not be
9024 forwarded to the other side of the proxy.
9025
9026 Please note that this has nothing to do with HTTP keep-alive.
9027
9028 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
9029 server side of a connection, which should help when session expirations are
9030 noticed between HAProxy and a server.
9031
9032 If this option has been enabled in a "defaults" section, it can be disabled
9033 in a specific instance by prepending the "no" keyword before it.
9034
9035 See also : "option clitcpka", "option tcpka"
9036
9037
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009038option ssl-hello-chk
9039 Use SSLv3 client hello health checks for server testing
9040 May be used in sections : defaults | frontend | listen | backend
9041 yes | no | yes | yes
9042 Arguments : none
9043
9044 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
9045 possible to test that the server correctly talks SSL instead of just testing
9046 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
9047 SSLv3 client hello messages are sent once the connection is established to
9048 the server, and the response is analyzed to find an SSL server hello message.
9049 The server is considered valid only when the response contains this server
9050 hello message.
9051
9052 All servers tested till there correctly reply to SSLv3 client hello messages,
9053 and most servers tested do not even log the requests containing only hello
9054 messages, which is appreciable.
9055
Willy Tarreau763a95b2012-10-04 23:15:39 +02009056 Note that this check works even when SSL support was not built into haproxy
9057 because it forges the SSL message. When SSL support is available, it is best
9058 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009059
Willy Tarreau763a95b2012-10-04 23:15:39 +02009060 See also: "option httpchk", "check-ssl"
9061
Willy Tarreaua453bdd2008-01-08 19:50:52 +01009062
Willy Tarreaued179852013-12-16 01:07:00 +01009063option tcp-check
9064 Perform health checks using tcp-check send/expect sequences
9065 May be used in sections: defaults | frontend | listen | backend
9066 yes | no | yes | yes
9067
9068 This health check method is intended to be combined with "tcp-check" command
9069 lists in order to support send/expect types of health check sequences.
9070
9071 TCP checks currently support 4 modes of operations :
9072 - no "tcp-check" directive : the health check only consists in a connection
9073 attempt, which remains the default mode.
9074
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009075 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01009076 used to send a string along with a connection opening. With some
9077 protocols, it helps sending a "QUIT" message for example that prevents
9078 the server from logging a connection error for each health check. The
9079 check result will still be based on the ability to open the connection
9080 only.
9081
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009082 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01009083 The connection is opened and haproxy waits for the server to present some
9084 contents which must validate some rules. The check result will be based
9085 on the matching between the contents and the rules. This is suited for
9086 POP, IMAP, SMTP, FTP, SSH, TELNET.
9087
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009088 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009089 used to test a hello-type protocol. HAProxy sends a message, the server
9090 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009091 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01009092 suited for protocols which require a binding or a request/response model.
9093 LDAP, MySQL, Redis and SSL are example of such protocols, though they
9094 already all have their dedicated checks with a deeper understanding of
9095 the respective protocols.
9096 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01009097 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01009098
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009099 A fifth mode can be used to insert comments in different steps of the script.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009100
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009101 For each tcp-check rule you create, you can add a "comment" directive,
9102 followed by a string. This string will be reported in the log and stderr in
9103 debug mode. It is useful to make user-friendly error reporting. The
9104 "comment" is of course optional.
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009105
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009106 During the execution of a health check, a variable scope is made available to
9107 store data samples, using the "tcp-check set-var" operation. Freeing those
9108 variable is possible using "tcp-check unset-var".
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +01009109
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009110
Willy Tarreaued179852013-12-16 01:07:00 +01009111 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01009112 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01009113 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009114 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01009115
Davor Ocelice9ed2812017-12-25 17:49:28 +01009116 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01009117 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009118 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01009119
9120 # look for the redis master server after ensuring it speaks well
9121 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009122 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01009123 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009124 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01009125 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009126 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009127 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01009128 tcp-check send info\ replication\r\n
9129 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009130 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01009131 tcp-check send QUIT\r\n
9132 tcp-check expect string +OK
9133
Davor Ocelice9ed2812017-12-25 17:49:28 +01009134 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01009135 (send many headers before analyzing)
9136 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009137 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01009138 tcp-check send HEAD\ /\ HTTP/1.1\r\n
9139 tcp-check send Host:\ www.mydomain.com\r\n
9140 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
9141 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02009142 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01009143
9144
Christopher Fauletc52ea4d2020-04-23 15:43:35 +02009145 See also : "tcp-check connect", "tcp-check expect" and "tcp-check send".
Willy Tarreaued179852013-12-16 01:07:00 +01009146
9147
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009148option tcp-smart-accept
9149no option tcp-smart-accept
9150 Enable or disable the saving of one ACK packet during the accept sequence
9151 May be used in sections : defaults | frontend | listen | backend
9152 yes | yes | yes | no
9153 Arguments : none
9154
9155 When an HTTP connection request comes in, the system acknowledges it on
9156 behalf of HAProxy, then the client immediately sends its request, and the
9157 system acknowledges it too while it is notifying HAProxy about the new
9158 connection. HAProxy then reads the request and responds. This means that we
9159 have one TCP ACK sent by the system for nothing, because the request could
9160 very well be acknowledged by HAProxy when it sends its response.
9161
9162 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
9163 sending this useless ACK on platforms which support it (currently at least
9164 Linux). It must not cause any problem, because the system will send it anyway
9165 after 40 ms if the response takes more time than expected to come.
9166
9167 During complex network debugging sessions, it may be desirable to disable
9168 this optimization because delayed ACKs can make troubleshooting more complex
9169 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01009170 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009171
9172 It is also possible to force it for non-HTTP proxies by simply specifying
9173 "option tcp-smart-accept". For instance, it can make sense with some services
9174 such as SMTP where the server speaks first.
9175
9176 It is recommended to avoid forcing this option in a defaults section. In case
9177 of doubt, consider setting it back to automatic values by prepending the
9178 "default" keyword before it, or disabling it using the "no" keyword.
9179
Willy Tarreaud88edf22009-06-14 15:48:17 +02009180 See also : "option tcp-smart-connect"
9181
9182
9183option tcp-smart-connect
9184no option tcp-smart-connect
9185 Enable or disable the saving of one ACK packet during the connect sequence
9186 May be used in sections : defaults | frontend | listen | backend
9187 yes | no | yes | yes
9188 Arguments : none
9189
9190 On certain systems (at least Linux), HAProxy can ask the kernel not to
9191 immediately send an empty ACK upon a connection request, but to directly
9192 send the buffer request instead. This saves one packet on the network and
9193 thus boosts performance. It can also be useful for some servers, because they
9194 immediately get the request along with the incoming connection.
9195
9196 This feature is enabled when "option tcp-smart-connect" is set in a backend.
9197 It is not enabled by default because it makes network troubleshooting more
9198 complex.
9199
9200 It only makes sense to enable it with protocols where the client speaks first
9201 such as HTTP. In other situations, if there is no data to send in place of
9202 the ACK, a normal ACK is sent.
9203
9204 If this option has been enabled in a "defaults" section, it can be disabled
9205 in a specific instance by prepending the "no" keyword before it.
9206
9207 See also : "option tcp-smart-accept"
9208
Willy Tarreau9ea05a72009-06-14 12:07:01 +02009209
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009210option tcpka
9211 Enable or disable the sending of TCP keepalive packets on both sides
9212 May be used in sections : defaults | frontend | listen | backend
9213 yes | yes | yes | yes
9214 Arguments : none
9215
9216 When there is a firewall or any session-aware component between a client and
9217 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01009218 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009219 components decides to expire a session which has remained idle for too long.
9220
9221 Enabling socket-level TCP keep-alives makes the system regularly send packets
9222 to the other end of the connection, leaving it active. The delay between
9223 keep-alive probes is controlled by the system only and depends both on the
9224 operating system and its tuning parameters.
9225
9226 It is important to understand that keep-alive packets are neither emitted nor
9227 received at the application level. It is only the network stacks which sees
9228 them. For this reason, even if one side of the proxy already uses keep-alives
9229 to maintain its connection alive, those keep-alive packets will not be
9230 forwarded to the other side of the proxy.
9231
9232 Please note that this has nothing to do with HTTP keep-alive.
9233
9234 Using option "tcpka" enables the emission of TCP keep-alive probes on both
9235 the client and server sides of a connection. Note that this is meaningful
9236 only in "defaults" or "listen" sections. If this option is used in a
9237 frontend, only the client side will get keep-alives, and if this option is
9238 used in a backend, only the server side will get keep-alives. For this
9239 reason, it is strongly recommended to explicitly use "option clitcpka" and
9240 "option srvtcpka" when the configuration is split between frontends and
9241 backends.
9242
9243 See also : "option clitcpka", "option srvtcpka"
9244
Willy Tarreau844e3c52008-01-11 16:28:18 +01009245
9246option tcplog
9247 Enable advanced logging of TCP connections with session state and timers
9248 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01009249 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01009250 Arguments : none
9251
9252 By default, the log output format is very poor, as it only contains the
9253 source and destination addresses, and the instance name. By specifying
9254 "option tcplog", each log line turns into a much richer format including, but
9255 not limited to, the connection timers, the session status, the connections
9256 numbers, the frontend, backend and server name, and of course the source
9257 address and ports. This option is useful for pure TCP proxies in order to
9258 find which of the client or server disconnects or times out. For normal HTTP
9259 proxies, it's better to use "option httplog" which is even more complete.
9260
Guillaume de Lafond29f45602017-03-31 19:52:15 +02009261 "option tcplog" overrides any previous "log-format" directive.
9262
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009263 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009264
9265
Willy Tarreau844e3c52008-01-11 16:28:18 +01009266option transparent
9267no option transparent
9268 Enable client-side transparent proxying
9269 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01009270 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01009271 Arguments : none
9272
9273 This option was introduced in order to provide layer 7 persistence to layer 3
9274 load balancers. The idea is to use the OS's ability to redirect an incoming
9275 connection for a remote address to a local process (here HAProxy), and let
9276 this process know what address was initially requested. When this option is
9277 used, sessions without cookies will be forwarded to the original destination
9278 IP address of the incoming request (which should match that of another
9279 equipment), while requests with cookies will still be forwarded to the
9280 appropriate server.
9281
9282 Note that contrary to a common belief, this option does NOT make HAProxy
9283 present the client's IP to the server when establishing the connection.
9284
Willy Tarreaua1146052011-03-01 09:51:54 +01009285 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009286 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01009287
Willy Tarreaubf1f8162007-12-28 17:42:56 +01009288
Simon Horman98637e52014-06-20 12:30:16 +09009289external-check command <command>
9290 Executable to run when performing an external-check
9291 May be used in sections : defaults | frontend | listen | backend
9292 yes | no | yes | yes
9293
9294 Arguments :
9295 <command> is the external command to run
9296
Simon Horman98637e52014-06-20 12:30:16 +09009297 The arguments passed to the to the command are:
9298
Cyril Bonté777be862014-12-02 21:21:35 +01009299 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09009300
Cyril Bonté777be862014-12-02 21:21:35 +01009301 The <proxy_address> and <proxy_port> are derived from the first listener
9302 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
9303 listener the proxy_address will be the path of the socket and the
9304 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
9305 possible to determine a listener, and both <proxy_address> and <proxy_port>
9306 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09009307
Cyril Bonté72cda2a2014-12-27 22:28:39 +01009308 Some values are also provided through environment variables.
9309
9310 Environment variables :
9311 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
9312 applicable, for example in a "backend" section).
9313
9314 HAPROXY_PROXY_ID The backend id.
9315
9316 HAPROXY_PROXY_NAME The backend name.
9317
9318 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
9319 applicable, for example in a "backend" section or
9320 for a UNIX socket).
9321
9322 HAPROXY_SERVER_ADDR The server address.
9323
9324 HAPROXY_SERVER_CURCONN The current number of connections on the server.
9325
9326 HAPROXY_SERVER_ID The server id.
9327
9328 HAPROXY_SERVER_MAXCONN The server max connections.
9329
9330 HAPROXY_SERVER_NAME The server name.
9331
9332 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
9333 socket).
9334
9335 PATH The PATH environment variable used when executing
9336 the command may be set using "external-check path".
9337
William Lallemand4d03e432019-06-14 15:35:37 +02009338 See also "2.3. Environment variables" for other variables.
9339
Simon Horman98637e52014-06-20 12:30:16 +09009340 If the command executed and exits with a zero status then the check is
9341 considered to have passed, otherwise the check is considered to have
9342 failed.
9343
9344 Example :
9345 external-check command /bin/true
9346
9347 See also : "external-check", "option external-check", "external-check path"
9348
9349
9350external-check path <path>
9351 The value of the PATH environment variable used when running an external-check
9352 May be used in sections : defaults | frontend | listen | backend
9353 yes | no | yes | yes
9354
9355 Arguments :
9356 <path> is the path used when executing external command to run
9357
9358 The default path is "".
9359
9360 Example :
9361 external-check path "/usr/bin:/bin"
9362
9363 See also : "external-check", "option external-check",
9364 "external-check command"
9365
9366
Emeric Brun647caf12009-06-30 17:57:00 +02009367persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02009368persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02009369 Enable RDP cookie-based persistence
9370 May be used in sections : defaults | frontend | listen | backend
9371 yes | no | yes | yes
9372 Arguments :
9373 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02009374 default cookie name "msts" will be used. There currently is no
9375 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02009376
9377 This statement enables persistence based on an RDP cookie. The RDP cookie
9378 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01009379 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02009380 and if an RDP cookie is found, it is decoded. If it matches a known server
9381 which is still UP (or if "option persist" is set), then the connection is
9382 forwarded to this server.
9383
9384 Note that this only makes sense in a TCP backend, but for this to work, the
9385 frontend must have waited long enough to ensure that an RDP cookie is present
9386 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009387 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02009388 a single "listen" section.
9389
Willy Tarreau61e28f22010-05-16 22:31:05 +02009390 Also, it is important to understand that the terminal server will emit this
9391 RDP cookie only if it is configured for "token redirection mode", which means
9392 that the "IP address redirection" option is disabled.
9393
Emeric Brun647caf12009-06-30 17:57:00 +02009394 Example :
9395 listen tse-farm
9396 bind :3389
9397 # wait up to 5s for an RDP cookie in the request
9398 tcp-request inspect-delay 5s
9399 tcp-request content accept if RDP_COOKIE
9400 # apply RDP cookie persistence
9401 persist rdp-cookie
9402 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02009403 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02009404 balance rdp-cookie
9405 server srv1 1.1.1.1:3389
9406 server srv2 1.1.1.2:3389
9407
Simon Hormanab814e02011-06-24 14:50:20 +09009408 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
9409 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02009410
9411
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009412rate-limit sessions <rate>
9413 Set a limit on the number of new sessions accepted per second on a frontend
9414 May be used in sections : defaults | frontend | listen | backend
9415 yes | yes | yes | no
9416 Arguments :
9417 <rate> The <rate> parameter is an integer designating the maximum number
9418 of new sessions per second to accept on the frontend.
9419
9420 When the frontend reaches the specified number of new sessions per second, it
9421 stops accepting new connections until the rate drops below the limit again.
9422 During this time, the pending sessions will be kept in the socket's backlog
9423 (in system buffers) and haproxy will not even be aware that sessions are
9424 pending. When applying very low limit on a highly loaded service, it may make
9425 sense to increase the socket's backlog using the "backlog" keyword.
9426
9427 This feature is particularly efficient at blocking connection-based attacks
9428 or service abuse on fragile servers. Since the session rate is measured every
9429 millisecond, it is extremely accurate. Also, the limit applies immediately,
9430 no delay is needed at all to detect the threshold.
9431
9432 Example : limit the connection rate on SMTP to 10 per second max
9433 listen smtp
9434 mode tcp
9435 bind :25
9436 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02009437 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009438
Willy Tarreaua17c2d92011-07-25 08:16:20 +02009439 Note : when the maximum rate is reached, the frontend's status is not changed
9440 but its sockets appear as "WAITING" in the statistics if the
9441 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01009442
9443 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
9444
9445
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009446redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
9447redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
9448redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009449 Return an HTTP redirection if/unless a condition is matched
9450 May be used in sections : defaults | frontend | listen | backend
9451 no | yes | yes | yes
9452
9453 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01009454 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009455
Willy Tarreau0140f252008-11-19 21:07:09 +01009456 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009457 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009458 the HTTP "Location" header. When used in an "http-request" rule,
9459 <loc> value follows the log-format rules and can include some
9460 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009461
9462 <pfx> With "redirect prefix", the "Location" header is built from the
9463 concatenation of <pfx> and the complete URI path, including the
9464 query string, unless the "drop-query" option is specified (see
9465 below). As a special case, if <pfx> equals exactly "/", then
9466 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009467 redirect to the same URL (for instance, to insert a cookie). When
9468 used in an "http-request" rule, <pfx> value follows the log-format
9469 rules and can include some dynamic values (see Custom Log Format
9470 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009471
9472 <sch> With "redirect scheme", then the "Location" header is built by
9473 concatenating <sch> with "://" then the first occurrence of the
9474 "Host" header, and then the URI path, including the query string
9475 unless the "drop-query" option is specified (see below). If no
9476 path is found or if the path is "*", then "/" is used instead. If
9477 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009478 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009479 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009480 HTTPS. When used in an "http-request" rule, <sch> value follows
9481 the log-format rules and can include some dynamic values (see
9482 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01009483
9484 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009485 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
9486 with 302 used by default if no code is specified. 301 means
9487 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02009488 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01009489 cache the redirection. 303 is equivalent to 302 except that the
9490 browser will fetch the location with a GET method. 307 is just
9491 like 302 but makes it clear that the same method must be reused.
9492 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01009493
9494 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009495 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01009496
9497 - "drop-query"
9498 When this keyword is used in a prefix-based redirection, then the
9499 location will be set without any possible query-string, which is useful
9500 for directing users to a non-secure page for instance. It has no effect
9501 with a location-type redirect.
9502
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009503 - "append-slash"
9504 This keyword may be used in conjunction with "drop-query" to redirect
9505 users who use a URL not ending with a '/' to the same one with the '/'.
9506 It can be useful to ensure that search engines will only see one URL.
9507 For this, a return code 301 is preferred.
9508
Willy Tarreau0140f252008-11-19 21:07:09 +01009509 - "set-cookie NAME[=value]"
9510 A "Set-Cookie" header will be added with NAME (and optionally "=value")
9511 to the response. This is sometimes used to indicate that a user has
9512 been seen, for instance to protect against some types of DoS. No other
9513 cookie option is added, so the cookie will be a session cookie. Note
9514 that for a browser, a sole cookie name without an equal sign is
9515 different from a cookie with an equal sign.
9516
9517 - "clear-cookie NAME[=]"
9518 A "Set-Cookie" header will be added with NAME (and optionally "="), but
9519 with the "Max-Age" attribute set to zero. This will tell the browser to
9520 delete this cookie. It is useful for instance on logout pages. It is
9521 important to note that clearing the cookie "NAME" will not remove a
9522 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
9523 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009524
9525 Example: move the login URL only to HTTPS.
9526 acl clear dst_port 80
9527 acl secure dst_port 8080
9528 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01009529 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01009530 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01009531 acl cookie_set hdr_sub(cookie) SEEN=1
9532
9533 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01009534 redirect prefix https://mysite.com if login_page !secure
9535 redirect prefix http://mysite.com drop-query if login_page !uid_given
9536 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01009537 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009538
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01009539 Example: send redirects for request for articles without a '/'.
9540 acl missing_slash path_reg ^/article/[^/]*$
9541 redirect code 301 prefix / drop-query append-slash if missing_slash
9542
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009543 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01009544 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02009545
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009546 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02009547 http-request redirect code 301 location \
9548 http://www.%[hdr(host)]%[capture.req.uri] \
9549 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01009550
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009551 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02009552
Willy Tarreau303c0352008-01-17 19:01:39 +01009553
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009554retries <value>
9555 Set the number of retries to perform on a server after a connection failure
9556 May be used in sections: defaults | frontend | listen | backend
9557 yes | no | yes | yes
9558 Arguments :
9559 <value> is the number of times a connection attempt should be retried on
9560 a server when a connection either is refused or times out. The
9561 default value is 3.
9562
9563 It is important to understand that this value applies to the number of
9564 connection attempts, not full requests. When a connection has effectively
9565 been established to a server, there will be no more retry.
9566
9567 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07009568 a turn-around timer of min("timeout connect", one second) is applied before
9569 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02009570
9571 When "option redispatch" is set, the last retry may be performed on another
9572 server even if a cookie references a different server.
9573
9574 See also : "option redispatch"
9575
9576
Olivier Houcharda254a372019-04-05 15:30:12 +02009577retry-on [list of keywords]
Jerome Magnin5ce3c142020-05-13 20:09:57 +02009578 Specify when to attempt to automatically retry a failed request.
9579 This setting is only valid when "mode" is set to http and is silently ignored
9580 otherwise.
Olivier Houcharda254a372019-04-05 15:30:12 +02009581 May be used in sections: defaults | frontend | listen | backend
9582 yes | no | yes | yes
9583 Arguments :
9584 <keywords> is a list of keywords or HTTP status codes, each representing a
9585 type of failure event on which an attempt to retry the request
9586 is desired. Please read the notes at the bottom before changing
9587 this setting. The following keywords are supported :
9588
9589 none never retry
9590
9591 conn-failure retry when the connection or the SSL handshake failed
9592 and the request could not be sent. This is the default.
9593
9594 empty-response retry when the server connection was closed after part
9595 of the request was sent, and nothing was received from
9596 the server. This type of failure may be caused by the
9597 request timeout on the server side, poor network
9598 condition, or a server crash or restart while
9599 processing the request.
9600
Olivier Houcharde3249a92019-05-03 23:01:47 +02009601 junk-response retry when the server returned something not looking
9602 like a complete HTTP response. This includes partial
9603 responses headers as well as non-HTTP contents. It
9604 usually is a bad idea to retry on such events, which
9605 may be caused a configuration issue (wrong server port)
9606 or by the request being harmful to the server (buffer
9607 overflow attack for example).
9608
Olivier Houcharda254a372019-04-05 15:30:12 +02009609 response-timeout the server timeout stroke while waiting for the server
9610 to respond to the request. This may be caused by poor
9611 network condition, the reuse of an idle connection
9612 which has expired on the path, or by the request being
9613 extremely expensive to process. It generally is a bad
9614 idea to retry on such events on servers dealing with
9615 heavy database processing (full scans, etc) as it may
9616 amplify denial of service attacks.
9617
Olivier Houchard865d8392019-05-03 22:46:27 +02009618 0rtt-rejected retry requests which were sent over early data and were
9619 rejected by the server. These requests are generally
9620 considered to be safe to retry.
9621
Julien Pivotto2de240a2020-11-12 11:14:05 +01009622 <status> any HTTP status code among "401" (Unauthorized), "403"
9623 (Forbidden), "404" (Not Found), "408" (Request Timeout),
9624 "425" (Too Early), "500" (Server Error), "501" (Not
9625 Implemented), "502" (Bad Gateway), "503" (Service
9626 Unavailable), "504" (Gateway Timeout).
Olivier Houcharda254a372019-04-05 15:30:12 +02009627
Olivier Houchardddf0e032019-05-10 18:05:40 +02009628 all-retryable-errors
9629 retry request for any error that are considered
9630 retryable. This currently activates "conn-failure",
9631 "empty-response", "junk-response", "response-timeout",
9632 "0rtt-rejected", "500", "502", "503", and "504".
9633
Olivier Houcharda254a372019-04-05 15:30:12 +02009634 Using this directive replaces any previous settings with the new ones; it is
9635 not cumulative.
9636
9637 Please note that using anything other than "none" and "conn-failure" requires
9638 to allocate a buffer and copy the whole request into it, so it has memory and
9639 performance impacts. Requests not fitting in a single buffer will never be
9640 retried (see the global tune.bufsize setting).
9641
9642 You have to make sure the application has a replay protection mechanism built
9643 in such as a unique transaction IDs passed in requests, or that replaying the
9644 same request has no consequence, or it is very dangerous to use any retry-on
9645 value beside "conn-failure" and "none". Static file servers and caches are
9646 generally considered safe against any type of retry. Using a status code can
9647 be useful to quickly leave a server showing an abnormal behavior (out of
9648 memory, file system issues, etc), but in this case it may be a good idea to
9649 immediately redispatch the connection to another server (please see "option
9650 redispatch" for this). Last, it is important to understand that most causes
9651 of failures are the requests themselves and that retrying a request causing a
9652 server to misbehave will often make the situation even worse for this server,
9653 or for the whole service in case of redispatch.
9654
9655 Unless you know exactly how the application deals with replayed requests, you
9656 should not use this directive.
9657
9658 The default is "conn-failure".
9659
9660 See also: "retries", "option redispatch", "tune.bufsize"
9661
David du Colombier486df472011-03-17 10:40:26 +01009662server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009663 Declare a server in a backend
9664 May be used in sections : defaults | frontend | listen | backend
9665 no | no | yes | yes
9666 Arguments :
9667 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009668 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05009669 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009670
David du Colombier486df472011-03-17 10:40:26 +01009671 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
9672 resolvable hostname is supported, but this name will be resolved
9673 during start-up. Address "0.0.0.0" or "*" has a special meaning.
9674 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02009675 address as the one from the client connection. This is useful in
9676 transparent proxy architectures where the client's connection is
9677 intercepted and haproxy must forward to the original destination
9678 address. This is more or less what the "transparent" keyword does
9679 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01009680 to report statistics. Optionally, an address family prefix may be
9681 used before the address to force the family regardless of the
9682 address format, which can be useful to specify a path to a unix
9683 socket with no slash ('/'). Currently supported prefixes are :
9684 - 'ipv4@' -> address is always IPv4
9685 - 'ipv6@' -> address is always IPv6
9686 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02009687 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02009688 - 'sockpair@' -> address is the FD of a connected unix
9689 socket or of a socketpair. During a connection, the
9690 backend creates a pair of connected sockets, and passes
9691 one of them over the FD. The bind part will use the
9692 received socket as the client FD. Should be used
9693 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02009694 You may want to reference some environment variables in the
9695 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01009696 variables. The "init-addr" setting can be used to modify the way
9697 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009698
Willy Tarreaub6205fd2012-09-24 12:27:33 +02009699 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009700 be sent to this port. If unset, the same port the client
9701 connected to will be used. The port may also be prefixed by a "+"
9702 or a "-". In this case, the server's port will be determined by
9703 adding this value to the client's port.
9704
9705 <param*> is a list of parameters for this server. The "server" keywords
9706 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009707 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009708
9709 Examples :
9710 server first 10.1.1.1:1080 cookie first check inter 1000
9711 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01009712 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02009713 server backup "${SRV_BACKUP}:1080" backup
9714 server www1_dc1 "${LAN_DC1}.101:80"
9715 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009716
Willy Tarreau55dcaf62015-09-27 15:03:15 +02009717 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
9718 sun_path length is used for the address length. Some other programs
9719 such as socat use the string length only by default. Pass the option
9720 ",unix-tightsocklen=0" to any abstract socket definition in socat to
9721 make it compatible with HAProxy's.
9722
Mark Lamourinec2247f02012-01-04 13:02:01 -05009723 See also: "default-server", "http-send-name-header" and section 5 about
9724 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009725
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009726server-state-file-name [<file>]
9727 Set the server state file to read, load and apply to servers available in
9728 this backend. It only applies when the directive "load-server-state-from-file"
9729 is set to "local". When <file> is not provided or if this directive is not
9730 set, then backend name is used. If <file> starts with a slash '/', then it is
9731 considered as an absolute path. Otherwise, <file> is concatenated to the
9732 global directive "server-state-file-base".
9733
9734 Example: the minimal configuration below would make HAProxy look for the
9735 state server file '/etc/haproxy/states/bk':
9736
9737 global
9738 server-state-file-base /etc/haproxy/states
9739
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +01009740 backend bk
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02009741 load-server-state-from-file
9742
9743 See also: "server-state-file-base", "load-server-state-from-file", and
9744 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009745
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02009746server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
9747 Set a template to initialize servers with shared parameters.
9748 The names of these servers are built from <prefix> and <num | range> parameters.
9749 May be used in sections : defaults | frontend | listen | backend
9750 no | no | yes | yes
9751
9752 Arguments:
9753 <prefix> A prefix for the server names to be built.
9754
9755 <num | range>
9756 If <num> is provided, this template initializes <num> servers
9757 with 1 up to <num> as server name suffixes. A range of numbers
9758 <num_low>-<num_high> may also be used to use <num_low> up to
9759 <num_high> as server name suffixes.
9760
9761 <fqdn> A FQDN for all the servers this template initializes.
9762
9763 <port> Same meaning as "server" <port> argument (see "server" keyword).
9764
9765 <params*>
9766 Remaining server parameters among all those supported by "server"
9767 keyword.
9768
9769 Examples:
9770 # Initializes 3 servers with srv1, srv2 and srv3 as names,
9771 # google.com as FQDN, and health-check enabled.
9772 server-template srv 1-3 google.com:80 check
9773
9774 # or
9775 server-template srv 3 google.com:80 check
9776
9777 # would be equivalent to:
9778 server srv1 google.com:80 check
9779 server srv2 google.com:80 check
9780 server srv3 google.com:80 check
9781
9782
9783
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009784source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02009785source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01009786source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009787 Set the source address for outgoing connections
9788 May be used in sections : defaults | frontend | listen | backend
9789 yes | no | yes | yes
9790 Arguments :
9791 <addr> is the IPv4 address HAProxy will bind to before connecting to a
9792 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01009793
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009794 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01009795 the most appropriate address to reach its destination. Optionally
9796 an address family prefix may be used before the address to force
9797 the family regardless of the address format, which can be useful
9798 to specify a path to a unix socket with no slash ('/'). Currently
9799 supported prefixes are :
9800 - 'ipv4@' -> address is always IPv4
9801 - 'ipv6@' -> address is always IPv6
9802 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02009803 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02009804 You may want to reference some environment variables in the
9805 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009806
9807 <port> is an optional port. It is normally not needed but may be useful
9808 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02009809 the system will select a free port. Note that port ranges are not
9810 supported in the backend. If you want to force port ranges, you
9811 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009812
9813 <addr2> is the IP address to present to the server when connections are
9814 forwarded in full transparent proxy mode. This is currently only
9815 supported on some patched Linux kernels. When this address is
9816 specified, clients connecting to the server will be presented
9817 with this address, while health checks will still use the address
9818 <addr>.
9819
9820 <port2> is the optional port to present to the server when connections
9821 are forwarded in full transparent proxy mode (see <addr2> above).
9822 The default value of zero means the system will select a free
9823 port.
9824
Willy Tarreaubce70882009-09-07 11:51:47 +02009825 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
9826 This is the name of a comma-separated header list which can
9827 contain multiple IP addresses. By default, the last occurrence is
9828 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01009829 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02009830 by previous proxy, typically Stunnel. In order to use another
9831 occurrence from the last one, please see the <occ> parameter
9832 below. When the header (or occurrence) is not found, no binding
9833 is performed so that the proxy's default IP address is used. Also
9834 keep in mind that the header name is case insensitive, as for any
9835 HTTP header.
9836
9837 <occ> is the occurrence number of a value to be used in a multi-value
9838 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009839 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02009840 address. Positive values indicate a position from the first
9841 occurrence, 1 being the first one. Negative values indicate
9842 positions relative to the last one, -1 being the last one. This
9843 is helpful for situations where an X-Forwarded-For header is set
9844 at the entry point of an infrastructure and must be used several
9845 proxy layers away. When this value is not specified, -1 is
9846 assumed. Passing a zero here disables the feature.
9847
Willy Tarreaud53f96b2009-02-04 18:46:54 +01009848 <name> is an optional interface name to which to bind to for outgoing
9849 traffic. On systems supporting this features (currently, only
9850 Linux), this allows one to bind all traffic to the server to
9851 this interface even if it is not the one the system would select
9852 based on routing tables. This should be used with extreme care.
9853 Note that using this option requires root privileges.
9854
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009855 The "source" keyword is useful in complex environments where a specific
9856 address only is allowed to connect to the servers. It may be needed when a
9857 private address must be used through a public gateway for instance, and it is
9858 known that the system cannot determine the adequate source address by itself.
9859
9860 An extension which is available on certain patched Linux kernels may be used
9861 through the "usesrc" optional keyword. It makes it possible to connect to the
9862 servers with an IP address which does not belong to the system itself. This
9863 is called "full transparent proxy mode". For this to work, the destination
9864 servers have to route their traffic back to this address through the machine
9865 running HAProxy, and IP forwarding must generally be enabled on this machine.
9866
9867 In this "full transparent proxy" mode, it is possible to force a specific IP
9868 address to be presented to the servers. This is not much used in fact. A more
9869 common use is to tell HAProxy to present the client's IP address. For this,
9870 there are two methods :
9871
9872 - present the client's IP and port addresses. This is the most transparent
9873 mode, but it can cause problems when IP connection tracking is enabled on
9874 the machine, because a same connection may be seen twice with different
9875 states. However, this solution presents the huge advantage of not
9876 limiting the system to the 64k outgoing address+port couples, because all
9877 of the client ranges may be used.
9878
9879 - present only the client's IP address and select a spare port. This
9880 solution is still quite elegant but slightly less transparent (downstream
9881 firewalls logs will not match upstream's). It also presents the downside
9882 of limiting the number of concurrent connections to the usual 64k ports.
9883 However, since the upstream and downstream ports are different, local IP
9884 connection tracking on the machine will not be upset by the reuse of the
9885 same session.
9886
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009887 This option sets the default source for all servers in the backend. It may
9888 also be specified in a "defaults" section. Finer source address specification
9889 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009890 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009891
Baptiste Assmann91bd3372015-07-17 21:59:42 +02009892 In order to work, "usesrc" requires root privileges.
9893
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009894 Examples :
9895 backend private
9896 # Connect to the servers using our 192.168.1.200 source address
9897 source 192.168.1.200
9898
9899 backend transparent_ssl1
9900 # Connect to the SSL farm from the client's source address
9901 source 192.168.1.200 usesrc clientip
9902
9903 backend transparent_ssl2
9904 # Connect to the SSL farm from the client's source address and port
9905 # not recommended if IP conntrack is present on the local machine.
9906 source 192.168.1.200 usesrc client
9907
9908 backend transparent_ssl3
9909 # Connect to the SSL farm from the client's source address. It
9910 # is more conntrack-friendly.
9911 source 192.168.1.200 usesrc clientip
9912
9913 backend transparent_smtp
9914 # Connect to the SMTP farm from the client's source address/port
9915 # with Tproxy version 4.
9916 source 0.0.0.0 usesrc clientip
9917
Willy Tarreaubce70882009-09-07 11:51:47 +02009918 backend transparent_http
9919 # Connect to the servers using the client's IP as seen by previous
9920 # proxy.
9921 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
9922
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009923 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009924 the Linux kernel on www.balabit.com, the "bind" keyword.
9925
Willy Tarreau844e3c52008-01-11 16:28:18 +01009926
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009927srvtcpka-cnt <count>
9928 Sets the maximum number of keepalive probes TCP should send before dropping
9929 the connection on the server side.
9930 May be used in sections : defaults | frontend | listen | backend
9931 yes | no | yes | yes
9932 Arguments :
9933 <count> is the maximum number of keepalive probes.
9934
9935 This keyword corresponds to the socket option TCP_KEEPCNT. If this keyword
9936 is not specified, system-wide TCP parameter (tcp_keepalive_probes) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009937 The availability of this setting depends on the operating system. It is
9938 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009939
9940 See also : "option srvtcpka", "srvtcpka-idle", "srvtcpka-intvl".
9941
9942
9943srvtcpka-idle <timeout>
9944 Sets the time the connection needs to remain idle before TCP starts sending
9945 keepalive probes, if enabled the sending of TCP keepalive packets on the
9946 server side.
9947 May be used in sections : defaults | frontend | listen | backend
9948 yes | no | yes | yes
9949 Arguments :
9950 <timeout> is the time the connection needs to remain idle before TCP starts
9951 sending keepalive probes. It is specified in seconds by default,
9952 but can be in any other unit if the number is suffixed by the
9953 unit, as explained at the top of this document.
9954
9955 This keyword corresponds to the socket option TCP_KEEPIDLE. If this keyword
9956 is not specified, system-wide TCP parameter (tcp_keepalive_time) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009957 The availability of this setting depends on the operating system. It is
9958 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009959
9960 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-intvl".
9961
9962
9963srvtcpka-intvl <timeout>
9964 Sets the time between individual keepalive probes on the server side.
9965 May be used in sections : defaults | frontend | listen | backend
9966 yes | no | yes | yes
9967 Arguments :
9968 <timeout> is the time between individual keepalive probes. It is specified
9969 in seconds by default, but can be in any other unit if the number
9970 is suffixed by the unit, as explained at the top of this
9971 document.
9972
9973 This keyword corresponds to the socket option TCP_KEEPINTVL. If this keyword
9974 is not specified, system-wide TCP parameter (tcp_keepalive_intvl) is used.
Willy Tarreau52543212020-07-09 05:58:51 +02009975 The availability of this setting depends on the operating system. It is
9976 known to work on Linux.
MIZUTA Takeshib24bc0d2020-07-09 11:13:20 +09009977
9978 See also : "option srvtcpka", "srvtcpka-cnt", "srvtcpka-idle".
9979
9980
Cyril Bonté66c327d2010-10-12 00:14:37 +02009981stats admin { if | unless } <cond>
9982 Enable statistics admin level if/unless a condition is matched
9983 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009984 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02009985
9986 This statement enables the statistics admin level if/unless a condition is
9987 matched.
9988
9989 The admin level allows to enable/disable servers from the web interface. By
9990 default, statistics page is read-only for security reasons.
9991
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009992 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9993 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009994 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009995
Cyril Bonté23b39d92011-02-10 22:54:44 +01009996 Currently, the POST request is limited to the buffer size minus the reserved
9997 buffer space, which means that if the list of servers is too long, the
9998 request won't be processed. It is recommended to alter few servers at a
9999 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +020010000
10001 Example :
10002 # statistics admin level only for localhost
10003 backend stats_localhost
10004 stats enable
10005 stats admin if LOCALHOST
10006
10007 Example :
10008 # statistics admin level always enabled because of the authentication
10009 backend stats_auth
10010 stats enable
10011 stats auth admin:AdMiN123
10012 stats admin if TRUE
10013
10014 Example :
10015 # statistics admin level depends on the authenticated user
10016 userlist stats-auth
10017 group admin users admin
10018 user admin insecure-password AdMiN123
10019 group readonly users haproxy
10020 user haproxy insecure-password haproxy
10021
10022 backend stats_auth
10023 stats enable
10024 acl AUTH http_auth(stats-auth)
10025 acl AUTH_ADMIN http_auth_group(stats-auth) admin
10026 stats http-request auth unless AUTH
10027 stats admin if AUTH_ADMIN
10028
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010029 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
10030 "bind-process", section 3.4 about userlists and section 7 about
10031 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +020010032
10033
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010034stats auth <user>:<passwd>
10035 Enable statistics with authentication and grant access to an account
10036 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010037 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010038 Arguments :
10039 <user> is a user name to grant access to
10040
10041 <passwd> is the cleartext password associated to this user
10042
10043 This statement enables statistics with default settings, and restricts access
10044 to declared users only. It may be repeated as many times as necessary to
10045 allow as many users as desired. When a user tries to access the statistics
10046 without a valid account, a "401 Forbidden" response will be returned so that
10047 the browser asks the user to provide a valid user and password. The real
10048 which will be returned to the browser is configurable using "stats realm".
10049
10050 Since the authentication method is HTTP Basic Authentication, the passwords
10051 circulate in cleartext on the network. Thus, it was decided that the
10052 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020010053 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010054
10055 It is also possible to reduce the scope of the proxies which appear in the
10056 report using "stats scope".
10057
10058 Though this statement alone is enough to enable statistics reporting, it is
10059 recommended to set all other settings in order to avoid relying on default
10060 unobvious parameters.
10061
10062 Example :
10063 # public access (limited to this backend only)
10064 backend public_www
10065 server srv1 192.168.0.1:80
10066 stats enable
10067 stats hide-version
10068 stats scope .
10069 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010070 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010071 stats auth admin1:AdMiN123
10072 stats auth admin2:AdMiN321
10073
10074 # internal monitoring access (unlimited)
10075 backend private_monitoring
10076 stats enable
10077 stats uri /admin?stats
10078 stats refresh 5s
10079
10080 See also : "stats enable", "stats realm", "stats scope", "stats uri"
10081
10082
10083stats enable
10084 Enable statistics reporting with default settings
10085 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010086 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010087 Arguments : none
10088
10089 This statement enables statistics reporting with default settings defined
10090 at build time. Unless stated otherwise, these settings are used :
10091 - stats uri : /haproxy?stats
10092 - stats realm : "HAProxy Statistics"
10093 - stats auth : no authentication
10094 - stats scope : no restriction
10095
10096 Though this statement alone is enough to enable statistics reporting, it is
10097 recommended to set all other settings in order to avoid relying on default
10098 unobvious parameters.
10099
10100 Example :
10101 # public access (limited to this backend only)
10102 backend public_www
10103 server srv1 192.168.0.1:80
10104 stats enable
10105 stats hide-version
10106 stats scope .
10107 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010108 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010109 stats auth admin1:AdMiN123
10110 stats auth admin2:AdMiN321
10111
10112 # internal monitoring access (unlimited)
10113 backend private_monitoring
10114 stats enable
10115 stats uri /admin?stats
10116 stats refresh 5s
10117
10118 See also : "stats auth", "stats realm", "stats uri"
10119
10120
Willy Tarreaud63335a2010-02-26 12:56:52 +010010121stats hide-version
10122 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010123 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010124 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010125 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010126
Willy Tarreaud63335a2010-02-26 12:56:52 +010010127 By default, the stats page reports some useful status information along with
10128 the statistics. Among them is HAProxy's version. However, it is generally
10129 considered dangerous to report precise version to anyone, as it can help them
10130 target known weaknesses with specific attacks. The "stats hide-version"
10131 statement removes the version from the statistics report. This is recommended
10132 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010133
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +020010134 Though this statement alone is enough to enable statistics reporting, it is
10135 recommended to set all other settings in order to avoid relying on default
10136 unobvious parameters.
10137
Willy Tarreaud63335a2010-02-26 12:56:52 +010010138 Example :
10139 # public access (limited to this backend only)
10140 backend public_www
10141 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +020010142 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +010010143 stats hide-version
10144 stats scope .
10145 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010146 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +010010147 stats auth admin1:AdMiN123
10148 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010149
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010150 # internal monitoring access (unlimited)
10151 backend private_monitoring
10152 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +010010153 stats uri /admin?stats
10154 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +010010155
Willy Tarreaud63335a2010-02-26 12:56:52 +010010156 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +020010157
Willy Tarreau983e01e2010-01-11 18:42:06 +010010158
Cyril Bonté2be1b3f2010-09-30 23:46:30 +020010159stats http-request { allow | deny | auth [realm <realm>] }
10160 [ { if | unless } <condition> ]
10161 Access control for statistics
10162
10163 May be used in sections: defaults | frontend | listen | backend
10164 no | no | yes | yes
10165
10166 As "http-request", these set of options allow to fine control access to
10167 statistics. Each option may be followed by if/unless and acl.
10168 First option with matched condition (or option without condition) is final.
10169 For "deny" a 403 error will be returned, for "allow" normal processing is
10170 performed, for "auth" a 401/407 error code is returned so the client
10171 should be asked to enter a username and password.
10172
10173 There is no fixed limit to the number of http-request statements per
10174 instance.
10175
10176 See also : "http-request", section 3.4 about userlists and section 7
10177 about ACL usage.
10178
10179
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010180stats realm <realm>
10181 Enable statistics and set authentication realm
10182 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010183 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010184 Arguments :
10185 <realm> is the name of the HTTP Basic Authentication realm reported to
10186 the browser. The browser uses it to display it in the pop-up
10187 inviting the user to enter a valid username and password.
10188
10189 The realm is read as a single word, so any spaces in it should be escaped
10190 using a backslash ('\').
10191
10192 This statement is useful only in conjunction with "stats auth" since it is
10193 only related to authentication.
10194
10195 Though this statement alone is enough to enable statistics reporting, it is
10196 recommended to set all other settings in order to avoid relying on default
10197 unobvious parameters.
10198
10199 Example :
10200 # public access (limited to this backend only)
10201 backend public_www
10202 server srv1 192.168.0.1:80
10203 stats enable
10204 stats hide-version
10205 stats scope .
10206 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010207 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010208 stats auth admin1:AdMiN123
10209 stats auth admin2:AdMiN321
10210
10211 # internal monitoring access (unlimited)
10212 backend private_monitoring
10213 stats enable
10214 stats uri /admin?stats
10215 stats refresh 5s
10216
10217 See also : "stats auth", "stats enable", "stats uri"
10218
10219
10220stats refresh <delay>
10221 Enable statistics with automatic refresh
10222 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010223 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010224 Arguments :
10225 <delay> is the suggested refresh delay, specified in seconds, which will
10226 be returned to the browser consulting the report page. While the
10227 browser is free to apply any delay, it will generally respect it
10228 and refresh the page this every seconds. The refresh interval may
10229 be specified in any other non-default time unit, by suffixing the
10230 unit after the value, as explained at the top of this document.
10231
10232 This statement is useful on monitoring displays with a permanent page
10233 reporting the load balancer's activity. When set, the HTML report page will
10234 include a link "refresh"/"stop refresh" so that the user can select whether
Jackie Tapia749f74c2020-07-22 18:59:40 -050010235 they want automatic refresh of the page or not.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010236
10237 Though this statement alone is enough to enable statistics reporting, it is
10238 recommended to set all other settings in order to avoid relying on default
10239 unobvious parameters.
10240
10241 Example :
10242 # public access (limited to this backend only)
10243 backend public_www
10244 server srv1 192.168.0.1:80
10245 stats enable
10246 stats hide-version
10247 stats scope .
10248 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010249 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010250 stats auth admin1:AdMiN123
10251 stats auth admin2:AdMiN321
10252
10253 # internal monitoring access (unlimited)
10254 backend private_monitoring
10255 stats enable
10256 stats uri /admin?stats
10257 stats refresh 5s
10258
10259 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10260
10261
10262stats scope { <name> | "." }
10263 Enable statistics and limit access scope
10264 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010265 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010266 Arguments :
10267 <name> is the name of a listen, frontend or backend section to be
10268 reported. The special name "." (a single dot) designates the
10269 section in which the statement appears.
10270
10271 When this statement is specified, only the sections enumerated with this
10272 statement will appear in the report. All other ones will be hidden. This
10273 statement may appear as many times as needed if multiple sections need to be
10274 reported. Please note that the name checking is performed as simple string
10275 comparisons, and that it is never checked that a give section name really
10276 exists.
10277
10278 Though this statement alone is enough to enable statistics reporting, it is
10279 recommended to set all other settings in order to avoid relying on default
10280 unobvious parameters.
10281
10282 Example :
10283 # public access (limited to this backend only)
10284 backend public_www
10285 server srv1 192.168.0.1:80
10286 stats enable
10287 stats hide-version
10288 stats scope .
10289 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010290 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010291 stats auth admin1:AdMiN123
10292 stats auth admin2:AdMiN321
10293
10294 # internal monitoring access (unlimited)
10295 backend private_monitoring
10296 stats enable
10297 stats uri /admin?stats
10298 stats refresh 5s
10299
10300 See also : "stats auth", "stats enable", "stats realm", "stats uri"
10301
Willy Tarreaud63335a2010-02-26 12:56:52 +010010302
Willy Tarreauc9705a12010-07-27 20:05:50 +020010303stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +010010304 Enable reporting of a description on the statistics page.
10305 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010306 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010307
Willy Tarreauc9705a12010-07-27 20:05:50 +020010308 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +010010309 description from global section is automatically used instead.
10310
10311 This statement is useful for users that offer shared services to their
10312 customers, where node or description should be different for each customer.
10313
10314 Though this statement alone is enough to enable statistics reporting, it is
10315 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010316 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010317
10318 Example :
10319 # internal monitoring access (unlimited)
10320 backend private_monitoring
10321 stats enable
10322 stats show-desc Master node for Europe, Asia, Africa
10323 stats uri /admin?stats
10324 stats refresh 5s
10325
10326 See also: "show-node", "stats enable", "stats uri" and "description" in
10327 global section.
10328
10329
10330stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +020010331 Enable reporting additional information on the statistics page
10332 May be used in sections : defaults | frontend | listen | backend
10333 yes | yes | yes | yes
10334 Arguments : none
10335
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010336 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +010010337 - cap: capabilities (proxy)
10338 - mode: one of tcp, http or health (proxy)
10339 - id: SNMP ID (proxy, socket, server)
10340 - IP (socket, server)
10341 - cookie (backend, server)
10342
10343 Though this statement alone is enough to enable statistics reporting, it is
10344 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +010010345 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010346
10347 See also: "stats enable", "stats uri".
10348
10349
Amaury Denoyelle0b70a8a2020-10-05 11:49:45 +020010350stats show-modules
10351 Enable display of extra statistics module on the statistics page
10352 May be used in sections : defaults | frontend | listen | backend
10353 yes | yes | yes | yes
10354 Arguments : none
10355
10356 New columns are added at the end of the line containing the extra statistics
10357 values as a tooltip.
10358
10359 Though this statement alone is enough to enable statistics reporting, it is
10360 recommended to set all other settings in order to avoid relying on default
10361 unobvious parameters. Default behavior is not to show this information.
10362
10363 See also: "stats enable", "stats uri".
10364
10365
Willy Tarreaud63335a2010-02-26 12:56:52 +010010366stats show-node [ <name> ]
10367 Enable reporting of a host name on the statistics page.
10368 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010369 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +010010370 Arguments:
10371 <name> is an optional name to be reported. If unspecified, the
10372 node name from global section is automatically used instead.
10373
10374 This statement is useful for users that offer shared services to their
10375 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +010010376 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +010010377
10378 Though this statement alone is enough to enable statistics reporting, it is
10379 recommended to set all other settings in order to avoid relying on default
10380 unobvious parameters.
10381
10382 Example:
10383 # internal monitoring access (unlimited)
10384 backend private_monitoring
10385 stats enable
10386 stats show-node Europe-1
10387 stats uri /admin?stats
10388 stats refresh 5s
10389
10390 See also: "show-desc", "stats enable", "stats uri", and "node" in global
10391 section.
10392
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010393
10394stats uri <prefix>
10395 Enable statistics and define the URI prefix to access them
10396 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +020010397 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010398 Arguments :
10399 <prefix> is the prefix of any URI which will be redirected to stats. This
10400 prefix may contain a question mark ('?') to indicate part of a
10401 query string.
10402
10403 The statistics URI is intercepted on the relayed traffic, so it appears as a
10404 page within the normal application. It is strongly advised to ensure that the
10405 selected URI will never appear in the application, otherwise it will never be
10406 possible to reach it in the application.
10407
10408 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010409 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010410 It is generally a good idea to include a question mark in the URI so that
10411 intermediate proxies refrain from caching the results. Also, since any string
10412 beginning with the prefix will be accepted as a stats request, the question
10413 mark helps ensuring that no valid URI will begin with the same words.
10414
10415 It is sometimes very convenient to use "/" as the URI prefix, and put that
10416 statement in a "listen" instance of its own. That makes it easy to dedicate
10417 an address or a port to statistics only.
10418
10419 Though this statement alone is enough to enable statistics reporting, it is
10420 recommended to set all other settings in order to avoid relying on default
10421 unobvious parameters.
10422
10423 Example :
10424 # public access (limited to this backend only)
10425 backend public_www
10426 server srv1 192.168.0.1:80
10427 stats enable
10428 stats hide-version
10429 stats scope .
10430 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +010010431 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010432 stats auth admin1:AdMiN123
10433 stats auth admin2:AdMiN321
10434
10435 # internal monitoring access (unlimited)
10436 backend private_monitoring
10437 stats enable
10438 stats uri /admin?stats
10439 stats refresh 5s
10440
10441 See also : "stats auth", "stats enable", "stats realm"
10442
10443
Willy Tarreaud63335a2010-02-26 12:56:52 +010010444stick match <pattern> [table <table>] [{if | unless} <cond>]
10445 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +010010446 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +010010447 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010448
10449 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010450 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010451 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010452 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010453 stickiness table. This rule is mandatory.
10454
10455 <table> is an optional stickiness table name. If unspecified, the same
10456 backend's table is used. A stickiness table is declared using
10457 the "stick-table" statement.
10458
10459 <cond> is an optional matching condition. It makes it possible to match
10460 on a certain criterion only when other conditions are met (or
10461 not met). For instance, it could be used to match on a source IP
10462 address except when a request passes through a known proxy, in
10463 which case we'd match on a header containing that IP address.
10464
10465 Some protocols or applications require complex stickiness rules and cannot
10466 always simply rely on cookies nor hashing. The "stick match" statement
10467 describes a rule to extract the stickiness criterion from an incoming request
10468 or connection. See section 7 for a complete list of possible patterns and
10469 transformation rules.
10470
10471 The table has to be declared using the "stick-table" statement. It must be of
10472 a type compatible with the pattern. By default it is the one which is present
10473 in the same backend. It is possible to share a table with other backends by
10474 referencing it using the "table" keyword. If another table is referenced,
10475 the server's ID inside the backends are used. By default, all server IDs
10476 start at 1 in each backend, so the server ordering is enough. But in case of
10477 doubt, it is highly recommended to force server IDs using their "id" setting.
10478
10479 It is possible to restrict the conditions where a "stick match" statement
10480 will apply, using "if" or "unless" followed by a condition. See section 7 for
10481 ACL based conditions.
10482
10483 There is no limit on the number of "stick match" statements. The first that
10484 applies and matches will cause the request to be directed to the same server
10485 as was used for the request which created the entry. That way, multiple
10486 matches can be used as fallbacks.
10487
10488 The stick rules are checked after the persistence cookies, so they will not
10489 affect stickiness if a cookie has already been used to select a server. That
10490 way, it becomes very easy to insert cookies and match on IP addresses in
10491 order to maintain stickiness between HTTP and HTTPS.
10492
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010493 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10494 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010495 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010496
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010497 Example :
10498 # forward SMTP users to the same server they just used for POP in the
10499 # last 30 minutes
10500 backend pop
10501 mode tcp
10502 balance roundrobin
10503 stick store-request src
10504 stick-table type ip size 200k expire 30m
10505 server s1 192.168.1.1:110
10506 server s2 192.168.1.1:110
10507
10508 backend smtp
10509 mode tcp
10510 balance roundrobin
10511 stick match src table pop
10512 server s1 192.168.1.1:25
10513 server s2 192.168.1.1:25
10514
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010515 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010516 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010517
10518
10519stick on <pattern> [table <table>] [{if | unless} <condition>]
10520 Define a request pattern to associate a user to a server
10521 May be used in sections : defaults | frontend | listen | backend
10522 no | no | yes | yes
10523
10524 Note : This form is exactly equivalent to "stick match" followed by
10525 "stick store-request", all with the same arguments. Please refer
10526 to both keywords for details. It is only provided as a convenience
10527 for writing more maintainable configurations.
10528
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010529 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10530 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010531 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010532
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010533 Examples :
10534 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +010010535 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010536
10537 # ...is strictly equivalent to this one :
10538 stick match src table pop if !localhost
10539 stick store-request src table pop if !localhost
10540
10541
10542 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
10543 # well as HTTP without cookie. Share the same table between both accesses.
10544 backend http
10545 mode http
10546 balance roundrobin
10547 stick on src table https
10548 cookie SRV insert indirect nocache
10549 server s1 192.168.1.1:80 cookie s1
10550 server s2 192.168.1.1:80 cookie s2
10551
10552 backend https
10553 mode tcp
10554 balance roundrobin
10555 stick-table type ip size 200k expire 30m
10556 stick on src
10557 server s1 192.168.1.1:443
10558 server s2 192.168.1.1:443
10559
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010560 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010561
10562
10563stick store-request <pattern> [table <table>] [{if | unless} <condition>]
10564 Define a request pattern used to create an entry in a stickiness table
10565 May be used in sections : defaults | frontend | listen | backend
10566 no | no | yes | yes
10567
10568 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010569 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010570 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +010010571 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010572 server is selected.
10573
10574 <table> is an optional stickiness table name. If unspecified, the same
10575 backend's table is used. A stickiness table is declared using
10576 the "stick-table" statement.
10577
10578 <cond> is an optional storage condition. It makes it possible to store
10579 certain criteria only when some conditions are met (or not met).
10580 For instance, it could be used to store the source IP address
10581 except when the request passes through a known proxy, in which
10582 case we'd store a converted form of a header containing that IP
10583 address.
10584
10585 Some protocols or applications require complex stickiness rules and cannot
10586 always simply rely on cookies nor hashing. The "stick store-request" statement
10587 describes a rule to decide what to extract from the request and when to do
10588 it, in order to store it into a stickiness table for further requests to
10589 match it using the "stick match" statement. Obviously the extracted part must
10590 make sense and have a chance to be matched in a further request. Storing a
10591 client's IP address for instance often makes sense. Storing an ID found in a
10592 URL parameter also makes sense. Storing a source port will almost never make
10593 any sense because it will be randomly matched. See section 7 for a complete
10594 list of possible patterns and transformation rules.
10595
10596 The table has to be declared using the "stick-table" statement. It must be of
10597 a type compatible with the pattern. By default it is the one which is present
10598 in the same backend. It is possible to share a table with other backends by
10599 referencing it using the "table" keyword. If another table is referenced,
10600 the server's ID inside the backends are used. By default, all server IDs
10601 start at 1 in each backend, so the server ordering is enough. But in case of
10602 doubt, it is highly recommended to force server IDs using their "id" setting.
10603
10604 It is possible to restrict the conditions where a "stick store-request"
10605 statement will apply, using "if" or "unless" followed by a condition. This
10606 condition will be evaluated while parsing the request, so any criteria can be
10607 used. See section 7 for ACL based conditions.
10608
10609 There is no limit on the number of "stick store-request" statements, but
10610 there is a limit of 8 simultaneous stores per request or response. This
10611 makes it possible to store up to 8 criteria, all extracted from either the
10612 request or the response, regardless of the number of rules. Only the 8 first
10613 ones which match will be kept. Using this, it is possible to feed multiple
10614 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010010615 another protocol or access method. Using multiple store-request rules with
10616 the same table is possible and may be used to find the best criterion to rely
10617 on, by arranging the rules by decreasing preference order. Only the first
10618 extracted criterion for a given table will be stored. All subsequent store-
10619 request rules referencing the same table will be skipped and their ACLs will
10620 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010621
10622 The "store-request" rules are evaluated once the server connection has been
10623 established, so that the table will contain the real server that processed
10624 the request.
10625
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010626 Note : Consider not using this feature in multi-process mode (nbproc > 1)
10627 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +010010628 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010629
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010630 Example :
10631 # forward SMTP users to the same server they just used for POP in the
10632 # last 30 minutes
10633 backend pop
10634 mode tcp
10635 balance roundrobin
10636 stick store-request src
10637 stick-table type ip size 200k expire 30m
10638 server s1 192.168.1.1:110
10639 server s2 192.168.1.1:110
10640
10641 backend smtp
10642 mode tcp
10643 balance roundrobin
10644 stick match src table pop
10645 server s1 192.168.1.1:25
10646 server s2 192.168.1.1:25
10647
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010648 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +020010649 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010650
10651
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010652stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Thayne McCombs92149f92020-11-20 01:28:26 -070010653 size <size> [expire <expire>] [nopurge] [peers <peersect>] [srvkey <srvkey>]
Emeric Brunf099e792010-09-27 12:05:28 +020010654 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +080010655 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010656 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +020010657 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010658
10659 Arguments :
10660 ip a table declared with "type ip" will only store IPv4 addresses.
10661 This form is very compact (about 50 bytes per entry) and allows
10662 very fast entry lookup and stores with almost no overhead. This
10663 is mainly used to store client source IP addresses.
10664
David du Colombier9a6d3c92011-03-17 10:40:24 +010010665 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
10666 This form is very compact (about 60 bytes per entry) and allows
10667 very fast entry lookup and stores with almost no overhead. This
10668 is mainly used to store client source IP addresses.
10669
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010670 integer a table declared with "type integer" will store 32bit integers
10671 which can represent a client identifier found in a request for
10672 instance.
10673
10674 string a table declared with "type string" will store substrings of up
10675 to <len> characters. If the string provided by the pattern
10676 extractor is larger than <len>, it will be truncated before
10677 being stored. During matching, at most <len> characters will be
10678 compared between the string in the table and the extracted
10679 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010680 to 32 characters.
10681
10682 binary a table declared with "type binary" will store binary blocks
10683 of <len> bytes. If the block provided by the pattern
10684 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +020010685 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010686 is shorter than <len>, it will be padded by 0. When not
10687 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010688
10689 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +020010690 "string" type table (See type "string" above). Or the number
10691 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010692 changing this parameter as memory usage will proportionally
10693 increase.
10694
10695 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +010010696 value directly impacts memory usage. Count approximately
10697 50 bytes per entry, plus the size of a string if any. The size
10698 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010699
10700 [nopurge] indicates that we refuse to purge older entries when the table
10701 is full. When not specified and the table is full when haproxy
10702 wants to store an entry in it, it will flush a few of the oldest
10703 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010704 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010705 be desirable to refuse new entries instead of purging the older
10706 ones. That may be the case when the amount of data to store is
10707 far above the hardware limits and we prefer not to offer access
10708 to new clients than to reject the ones already connected. When
10709 using this parameter, be sure to properly set the "expire"
10710 parameter (see below).
10711
Emeric Brunf099e792010-09-27 12:05:28 +020010712 <peersect> is the name of the peers section to use for replication. Entries
10713 which associate keys to server IDs are kept synchronized with
10714 the remote peers declared in this section. All entries are also
10715 automatically learned from the local peer (old process) during a
10716 soft restart.
10717
Willy Tarreau1abc6732015-05-01 19:21:02 +020010718 NOTE : each peers section may be referenced only by tables
10719 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +010010720
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010721 <expire> defines the maximum duration of an entry in the table since it
10722 was last created, refreshed or matched. The expiration delay is
10723 defined using the standard time format, similarly as the various
10724 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +030010725 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +020010726 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010727 be removed once full. Be sure not to use the "nopurge" parameter
10728 if not expiration delay is specified.
10729
Thayne McCombs92149f92020-11-20 01:28:26 -070010730 <srvkey> specifies how each server is identified for the purposes of the
10731 stick table. The valid values are "name" and "addr". If "name" is
10732 given, then <name> argument for the server (may be generated by
10733 a template). If "addr" is given, then the server is identified
10734 by its current network address, including the port. "addr" is
10735 especially useful if you are using service discovery to generate
10736 the addresses for servers with peered stick-tables and want
10737 to consistently use the same host across peers for a stickiness
10738 token.
10739
Willy Tarreau08d5f982010-06-06 13:34:54 +020010740 <data_type> is used to store additional information in the stick-table. This
10741 may be used by ACLs in order to control various criteria related
10742 to the activity of the client matching the stick-table. For each
10743 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +020010744 that the additional data can fit. Several data types may be
10745 stored with an entry. Multiple data types may be specified after
10746 the "store" keyword, as a comma-separated list. Alternatively,
10747 it is possible to repeat the "store" keyword followed by one or
10748 several data types. Except for the "server_id" type which is
10749 automatically detected and enabled, all data types must be
10750 explicitly declared to be stored. If an ACL references a data
10751 type which is not stored, the ACL will simply not match. Some
10752 data types require an argument which must be passed just after
10753 the type between parenthesis. See below for the supported data
10754 types and their arguments.
10755
10756 The data types that can be stored with an entry are the following :
10757 - server_id : this is an integer which holds the numeric ID of the server a
10758 request was assigned to. It is used by the "stick match", "stick store",
10759 and "stick on" rules. It is automatically enabled when referenced.
10760
10761 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
10762 integer which may be used for anything. Most of the time it will be used
10763 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010764 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +020010765
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010766 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
10767 over a period. It is a positive 32-bit integer integer which may be used
10768 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +010010769 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010770 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +010010771 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +020010772
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010773 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
10774 integer which may be used for anything. Most of the time it will be used
10775 to put a special tag on some entries, for instance to note that a
10776 specific behavior was detected and must be known for future matches.
10777
10778 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
10779 over a period. It is a positive 32-bit integer integer which may be used
10780 for anything. Just like <gpc1>, it counts events, but instead of keeping
10781 a cumulative number, it maintains the rate at which the counter is
10782 incremented. Most of the time it will be used to measure the frequency of
10783 occurrence of certain events (e.g. requests to a specific URL).
10784
Willy Tarreauc9705a12010-07-27 20:05:50 +020010785 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
10786 the absolute number of connections received from clients which matched
10787 this entry. It does not mean the connections were accepted, just that
10788 they were received.
10789
10790 - conn_cur : Current Connections. It is a positive 32-bit integer which
10791 stores the concurrent connection counts for the entry. It is incremented
10792 once an incoming connection matches the entry, and decremented once the
10793 connection leaves. That way it is possible to know at any time the exact
10794 number of concurrent connections for an entry.
10795
10796 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10797 integer parameter <period> which indicates in milliseconds the length
10798 of the period over which the average is measured. It reports the average
10799 incoming connection rate over that period, in connections per period. The
10800 result is an integer which can be matched using ACLs.
10801
10802 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
10803 the absolute number of sessions received from clients which matched this
10804 entry. A session is a connection that was accepted by the layer 4 rules.
10805
10806 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10807 integer parameter <period> which indicates in milliseconds the length
10808 of the period over which the average is measured. It reports the average
10809 incoming session rate over that period, in sessions per period. The
10810 result is an integer which can be matched using ACLs.
10811
10812 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
10813 counts the absolute number of HTTP requests received from clients which
10814 matched this entry. It does not matter whether they are valid requests or
10815 not. Note that this is different from sessions when keep-alive is used on
10816 the client side.
10817
10818 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10819 integer parameter <period> which indicates in milliseconds the length
10820 of the period over which the average is measured. It reports the average
10821 HTTP request rate over that period, in requests per period. The result is
10822 an integer which can be matched using ACLs. It does not matter whether
10823 they are valid requests or not. Note that this is different from sessions
10824 when keep-alive is used on the client side.
10825
10826 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
10827 counts the absolute number of HTTP requests errors induced by clients
10828 which matched this entry. Errors are counted on invalid and truncated
10829 requests, as well as on denied or tarpitted requests, and on failed
10830 authentications. If the server responds with 4xx, then the request is
10831 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +010010832 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +020010833
10834 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10835 integer parameter <period> which indicates in milliseconds the length
10836 of the period over which the average is measured. It reports the average
10837 HTTP request error rate over that period, in requests per period (see
10838 http_err_cnt above for what is accounted as an error). The result is an
10839 integer which can be matched using ACLs.
10840
10841 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010010842 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +020010843 which matched this entry. Headers are included in the count. This may be
10844 used to limit abuse of upload features on photo or video servers.
10845
10846 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
10847 integer parameter <period> which indicates in milliseconds the length
10848 of the period over which the average is measured. It reports the average
10849 incoming bytes rate over that period, in bytes per period. It may be used
10850 to detect users which upload too much and too fast. Warning: with large
10851 uploads, it is possible that the amount of uploaded data will be counted
10852 once upon termination, thus causing spikes in the average transfer speed
10853 instead of having a smooth one. This may partially be smoothed with
10854 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
10855 recommended for better fairness.
10856
10857 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +010010858 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +020010859 matched this entry. Headers are included in the count. This may be used
10860 to limit abuse of bots sucking the whole site.
10861
10862 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
10863 an integer parameter <period> which indicates in milliseconds the length
10864 of the period over which the average is measured. It reports the average
10865 outgoing bytes rate over that period, in bytes per period. It may be used
10866 to detect users which download too much and too fast. Warning: with large
10867 transfers, it is possible that the amount of transferred data will be
10868 counted once upon termination, thus causing spikes in the average
10869 transfer speed instead of having a smooth one. This may partially be
10870 smoothed with "option contstats" though this is not perfect yet. Use of
10871 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +020010872
Willy Tarreauc00cdc22010-06-06 16:48:26 +020010873 There is only one stick-table per proxy. At the moment of writing this doc,
10874 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010875 to be required, simply create a dummy backend with a stick-table in it and
10876 reference it.
10877
10878 It is important to understand that stickiness based on learning information
10879 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +010010880 lost upon restart unless peers are properly configured to transfer such
10881 information upon restart (recommended). In general it can be good as a
10882 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010883
Willy Tarreauc9705a12010-07-27 20:05:50 +020010884 Last, memory requirements may be important when storing many data types.
10885 Indeed, storing all indicators above at once in each entry requires 116 bytes
10886 per entry, or 116 MB for a 1-million entries table. This is definitely not
10887 something that can be ignored.
10888
10889 Example:
10890 # Keep track of counters of up to 1 million IP addresses over 5 minutes
10891 # and store a general purpose counter and the average connection rate
10892 # computed over a sliding window of 30 seconds.
10893 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
10894
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +030010895 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +010010896 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +010010897
10898
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010899stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +010010900 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010901 May be used in sections : defaults | frontend | listen | backend
10902 no | no | yes | yes
10903
10904 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010905 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010906 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +010010907 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010908 server is selected.
10909
10910 <table> is an optional stickiness table name. If unspecified, the same
10911 backend's table is used. A stickiness table is declared using
10912 the "stick-table" statement.
10913
10914 <cond> is an optional storage condition. It makes it possible to store
10915 certain criteria only when some conditions are met (or not met).
10916 For instance, it could be used to store the SSL session ID only
10917 when the response is a SSL server hello.
10918
10919 Some protocols or applications require complex stickiness rules and cannot
10920 always simply rely on cookies nor hashing. The "stick store-response"
10921 statement describes a rule to decide what to extract from the response and
10922 when to do it, in order to store it into a stickiness table for further
10923 requests to match it using the "stick match" statement. Obviously the
10924 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010925 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010926 See section 7 for a complete list of possible patterns and transformation
10927 rules.
10928
10929 The table has to be declared using the "stick-table" statement. It must be of
10930 a type compatible with the pattern. By default it is the one which is present
10931 in the same backend. It is possible to share a table with other backends by
10932 referencing it using the "table" keyword. If another table is referenced,
10933 the server's ID inside the backends are used. By default, all server IDs
10934 start at 1 in each backend, so the server ordering is enough. But in case of
10935 doubt, it is highly recommended to force server IDs using their "id" setting.
10936
10937 It is possible to restrict the conditions where a "stick store-response"
10938 statement will apply, using "if" or "unless" followed by a condition. This
10939 condition will be evaluated while parsing the response, so any criteria can
10940 be used. See section 7 for ACL based conditions.
10941
10942 There is no limit on the number of "stick store-response" statements, but
10943 there is a limit of 8 simultaneous stores per request or response. This
10944 makes it possible to store up to 8 criteria, all extracted from either the
10945 request or the response, regardless of the number of rules. Only the 8 first
10946 ones which match will be kept. Using this, it is possible to feed multiple
10947 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +010010948 another protocol or access method. Using multiple store-response rules with
10949 the same table is possible and may be used to find the best criterion to rely
10950 on, by arranging the rules by decreasing preference order. Only the first
10951 extracted criterion for a given table will be stored. All subsequent store-
10952 response rules referencing the same table will be skipped and their ACLs will
10953 not be evaluated. However, even if a store-request rule references a table, a
10954 store-response rule may also use the same table. This means that each table
10955 may learn exactly one element from the request and one element from the
10956 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010957
10958 The table will contain the real server that processed the request.
10959
10960 Example :
10961 # Learn SSL session ID from both request and response and create affinity.
10962 backend https
10963 mode tcp
10964 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +020010965 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010966 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010967
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010968 acl clienthello req_ssl_hello_type 1
10969 acl serverhello rep_ssl_hello_type 2
10970
10971 # use tcp content accepts to detects ssl client and server hello.
10972 tcp-request inspect-delay 5s
10973 tcp-request content accept if clienthello
10974
10975 # no timeout on response inspect delay by default.
10976 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010977
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010978 # SSL session ID (SSLID) may be present on a client or server hello.
10979 # Its length is coded on 1 byte at offset 43 and its value starts
10980 # at offset 44.
10981
10982 # Match and learn on request if client hello.
10983 stick on payload_lv(43,1) if clienthello
10984
10985 # Learn on response if server hello.
10986 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +020010987
Emeric Brun6a1cefa2010-09-24 18:15:17 +020010988 server s1 192.168.1.1:443
10989 server s2 192.168.1.1:443
10990
10991 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
10992 extraction.
10993
10994
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020010995tcp-check comment <string>
10996 Defines a comment for the following the tcp-check rule, reported in logs if
10997 it fails.
10998 May be used in sections : defaults | frontend | listen | backend
10999 yes | no | yes | yes
11000
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011001 Arguments :
11002 <string> is the comment message to add in logs if the following tcp-check
11003 rule fails.
11004
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011005 It only works for connect, send and expect rules. It is useful to make
11006 user-friendly error reporting.
11007
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011008 See also : "option tcp-check", "tcp-check connect", "tcp-check send" and
11009 "tcp-check expect".
11010
11011
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011012tcp-check connect [default] [port <expr>] [addr <ip>] [send-proxy] [via-socks4]
11013 [ssl] [sni <sni>] [alpn <alpn>] [linger]
Christopher Fauletedc6ed92020-04-23 16:27:59 +020011014 [proto <name>] [comment <msg>]
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011015 Opens a new connection
11016 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011017 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011018
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011019 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011020 comment <msg> defines a message to report if the rule evaluation fails.
11021
Christopher Faulet4dce5922020-03-30 13:54:42 +020011022 default Use default options of the server line to do the health
Daniel Corbett67a82712020-07-06 23:01:19 -040011023 checks. The server options are used only if not redefined.
Christopher Faulet4dce5922020-03-30 13:54:42 +020011024
Christopher Fauletb7d30092020-03-30 15:19:03 +020011025 port <expr> if not set, check port or server port is used.
Christopher Faulet5c288742020-03-31 08:15:58 +020011026 It tells HAProxy where to open the connection to.
11027 <port> must be a valid TCP port source integer, from 1 to
Christopher Fauletb7d30092020-03-30 15:19:03 +020011028 65535 or an sample-fetch expression.
Christopher Faulet5c288742020-03-31 08:15:58 +020011029
11030 addr <ip> defines the IP address to do the health check.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011031
11032 send-proxy send a PROXY protocol string
11033
Christopher Faulet085426a2020-03-30 13:07:02 +020011034 via-socks4 enables outgoing health checks using upstream socks4 proxy.
11035
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011036 ssl opens a ciphered connection
11037
Christopher Faulet79b31d42020-03-30 13:00:05 +020011038 sni <sni> specifies the SNI to use to do health checks over SSL.
11039
Christopher Faulet98572322020-03-30 13:16:44 +020011040 alpn <alpn> defines which protocols to advertise with ALPN. The protocol
11041 list consists in a comma-delimited list of protocol names,
11042 for instance: "http/1.1,http/1.0" (without quotes).
11043 If it is not set, the server ALPN is used.
11044
Christopher Fauletedc6ed92020-04-23 16:27:59 +020011045 proto <name> forces the multiplexer's protocol to use for this connection.
11046 It must be a TCP mux protocol and it must be usable on the
11047 backend side. The list of available protocols is reported in
11048 haproxy -vv.
11049
Christopher Faulet5c288742020-03-31 08:15:58 +020011050 linger cleanly close the connection instead of using a single RST.
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010011051
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011052 When an application lies on more than a single TCP port or when HAProxy
11053 load-balance many services in a single backend, it makes sense to probe all
11054 the services individually before considering a server as operational.
11055
11056 When there are no TCP port configured on the server line neither server port
11057 directive, then the 'tcp-check connect port <port>' must be the first step
11058 of the sequence.
11059
11060 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
11061 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
11062 do.
11063
11064 When a connect must start the ruleset, if may still be preceded by set-var,
11065 unset-var or comment rules.
11066
11067 Examples :
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011068 # check HTTP and HTTPs services on a server.
11069 # first open port 80 thanks to server line port directive, then
11070 # tcp-check opens port 443, ciphered and run a request on it:
11071 option tcp-check
11072 tcp-check connect
11073 tcp-check send GET\ /\ HTTP/1.0\r\n
11074 tcp-check send Host:\ haproxy.1wt.eu\r\n
11075 tcp-check send \r\n
11076 tcp-check expect rstring (2..|3..)
11077 tcp-check connect port 443 ssl
11078 tcp-check send GET\ /\ HTTP/1.0\r\n
11079 tcp-check send Host:\ haproxy.1wt.eu\r\n
11080 tcp-check send \r\n
11081 tcp-check expect rstring (2..|3..)
11082 server www 10.0.0.1 check port 80
11083
11084 # check both POP and IMAP from a single server:
11085 option tcp-check
Gaetan Rivetf8ba6772020-02-07 15:37:17 +010011086 tcp-check connect port 110 linger
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011087 tcp-check expect string +OK\ POP3\ ready
11088 tcp-check connect port 143
11089 tcp-check expect string *\ OK\ IMAP4\ ready
11090 server mail 10.0.0.1 check
11091
11092 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
11093
11094
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011095tcp-check expect [min-recv <int>] [comment <msg>]
Christopher Fauletec07e382020-04-07 14:56:26 +020011096 [ok-status <st>] [error-status <st>] [tout-status <st>]
Christopher Faulet98cc57c2020-04-01 20:52:31 +020011097 [on-success <fmt>] [on-error <fmt>] [status-code <expr>]
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011098 [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +010011099 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011100 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011101 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011102
11103 Arguments :
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011104 comment <msg> defines a message to report if the rule evaluation fails.
11105
Gaetan Rivet1afd8262020-02-07 15:37:17 +010011106 min-recv is optional and can define the minimum amount of data required to
11107 evaluate the current expect rule. If the number of received bytes
11108 is under this limit, the check will wait for more data. This
11109 option can be used to resolve some ambiguous matching rules or to
11110 avoid executing costly regex matches on content known to be still
11111 incomplete. If an exact string (string or binary) is used, the
11112 minimum between the string length and this parameter is used.
11113 This parameter is ignored if it is set to -1. If the expect rule
11114 does not match, the check will wait for more data. If set to 0,
11115 the evaluation result is always conclusive.
11116
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011117 <match> is a keyword indicating how to look for a specific pattern in the
Gaetan Rivetefab6c62020-02-07 15:37:17 +010011118 response. The keyword may be one of "string", "rstring", "binary" or
11119 "rbinary".
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011120 The keyword may be preceded by an exclamation mark ("!") to negate
11121 the match. Spaces are allowed between the exclamation mark and the
11122 keyword. See below for more details on the supported keywords.
11123
Christopher Fauletec07e382020-04-07 14:56:26 +020011124 ok-status <st> is optional and can be used to set the check status if
11125 the expect rule is successfully evaluated and if it is
11126 the last rule in the tcp-check ruleset. "L7OK", "L7OKC",
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011127 "L6OK" and "L4OK" are supported :
11128 - L7OK : check passed on layer 7
Christopher Faulet83662b52020-11-20 17:47:47 +010011129 - L7OKC : check conditionally passed on layer 7, set
11130 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011131 - L6OK : check passed on layer 6
11132 - L4OK : check passed on layer 4
Christopher Fauletec07e382020-04-07 14:56:26 +020011133 By default "L7OK" is used.
11134
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011135 error-status <st> is optional and can be used to set the check status if
11136 an error occurred during the expect rule evaluation.
Christopher Faulet83662b52020-11-20 17:47:47 +010011137 "L7OKC", "L7RSP", "L7STS", "L6RSP" and "L4CON" are
11138 supported :
11139 - L7OKC : check conditionally passed on layer 7, set
11140 server to NOLB state.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011141 - L7RSP : layer 7 invalid response - protocol error
11142 - L7STS : layer 7 response error, for example HTTP 5xx
11143 - L6RSP : layer 6 invalid response - protocol error
11144 - L4CON : layer 1-4 connection problem
11145 By default "L7RSP" is used.
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011146
Christopher Fauletec07e382020-04-07 14:56:26 +020011147 tout-status <st> is optional and can be used to set the check status if
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011148 a timeout occurred during the expect rule evaluation.
Christopher Fauletd888f0f2020-05-07 07:40:17 +020011149 "L7TOUT", "L6TOUT", and "L4TOUT" are supported :
11150 - L7TOUT : layer 7 (HTTP/SMTP) timeout
11151 - L6TOUT : layer 6 (SSL) timeout
11152 - L4TOUT : layer 1-4 timeout
Christopher Fauletcf80f2f2020-04-01 11:04:52 +020011153 By default "L7TOUT" is used.
11154
Christopher Fauletbe52b4d2020-04-01 16:30:22 +020011155 on-success <fmt> is optional and can be used to customize the
11156 informational message reported in logs if the expect
11157 rule is successfully evaluated and if it is the last rule
11158 in the tcp-check ruleset. <fmt> is a log-format string.
11159
11160 on-error <fmt> is optional and can be used to customize the
11161 informational message reported in logs if an error
11162 occurred during the expect rule evaluation. <fmt> is a
11163 log-format string.
11164
Christopher Faulet98cc57c2020-04-01 20:52:31 +020011165 status-code <expr> is optional and can be used to set the check status code
11166 reported in logs, on success or on error. <expr> is a
11167 standard HAProxy expression formed by a sample-fetch
11168 followed by some converters.
11169
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011170 <pattern> is the pattern to look for. It may be a string or a regular
11171 expression. If the pattern contains spaces, they must be escaped
11172 with the usual backslash ('\').
11173 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +010011174 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011175 two digits will represent a byte. The hexadecimal digits may be
11176 used upper or lower case.
11177
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011178 The available matches are intentionally similar to their http-check cousins :
11179
11180 string <string> : test the exact string matches in the response buffer.
11181 A health check response will be considered valid if the
11182 response's buffer contains this exact string. If the
11183 "string" keyword is prefixed with "!", then the response
11184 will be considered invalid if the body contains this
11185 string. This can be used to look for a mandatory pattern
11186 in a protocol response, or to detect a failure when a
11187 specific error appears in a protocol banner.
11188
11189 rstring <regex> : test a regular expression on the response buffer.
11190 A health check response will be considered valid if the
11191 response's buffer matches this expression. If the
11192 "rstring" keyword is prefixed with "!", then the response
11193 will be considered invalid if the body matches the
11194 expression.
11195
Christopher Fauletaaab0832020-05-05 15:54:22 +020011196 string-lf <fmt> : test a log-format string match in the response's buffer.
11197 A health check response will be considered valid if the
11198 response's buffer contains the string resulting of the
11199 evaluation of <fmt>, which follows the log-format rules.
11200 If prefixed with "!", then the response will be
11201 considered invalid if the buffer contains the string.
11202
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011203 binary <hexstring> : test the exact string in its hexadecimal form matches
11204 in the response buffer. A health check response will
11205 be considered valid if the response's buffer contains
11206 this exact hexadecimal string.
11207 Purpose is to match data on binary protocols.
11208
Gaetan Rivetefab6c62020-02-07 15:37:17 +010011209 rbinary <regex> : test a regular expression on the response buffer, like
11210 "rstring". However, the response buffer is transformed
11211 into its hexadecimal form, including NUL-bytes. This
11212 allows using all regex engines to match any binary
11213 content. The hexadecimal transformation takes twice the
11214 size of the original response. As such, the expected
11215 pattern should work on at-most half the response buffer
11216 size.
11217
Christopher Fauletaaab0832020-05-05 15:54:22 +020011218 binary-lf <hexfmt> : test a log-format string in its hexadecimal form
11219 match in the response's buffer. A health check response
11220 will be considered valid if the response's buffer
11221 contains the hexadecimal string resulting of the
11222 evaluation of <fmt>, which follows the log-format
11223 rules. If prefixed with "!", then the response will be
11224 considered invalid if the buffer contains the
11225 hexadecimal string. The hexadecimal string is converted
11226 in a binary string before matching the response's
11227 buffer.
11228
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011229 It is important to note that the responses will be limited to a certain size
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011230 defined by the global "tune.bufsize" option, which defaults to 16384 bytes.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011231 Thus, too large responses may not contain the mandatory pattern when using
11232 "string", "rstring" or binary. If a large response is absolutely required, it
11233 is possible to change the default max size by setting the global variable.
11234 However, it is worth keeping in mind that parsing very large responses can
11235 waste some CPU cycles, especially when regular expressions are used, and that
11236 it is always better to focus the checks on smaller resources. Also, in its
11237 current state, the check will not find any string nor regex past a null
11238 character in the response. Similarly it is not possible to request matching
11239 the null character.
11240
11241 Examples :
11242 # perform a POP check
11243 option tcp-check
11244 tcp-check expect string +OK\ POP3\ ready
11245
11246 # perform an IMAP check
11247 option tcp-check
11248 tcp-check expect string *\ OK\ IMAP4\ ready
11249
11250 # look for the redis master server
11251 option tcp-check
11252 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +020011253 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011254 tcp-check send info\ replication\r\n
11255 tcp-check expect string role:master
11256 tcp-check send QUIT\r\n
11257 tcp-check expect string +OK
11258
11259
11260 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011261 "tcp-check send-binary", "http-check expect", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011262
11263
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011264tcp-check send <data> [comment <msg>]
11265tcp-check send-lf <fmt> [comment <msg>]
11266 Specify a string or a log-format string to be sent as a question during a
11267 generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011268 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011269 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011270
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011271 Arguments :
11272 comment <msg> defines a message to report if the rule evaluation fails.
11273
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011274 <data> is the string that will be sent during a generic health
11275 check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020011276
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011277 <fmt> is the log-format string that will be sent, once evaluated,
11278 during a generic health check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011279
11280 Examples :
11281 # look for the redis master server
11282 option tcp-check
11283 tcp-check send info\ replication\r\n
11284 tcp-check expect string role:master
11285
11286 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011287 "tcp-check send-binary", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011288
11289
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011290tcp-check send-binary <hexstring> [comment <msg>]
11291tcp-check send-binary-lf <hexfmt> [comment <msg>]
11292 Specify an hex digits string or an hex digits log-format string to be sent as
11293 a binary question during a raw tcp health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011294 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011295 yes | no | yes | yes
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011296
Christopher Faulet4f5c2e22020-04-23 15:22:33 +020011297 Arguments :
11298 comment <msg> defines a message to report if the rule evaluation fails.
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011299
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011300 <hexstring> is the hexadecimal string that will be send, once converted
11301 to binary, during a generic health check session.
Christopher Faulet16fff672020-04-30 07:50:54 +020011302
Christopher Fauletb50b3e62020-05-05 18:43:43 +020011303 <hexfmt> is the hexadecimal log-format string that will be send, once
11304 evaluated and converted to binary, during a generic health
11305 check session.
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011306
11307 Examples :
11308 # redis check in binary
11309 option tcp-check
11310 tcp-check send-binary 50494e470d0a # PING\r\n
11311 tcp-check expect binary 2b504F4e47 # +PONG
11312
11313
11314 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
Christopher Fauletbb9fb8b2020-11-25 17:20:57 +010011315 "tcp-check send", tune.bufsize
Willy Tarreau938c7fe2014-04-25 14:21:39 +020011316
11317
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011318tcp-check set-var(<var-name>) <expr>
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011319 This operation sets the content of a variable. The variable is declared inline.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011320 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011321 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011322
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011323 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011324 <var-name> The name of the variable starts with an indication about its
11325 scope. The scopes allowed for tcp-check are:
11326 "proc" : the variable is shared with the whole process.
11327 "sess" : the variable is shared with the tcp-check session.
11328 "check": the variable is declared for the lifetime of the tcp-check.
11329 This prefix is followed by a name. The separator is a '.'.
11330 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11331 and '-'.
11332
11333 <expr> Is a sample-fetch expression potentially followed by converters.
11334
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011335 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011336 tcp-check set-var(check.port) int(1234)
11337
11338
11339tcp-check unset-var(<var-name>)
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011340 Free a reference to a variable within its scope.
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011341 May be used in sections: defaults | frontend | listen | backend
Christopher Faulet404f9192020-04-09 23:13:54 +020011342 yes | no | yes | yes
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011343
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011344 Arguments :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011345 <var-name> The name of the variable starts with an indication about its
11346 scope. The scopes allowed for tcp-check are:
11347 "proc" : the variable is shared with the whole process.
11348 "sess" : the variable is shared with the tcp-check session.
11349 "check": the variable is declared for the lifetime of the tcp-check.
11350 This prefix is followed by a name. The separator is a '.'.
11351 The name may only contain characters 'a-z', 'A-Z', '0-9', '.',
11352 and '-'.
11353
Christopher Fauletc52ea4d2020-04-23 15:43:35 +020011354 Examples :
Gaetan Rivet0c39ecc2020-02-24 17:34:11 +010011355 tcp-check unset-var(check.port)
11356
11357
Willy Tarreaue9656522010-08-17 15:40:09 +020011358tcp-request connection <action> [{if | unless} <condition>]
11359 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +020011360 May be used in sections : defaults | frontend | listen | backend
11361 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +020011362 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011363 <action> defines the action to perform if the condition applies. See
11364 below.
Willy Tarreau1a687942010-05-23 22:40:30 +020011365
Willy Tarreaue9656522010-08-17 15:40:09 +020011366 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011367
11368 Immediately after acceptance of a new incoming connection, it is possible to
11369 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +020011370 or dropped or have its counters tracked. Those conditions cannot make use of
11371 any data contents because the connection has not been read from yet, and the
11372 buffers are not yet allocated. This is used to selectively and very quickly
11373 accept or drop connections from various sources with a very low overhead. If
11374 some contents need to be inspected in order to take the decision, the
11375 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011376
Willy Tarreaue9656522010-08-17 15:40:09 +020011377 The "tcp-request connection" rules are evaluated in their exact declaration
11378 order. If no rule matches or if there is no rule, the default action is to
11379 accept the incoming connection. There is no specific limit to the number of
11380 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011381
Willy Tarreaua9083d02015-05-08 15:27:59 +020011382 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +020011383 - accept :
11384 accepts the connection if the condition is true (when used with "if")
11385 or false (when used with "unless"). The first such rule executed ends
11386 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011387
Willy Tarreaue9656522010-08-17 15:40:09 +020011388 - reject :
11389 rejects the connection if the condition is true (when used with "if")
11390 or false (when used with "unless"). The first such rule executed ends
11391 the rules evaluation. Rejected connections do not even become a
11392 session, which is why they are accounted separately for in the stats,
11393 as "denied connections". They are not considered for the session
11394 rate-limit and are not logged either. The reason is that these rules
11395 should only be used to filter extremely high connection rates such as
11396 the ones encountered during a massive DDoS attack. Under these extreme
11397 conditions, the simple action of logging each event would make the
11398 system collapse and would considerably lower the filtering capacity. If
11399 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +020011400 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011401
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011402 - expect-proxy layer4 :
11403 configures the client-facing connection to receive a PROXY protocol
11404 header before any byte is read from the socket. This is equivalent to
11405 having the "accept-proxy" keyword on the "bind" line, except that using
11406 the TCP rule allows the PROXY protocol to be accepted only for certain
11407 IP address ranges using an ACL. This is convenient when multiple layers
11408 of load balancers are passed through by traffic coming from public
11409 hosts.
11410
Bertrand Jacquin90759682016-06-06 15:35:39 +010011411 - expect-netscaler-cip layer4 :
11412 configures the client-facing connection to receive a NetScaler Client
11413 IP insertion protocol header before any byte is read from the socket.
11414 This is equivalent to having the "accept-netscaler-cip" keyword on the
11415 "bind" line, except that using the TCP rule allows the PROXY protocol
11416 to be accepted only for certain IP address ranges using an ACL. This
11417 is convenient when multiple layers of load balancers are passed
11418 through by traffic coming from public hosts.
11419
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011420 - capture <sample> len <length> :
11421 This only applies to "tcp-request content" rules. It captures sample
11422 expression <sample> from the request buffer, and converts it to a
11423 string of at most <len> characters. The resulting string is stored into
11424 the next request "capture" slot, so it will possibly appear next to
11425 some captured HTTP headers. It will then automatically appear in the
11426 logs, and it will be possible to extract it using sample fetch rules to
11427 feed it into headers or anything. The length should be limited given
11428 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +020011429 session life. Please check section 7.3 (Fetching samples) and "capture
11430 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011431
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011432 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +020011433 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011434 rules do not stop evaluation and do not change default action. The
11435 number of counters that may be simultaneously tracked by the same
11436 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roeslerfb2fce12019-07-10 15:45:51 -050011437 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Matteo Contrini1857b8c2020-10-16 17:35:54 +020011438 and (MAX_SESS_STKCTR-1). The first "track-sc0" rule executed enables
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020011439 tracking of the counters of the specified table as the first set. The
11440 first "track-sc1" rule executed enables tracking of the counters of the
11441 specified table as the second set. The first "track-sc2" rule executed
11442 enables tracking of the counters of the specified table as the third
11443 set. It is a recommended practice to use the first set of counters for
11444 the per-frontend counters and the second set for the per-backend ones.
11445 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011446
Willy Tarreaue9656522010-08-17 15:40:09 +020011447 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020011448 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +020011449 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +010011450 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011451 and used to select which table entry to update the counters.
11452 Note that "tcp-request connection" cannot use content-based
11453 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011454
Willy Tarreaue9656522010-08-17 15:40:09 +020011455 <table> is an optional table to be used instead of the default one,
11456 which is the stick-table declared in the current proxy. All
11457 the counters for the matches and updates for the key will
11458 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011459
Willy Tarreaue9656522010-08-17 15:40:09 +020011460 Once a "track-sc*" rule is executed, the key is looked up in the table
11461 and if it is not found, an entry is allocated for it. Then a pointer to
11462 that entry is kept during all the session's life, and this entry's
11463 counters are updated as often as possible, every time the session's
11464 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011465 Counters are only updated for events that happen after the tracking has
11466 been started. For example, connection counters will not be updated when
11467 tracking layer 7 information, since the connection event happens before
11468 layer7 information is extracted.
11469
Willy Tarreaue9656522010-08-17 15:40:09 +020011470 If the entry tracks concurrent connection counters, one connection is
11471 counted for as long as the entry is tracked, and the entry will not
11472 expire during that time. Tracking counters also provides a performance
11473 advantage over just checking the keys, because only one table lookup is
11474 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011475
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011476 - sc-inc-gpc0(<sc-id>):
11477 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
11478 counter designated by <sc-id>. If an error occurs, this action silently
11479 fails and the actions evaluation continues.
11480
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011481 - sc-inc-gpc1(<sc-id>):
11482 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
11483 counter designated by <sc-id>. If an error occurs, this action silently
11484 fails and the actions evaluation continues.
11485
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011486 - sc-set-gpt0(<sc-id>) { <int> | <expr> }:
11487 This action sets the 32-bit unsigned GPT0 tag according to the sticky
11488 counter designated by <sc-id> and the value of <int>/<expr>. The
11489 expected result is a boolean. If an error occurs, this action silently
11490 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011491
William Lallemand2e785f22016-05-25 01:48:42 +020011492 - set-src <expr> :
11493 Is used to set the source IP address to the value of specified
11494 expression. Useful if you want to mask source IP for privacy.
11495 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011496 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +020011497
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011498 Arguments:
11499 <expr> Is a standard HAProxy expression formed by a sample-fetch
11500 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +020011501
11502 Example:
William Lallemand2e785f22016-05-25 01:48:42 +020011503 tcp-request connection set-src src,ipmask(24)
11504
Willy Tarreau0c630532016-10-21 17:52:58 +020011505 When possible, set-src preserves the original source port as long as the
11506 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +020011507
William Lallemand44be6402016-05-25 01:51:35 +020011508 - set-src-port <expr> :
11509 Is used to set the source port address to the value of specified
11510 expression.
11511
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020011512 Arguments:
11513 <expr> Is a standard HAProxy expression formed by a sample-fetch
11514 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +020011515
11516 Example:
William Lallemand44be6402016-05-25 01:51:35 +020011517 tcp-request connection set-src-port int(4000)
11518
Willy Tarreau0c630532016-10-21 17:52:58 +020011519 When possible, set-src-port preserves the original source address as long
11520 as the address family supports a port, otherwise it forces the source
11521 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +020011522
William Lallemand13e9b0c2016-05-25 02:34:07 +020011523 - set-dst <expr> :
11524 Is used to set the destination IP address to the value of specified
11525 expression. Useful if you want to mask IP for privacy in log.
11526 If you want to provide an IP from a HTTP header use "http-request
11527 set-dst". If you want to connect to the new address/port, use
11528 '0.0.0.0:0' as a server address in the backend.
11529
11530 <expr> Is a standard HAProxy expression formed by a sample-fetch
11531 followed by some converters.
11532
11533 Example:
11534
11535 tcp-request connection set-dst dst,ipmask(24)
11536 tcp-request connection set-dst ipv4(10.0.0.1)
11537
Willy Tarreau0c630532016-10-21 17:52:58 +020011538 When possible, set-dst preserves the original destination port as long as
11539 the address family allows it, otherwise the destination port is set to 0.
11540
William Lallemand13e9b0c2016-05-25 02:34:07 +020011541 - set-dst-port <expr> :
11542 Is used to set the destination port address to the value of specified
11543 expression. If you want to connect to the new address/port, use
11544 '0.0.0.0:0' as a server address in the backend.
11545
11546
11547 <expr> Is a standard HAProxy expression formed by a sample-fetch
11548 followed by some converters.
11549
11550 Example:
11551
11552 tcp-request connection set-dst-port int(4000)
11553
Willy Tarreau0c630532016-10-21 17:52:58 +020011554 When possible, set-dst-port preserves the original destination address as
11555 long as the address family supports a port, otherwise it forces the
11556 destination address to IPv4 "0.0.0.0" before rewriting the port.
11557
Willy Tarreau2d392c22015-08-24 01:43:45 +020011558 - "silent-drop" :
11559 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011560 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020011561 to prevent the client from being notified. The effect it then that the
11562 client still sees an established connection while there's none on
11563 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
11564 except that it doesn't use any local resource at all on the machine
11565 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010011566 slow down stronger attackers. It is important to understand the impact
11567 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011568 client and HAProxy (firewalls, proxies, load balancers) will also keep
11569 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011570 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011571 TCP_REPAIR socket option is used to block the emission of a TCP
11572 reset. On other systems, the socket's TTL is reduced to 1 so that the
11573 TCP reset doesn't pass the first router, though it's still delivered to
11574 local networks. Do not use it unless you fully understand how it works.
11575
Willy Tarreaue9656522010-08-17 15:40:09 +020011576 Note that the "if/unless" condition is optional. If no condition is set on
11577 the action, it is simply performed unconditionally. That can be useful for
11578 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011579
Willy Tarreaue9656522010-08-17 15:40:09 +020011580 Example: accept all connections from white-listed hosts, reject too fast
11581 connection without counting them, and track accepted connections.
11582 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011583
Willy Tarreaue9656522010-08-17 15:40:09 +020011584 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011585 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011586 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011587
Willy Tarreaue9656522010-08-17 15:40:09 +020011588 Example: accept all connections from white-listed hosts, count all other
11589 connections and reject too fast ones. This results in abusive ones
11590 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011591
Willy Tarreaue9656522010-08-17 15:40:09 +020011592 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011593 tcp-request connection track-sc0 src
11594 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011595
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011596 Example: enable the PROXY protocol for traffic coming from all known proxies.
11597
11598 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
11599
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011600 See section 7 about ACL usage.
11601
Willy Tarreau4f614292016-10-21 17:49:36 +020011602 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011603
11604
Willy Tarreaue9656522010-08-17 15:40:09 +020011605tcp-request content <action> [{if | unless} <condition>]
11606 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +020011607 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020011608 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +020011609 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011610 <action> defines the action to perform if the condition applies. See
11611 below.
Willy Tarreau62644772008-07-16 18:36:06 +020011612
Willy Tarreaue9656522010-08-17 15:40:09 +020011613 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +020011614
Davor Ocelice9ed2812017-12-25 17:49:28 +010011615 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +020011616 called "TCP content inspection". During this stage, ACL-based rules are
11617 evaluated every time the request contents are updated, until either an
11618 "accept" or a "reject" rule matches, or the TCP request inspection delay
11619 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +020011620
Willy Tarreaue9656522010-08-17 15:40:09 +020011621 The first difference between these rules and "tcp-request connection" rules
11622 is that "tcp-request content" rules can make use of contents to take a
11623 decision. Most often, these decisions will consider a protocol recognition or
11624 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +010011625 both frontends and backends. In case of HTTP keep-alive with the client, all
11626 tcp-request content rules are evaluated again, so haproxy keeps a record of
11627 what sticky counters were assigned by a "tcp-request connection" versus a
11628 "tcp-request content" rule, and flushes all the content-related ones after
11629 processing an HTTP request, so that they may be evaluated again by the rules
11630 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011631 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +010011632 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +020011633
Willy Tarreaue9656522010-08-17 15:40:09 +020011634 Content-based rules are evaluated in their exact declaration order. If no
11635 rule matches or if there is no rule, the default action is to accept the
11636 contents. There is no specific limit to the number of rules which may be
11637 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +020011638
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011639 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011640 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +010011641 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +020011642 - reject : the request is rejected and the connection is closed
11643 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -040011644 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011645 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011646 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011647 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011648 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011649 - set-dst <expr>
11650 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011651 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010011652 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +020011653 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +010011654 - send-spoe-group <engine-name> <group-name>
Christopher Faulet579d83b2019-11-22 15:34:17 +010011655 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +020011656
Willy Tarreaue9656522010-08-17 15:40:09 +020011657 They have the same meaning as their counter-parts in "tcp-request connection"
11658 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +010011659 For "do-resolve" action, please check the "http-request do-resolve"
11660 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +020011661
Willy Tarreauf3338342014-01-28 21:40:28 +010011662 While there is nothing mandatory about it, it is recommended to use the
11663 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
11664 content" rules in the frontend, and track-sc2 for "tcp-request content"
11665 rules in the backend, because that makes the configuration more readable
11666 and easier to troubleshoot, but this is just a guideline and all counters
11667 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +020011668
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011669 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +020011670 the action, it is simply performed unconditionally. That can be useful for
11671 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +020011672
Christopher Faulet2079a4a2020-10-02 11:48:57 +020011673 Note also that it is recommended to use a "tcp-request session" rule to track
11674 information that does *not* depend on Layer 7 contents, especially for HTTP
11675 frontends. Some HTTP processing are performed at the session level and may
11676 lead to an early rejection of the requests. Thus, the tracking at the content
11677 level may be disturbed in such case. A warning is emitted during startup to
11678 prevent, as far as possible, such unreliable usage.
11679
Willy Tarreaue9656522010-08-17 15:40:09 +020011680 It is perfectly possible to match layer 7 contents with "tcp-request content"
Christopher Faulet7ea509e2020-10-02 11:38:46 +020011681 rules from a TCP proxy, since HTTP-specific ACL matches are able to
11682 preliminarily parse the contents of a buffer before extracting the required
11683 data. If the buffered contents do not parse as a valid HTTP message, then the
11684 ACL does not match. The parser which is involved there is exactly the same
11685 as for all other HTTP processing, so there is no risk of parsing something
11686 differently. In an HTTP frontend or an HTTP backend, it is guaranteed that
11687 HTTP contents will always be immediately present when the rule is evaluated
11688 first because the HTTP parsing is performed in the early stages of the
11689 connection processing, at the session level. But for such proxies, using
11690 "http-request" rules is much more natural and recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020011691
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011692 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011693 are present when the rule is processed. The rule processing engine is able to
11694 wait until the inspect delay expires when the data to be tracked is not yet
11695 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011696
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020011697 The "set-dst" and "set-dst-port" are used to set respectively the destination
11698 IP and port. More information on how to use it at "http-request set-dst".
11699
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011700 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +020011701 declared inline. For "tcp-request session" rules, only session-level
11702 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011703
Daniel Schneller0b547052016-03-21 20:46:57 +010011704 <var-name> The name of the variable starts with an indication about
11705 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010011706 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010011707 "sess" : the variable is shared with the whole session
11708 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011709 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010011710 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011711 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010011712 "res" : the variable is shared only during response
11713 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011714 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010011715 The name may only contain characters 'a-z', 'A-Z', '0-9',
11716 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011717
11718 <expr> Is a standard HAProxy expression formed by a sample-fetch
11719 followed by some converters.
11720
Christopher Faulet85d79c92016-11-09 16:54:56 +010011721 The "unset-var" is used to unset a variable. See above for details about
11722 <var-name>.
11723
Patrick Hemmer268a7072018-05-11 12:52:31 -040011724 The "set-priority-class" is used to set the queue priority class of the
11725 current request. The value must be a sample expression which converts to an
11726 integer in the range -2047..2047. Results outside this range will be
11727 truncated. The priority class determines the order in which queued requests
11728 are processed. Lower values have higher priority.
11729
11730 The "set-priority-offset" is used to set the queue priority timestamp offset
11731 of the current request. The value must be a sample expression which converts
11732 to an integer in the range -524287..524287. Results outside this range will be
11733 truncated. When a request is queued, it is ordered first by the priority
11734 class, then by the current timestamp adjusted by the given offset in
11735 milliseconds. Lower values have higher priority.
11736 Note that the resulting timestamp is is only tracked with enough precision for
11737 524,287ms (8m44s287ms). If the request is queued long enough to where the
11738 adjusted timestamp exceeds this value, it will be misidentified as highest
11739 priority. Thus it is important to set "timeout queue" to a value, where when
11740 combined with the offset, does not exceed this limit.
11741
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011742 The "send-spoe-group" is used to trigger sending of a group of SPOE
11743 messages. To do so, the SPOE engine used to send messages must be defined, as
11744 well as the SPOE group to send. Of course, the SPOE engine must refer to an
11745 existing SPOE filter. If not engine name is provided on the SPOE filter line,
11746 the SPOE agent name must be used.
11747
11748 <engine-name> The SPOE engine name.
11749
11750 <group-name> The SPOE group name as specified in the engine configuration.
11751
Christopher Faulet579d83b2019-11-22 15:34:17 +010011752 The "use-service" is used to executes a TCP service which will reply to the
11753 request and stop the evaluation of the rules. This service may choose to
11754 reply by sending any valid response or it may immediately close the
11755 connection without sending anything. Outside natives services, it is possible
11756 to write your own services in Lua. No further "tcp-request" rules are
11757 evaluated.
11758
11759 Example:
11760 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
11761
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011762 Example:
11763
11764 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +010011765 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011766
Willy Tarreau62644772008-07-16 18:36:06 +020011767 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +020011768 # Accept HTTP requests containing a Host header saying "example.com"
11769 # and reject everything else.
11770 acl is_host_com hdr(Host) -i example.com
11771 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +020011772 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +020011773 tcp-request content reject
11774
11775 Example:
Willy Tarreau62644772008-07-16 18:36:06 +020011776 # reject SMTP connection if client speaks first
11777 tcp-request inspect-delay 30s
11778 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011779 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +020011780
11781 # Forward HTTPS connection only if client speaks
11782 tcp-request inspect-delay 30s
11783 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020011784 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +020011785 tcp-request content reject
11786
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011787 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011788 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011789 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011790 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011791 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
11792 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011793
11794 Example:
11795 # track request counts per "base" (concatenation of Host+URL)
11796 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020011797 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010011798
Willy Tarreaue9656522010-08-17 15:40:09 +020011799 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011800 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +020011801
11802 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +010011803 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +020011804 # protecting all our sites
11805 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011806 tcp-request connection track-sc0 src
11807 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +020011808 ...
11809 use_backend http_dynamic if { path_end .php }
11810
11811 backend http_dynamic
11812 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011813 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +020011814 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011815 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011816 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020011817 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +020011818 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +020011819
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011820 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +020011821
Jarno Huuskonen95b012b2017-04-06 13:59:14 +030011822 See also : "tcp-request connection", "tcp-request session",
11823 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +020011824
11825
11826tcp-request inspect-delay <timeout>
11827 Set the maximum allowed time to wait for data during content inspection
11828 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020011829 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +020011830 Arguments :
11831 <timeout> is the timeout value specified in milliseconds by default, but
11832 can be in any other unit if the number is suffixed by the unit,
11833 as explained at the top of this document.
11834
11835 People using haproxy primarily as a TCP relay are often worried about the
11836 risk of passing any type of protocol to a server without any analysis. In
11837 order to be able to analyze the request contents, we must first withhold
11838 the data then analyze them. This statement simply enables withholding of
11839 data for at most the specified amount of time.
11840
Willy Tarreaufb356202010-08-03 14:02:05 +020011841 TCP content inspection applies very early when a connection reaches a
11842 frontend, then very early when the connection is forwarded to a backend. This
11843 means that a connection may experience a first delay in the frontend and a
11844 second delay in the backend if both have tcp-request rules.
11845
Willy Tarreau62644772008-07-16 18:36:06 +020011846 Note that when performing content inspection, haproxy will evaluate the whole
11847 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011848 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +020011849 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +010011850 contents are definitive. If no delay is set, haproxy will not wait at all
11851 and will immediately apply a verdict based on the available information.
11852 Obviously this is unlikely to be very useful and might even be racy, so such
11853 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020011854
11855 As soon as a rule matches, the request is released and continues as usual. If
11856 the timeout is reached and no rule matches, the default policy will be to let
11857 it pass through unaffected.
11858
11859 For most protocols, it is enough to set it to a few seconds, as most clients
11860 send the full request immediately upon connection. Add 3 or more seconds to
11861 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +010011862 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +010011863 before the server (e.g. SMTP), or to wait for a client to talk before passing
11864 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +020011865 least the inspection delay, otherwise it will expire first. If the client
11866 closes the connection or if the buffer is full, the delay immediately expires
11867 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +020011868
Willy Tarreau55165fe2009-05-10 12:02:55 +020011869 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020011870 "timeout client".
11871
11872
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011873tcp-response content <action> [{if | unless} <condition>]
11874 Perform an action on a session response depending on a layer 4-7 condition
11875 May be used in sections : defaults | frontend | listen | backend
11876 no | no | yes | yes
11877 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020011878 <action> defines the action to perform if the condition applies. See
11879 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011880
11881 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
11882
Davor Ocelice9ed2812017-12-25 17:49:28 +010011883 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011884 called "TCP content inspection". During this stage, ACL-based rules are
11885 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011886 "accept", "close" or a "reject" rule matches, or a TCP response inspection
11887 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011888
11889 Most often, these decisions will consider a protocol recognition or validity.
11890
11891 Content-based rules are evaluated in their exact declaration order. If no
11892 rule matches or if there is no rule, the default action is to accept the
11893 contents. There is no specific limit to the number of rules which may be
11894 inserted.
11895
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011896 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011897 - accept :
11898 accepts the response if the condition is true (when used with "if")
11899 or false (when used with "unless"). The first such rule executed ends
11900 the rules evaluation.
11901
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011902 - close :
11903 immediately closes the connection with the server if the condition is
11904 true (when used with "if"), or false (when used with "unless"). The
11905 first such rule executed ends the rules evaluation. The main purpose of
11906 this action is to force a connection to be finished between a client
11907 and a server after an exchange when the application protocol expects
11908 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011909 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020011910 protocols.
11911
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011912 - reject :
11913 rejects the response if the condition is true (when used with "if")
11914 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011915 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011916
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011917 - set-var(<var-name>) <expr>
11918 Sets a variable.
11919
Christopher Faulet85d79c92016-11-09 16:54:56 +010011920 - unset-var(<var-name>)
11921 Unsets a variable.
11922
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020011923 - sc-inc-gpc0(<sc-id>):
11924 This action increments the GPC0 counter according to the sticky
11925 counter designated by <sc-id>. If an error occurs, this action fails
11926 silently and the actions evaluation continues.
11927
Frédéric Lécaille6778b272018-01-29 15:22:53 +010011928 - sc-inc-gpc1(<sc-id>):
11929 This action increments the GPC1 counter according to the sticky
11930 counter designated by <sc-id>. If an error occurs, this action fails
11931 silently and the actions evaluation continues.
11932
Cédric Dufour0d7712d2019-11-06 18:38:53 +010011933 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
11934 This action sets the 32-bit unsigned GPT0 tag according to the sticky
11935 counter designated by <sc-id> and the value of <int>/<expr>. The
11936 expected result is a boolean. If an error occurs, this action silently
11937 fails and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +020011938
Willy Tarreau2d392c22015-08-24 01:43:45 +020011939 - "silent-drop" :
11940 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011941 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020011942 to prevent the client from being notified. The effect it then that the
11943 client still sees an established connection while there's none on
11944 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
11945 except that it doesn't use any local resource at all on the machine
11946 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010011947 slow down stronger attackers. It is important to understand the impact
11948 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011949 client and HAProxy (firewalls, proxies, load balancers) will also keep
11950 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010011951 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020011952 TCP_REPAIR socket option is used to block the emission of a TCP
11953 reset. On other systems, the socket's TTL is reduced to 1 so that the
11954 TCP reset doesn't pass the first router, though it's still delivered to
11955 local networks. Do not use it unless you fully understand how it works.
11956
Christopher Faulet76c09ef2017-09-21 11:03:52 +020011957 - send-spoe-group <engine-name> <group-name>
11958 Send a group of SPOE messages.
11959
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011960 Note that the "if/unless" condition is optional. If no condition is set on
11961 the action, it is simply performed unconditionally. That can be useful for
11962 for changing the default action to a reject.
11963
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011964 It is perfectly possible to match layer 7 contents with "tcp-response
11965 content" rules, but then it is important to ensure that a full response has
11966 been buffered, otherwise no contents will match. In order to achieve this,
11967 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020011968 period.
11969
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011970 The "set-var" is used to set the content of a variable. The variable is
11971 declared inline.
11972
Daniel Schneller0b547052016-03-21 20:46:57 +010011973 <var-name> The name of the variable starts with an indication about
11974 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010011975 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010011976 "sess" : the variable is shared with the whole session
11977 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011978 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010011979 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011980 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010011981 "res" : the variable is shared only during response
11982 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011983 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010011984 The name may only contain characters 'a-z', 'A-Z', '0-9',
11985 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020011986
11987 <expr> Is a standard HAProxy expression formed by a sample-fetch
11988 followed by some converters.
11989
11990 Example:
11991
11992 tcp-request content set-var(sess.my_var) src
11993
Christopher Faulet85d79c92016-11-09 16:54:56 +010011994 The "unset-var" is used to unset a variable. See above for details about
11995 <var-name>.
11996
11997 Example:
11998
11999 tcp-request content unset-var(sess.my_var)
12000
Christopher Faulet76c09ef2017-09-21 11:03:52 +020012001 The "send-spoe-group" is used to trigger sending of a group of SPOE
12002 messages. To do so, the SPOE engine used to send messages must be defined, as
12003 well as the SPOE group to send. Of course, the SPOE engine must refer to an
12004 existing SPOE filter. If not engine name is provided on the SPOE filter line,
12005 the SPOE agent name must be used.
12006
12007 <engine-name> The SPOE engine name.
12008
12009 <group-name> The SPOE group name as specified in the engine configuration.
12010
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012011 See section 7 about ACL usage.
12012
12013 See also : "tcp-request content", "tcp-response inspect-delay"
12014
12015
Willy Tarreau4f614292016-10-21 17:49:36 +020012016tcp-request session <action> [{if | unless} <condition>]
12017 Perform an action on a validated session depending on a layer 5 condition
12018 May be used in sections : defaults | frontend | listen | backend
12019 no | yes | yes | no
12020 Arguments :
12021 <action> defines the action to perform if the condition applies. See
12022 below.
12023
12024 <condition> is a standard layer5-only ACL-based condition (see section 7).
12025
Davor Ocelice9ed2812017-12-25 17:49:28 +010012026 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020012027 it is possible to evaluate some conditions to decide whether this session
12028 must be accepted or dropped or have its counters tracked. Those conditions
12029 cannot make use of any data contents because no buffers are allocated yet and
12030 the processing cannot wait at this stage. The main use case it to copy some
12031 early information into variables (since variables are accessible in the
12032 session), or to keep track of some information collected after the handshake,
12033 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010012034 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020012035 extracted information can thus be copied to a variable or tracked using
12036 "track-sc" rules. Of course it is also possible to decide to accept/reject as
12037 with other rulesets. Most operations performed here could also be performed
12038 in "tcp-request content" rules, except that in HTTP these rules are evaluated
12039 for each new request, and that might not always be acceptable. For example a
12040 rule might increment a counter on each evaluation. It would also be possible
12041 that a country is resolved by geolocation from the source IP address,
12042 assigned to a session-wide variable, then the source address rewritten from
12043 an HTTP header for all requests. If some contents need to be inspected in
12044 order to take the decision, the "tcp-request content" statements must be used
12045 instead.
12046
12047 The "tcp-request session" rules are evaluated in their exact declaration
12048 order. If no rule matches or if there is no rule, the default action is to
12049 accept the incoming session. There is no specific limit to the number of
12050 rules which may be inserted.
12051
12052 Several types of actions are supported :
12053 - accept : the request is accepted
12054 - reject : the request is rejected and the connection is closed
12055 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
12056 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010012057 - sc-inc-gpc1(<sc-id>)
Cédric Dufour0d7712d2019-11-06 18:38:53 +010012058 - sc-set-gpt0(<sc-id>) { <int> | <expr> }
Willy Tarreau4f614292016-10-21 17:49:36 +020012059 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010012060 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020012061 - silent-drop
12062
12063 These actions have the same meaning as their respective counter-parts in
12064 "tcp-request connection" and "tcp-request content", so please refer to these
12065 sections for a complete description.
12066
12067 Note that the "if/unless" condition is optional. If no condition is set on
12068 the action, it is simply performed unconditionally. That can be useful for
12069 "track-sc*" actions as well as for changing the default action to a reject.
12070
12071 Example: track the original source address by default, or the one advertised
12072 in the PROXY protocol header for connection coming from the local
12073 proxies. The first connection-level rule enables receipt of the
12074 PROXY protocol for these ones, the second rule tracks whatever
12075 address we decide to keep after optional decoding.
12076
12077 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
12078 tcp-request session track-sc0 src
12079
12080 Example: accept all sessions from white-listed hosts, reject too fast
12081 sessions without counting them, and track accepted sessions.
12082 This results in session rate being capped from abusive sources.
12083
12084 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
12085 tcp-request session reject if { src_sess_rate gt 10 }
12086 tcp-request session track-sc0 src
12087
12088 Example: accept all sessions from white-listed hosts, count all other
12089 sessions and reject too fast ones. This results in abusive ones
12090 being blocked as long as they don't slow down.
12091
12092 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
12093 tcp-request session track-sc0 src
12094 tcp-request session reject if { sc0_sess_rate gt 10 }
12095
12096 See section 7 about ACL usage.
12097
12098 See also : "tcp-request connection", "tcp-request content", "stick-table"
12099
12100
Emeric Brun0a3b67f2010-09-24 15:34:53 +020012101tcp-response inspect-delay <timeout>
12102 Set the maximum allowed time to wait for a response during content inspection
12103 May be used in sections : defaults | frontend | listen | backend
12104 no | no | yes | yes
12105 Arguments :
12106 <timeout> is the timeout value specified in milliseconds by default, but
12107 can be in any other unit if the number is suffixed by the unit,
12108 as explained at the top of this document.
12109
12110 See also : "tcp-response content", "tcp-request inspect-delay".
12111
12112
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012113timeout check <timeout>
12114 Set additional check timeout, but only after a connection has been already
12115 established.
12116
12117 May be used in sections: defaults | frontend | listen | backend
12118 yes | no | yes | yes
12119 Arguments:
12120 <timeout> is the timeout value specified in milliseconds by default, but
12121 can be in any other unit if the number is suffixed by the unit,
12122 as explained at the top of this document.
12123
12124 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
12125 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010012126 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012127 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010012128 (Please also note that there is no valid reason to have such long connect
12129 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
12130 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012131
12132 If "timeout check" is not set haproxy uses "inter" for complete check
12133 timeout (connect + read) exactly like all <1.3.15 version.
12134
12135 In most cases check request is much simpler and faster to handle than normal
12136 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010012137 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012138
12139 This parameter is specific to backends, but can be specified once for all in
12140 "defaults" sections. This is in fact one of the easiest solutions not to
12141 forget about it.
12142
Willy Tarreau41a340d2008-01-22 12:25:31 +010012143 See also: "timeout connect", "timeout queue", "timeout server",
12144 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012145
12146
Willy Tarreau0ba27502007-12-24 16:55:16 +010012147timeout client <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010012148 Set the maximum inactivity time on the client side.
12149 May be used in sections : defaults | frontend | listen | backend
12150 yes | yes | yes | no
12151 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012152 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010012153 can be in any other unit if the number is suffixed by the unit,
12154 as explained at the top of this document.
12155
12156 The inactivity timeout applies when the client is expected to acknowledge or
12157 send data. In HTTP mode, this timeout is particularly important to consider
12158 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010012159 response while it is reading data sent by the server. That said, for the
12160 first phase, it is preferable to set the "timeout http-request" to better
12161 protect HAProxy from Slowloris like attacks. The value is specified in
12162 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010012163 suffixed by the unit, as specified at the top of this document. In TCP mode
12164 (and to a lesser extent, in HTTP mode), it is highly recommended that the
12165 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012166 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010012167 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010012168 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
12169 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020012170 which overrides "timeout client" and "timeout server" for tunnels, as well as
12171 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012172
12173 This parameter is specific to frontends, but can be specified once for all in
12174 "defaults" sections. This is in fact one of the easiest solutions not to
12175 forget about it. An unspecified timeout results in an infinite timeout, which
12176 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012177 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010012178 the system if the system's timeouts are not configured either.
12179
Willy Tarreau95c4e142017-11-26 12:18:55 +010012180 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010012181
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012182 See also : "timeout server", "timeout tunnel", "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012183
Willy Tarreau0ba27502007-12-24 16:55:16 +010012184
Willy Tarreau05cdd962014-05-10 14:30:07 +020012185timeout client-fin <timeout>
12186 Set the inactivity timeout on the client side for half-closed connections.
12187 May be used in sections : defaults | frontend | listen | backend
12188 yes | yes | yes | no
12189 Arguments :
12190 <timeout> is the timeout value specified in milliseconds by default, but
12191 can be in any other unit if the number is suffixed by the unit,
12192 as explained at the top of this document.
12193
12194 The inactivity timeout applies when the client is expected to acknowledge or
12195 send data while one direction is already shut down. This timeout is different
12196 from "timeout client" in that it only applies to connections which are closed
12197 in one direction. This is particularly useful to avoid keeping connections in
12198 FIN_WAIT state for too long when clients do not disconnect cleanly. This
12199 problem is particularly common long connections such as RDP or WebSocket.
12200 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010012201 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
12202 frame was sent, often indicating an expectation that the connection quickly
12203 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020012204
12205 This parameter is specific to frontends, but can be specified once for all in
12206 "defaults" sections. By default it is not set, so half-closed connections
12207 will use the other timeouts (timeout.client or timeout.tunnel).
12208
12209 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
12210
12211
Willy Tarreau0ba27502007-12-24 16:55:16 +010012212timeout connect <timeout>
Willy Tarreau0ba27502007-12-24 16:55:16 +010012213 Set the maximum time to wait for a connection attempt to a server to succeed.
12214 May be used in sections : defaults | frontend | listen | backend
12215 yes | no | yes | yes
12216 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012217 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010012218 can be in any other unit if the number is suffixed by the unit,
12219 as explained at the top of this document.
12220
12221 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012222 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010012223 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010012224 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010012225 connect timeout also presets both queue and tarpit timeouts to the same value
12226 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012227
12228 This parameter is specific to backends, but can be specified once for all in
12229 "defaults" sections. This is in fact one of the easiest solutions not to
12230 forget about it. An unspecified timeout results in an infinite timeout, which
12231 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012232 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010012233 the system if the system's timeouts are not configured either.
12234
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012235 See also: "timeout check", "timeout queue", "timeout server", "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012236
Willy Tarreau0ba27502007-12-24 16:55:16 +010012237
Willy Tarreaub16a5742010-01-10 14:46:16 +010012238timeout http-keep-alive <timeout>
12239 Set the maximum allowed time to wait for a new HTTP request to appear
12240 May be used in sections : defaults | frontend | listen | backend
12241 yes | yes | yes | yes
12242 Arguments :
12243 <timeout> is the timeout value specified in milliseconds by default, but
12244 can be in any other unit if the number is suffixed by the unit,
12245 as explained at the top of this document.
12246
12247 By default, the time to wait for a new request in case of keep-alive is set
12248 by "timeout http-request". However this is not always convenient because some
12249 people want very short keep-alive timeouts in order to release connections
12250 faster, and others prefer to have larger ones but still have short timeouts
12251 once the request has started to present itself.
12252
12253 The "http-keep-alive" timeout covers these needs. It will define how long to
12254 wait for a new HTTP request to start coming after a response was sent. Once
12255 the first byte of request has been seen, the "http-request" timeout is used
12256 to wait for the complete request to come. Note that empty lines prior to a
12257 new request do not refresh the timeout and are not counted as a new request.
12258
12259 There is also another difference between the two timeouts : when a connection
12260 expires during timeout http-keep-alive, no error is returned, the connection
12261 just closes. If the connection expires in "http-request" while waiting for a
12262 connection to complete, a HTTP 408 error is returned.
12263
12264 In general it is optimal to set this value to a few tens to hundreds of
12265 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010012266 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010012267 1 millisecond) it will probably only accept pipelined requests but not the
12268 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020012269 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010012270
12271 If this parameter is not set, the "http-request" timeout applies, and if both
12272 are not set, "timeout client" still applies at the lower level. It should be
12273 set in the frontend to take effect, unless the frontend is in TCP mode, in
12274 which case the HTTP backend's timeout will be used.
12275
Willy Tarreau95c4e142017-11-26 12:18:55 +010012276 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
12277 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010012278 (where we only have one connection per client and a connection setup).
12279
Willy Tarreaub16a5742010-01-10 14:46:16 +010012280 See also : "timeout http-request", "timeout client".
12281
12282
Willy Tarreau036fae02008-01-06 13:24:40 +010012283timeout http-request <timeout>
12284 Set the maximum allowed time to wait for a complete HTTP request
12285 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012286 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010012287 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010012288 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010012289 can be in any other unit if the number is suffixed by the unit,
12290 as explained at the top of this document.
12291
12292 In order to offer DoS protection, it may be required to lower the maximum
12293 accepted time to receive a complete HTTP request without affecting the client
12294 timeout. This helps protecting against established connections on which
12295 nothing is sent. The client timeout cannot offer a good protection against
12296 this abuse because it is an inactivity timeout, which means that if the
12297 attacker sends one character every now and then, the timeout will not
12298 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020012299 types, the request will be aborted if it does not complete in time. When the
12300 timeout expires, an HTTP 408 response is sent to the client to inform it
12301 about the problem, and the connection is closed. The logs will report
12302 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010012303 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020012304 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
12305 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010012306
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012307 By default, this timeout only applies to the header part of the request,
12308 and not to any data. As soon as the empty line is received, this timeout is
12309 not used anymore. When combined with "option http-buffer-request", this
12310 timeout also applies to the body of the request..
12311 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010012312 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010012313
12314 Generally it is enough to set it to a few seconds, as most clients send the
12315 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010012316 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010012317 generally work on local networks as long as there are no packet losses. This
12318 will prevent people from sending bare HTTP requests using telnet.
12319
12320 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020012321 chunk of the incoming request. It should be set in the frontend to take
12322 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
12323 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010012324
Willy Tarreau0f228a02015-05-01 15:37:53 +020012325 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010012326 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010012327
Willy Tarreau844e3c52008-01-11 16:28:18 +010012328
12329timeout queue <timeout>
12330 Set the maximum time to wait in the queue for a connection slot to be free
12331 May be used in sections : defaults | frontend | listen | backend
12332 yes | no | yes | yes
12333 Arguments :
12334 <timeout> is the timeout value specified in milliseconds by default, but
12335 can be in any other unit if the number is suffixed by the unit,
12336 as explained at the top of this document.
12337
12338 When a server's maxconn is reached, connections are left pending in a queue
12339 which may be server-specific or global to the backend. In order not to wait
12340 indefinitely, a timeout is applied to requests pending in the queue. If the
12341 timeout is reached, it is considered that the request will almost never be
12342 served, so it is dropped and a 503 error is returned to the client.
12343
12344 The "timeout queue" statement allows to fix the maximum time for a request to
12345 be left pending in a queue. If unspecified, the same value as the backend's
12346 connection timeout ("timeout connect") is used, for backwards compatibility
12347 with older versions with no "timeout queue" parameter.
12348
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012349 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012350
12351
12352timeout server <timeout>
Willy Tarreau844e3c52008-01-11 16:28:18 +010012353 Set the maximum inactivity time on the server side.
12354 May be used in sections : defaults | frontend | listen | backend
12355 yes | no | yes | yes
12356 Arguments :
12357 <timeout> is the timeout value specified in milliseconds by default, but
12358 can be in any other unit if the number is suffixed by the unit,
12359 as explained at the top of this document.
12360
12361 The inactivity timeout applies when the server is expected to acknowledge or
12362 send data. In HTTP mode, this timeout is particularly important to consider
12363 during the first phase of the server's response, when it has to send the
12364 headers, as it directly represents the server's processing time for the
12365 request. To find out what value to put there, it's often good to start with
12366 what would be considered as unacceptable response times, then check the logs
12367 to observe the response time distribution, and adjust the value accordingly.
12368
12369 The value is specified in milliseconds by default, but can be in any other
12370 unit if the number is suffixed by the unit, as specified at the top of this
12371 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
12372 recommended that the client timeout remains equal to the server timeout in
12373 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010012374 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010012375 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010012376 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
12377 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020012378 "timeout tunnel", which overrides "timeout client" and "timeout server" for
12379 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012380
12381 This parameter is specific to backends, but can be specified once for all in
12382 "defaults" sections. This is in fact one of the easiest solutions not to
12383 forget about it. An unspecified timeout results in an infinite timeout, which
12384 is not recommended. Such a usage is accepted and works but reports a warning
John Roeslerfb2fce12019-07-10 15:45:51 -050012385 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010012386 the system if the system's timeouts are not configured either.
12387
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012388 See also : "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012389
Willy Tarreau05cdd962014-05-10 14:30:07 +020012390
12391timeout server-fin <timeout>
12392 Set the inactivity timeout on the server side for half-closed connections.
12393 May be used in sections : defaults | frontend | listen | backend
12394 yes | no | yes | yes
12395 Arguments :
12396 <timeout> is the timeout value specified in milliseconds by default, but
12397 can be in any other unit if the number is suffixed by the unit,
12398 as explained at the top of this document.
12399
12400 The inactivity timeout applies when the server is expected to acknowledge or
12401 send data while one direction is already shut down. This timeout is different
12402 from "timeout server" in that it only applies to connections which are closed
12403 in one direction. This is particularly useful to avoid keeping connections in
12404 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
12405 This problem is particularly common long connections such as RDP or WebSocket.
12406 Note that this timeout can override "timeout tunnel" when a connection shuts
12407 down in one direction. This setting was provided for completeness, but in most
12408 situations, it should not be needed.
12409
12410 This parameter is specific to backends, but can be specified once for all in
12411 "defaults" sections. By default it is not set, so half-closed connections
12412 will use the other timeouts (timeout.server or timeout.tunnel).
12413
12414 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
12415
Willy Tarreau844e3c52008-01-11 16:28:18 +010012416
12417timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010012418 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010012419 May be used in sections : defaults | frontend | listen | backend
12420 yes | yes | yes | yes
12421 Arguments :
12422 <timeout> is the tarpit duration specified in milliseconds by default, but
12423 can be in any other unit if the number is suffixed by the unit,
12424 as explained at the top of this document.
12425
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020012426 When a connection is tarpitted using "http-request tarpit", it is maintained
12427 open with no activity for a certain amount of time, then closed. "timeout
12428 tarpit" defines how long it will be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012429
12430 The value is specified in milliseconds by default, but can be in any other
12431 unit if the number is suffixed by the unit, as specified at the top of this
12432 document. If unspecified, the same value as the backend's connection timeout
12433 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010012434 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012435
Tim Duesterhus86e6b6e2019-05-14 20:57:59 +020012436 See also : "timeout connect".
Willy Tarreau844e3c52008-01-11 16:28:18 +010012437
12438
Willy Tarreauce887fd2012-05-12 12:50:00 +020012439timeout tunnel <timeout>
12440 Set the maximum inactivity time on the client and server side for tunnels.
12441 May be used in sections : defaults | frontend | listen | backend
12442 yes | no | yes | yes
12443 Arguments :
12444 <timeout> is the timeout value specified in milliseconds by default, but
12445 can be in any other unit if the number is suffixed by the unit,
12446 as explained at the top of this document.
12447
Jamie Gloudonaaa21002012-08-25 00:18:33 -040012448 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020012449 between a client and a server, and the connection remains inactive in both
12450 directions. This timeout supersedes both the client and server timeouts once
12451 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012452 analyzer remains attached to either connection (e.g. tcp content rules are
12453 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020012454 when switching to the WebSocket protocol, or forwarding a CONNECT request
12455 to a proxy), or after the first response when no keepalive/close option is
12456 specified.
12457
Willy Tarreau05cdd962014-05-10 14:30:07 +020012458 Since this timeout is usually used in conjunction with long-lived connections,
12459 it usually is a good idea to also set "timeout client-fin" to handle the
12460 situation where a client suddenly disappears from the net and does not
12461 acknowledge a close, or sends a shutdown and does not acknowledge pending
12462 data anymore. This can happen in lossy networks where firewalls are present,
12463 and is detected by the presence of large amounts of sessions in a FIN_WAIT
12464 state.
12465
Willy Tarreauce887fd2012-05-12 12:50:00 +020012466 The value is specified in milliseconds by default, but can be in any other
12467 unit if the number is suffixed by the unit, as specified at the top of this
12468 document. Whatever the expected normal idle time, it is a good practice to
12469 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010012470 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020012471
12472 This parameter is specific to backends, but can be specified once for all in
12473 "defaults" sections. This is in fact one of the easiest solutions not to
12474 forget about it.
12475
12476 Example :
12477 defaults http
12478 option http-server-close
12479 timeout connect 5s
12480 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020012481 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020012482 timeout server 30s
12483 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
12484
Willy Tarreau05cdd962014-05-10 14:30:07 +020012485 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020012486
12487
Willy Tarreau844e3c52008-01-11 16:28:18 +010012488transparent (deprecated)
12489 Enable client-side transparent proxying
12490 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010012491 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010012492 Arguments : none
12493
12494 This keyword was introduced in order to provide layer 7 persistence to layer
12495 3 load balancers. The idea is to use the OS's ability to redirect an incoming
12496 connection for a remote address to a local process (here HAProxy), and let
12497 this process know what address was initially requested. When this option is
12498 used, sessions without cookies will be forwarded to the original destination
12499 IP address of the incoming request (which should match that of another
12500 equipment), while requests with cookies will still be forwarded to the
12501 appropriate server.
12502
12503 The "transparent" keyword is deprecated, use "option transparent" instead.
12504
12505 Note that contrary to a common belief, this option does NOT make HAProxy
12506 present the client's IP to the server when establishing the connection.
12507
Willy Tarreau844e3c52008-01-11 16:28:18 +010012508 See also: "option transparent"
12509
William Lallemanda73203e2012-03-12 12:48:57 +010012510unique-id-format <string>
12511 Generate a unique ID for each request.
12512 May be used in sections : defaults | frontend | listen | backend
12513 yes | yes | yes | no
12514 Arguments :
12515 <string> is a log-format string.
12516
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012517 This keyword creates a ID for each request using the custom log format. A
12518 unique ID is useful to trace a request passing through many components of
12519 a complex infrastructure. The newly created ID may also be logged using the
12520 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010012521
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012522 The format should be composed from elements that are guaranteed to be
12523 unique when combined together. For instance, if multiple haproxy instances
12524 are involved, it might be important to include the node name. It is often
12525 needed to log the incoming connection's source and destination addresses
12526 and ports. Note that since multiple requests may be performed over the same
12527 connection, including a request counter may help differentiate them.
12528 Similarly, a timestamp may protect against a rollover of the counter.
12529 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010012530
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012531 It is recommended to use hexadecimal notation for many fields since it
12532 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010012533
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012534 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012535
Julien Vehentf21be322014-03-07 08:27:34 -050012536 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012537
12538 will generate:
12539
12540 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12541
12542 See also: "unique-id-header"
12543
12544unique-id-header <name>
12545 Add a unique ID header in the HTTP request.
12546 May be used in sections : defaults | frontend | listen | backend
12547 yes | yes | yes | no
12548 Arguments :
12549 <name> is the name of the header.
12550
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012551 Add a unique-id header in the HTTP request sent to the server, using the
12552 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010012553
Cyril Bonté108cf6e2012-04-21 23:30:29 +020012554 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010012555
Julien Vehentf21be322014-03-07 08:27:34 -050012556 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010012557 unique-id-header X-Unique-ID
12558
12559 will generate:
12560
12561 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
12562
12563 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010012564
Willy Tarreauf51658d2014-04-23 01:21:56 +020012565use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012566 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012567 May be used in sections : defaults | frontend | listen | backend
12568 no | yes | yes | no
12569 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012570 <backend> is the name of a valid backend or "listen" section, or a
12571 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012572
Willy Tarreauf51658d2014-04-23 01:21:56 +020012573 <condition> is a condition composed of ACLs, as described in section 7. If
12574 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012575
12576 When doing content-switching, connections arrive on a frontend and are then
12577 dispatched to various backends depending on a number of conditions. The
12578 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012579 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010012580 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020012581 source address validation) or combined with a "tcp-request" rule to wait for
12582 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010012583
12584 There may be as many "use_backend" rules as desired. All of these rules are
12585 evaluated in their declaration order, and the first one which matches will
12586 assign the backend.
12587
12588 In the first form, the backend will be used if the condition is met. In the
12589 second form, the backend will be used if the condition is not met. If no
12590 condition is valid, the backend defined with "default_backend" will be used.
12591 If no default backend is defined, either the servers in the same section are
12592 used (in case of a "listen" section) or, in case of a frontend, no server is
12593 used and a 503 service unavailable response is returned.
12594
Willy Tarreau51aecc72009-07-12 09:47:04 +020012595 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012596 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020012597 and backend processing will immediately follow, or the backend will wait for
12598 a complete HTTP request to get in. This feature is useful when a frontend
12599 must decode several protocols on a unique port, one of them being HTTP.
12600
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012601 When <backend> is a simple name, it is resolved at configuration time, and an
12602 error is reported if the specified backend does not exist. If <backend> is
12603 a log-format string instead, no check may be done at configuration time, so
12604 the backend name is resolved dynamically at run time. If the resulting
12605 backend name does not correspond to any valid backend, no other rule is
12606 evaluated, and the default_backend directive is applied instead. Note that
12607 when using dynamic backend names, it is highly recommended to use a prefix
12608 that no other backend uses in order to ensure that an unauthorized backend
12609 cannot be forced from the request.
12610
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012611 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010012612 used to detect the association between frontends and backends to compute the
12613 backend's "fullconn" setting. This cannot be done for dynamic names.
12614
12615 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
12616 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010012617
Christopher Fauletb30b3102019-09-12 23:03:09 +020012618use-fcgi-app <name>
12619 Defines the FastCGI application to use for the backend.
12620 May be used in sections : defaults | frontend | listen | backend
12621 no | no | yes | yes
12622 Arguments :
12623 <name> is the name of the FastCGI application to use.
12624
12625 See section 10.1 about FastCGI application setup for details.
Willy Tarreau036fae02008-01-06 13:24:40 +010012626
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012627use-server <server> if <condition>
12628use-server <server> unless <condition>
12629 Only use a specific server if/unless an ACL-based condition is matched.
12630 May be used in sections : defaults | frontend | listen | backend
12631 no | no | yes | yes
12632 Arguments :
Jerome Magnin824186b2020-03-29 09:37:12 +020012633 <server> is the name of a valid server in the same backend section
12634 or a "log-format" string resolving to a server name.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012635
12636 <condition> is a condition composed of ACLs, as described in section 7.
12637
12638 By default, connections which arrive to a backend are load-balanced across
12639 the available servers according to the configured algorithm, unless a
12640 persistence mechanism such as a cookie is used and found in the request.
12641
12642 Sometimes it is desirable to forward a particular request to a specific
12643 server without having to declare a dedicated backend for this server. This
12644 can be achieved using the "use-server" rules. These rules are evaluated after
12645 the "redirect" rules and before evaluating cookies, and they have precedence
12646 on them. There may be as many "use-server" rules as desired. All of these
12647 rules are evaluated in their declaration order, and the first one which
12648 matches will assign the server.
12649
12650 If a rule designates a server which is down, and "option persist" is not used
12651 and no force-persist rule was validated, it is ignored and evaluation goes on
12652 with the next rules until one matches.
12653
12654 In the first form, the server will be used if the condition is met. In the
12655 second form, the server will be used if the condition is not met. If no
12656 condition is valid, the processing continues and the server will be assigned
12657 according to other persistence mechanisms.
12658
12659 Note that even if a rule is matched, cookie processing is still performed but
12660 does not assign the server. This allows prefixed cookies to have their prefix
12661 stripped.
12662
12663 The "use-server" statement works both in HTTP and TCP mode. This makes it
12664 suitable for use with content-based inspection. For instance, a server could
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012665 be selected in a farm according to the TLS SNI field when using protocols with
12666 implicit TLS (also see "req_ssl_sni"). And if these servers have their weight
12667 set to zero, they will not be used for other traffic.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012668
12669 Example :
12670 # intercept incoming TLS requests based on the SNI field
12671 use-server www if { req_ssl_sni -i www.example.com }
12672 server www 192.168.0.1:443 weight 0
12673 use-server mail if { req_ssl_sni -i mail.example.com }
Lukas Tribusa267b5d2020-07-19 00:25:06 +020012674 server mail 192.168.0.1:465 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012675 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000012676 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012677 # all the rest is forwarded to this server
12678 server default 192.168.0.2:443 check
12679
Jerome Magnin824186b2020-03-29 09:37:12 +020012680 When <server> is a simple name, it is checked against existing servers in the
12681 configuration and an error is reported if the specified server does not exist.
12682 If it is a log-format, no check is performed when parsing the configuration,
12683 and if we can't resolve a valid server name at runtime but the use-server rule
Ilya Shipitsin11057a32020-06-21 21:18:27 +050012684 was conditioned by an ACL returning true, no other use-server rule is applied
Jerome Magnin824186b2020-03-29 09:37:12 +020012685 and we fall back to load balancing.
12686
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012687 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012688
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012689
Davor Ocelice9ed2812017-12-25 17:49:28 +0100126905. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012691--------------------------
12692
12693The "bind", "server" and "default-server" keywords support a number of settings
12694depending on some build options and on the system HAProxy was built on. These
12695settings generally each consist in one word sometimes followed by a value,
12696written on the same line as the "bind" or "server" line. All these options are
12697described in this section.
12698
12699
127005.1. Bind options
12701-----------------
12702
12703The "bind" keyword supports a certain number of settings which are all passed
12704as arguments on the same line. The order in which those arguments appear makes
12705no importance, provided that they appear after the bind address. All of these
12706parameters are optional. Some of them consist in a single words (booleans),
12707while other ones expect a value after them. In this case, the value must be
12708provided immediately after the setting name.
12709
12710The currently supported settings are the following ones.
12711
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012712accept-netscaler-cip <magic number>
12713 Enforces the use of the NetScaler Client IP insertion protocol over any
12714 connection accepted by any of the TCP sockets declared on the same line. The
12715 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
12716 the incoming connection to be used everywhere an address is used, with the
12717 only exception of "tcp-request connection" rules which will only see the
12718 real connection address. Logs will reflect the addresses indicated in the
12719 protocol, unless it is violated, in which case the real address will still
12720 be used. This keyword combined with support from external components can be
12721 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010012722 mechanism which is not always reliable and not even always usable. See also
12723 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
12724 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012725
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012726accept-proxy
12727 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020012728 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
12729 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012730 3/4 addresses of the incoming connection to be used everywhere an address is
12731 used, with the only exception of "tcp-request connection" rules which will
12732 only see the real connection address. Logs will reflect the addresses
12733 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010012734 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012735 components can be used as an efficient and reliable alternative to the
12736 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020012737 usable. See also "tcp-request connection expect-proxy" for a finer-grained
12738 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012739
Olivier Houchardc2aae742017-09-22 18:26:28 +020012740allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010012741 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010012742 due to security considerations. Because it is vulnerable to replay attacks,
John Roeslerfb2fce12019-07-10 15:45:51 -050012743 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010012744 that are idempotent. You can use the "wait-for-handshake" action for any
12745 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020012746
Willy Tarreauab861d32013-04-02 02:30:41 +020012747alpn <protocols>
12748 This enables the TLS ALPN extension and advertises the specified protocol
12749 list as supported on top of ALPN. The protocol list consists in a comma-
12750 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050012751 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020012752 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010012753 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
12754 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
12755 now obsolete NPN extension. At the time of writing this, most browsers still
12756 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
12757 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
12758 are expected to be supported, both versions can be advertised, in order of
12759 preference, like below :
12760
12761 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020012762
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012763backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010012764 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012765 backlog is used instead, which generally defaults to the maxconn value.
12766
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010012767curves <curves>
12768 This setting is only available when support for OpenSSL was built in. It sets
12769 the string describing the list of elliptic curves algorithms ("curve suite")
12770 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
12771 string is a colon-delimited list of curve name.
12772 Example: "X25519:P-256" (without quote)
12773 When "curves" is set, "ecdhe" parameter is ignored.
12774
Emeric Brun7fb34422012-09-28 15:26:15 +020012775ecdhe <named curve>
12776 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010012777 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
12778 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020012779
Emeric Brunfd33a262012-10-11 16:28:27 +020012780ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020012781 This setting is only available when support for OpenSSL was built in. It
12782 designates a PEM file from which to load CA certificates used to verify
12783 client's certificate.
12784
Emeric Brunb6dc9342012-09-28 17:55:37 +020012785ca-ignore-err [all|<errorID>,...]
12786 This setting is only available when support for OpenSSL was built in.
12787 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
12788 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
12789 error is ignored.
12790
Christopher Faulet31af49d2015-06-09 17:29:50 +020012791ca-sign-file <cafile>
12792 This setting is only available when support for OpenSSL was built in. It
12793 designates a PEM file containing both the CA certificate and the CA private
12794 key used to create and sign server's certificates. This is a mandatory
12795 setting when the dynamic generation of certificates is enabled. See
12796 'generate-certificates' for details.
12797
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000012798ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020012799 This setting is only available when support for OpenSSL was built in. It is
12800 the CA private key passphrase. This setting is optional and used only when
12801 the dynamic generation of certificates is enabled. See
12802 'generate-certificates' for details.
12803
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010012804ca-verify-file <cafile>
12805 This setting designates a PEM file from which to load CA certificates used to
12806 verify client's certificate. It designates CA certificates which must not be
12807 included in CA names sent in server hello message. Typically, "ca-file" must
12808 be defined with intermediate certificates, and "ca-verify-file" with
12809 certificates to ending the chain, like root CA.
12810
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012811ciphers <ciphers>
12812 This setting is only available when support for OpenSSL was built in. It sets
12813 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000012814 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012815 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012816 information and recommendations see e.g.
12817 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
12818 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
12819 cipher configuration, please check the "ciphersuites" keyword.
12820
12821ciphersuites <ciphersuites>
12822 This setting is only available when support for OpenSSL was built in and
12823 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
12824 the list of cipher algorithms ("cipher suite") that are negotiated during the
12825 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012826 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
12827 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012828
Emeric Brunfd33a262012-10-11 16:28:27 +020012829crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020012830 This setting is only available when support for OpenSSL was built in. It
12831 designates a PEM file from which to load certificate revocation list used
12832 to verify client's certificate.
12833
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012834crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000012835 This setting is only available when support for OpenSSL was built in. It
12836 designates a PEM file containing both the required certificates and any
12837 associated private keys. This file can be built by concatenating multiple
12838 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
12839 requires an intermediate certificate, this can also be concatenated into this
Emmanuel Hocdet70df7bf2019-01-04 11:08:20 +010012840 file. Intermediate certificate can also be shared in a directory via
12841 "issuers-chain-path" directive.
Alex Davies0fbf0162013-03-02 16:04:50 +000012842
William Lallemand4c5adbf2020-02-24 14:23:22 +010012843 If the file does not contain a private key, HAProxy will try to load
12844 the key at the same path suffixed by a ".key".
12845
Alex Davies0fbf0162013-03-02 16:04:50 +000012846 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
12847 are loaded.
12848
12849 If a directory name is used instead of a PEM file, then all files found in
William Lallemand3f25ae32020-02-24 16:30:12 +010012850 that directory will be loaded in alphabetic order unless their name ends
12851 with '.key', '.issuer', '.ocsp' or '.sctl' (reserved extensions). This
12852 directive may be specified multiple times in order to load certificates from
12853 multiple files or directories. The certificates will be presented to clients
12854 who provide a valid TLS Server Name Indication field matching one of their
12855 CN or alt subjects. Wildcards are supported, where a wildcard character '*'
12856 is used instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010012857 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000012858
12859 If no SNI is provided by the client or if the SSL library does not support
12860 TLS extensions, or if the client provides an SNI hostname which does not
12861 match any certificate, then the first loaded certificate will be presented.
12862 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010012863 recommended to load the default one first as a file or to ensure that it will
12864 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000012865
Emeric Brune032bfa2012-09-28 13:01:45 +020012866 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012867
Davor Ocelice9ed2812017-12-25 17:49:28 +010012868 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000012869 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010012870 choose a web server that the CA believes requires an intermediate CA (for
12871 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000012872 others, e.g. nginx, result in a wrong bundle that will not work for some
12873 clients).
12874
Emeric Brun4147b2e2014-06-16 18:36:30 +020012875 For each PEM file, haproxy checks for the presence of file at the same path
12876 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
12877 Status Request extension (also known as "OCSP stapling") is automatically
12878 enabled. The content of this file is optional. If not empty, it must contain
12879 a valid OCSP Response in DER format. In order to be valid an OCSP Response
12880 must comply with the following rules: it has to indicate a good status,
12881 it has to be a single response for the certificate of the PEM file, and it
12882 has to be valid at the moment of addition. If these rules are not respected
12883 the OCSP Response is ignored and a warning is emitted. In order to identify
12884 which certificate an OCSP Response applies to, the issuer's certificate is
12885 necessary. If the issuer's certificate is not found in the PEM file, it will
12886 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
12887 if it exists otherwise it will fail with an error.
12888
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010012889 For each PEM file, haproxy also checks for the presence of file at the same
12890 path suffixed by ".sctl". If such file is found, support for Certificate
12891 Transparency (RFC6962) TLS extension is enabled. The file must contain a
12892 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
12893 to check basic syntax, but no signatures are verified.
12894
yanbzhu6c25e9e2016-01-05 12:52:02 -050012895 There are cases where it is desirable to support multiple key types, e.g. RSA
12896 and ECDSA in the cipher suites offered to the clients. This allows clients
12897 that support EC certificates to be able to use EC ciphers, while
12898 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050012899
William Lallemandf9ff3ec2020-10-02 17:57:44 +020012900 To achieve this, OpenSSL 1.1.1 is required, you can configure this behavior
12901 by providing one crt entry per certificate type, or by configuring a "cert
12902 bundle" like it was required before HAProxy 1.8. See "ssl-load-extra-files".
yanbzhud19630c2015-12-14 15:10:25 -050012903
Emeric Brunb6dc9342012-09-28 17:55:37 +020012904crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000012905 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010012906 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012907 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000012908 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020012909
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012910crt-list <file>
12911 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012912 designates a list of PEM file with an optional ssl configuration and a SNI
12913 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012914
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012915 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
12916
William Lallemand5d036392020-06-30 16:11:36 +020012917 sslbindconf supports "allow-0rtt", "alpn", "ca-file", "ca-verify-file",
12918 "ciphers", "ciphersuites", "crl-file", "curves", "ecdhe", "no-ca-names",
12919 "npn", "verify" configuration. With BoringSSL and Openssl >= 1.1.1
12920 "ssl-min-ver" and "ssl-max-ver" are also supported. It overrides the
12921 configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012922
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020012923 Wildcards are supported in the SNI filter. Negative filter are also supported,
Joao Moraise51fab02020-11-21 07:42:20 -030012924 useful in combination with a wildcard filter to exclude a particular SNI, or
12925 after the first certificate to exclude a pattern from its CN or Subject Alt
12926 Name (SAN). The certificates will be presented to clients who provide a valid
12927 TLS Server Name Indication field matching one of the SNI filters. If no SNI
12928 filter is specified, the CN and SAN are used. This directive may be specified
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020012929 multiple times. See the "crt" option for more information. The default
12930 certificate is still needed to meet OpenSSL expectations. If it is not used,
12931 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010012932
William Lallemandf9ff3ec2020-10-02 17:57:44 +020012933 Multi-cert bundling (see "ssl-load-extra-files") is supported with crt-list,
12934 as long as only the base name is given in the crt-list. SNI filter will do
12935 the same work on all bundled certificates.
yanbzhud19630c2015-12-14 15:10:25 -050012936
William Lallemand7c26ed72020-06-03 17:34:48 +020012937 Empty lines as well as lines beginning with a hash ('#') will be ignored.
12938
Joao Moraisaa8fcc42020-11-24 08:24:30 -030012939 The first declared certificate of a bind line is used as the default
12940 certificate, either from crt or crt-list option, which haproxy should use in
12941 the TLS handshake if no other certificate matches. This certificate will also
12942 be used if the provided SNI matches its CN or SAN, even if a matching SNI
12943 filter is found on any crt-list. The SNI filter !* can be used after the first
12944 declared certificate to not include its CN and SAN in the SNI tree, so it will
12945 never match except if no other certificate matches. This way the first
12946 declared certificate act as a fallback.
Joao Moraise51fab02020-11-21 07:42:20 -030012947
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012948 crt-list file example:
Joao Moraise51fab02020-11-21 07:42:20 -030012949 cert1.pem !*
William Lallemand7c26ed72020-06-03 17:34:48 +020012950 # comment
Emmanuel Hocdet05942112017-02-20 16:11:50 +010012951 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012952 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010012953 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010012954
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012955defer-accept
12956 Is an optional keyword which is supported only on certain Linux kernels. It
12957 states that a connection will only be accepted once some data arrive on it,
12958 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010012959 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012960 performance by ensuring that most of the request is already available when
12961 the connection is accepted. On the other hand, it will not be able to detect
12962 connections which don't talk. It is important to note that this option is
12963 broken in all kernels up to 2.6.31, as the connection is never accepted until
12964 the client talks. This can cause issues with front firewalls which would see
12965 an established connection while the proxy will only see it in SYN_RECV. This
12966 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
12967
William Lallemandf6975e92017-05-26 17:42:10 +020012968expose-fd listeners
12969 This option is only usable with the stats socket. It gives your stats socket
12970 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020012971 During a reload with the master-worker mode, the process is automatically
12972 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012973 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020012974
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012975force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012976 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012977 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012978 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012979 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012980
12981force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012982 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012983 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012984 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012985
12986force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012987 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012988 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012989 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012990
12991force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012992 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012993 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012994 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020012995
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012996force-tlsv13
12997 This option enforces use of TLSv1.3 only on SSL connections instantiated from
12998 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012999 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013000
Christopher Faulet31af49d2015-06-09 17:29:50 +020013001generate-certificates
13002 This setting is only available when support for OpenSSL was built in. It
13003 enables the dynamic SSL certificates generation. A CA certificate and its
13004 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
13005 a transparent forward proxy, SSL requests generate errors because of a common
13006 name mismatch on the certificate presented to the client. With this option
13007 enabled, HAProxy will try to forge a certificate using the SNI hostname
13008 indicated by the client. This is done only if no certificate matches the SNI
13009 hostname (see 'crt-list'). If an error occurs, the default certificate is
13010 used, else the 'strict-sni' option is set.
13011 It can also be used when HAProxy is configured as a reverse proxy to ease the
13012 deployment of an architecture with many backends.
13013
13014 Creating a SSL certificate is an expensive operation, so a LRU cache is used
13015 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010013016 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020013017 certificate is used many times.
13018
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013019gid <gid>
13020 Sets the group of the UNIX sockets to the designated system gid. It can also
13021 be set by default in the global section's "unix-bind" statement. Note that
13022 some platforms simply ignore this. This setting is equivalent to the "group"
13023 setting except that the group ID is used instead of its name. This setting is
13024 ignored by non UNIX sockets.
13025
13026group <group>
13027 Sets the group of the UNIX sockets to the designated system group. It can
13028 also be set by default in the global section's "unix-bind" statement. Note
13029 that some platforms simply ignore this. This setting is equivalent to the
13030 "gid" setting except that the group name is used instead of its gid. This
13031 setting is ignored by non UNIX sockets.
13032
13033id <id>
13034 Fixes the socket ID. By default, socket IDs are automatically assigned, but
13035 sometimes it is more convenient to fix them to ease monitoring. This value
13036 must be strictly positive and unique within the listener/frontend. This
13037 option can only be used when defining only a single socket.
13038
13039interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010013040 Restricts the socket to a specific interface. When specified, only packets
13041 received from that particular interface are processed by the socket. This is
13042 currently only supported on Linux. The interface must be a primary system
13043 interface, not an aliased interface. It is also possible to bind multiple
13044 frontends to the same address if they are bound to different interfaces. Note
13045 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010013046 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
13047 uses the same interface as inbound traffic, and its associated routing table,
13048 even if there are explicit routes through different interfaces configured.
13049 This can prove useful to address asymmetric routing issues when the same
13050 client IP addresses need to be able to reach frontends hosted on different
13051 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013052
Willy Tarreauabb175f2012-09-24 12:43:26 +020013053level <level>
13054 This setting is used with the stats sockets only to restrict the nature of
13055 the commands that can be issued on the socket. It is ignored by other
13056 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013057 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020013058 read, and no change is allowed. It would make sense on systems where it
13059 is not easy to restrict access to the socket.
13060 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010013061 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020013062 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010013063 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020013064 all counters).
13065
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020013066severity-output <format>
13067 This setting is used with the stats sockets only to configure severity
13068 level output prepended to informational feedback messages. Severity
13069 level of messages can range between 0 and 7, conforming to syslog
13070 rfc5424. Valid and successful socket commands requesting data
13071 (i.e. "show map", "get acl foo" etc.) will never have a severity level
13072 prepended. It is ignored by other sockets. <format> can be one of :
13073 - "none" (default) no severity level is prepended to feedback messages.
13074 - "number" severity level is prepended as a number.
13075 - "string" severity level is prepended as a string following the
13076 rfc5424 convention.
13077
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013078maxconn <maxconn>
13079 Limits the sockets to this number of concurrent connections. Extraneous
13080 connections will remain in the system's backlog until a connection is
13081 released. If unspecified, the limit will be the same as the frontend's
13082 maxconn. Note that in case of port ranges or multiple addresses, the same
13083 value will be applied to each socket. This setting enables different
13084 limitations on expensive sockets, for instance SSL entries which may easily
13085 eat all memory.
13086
13087mode <mode>
13088 Sets the octal mode used to define access permissions on the UNIX socket. It
13089 can also be set by default in the global section's "unix-bind" statement.
13090 Note that some platforms simply ignore this. This setting is ignored by non
13091 UNIX sockets.
13092
13093mss <maxseg>
13094 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
13095 connections. This can be used to force a lower MSS for certain specific
13096 ports, for instance for connections passing through a VPN. Note that this
13097 relies on a kernel feature which is theoretically supported under Linux but
13098 was buggy in all versions prior to 2.6.28. It may or may not work on other
13099 operating systems. It may also not change the advertised value but change the
13100 effective size of outgoing segments. The commonly advertised value for TCPv4
13101 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
13102 positive, it will be used as the advertised MSS. If it is negative, it will
13103 indicate by how much to reduce the incoming connection's advertised MSS for
13104 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
13105
13106name <name>
13107 Sets an optional name for these sockets, which will be reported on the stats
13108 page.
13109
Willy Tarreaud72f0f32015-10-13 14:50:22 +020013110namespace <name>
13111 On Linux, it is possible to specify which network namespace a socket will
13112 belong to. This directive makes it possible to explicitly bind a listener to
13113 a namespace different from the default one. Please refer to your operating
13114 system's documentation to find more details about network namespaces.
13115
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013116nice <nice>
13117 Sets the 'niceness' of connections initiated from the socket. Value must be
13118 in the range -1024..1024 inclusive, and defaults to zero. Positive values
13119 means that such connections are more friendly to others and easily offer
13120 their place in the scheduler. On the opposite, negative values mean that
13121 connections want to run with a higher priority than others. The difference
13122 only happens under high loads when the system is close to saturation.
13123 Negative values are appropriate for low-latency or administration services,
13124 and high values are generally recommended for CPU intensive tasks such as SSL
13125 processing or bulk transfers which are less sensible to latency. For example,
13126 it may make sense to use a positive value for an SMTP socket and a negative
13127 one for an RDP socket.
13128
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020013129no-ca-names
13130 This setting is only available when support for OpenSSL was built in. It
13131 prevents from send CA names in server hello message when ca-file is used.
Emmanuel Hocdet842e94e2019-12-16 16:39:17 +010013132 Use "ca-verify-file" instead of "ca-file" with "no-ca-names".
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020013133
Emeric Brun9b3009b2012-10-05 11:55:06 +020013134no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013135 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013136 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013137 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013138 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013139 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
13140 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013141
Emeric Brun90ad8722012-10-02 14:00:59 +020013142no-tls-tickets
13143 This setting is only available when support for OpenSSL was built in. It
13144 disables the stateless session resumption (RFC 5077 TLS Ticket
13145 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013146 session resumption is more expensive in CPU usage. This option is also
13147 available on global statement "ssl-default-bind-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010013148 The TLS ticket mechanism is only used up to TLS 1.2.
13149 Forward Secrecy is compromised with TLS tickets, unless ticket keys
13150 are periodically rotated (via reload or by using "tls-ticket-keys").
Emeric Brun90ad8722012-10-02 14:00:59 +020013151
Emeric Brun9b3009b2012-10-05 11:55:06 +020013152no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013153 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013154 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013155 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013156 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013157 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13158 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013159
Emeric Brun9b3009b2012-10-05 11:55:06 +020013160no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020013161 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013162 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013163 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013164 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013165 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13166 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020013167
Emeric Brun9b3009b2012-10-05 11:55:06 +020013168no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020013169 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013170 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020013171 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013172 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013173 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13174 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020013175
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013176no-tlsv13
13177 This setting is only available when support for OpenSSL was built in. It
13178 disables support for TLSv1.3 on any sockets instantiated from the listener
13179 when SSL is supported. Note that SSLv2 is forced disabled in the code and
13180 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013181 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
13182 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013183
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020013184npn <protocols>
13185 This enables the NPN TLS extension and advertises the specified protocol list
13186 as supported on top of NPN. The protocol list consists in a comma-delimited
13187 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050013188 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020013189 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010013190 replaced with the ALPN extension (see the "alpn" keyword), though this one is
13191 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
13192 version of OpenSSL, NPN might still be used as most clients still support it
13193 at the time of writing this. It is possible to enable both NPN and ALPN
13194 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020013195
Lukas Tribus53ae85c2017-05-04 15:45:40 +000013196prefer-client-ciphers
13197 Use the client's preference when selecting the cipher suite, by default
13198 the server's preference is enforced. This option is also available on
13199 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020013200 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
13201 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
13202 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000013203
Christopher Fauletc644fa92017-11-23 22:44:11 +010013204process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010013205 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010013206 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010013207 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010013208 between the two is applied. If in the end the listener is not allowed to run
13209 on any remaining process, a warning is emitted, and the listener will either
13210 run on the first process of the listener if a single process was specified,
13211 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010013212 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010013213 connections for this listener, for the the process set. If multiple processes
13214 and threads are configured, a warning is emitted, as it either results from a
13215 configuration error or a misunderstanding of these models. For the unlikely
13216 case where several ranges are needed, this directive may be repeated.
13217 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010013218
13219 all | odd | even | number[-[number]]
13220
13221 Ranges can be partially defined. The higher bound can be omitted. In such
13222 case, it is replaced by the corresponding maximum value. The main purpose of
13223 this directive is to be used with the stats sockets and have one different
13224 socket per process. The second purpose is to have multiple bind lines sharing
13225 the same IP:port but not the same process in a listener, so that the system
13226 can distribute the incoming connections into multiple queues and allow a
13227 smoother inter-process load balancing. Currently Linux 3.9 and above is known
13228 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020013229
Christopher Fauleta717b992018-04-10 14:43:00 +020013230proto <name>
13231 Forces the multiplexer's protocol to use for the incoming connections. It
13232 must be compatible with the mode of the frontend (TCP or HTTP). It must also
13233 be usable on the frontend side. The list of available protocols is reported
13234 in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040013235 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauleta717b992018-04-10 14:43:00 +020013236 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080013237 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020013238 h2" on the bind line.
13239
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013240ssl
13241 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013242 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013243 certificate is necessary (see "crt" above). All contents in the buffers will
13244 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020013245 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
13246 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013247
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013248ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
13249 This option enforces use of <version> or lower on SSL connections instantiated
William Lallemand50df1cb2020-06-02 10:52:24 +020013250 from this listener. Using this setting without "ssl-min-ver" can be
13251 ambiguous because the default ssl-min-ver value could change in future HAProxy
13252 versions. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013253 "ssl-default-bind-options". See also "ssl-min-ver".
13254
13255ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
William Lallemand50df1cb2020-06-02 10:52:24 +020013256 This option enforces use of <version> or upper on SSL connections
13257 instantiated from this listener. The default value is "TLSv1.2". This option
13258 is also available on global statement "ssl-default-bind-options".
13259 See also "ssl-max-ver".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013260
Emmanuel Hocdet65623372013-01-24 17:17:15 +010013261strict-sni
13262 This setting is only available when support for OpenSSL was built in. The
13263 SSL/TLS negotiation is allow only if the client provided an SNI which match
13264 a certificate. The default certificate is not used.
13265 See the "crt" option for more information.
13266
Willy Tarreau2af207a2015-02-04 00:45:58 +010013267tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010013268 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010013269 listening socket. This option is available on Linux since version 2.6.37. It
13270 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010013271 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010013272 useful on long-lived connections experiencing long idle periods such as
13273 remote terminals or database connection pools, where the client and server
13274 timeouts must remain high to allow a long period of idle, but where it is
13275 important to detect that the client has disappeared in order to release all
13276 resources associated with its connection (and the server's session). The
13277 argument is a delay expressed in milliseconds by default. This only works
13278 for regular TCP connections, and is ignored for other protocols.
13279
Willy Tarreau1c862c52012-10-05 16:21:00 +020013280tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010013281 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020013282 enables TCP Fast Open on the listening socket, which means that clients which
13283 support this feature will be able to send a request and receive a response
13284 during the 3-way handshake starting from second connection, thus saving one
13285 round-trip after the first connection. This only makes sense with protocols
13286 that use high connection rates and where each round trip matters. This can
13287 possibly cause issues with many firewalls which do not accept data on SYN
13288 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020013289 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
13290 need to build HAProxy with USE_TFO=1 if your libc doesn't define
13291 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020013292
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010013293tls-ticket-keys <keyfile>
13294 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010013295 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
13296 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
13297 The first key determines the key length used for next keys: you can't mix
13298 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
13299 build option (default 3) and at least as many keys need to be present in
13300 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
13301 penultimate one for encryption. This enables easy key rotation by just
13302 appending new key to the file and reloading the process. Keys must be
13303 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
13304 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010013305 storage such as hard drives (hint: use tmpfs and don't swap those files).
13306 Lifetime hint can be changed using tune.ssl.timeout.
13307
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013308transparent
13309 Is an optional keyword which is supported only on certain Linux kernels. It
13310 indicates that the addresses will be bound even if they do not belong to the
13311 local machine, and that packets targeting any of these addresses will be
13312 intercepted just as if the addresses were locally configured. This normally
13313 requires that IP forwarding is enabled. Caution! do not use this with the
13314 default address '*', as it would redirect any traffic for the specified port.
13315 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
13316 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
13317 kernel version. Some distribution kernels include backports of the feature,
13318 so check for support with your vendor.
13319
Willy Tarreau77e3af92012-11-24 15:07:23 +010013320v4v6
13321 Is an optional keyword which is supported only on most recent systems
13322 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
13323 and IPv6 when it uses the default address. Doing so is sometimes necessary
13324 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013325 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010013326
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013327v6only
13328 Is an optional keyword which is supported only on most recent systems
13329 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
13330 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010013331 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
13332 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010013333
Willy Tarreaub6205fd2012-09-24 12:27:33 +020013334uid <uid>
13335 Sets the owner of the UNIX sockets to the designated system uid. It can also
13336 be set by default in the global section's "unix-bind" statement. Note that
13337 some platforms simply ignore this. This setting is equivalent to the "user"
13338 setting except that the user numeric ID is used instead of its name. This
13339 setting is ignored by non UNIX sockets.
13340
13341user <user>
13342 Sets the owner of the UNIX sockets to the designated system user. It can also
13343 be set by default in the global section's "unix-bind" statement. Note that
13344 some platforms simply ignore this. This setting is equivalent to the "uid"
13345 setting except that the user name is used instead of its uid. This setting is
13346 ignored by non UNIX sockets.
13347
Emeric Brun1a073b42012-09-28 17:07:34 +020013348verify [none|optional|required]
13349 This setting is only available when support for OpenSSL was built in. If set
13350 to 'none', client certificate is not requested. This is the default. In other
13351 cases, a client certificate is requested. If the client does not provide a
13352 certificate after the request and if 'verify' is set to 'required', then the
13353 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020013354 certificate provided by the client is always verified using CAs from
13355 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
13356 is aborted, regardless of the 'verify' option, unless the error code exactly
13357 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020013358
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200133595.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010013360------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013361
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013362The "server" and "default-server" keywords support a certain number of settings
13363which are all passed as arguments on the server line. The order in which those
13364arguments appear does not count, and they are all optional. Some of those
13365settings are single words (booleans) while others expect one or several values
13366after them. In this case, the values must immediately follow the setting name.
13367Except default-server, all those settings must be specified after the server's
13368address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013369
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013370 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010013371 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020013372
Frédéric Lécailled2376272017-03-21 18:52:12 +010013373Note that all these settings are supported both by "server" and "default-server"
13374keywords, except "id" which is only supported by "server".
13375
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013376The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013377
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013378addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013379 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010013380 to send health-checks or to probe the agent-check. On some servers, it may be
13381 desirable to dedicate an IP address to specific component able to perform
13382 complex tests which are more suitable to health-checks than the application.
13383 This parameter is ignored if the "check" parameter is not set. See also the
13384 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013385
Simon Hormand60d6912013-11-25 10:46:36 +090013386agent-check
13387 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010013388 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010013389 to the port set by the "agent-port" parameter and reading an ASCII string
13390 terminated by the first '\r' or '\n' met. The string is made of a series of
13391 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090013392
Willy Tarreau81f5d942013-12-09 20:51:51 +010013393 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090013394 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020013395 weight of a server as configured when haproxy starts. Note that a zero
13396 weight is reported on the stats page as "DRAIN" since it has the same
13397 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090013398
Davor Ocelice9ed2812017-12-25 17:49:28 +010013399 - The string "maxconn:" followed by an integer (no space between). Values
13400 in this format will set the maxconn of a server. The maximum number of
13401 connections advertised needs to be multiplied by the number of load
13402 balancers and different backends that use this health check to get the
13403 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020013404
Willy Tarreau81f5d942013-12-09 20:51:51 +010013405 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013406 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090013407
Willy Tarreau81f5d942013-12-09 20:51:51 +010013408 - The word "drain". This will turn the server's administrative state to the
13409 DRAIN mode, thus it will not accept any new connections other than those
13410 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090013411
Willy Tarreau81f5d942013-12-09 20:51:51 +010013412 - The word "maint". This will turn the server's administrative state to the
13413 MAINT mode, thus it will not accept any new connections at all, and health
13414 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090013415
William Dauchyf8e795c2020-09-26 13:35:51 +020013416 - The words "down", "fail", or "stopped", optionally followed by a
Willy Tarreau81f5d942013-12-09 20:51:51 +010013417 description string after a sharp ('#'). All of these mark the server's
13418 operating state as DOWN, but since the word itself is reported on the stats
13419 page, the difference allows an administrator to know if the situation was
13420 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010013421 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010013422 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090013423
Willy Tarreau81f5d942013-12-09 20:51:51 +010013424 - The word "up" sets back the server's operating state as UP if health checks
13425 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090013426
Willy Tarreau81f5d942013-12-09 20:51:51 +010013427 Parameters which are not advertised by the agent are not changed. For
13428 example, an agent might be designed to monitor CPU usage and only report a
13429 relative weight and never interact with the operating status. Similarly, an
13430 agent could be designed as an end-user interface with 3 radio buttons
13431 allowing an administrator to change only the administrative state. However,
13432 it is important to consider that only the agent may revert its own actions,
13433 so if a server is set to DRAIN mode or to DOWN state using the agent, the
13434 agent must implement the other equivalent actions to bring the service into
13435 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090013436
Simon Horman2f1f9552013-11-25 10:46:37 +090013437 Failure to connect to the agent is not considered an error as connectivity
13438 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010013439 parameter. Warning though, it is not a good idea to stop an agent after it
13440 reports "down", since only an agent reporting "up" will be able to turn the
13441 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010013442 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090013443
Willy Tarreau81f5d942013-12-09 20:51:51 +010013444 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010013445 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090013446
James Brown55f9ff12015-10-21 18:19:05 -070013447agent-send <string>
13448 If this option is specified, haproxy will send the given string (verbatim)
13449 to the agent server upon connection. You could, for example, encode
13450 the backend name into this string, which would enable your agent to send
13451 different responses based on the backend. Make sure to include a '\n' if
13452 you want to terminate your request with a newline.
13453
Simon Hormand60d6912013-11-25 10:46:36 +090013454agent-inter <delay>
13455 The "agent-inter" parameter sets the interval between two agent checks
13456 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
13457
13458 Just as with every other time-based parameter, it may be entered in any
13459 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
13460 parameter also serves as a timeout for agent checks "timeout check" is
13461 not set. In order to reduce "resonance" effects when multiple servers are
13462 hosted on the same hardware, the agent and health checks of all servers
13463 are started with a small time offset between them. It is also possible to
13464 add some random noise in the agent and health checks interval using the
13465 global "spread-checks" keyword. This makes sense for instance when a lot
13466 of backends use the same servers.
13467
13468 See also the "agent-check" and "agent-port" parameters.
13469
Misiek768d8602017-01-09 09:52:43 +010013470agent-addr <addr>
13471 The "agent-addr" parameter sets address for agent check.
13472
13473 You can offload agent-check to another target, so you can make single place
13474 managing status and weights of servers defined in haproxy in case you can't
13475 make self-aware and self-managing services. You can specify both IP or
13476 hostname, it will be resolved.
13477
Simon Hormand60d6912013-11-25 10:46:36 +090013478agent-port <port>
13479 The "agent-port" parameter sets the TCP port used for agent checks.
13480
13481 See also the "agent-check" and "agent-inter" parameters.
13482
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013483allow-0rtt
13484 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020013485 Note that early data will be sent only if the client used early data, or
13486 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020013487
Olivier Houchardc7566002018-11-20 23:33:50 +010013488alpn <protocols>
13489 This enables the TLS ALPN extension and advertises the specified protocol
13490 list as supported on top of ALPN. The protocol list consists in a comma-
13491 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roeslerfb2fce12019-07-10 15:45:51 -050013492 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010013493 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
13494 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
13495 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
13496 now obsolete NPN extension.
13497 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
13498 be advertised, in order of preference, like below :
13499
13500 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
13501
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013502backup
13503 When "backup" is present on a server line, the server is only used in load
13504 balancing when all other non-backup servers are unavailable. Requests coming
13505 with a persistence cookie referencing the server will always be served
13506 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010013507 the "allbackups" option is set in the backend. See also the "no-backup" and
13508 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013509
Emeric Brunef42d922012-10-11 16:11:36 +020013510ca-file <cafile>
13511 This setting is only available when support for OpenSSL was built in. It
13512 designates a PEM file from which to load CA certificates used to verify
13513 server's certificate.
13514
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013515check
Jerome Magnin90702bc2020-04-26 14:23:04 +020013516 This option enables health checks on a server:
13517 - when not set, no health checking is performed, and the server is always
13518 considered available.
13519 - when set and no other check method is configured, the server is considered
13520 available when a connection can be established at the highest configured
13521 transport layer. This means TCP by default, or SSL/TLS when "ssl" or
13522 "check-ssl" are set, both possibly combined with connection prefixes such
13523 as a PROXY protocol header when "send-proxy" or "check-send-proxy" are
13524 set.
13525 - when set and an application-level health check is defined, the
13526 application-level exchanges are performed on top of the configured
13527 transport layer and the server is considered available if all of the
13528 exchanges succeed.
13529
13530 By default, health checks are performed on the same address and port as
13531 configured on the server, using the same encapsulation parameters (SSL/TLS,
13532 proxy-protocol header, etc... ). It is possible to change the destination
13533 address using "addr" and the port using "port". When done, it is assumed the
13534 server isn't checked on the service port, and configured encapsulation
Ilya Shipitsin4329a9a2020-05-05 21:17:10 +050013535 parameters are not reused. One must explicitly set "check-send-proxy" to send
Jerome Magnin90702bc2020-04-26 14:23:04 +020013536 connection headers, "check-ssl" to use SSL/TLS.
13537
13538 When "sni" or "alpn" are set on the server line, their value is not used for
13539 health checks and one must use "check-sni" or "check-alpn".
13540
13541 The default source address for health check traffic is the same as the one
13542 defined in the backend. It can be changed with the "source" keyword.
13543
13544 The interval between checks can be set using the "inter" keyword, and the
13545 "rise" and "fall" keywords can be used to define how many successful or
13546 failed health checks are required to flag a server available or not
13547 available.
13548
13549 Optional application-level health checks can be configured with "option
13550 httpchk", "option mysql-check" "option smtpchk", "option pgsql-check",
13551 "option ldap-check", or "option redis-check".
13552
13553 Example:
13554 # simple tcp check
13555 backend foo
13556 server s1 192.168.0.1:80 check
13557 # this does a tcp connect + tls handshake
13558 backend foo
13559 server s1 192.168.0.1:443 ssl check
13560 # simple tcp check is enough for check success
13561 backend foo
13562 option tcp-check
13563 tcp-check connect
13564 server s1 192.168.0.1:443 ssl check
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013565
Willy Tarreau6c16adc2012-10-05 00:04:16 +020013566check-send-proxy
13567 This option forces emission of a PROXY protocol line with outgoing health
13568 checks, regardless of whether the server uses send-proxy or not for the
13569 normal traffic. By default, the PROXY protocol is enabled for health checks
13570 if it is already enabled for normal traffic and if no "port" nor "addr"
13571 directive is present. However, if such a directive is present, the
13572 "check-send-proxy" option needs to be used to force the use of the
13573 protocol. See also the "send-proxy" option for more information.
13574
Olivier Houchard92150142018-12-21 19:47:01 +010013575check-alpn <protocols>
13576 Defines which protocols to advertise with ALPN. The protocol list consists in
13577 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
13578 (without quotes). If it is not set, the server ALPN is used.
13579
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013580check-proto <name>
13581 Forces the multiplexer's protocol to use for the server's health-check
13582 connections. It must be compatible with the health-check type (TCP or
13583 HTTP). It must also be usable on the backend side. The list of available
13584 protocols is reported in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040013585 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Fauletedc6ed92020-04-23 16:27:59 +020013586 protocol for health-check connections established to this server.
13587 If not defined, the server one will be used, if set.
13588
Jérôme Magninae9bb762018-12-09 16:08:26 +010013589check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020013590 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010013591 over SSL. It is only possible to use a string to set <sni>. If you want to
13592 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020013593
Willy Tarreau763a95b2012-10-04 23:15:39 +020013594check-ssl
13595 This option forces encryption of all health checks over SSL, regardless of
13596 whether the server uses SSL or not for the normal traffic. This is generally
13597 used when an explicit "port" or "addr" directive is specified and SSL health
13598 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013599 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020013600 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
13601 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013602 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010013603 See the "ssl" option for more information and "no-check-ssl" to disable
13604 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013605
Alexander Liu2a54bb72019-05-22 19:44:48 +080013606check-via-socks4
John Roeslerfb2fce12019-07-10 15:45:51 -050013607 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080013608 default, the health checks won't go through socks tunnel even it was enabled
13609 for normal traffic.
13610
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013611ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013612 This setting is only available when support for OpenSSL was built in. This
13613 option sets the string describing the list of cipher algorithms that is
13614 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013615 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
13616 information and recommendations see e.g.
13617 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
13618 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
13619 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013620
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013621ciphersuites <ciphersuites>
13622 This setting is only available when support for OpenSSL was built in and
13623 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
13624 describing the list of cipher algorithms that is negotiated during the TLS
13625 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000013626 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
13627 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
13628 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020013629
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013630cookie <value>
13631 The "cookie" parameter sets the cookie value assigned to the server to
13632 <value>. This value will be checked in incoming requests, and the first
13633 operational server possessing the same value will be selected. In return, in
13634 cookie insertion or rewrite modes, this value will be assigned to the cookie
13635 sent to the client. There is nothing wrong in having several servers sharing
13636 the same cookie value, and it is in fact somewhat common between normal and
13637 backup servers. See also the "cookie" keyword in backend section.
13638
Emeric Brunef42d922012-10-11 16:11:36 +020013639crl-file <crlfile>
13640 This setting is only available when support for OpenSSL was built in. It
13641 designates a PEM file from which to load certificate revocation list used
13642 to verify server's certificate.
13643
Emeric Bruna7aa3092012-10-26 12:58:00 +020013644crt <cert>
13645 This setting is only available when support for OpenSSL was built in.
13646 It designates a PEM file from which to load both a certificate and the
13647 associated private key. This file can be built by concatenating both PEM
13648 files into one. This certificate will be sent if the server send a client
13649 certificate request.
13650
Willy Tarreau96839092010-03-29 10:02:24 +020013651disabled
13652 The "disabled" keyword starts the server in the "disabled" state. That means
13653 that it is marked down in maintenance mode, and no connection other than the
13654 ones allowed by persist mode will reach it. It is very well suited to setup
13655 new servers, because normal traffic will never reach them, while it is still
13656 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010013657 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020013658
Frédéric Lécailled2376272017-03-21 18:52:12 +010013659enabled
13660 This option may be used as 'server' setting to reset any 'disabled'
13661 setting which would have been inherited from 'default-server' directive as
13662 default value.
13663 It may also be used as 'default-server' setting to reset any previous
13664 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020013665
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013666error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010013667 If health observing is enabled, the "error-limit" parameter specifies the
13668 number of consecutive errors that triggers event selected by the "on-error"
13669 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013670
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013671 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013672
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013673fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013674 The "fall" parameter states that a server will be considered as dead after
13675 <count> consecutive unsuccessful health checks. This value defaults to 3 if
13676 unspecified. See also the "check", "inter" and "rise" parameters.
13677
Emeric Brun8694b9a2012-10-05 14:39:07 +020013678force-sslv3
13679 This option enforces use of SSLv3 only when SSL is used to communicate with
13680 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013681 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013682 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013683
13684force-tlsv10
13685 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013686 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013687 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013688
13689force-tlsv11
13690 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013691 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013692 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013693
13694force-tlsv12
13695 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013696 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013697 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020013698
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013699force-tlsv13
13700 This option enforces use of TLSv1.3 only when SSL is used to communicate with
13701 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013702 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013703
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013704id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020013705 Set a persistent ID for the server. This ID must be positive and unique for
13706 the proxy. An unused ID will automatically be assigned if unset. The first
13707 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013708
Willy Tarreau6a031d12016-11-07 19:42:35 +010013709init-addr {last | libc | none | <ip>},[...]*
13710 Indicate in what order the server's address should be resolved upon startup
13711 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010013712 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010013713 method which succeeds is used. If the end of the list is reached without
13714 finding a working method, an error is thrown. Method "last" suggests to pick
13715 the address which appears in the state file (see "server-state-file"). Method
13716 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
13717 depending on the operating system and build options). Method "none"
13718 specifically indicates that the server should start without any valid IP
13719 address in a down state. It can be useful to ignore some DNS issues upon
13720 startup, waiting for the situation to get fixed later. Finally, an IP address
13721 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013722 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010013723 server used to catch old sessions and present them with a decent error
13724 message for example. When the "first" load balancing algorithm is used, this
13725 IP address could point to a fake server used to trigger the creation of new
13726 instances on the fly. This option defaults to "last,libc" indicating that the
13727 previous address found in the state file (if any) is used first, otherwise
13728 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013729 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010013730
13731 Example:
13732 defaults
13733 # never fail on address resolution
13734 default-server init-addr last,libc,none
13735
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013736inter <delay>
13737fastinter <delay>
13738downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013739 The "inter" parameter sets the interval between two consecutive health checks
13740 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
13741 It is also possible to use "fastinter" and "downinter" to optimize delays
13742 between checks depending on the server state :
13743
Pieter Baauw44fc9df2015-09-17 21:30:46 +020013744 Server state | Interval used
13745 ----------------------------------------+----------------------------------
13746 UP 100% (non-transitional) | "inter"
13747 ----------------------------------------+----------------------------------
13748 Transitionally UP (going down "fall"), | "fastinter" if set,
13749 Transitionally DOWN (going up "rise"), | "inter" otherwise.
13750 or yet unchecked. |
13751 ----------------------------------------+----------------------------------
13752 DOWN 100% (non-transitional) | "downinter" if set,
13753 | "inter" otherwise.
13754 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010013755
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013756 Just as with every other time-based parameter, they can be entered in any
13757 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
13758 serves as a timeout for health checks sent to servers if "timeout check" is
13759 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090013760 hosted on the same hardware, the agent and health checks of all servers
13761 are started with a small time offset between them. It is also possible to
13762 add some random noise in the agent and health checks interval using the
13763 global "spread-checks" keyword. This makes sense for instance when a lot
13764 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013765
Emeric Brun97556472020-05-30 01:42:45 +020013766log-proto <logproto>
13767 The "log-proto" specifies the protocol used to forward event messages to
13768 a server configured in a ring section. Possible values are "legacy"
13769 and "octet-count" corresponding respectively to "Non-transparent-framing"
13770 and "Octet counting" in rfc6587. "legacy" is the default.
13771
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013772maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013773 The "maxconn" parameter specifies the maximal number of concurrent
13774 connections that will be sent to this server. If the number of incoming
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010013775 concurrent connections goes higher than this value, they will be queued,
13776 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013777 save fragile servers from going down under extreme loads. If a "minconn"
13778 parameter is specified, the limit becomes dynamic. The default value is "0"
13779 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
13780 the backend's "fullconn" keyword.
13781
Tim Duesterhuscefbbd92019-11-27 22:35:27 +010013782 In HTTP mode this parameter limits the number of concurrent requests instead
13783 of the number of connections. Multiple requests might be multiplexed over a
13784 single TCP connection to the server. As an example if you specify a maxconn
13785 of 50 you might see between 1 and 50 actual server connections, but no more
13786 than 50 concurrent requests.
13787
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013788maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013789 The "maxqueue" parameter specifies the maximal number of connections which
13790 will wait in the queue for this server. If this limit is reached, next
13791 requests will be redispatched to other servers instead of indefinitely
13792 waiting to be served. This will break persistence but may allow people to
Willy Tarreau8ae8c482020-10-22 17:19:07 +020013793 quickly re-log in when the server they try to connect to is dying. Some load
13794 balancing algorithms such as leastconn take this into account and accept to
13795 add requests into a server's queue up to this value if it is explicitly set
13796 to a value greater than zero, which often allows to better smooth the load
13797 when dealing with single-digit maxconn values. The default value is "0" which
13798 means the queue is unlimited. See also the "maxconn" and "minconn" parameters
13799 and "balance leastconn".
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013800
Willy Tarreau9c538e02019-01-23 10:21:49 +010013801max-reuse <count>
13802 The "max-reuse" argument indicates the HTTP connection processors that they
13803 should not reuse a server connection more than this number of times to send
13804 new requests. Permitted values are -1 (the default), which disables this
13805 limit, or any positive value. Value zero will effectively disable keep-alive.
13806 This is only used to work around certain server bugs which cause them to leak
13807 resources over time. The argument is not necessarily respected by the lower
13808 layers as there might be technical limitations making it impossible to
13809 enforce. At least HTTP/2 connections to servers will respect it.
13810
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013811minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013812 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
13813 limit following the backend's load. The server will always accept at least
13814 <minconn> connections, never more than <maxconn>, and the limit will be on
13815 the ramp between both values when the backend has less than <fullconn>
13816 concurrent connections. This makes it possible to limit the load on the
13817 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013818 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013819 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013820
Willy Tarreaud72f0f32015-10-13 14:50:22 +020013821namespace <name>
13822 On Linux, it is possible to specify which network namespace a socket will
13823 belong to. This directive makes it possible to explicitly bind a server to
13824 a namespace different from the default one. Please refer to your operating
13825 system's documentation to find more details about network namespaces.
13826
Frédéric Lécailled2376272017-03-21 18:52:12 +010013827no-agent-check
13828 This option may be used as "server" setting to reset any "agent-check"
13829 setting which would have been inherited from "default-server" directive as
13830 default value.
13831 It may also be used as "default-server" setting to reset any previous
13832 "default-server" "agent-check" setting.
13833
13834no-backup
13835 This option may be used as "server" setting to reset any "backup"
13836 setting which would have been inherited from "default-server" directive as
13837 default value.
13838 It may also be used as "default-server" setting to reset any previous
13839 "default-server" "backup" setting.
13840
13841no-check
13842 This option may be used as "server" setting to reset any "check"
13843 setting which would have been inherited from "default-server" directive as
13844 default value.
13845 It may also be used as "default-server" setting to reset any previous
13846 "default-server" "check" setting.
13847
13848no-check-ssl
13849 This option may be used as "server" setting to reset any "check-ssl"
13850 setting which would have been inherited from "default-server" directive as
13851 default value.
13852 It may also be used as "default-server" setting to reset any previous
13853 "default-server" "check-ssl" setting.
13854
Frédéric Lécailled2376272017-03-21 18:52:12 +010013855no-send-proxy
13856 This option may be used as "server" setting to reset any "send-proxy"
13857 setting which would have been inherited from "default-server" directive as
13858 default value.
13859 It may also be used as "default-server" setting to reset any previous
13860 "default-server" "send-proxy" setting.
13861
13862no-send-proxy-v2
13863 This option may be used as "server" setting to reset any "send-proxy-v2"
13864 setting which would have been inherited from "default-server" directive as
13865 default value.
13866 It may also be used as "default-server" setting to reset any previous
13867 "default-server" "send-proxy-v2" setting.
13868
13869no-send-proxy-v2-ssl
13870 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
13871 setting which would have been inherited from "default-server" directive as
13872 default value.
13873 It may also be used as "default-server" setting to reset any previous
13874 "default-server" "send-proxy-v2-ssl" setting.
13875
13876no-send-proxy-v2-ssl-cn
13877 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
13878 setting which would have been inherited from "default-server" directive as
13879 default value.
13880 It may also be used as "default-server" setting to reset any previous
13881 "default-server" "send-proxy-v2-ssl-cn" setting.
13882
13883no-ssl
13884 This option may be used as "server" setting to reset any "ssl"
13885 setting which would have been inherited from "default-server" directive as
13886 default value.
13887 It may also be used as "default-server" setting to reset any previous
13888 "default-server" "ssl" setting.
13889
William Dauchyf6370442020-11-14 19:25:33 +010013890 Note that using `default-server ssl` setting and `no-ssl` on server will
13891 however init SSL connection, so it can be later be enabled through the
13892 runtime API: see `set server` commands in management doc.
13893
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010013894no-ssl-reuse
13895 This option disables SSL session reuse when SSL is used to communicate with
13896 the server. It will force the server to perform a full handshake for every
13897 new connection. It's probably only useful for benchmarking, troubleshooting,
13898 and for paranoid users.
13899
Emeric Brun9b3009b2012-10-05 11:55:06 +020013900no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013901 This option disables support for SSLv3 when SSL is used to communicate with
13902 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013903 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013904
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013905 Supported in default-server: No
13906
Emeric Brunf9c5c472012-10-11 15:28:34 +020013907no-tls-tickets
13908 This setting is only available when support for OpenSSL was built in. It
13909 disables the stateless session resumption (RFC 5077 TLS Ticket
13910 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013911 session resumption is more expensive in CPU usage for servers. This option
13912 is also available on global statement "ssl-default-server-options".
Lukas Tribusbdb386d2020-03-10 00:56:09 +010013913 The TLS ticket mechanism is only used up to TLS 1.2.
13914 Forward Secrecy is compromised with TLS tickets, unless ticket keys
13915 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010013916 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020013917
Emeric Brun9b3009b2012-10-05 11:55:06 +020013918no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020013919 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020013920 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13921 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013922 often makes sense to disable it when communicating with local servers. This
13923 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013924 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013925
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013926 Supported in default-server: No
13927
Emeric Brun9b3009b2012-10-05 11:55:06 +020013928no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020013929 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020013930 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13931 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013932 often makes sense to disable it when communicating with local servers. This
13933 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013934 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013935
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013936 Supported in default-server: No
13937
Emeric Brun9b3009b2012-10-05 11:55:06 +020013938no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020013939 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013940 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13941 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010013942 often makes sense to disable it when communicating with local servers. This
13943 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013944 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020013945
13946 Supported in default-server: No
13947
13948no-tlsv13
13949 This option disables support for TLSv1.3 when SSL is used to communicate with
13950 the server. Note that SSLv2 is disabled in the code and cannot be enabled
13951 using any configuration option. TLSv1 is more expensive than SSLv3 so it
13952 often makes sense to disable it when communicating with local servers. This
13953 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020013954 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020013955
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020013956 Supported in default-server: No
13957
Frédéric Lécailled2376272017-03-21 18:52:12 +010013958no-verifyhost
13959 This option may be used as "server" setting to reset any "verifyhost"
13960 setting which would have been inherited from "default-server" directive as
13961 default value.
13962 It may also be used as "default-server" setting to reset any previous
13963 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020013964
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020013965no-tfo
13966 This option may be used as "server" setting to reset any "tfo"
13967 setting which would have been inherited from "default-server" directive as
13968 default value.
13969 It may also be used as "default-server" setting to reset any previous
13970 "default-server" "tfo" setting.
13971
Simon Hormanfa461682011-06-25 09:39:49 +090013972non-stick
13973 Never add connections allocated to this sever to a stick-table.
13974 This may be used in conjunction with backup to ensure that
13975 stick-table persistence is disabled for backup servers.
13976
Olivier Houchardc7566002018-11-20 23:33:50 +010013977npn <protocols>
13978 This enables the NPN TLS extension and advertises the specified protocol list
13979 as supported on top of NPN. The protocol list consists in a comma-delimited
13980 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roeslerfb2fce12019-07-10 15:45:51 -050013981 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010013982 enabled (check with haproxy -vv). Note that the NPN extension has been
13983 replaced with the ALPN extension (see the "alpn" keyword), though this one is
13984 only available starting with OpenSSL 1.0.2.
13985
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013986observe <mode>
13987 This option enables health adjusting based on observing communication with
13988 the server. By default this functionality is disabled and enabling it also
13989 requires to enable health checks. There are two supported modes: "layer4" and
13990 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
13991 significant. In layer7, which is only allowed for http proxies, responses
13992 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010013993 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013994
13995 See also the "check", "on-error" and "error-limit".
13996
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010013997on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010013998 Select what should happen when enough consecutive errors are detected.
13999 Currently, four modes are available:
14000 - fastinter: force fastinter
14001 - fail-check: simulate a failed check, also forces fastinter (default)
14002 - sudden-death: simulate a pre-fatal failed health check, one more failed
14003 check will mark a server down, forces fastinter
14004 - mark-down: mark the server immediately down and force fastinter
14005
14006 See also the "check", "observe" and "error-limit".
14007
Simon Hormane0d1bfb2011-06-21 14:34:58 +090014008on-marked-down <action>
14009 Modify what occurs when a server is marked down.
14010 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014011 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
14012 all connections to the server are immediately terminated when the server
14013 goes down. It might be used if the health check detects more complex cases
14014 than a simple connection status, and long timeouts would cause the service
14015 to remain unresponsive for too long a time. For instance, a health check
14016 might detect that a database is stuck and that there's no chance to reuse
14017 existing connections anymore. Connections killed this way are logged with
14018 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090014019
14020 Actions are disabled by default
14021
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014022on-marked-up <action>
14023 Modify what occurs when a server is marked up.
14024 Currently one action is available:
14025 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
14026 done only if the server is not in backup state and if it is not disabled
14027 (it must have an effective weight > 0). This can be used sometimes to force
14028 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014029 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
14030 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070014031 with extreme care. Sessions killed because a server comes up are logged
14032 with an 'U' termination code (for "Up").
14033
14034 Actions are disabled by default
14035
Willy Tarreau2f3f4d32020-07-01 07:43:51 +020014036pool-low-conn <max>
14037 Set a low threshold on the number of idling connections for a server, below
14038 which a thread will not try to steal a connection from another thread. This
14039 can be useful to improve CPU usage patterns in scenarios involving many very
14040 fast servers, in order to ensure all threads will keep a few idle connections
14041 all the time instead of letting them accumulate over one thread and migrating
14042 them from thread to thread. Typical values of twice the number of threads
14043 seem to show very good performance already with sub-millisecond response
14044 times. The default is zero, indicating that any idle connection can be used
14045 at any time. It is the recommended setting for normal use. This only applies
14046 to connections that can be shared according to the same principles as those
14047 applying to "http-reuse".
14048
Olivier Houchard006e3102018-12-10 18:30:32 +010014049pool-max-conn <max>
14050 Set the maximum number of idling connections for a server. -1 means unlimited
14051 connections, 0 means no idle connections. The default is -1. When idle
14052 connections are enabled, orphaned idle connections which do not belong to any
14053 client session anymore are moved to a dedicated pool so that they remain
14054 usable by future clients. This only applies to connections that can be shared
14055 according to the same principles as those applying to "http-reuse".
14056
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010014057pool-purge-delay <delay>
14058 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010014059 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020014060 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010014061
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014062port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014063 Using the "port" parameter, it becomes possible to use a different port to
14064 send health-checks. On some servers, it may be desirable to dedicate a port
14065 to a specific component able to perform complex tests which are more suitable
14066 to health-checks than the application. It is common to run a simple script in
14067 inetd for instance. This parameter is ignored if the "check" parameter is not
14068 set. See also the "addr" parameter.
14069
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014070proto <name>
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014071 Forces the multiplexer's protocol to use for the outgoing connections to this
14072 server. It must be compatible with the mode of the backend (TCP or HTTP). It
14073 must also be usable on the backend side. The list of available protocols is
14074 reported in haproxy -vv.
Daniel Corbett67a82712020-07-06 23:01:19 -040014075 Idea behind this option is to bypass the selection of the best multiplexer's
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020014076 protocol for all connections established to this server.
14077
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014078redir <prefix>
14079 The "redir" parameter enables the redirection mode for all GET and HEAD
14080 requests addressing this server. This means that instead of having HAProxy
14081 forward the request to the server, it will send an "HTTP 302" response with
14082 the "Location" header composed of this prefix immediately followed by the
14083 requested URI beginning at the leading '/' of the path component. That means
14084 that no trailing slash should be used after <prefix>. All invalid requests
14085 will be rejected, and all non-GET or HEAD requests will be normally served by
14086 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010014087 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014088 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014089 users to a remote location in case of local disaster. Main use consists in
14090 increasing bandwidth for static servers by having the clients directly
14091 connect to them. Note: never use a relative location here, it would cause a
14092 loop between the client and HAProxy!
14093
14094 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
14095
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014096rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014097 The "rise" parameter states that a server will be considered as operational
14098 after <count> consecutive successful health checks. This value defaults to 2
14099 if unspecified. See also the "check", "inter" and "fall" parameters.
14100
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020014101resolve-opts <option>,<option>,...
14102 Comma separated list of options to apply to DNS resolution linked to this
14103 server.
14104
14105 Available options:
14106
14107 * allow-dup-ip
14108 By default, HAProxy prevents IP address duplication in a backend when DNS
14109 resolution at runtime is in operation.
14110 That said, for some cases, it makes sense that two servers (in the same
14111 backend, being resolved by the same FQDN) have the same IP address.
14112 For such case, simply enable this option.
14113 This is the opposite of prevent-dup-ip.
14114
Daniel Corbettf8716912019-11-17 09:48:56 -050014115 * ignore-weight
14116 Ignore any weight that is set within an SRV record. This is useful when
14117 you would like to control the weights using an alternate method, such as
14118 using an "agent-check" or through the runtime api.
14119
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020014120 * prevent-dup-ip
14121 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
14122 an IP address already set to a server in the same backend and sharing the
14123 same fqdn.
14124 This is the opposite of allow-dup-ip.
14125
14126 Example:
14127 backend b_myapp
14128 default-server init-addr none resolvers dns
14129 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
14130 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
14131
14132 With the option allow-dup-ip set:
14133 * if the nameserver returns a single IP address, then both servers will use
14134 it
14135 * If the nameserver returns 2 IP addresses, then each server will pick up a
14136 different address
14137
14138 Default value: not set
14139
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014140resolve-prefer <family>
14141 When DNS resolution is enabled for a server and multiple IP addresses from
14142 different families are returned, HAProxy will prefer using an IP address
14143 from the family mentioned in the "resolve-prefer" parameter.
14144 Available families: "ipv4" and "ipv6"
14145
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020014146 Default value: ipv6
14147
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014148 Example:
14149
14150 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014151
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014152resolve-net <network>[,<network[,...]]
John Roeslerfb2fce12019-07-10 15:45:51 -050014153 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014154 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010014155 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010014156 different datacenters. The latency between datacenter is not negligible, so
14157 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014158 configured network, another address is selected.
14159
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014160 Example:
14161
14162 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010014163
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014164resolvers <id>
14165 Points to an existing "resolvers" section to resolve current server's
14166 hostname.
14167
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014168 Example:
14169
14170 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014171
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014172 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014173
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010014174send-proxy
14175 The "send-proxy" parameter enforces use of the PROXY protocol over any
14176 connection established to this server. The PROXY protocol informs the other
14177 end about the layer 3/4 addresses of the incoming connection, so that it can
14178 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014179 upper layer protocol. For connections accepted by an "accept-proxy" or
14180 "accept-netscaler-cip" listener, the advertised address will be used. Only
14181 TCPv4 and TCPv6 address families are supported. Other families such as
14182 Unix sockets, will report an UNKNOWN family. Servers using this option can
14183 fully be chained to another instance of haproxy listening with an
14184 "accept-proxy" setting. This setting must not be used if the server isn't
14185 aware of the protocol. When health checks are sent to the server, the PROXY
14186 protocol is automatically used when this option is set, unless there is an
14187 explicit "port" or "addr" directive, in which case an explicit
14188 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014189 See also the "no-send-proxy" option of this section and "accept-proxy" and
14190 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010014191
David Safb76832014-05-08 23:42:08 -040014192send-proxy-v2
14193 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
14194 over any connection established to this server. The PROXY protocol informs
14195 the other end about the layer 3/4 addresses of the incoming connection, so
14196 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020014197 whatever the upper layer protocol. It also send ALPN information if an alpn
14198 have been negotiated. This setting must not be used if the server isn't aware
14199 of this version of the protocol. See also the "no-send-proxy-v2" option of
14200 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014201
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010014202proxy-v2-options <option>[,<option>]*
Tim Duesterhuscf6e0c82020-03-13 12:34:24 +010014203 The "proxy-v2-options" parameter add options to send in PROXY protocol
14204 version 2 when "send-proxy-v2" is used. Options available are:
14205
14206 - ssl : See also "send-proxy-v2-ssl".
14207 - cert-cn : See also "send-proxy-v2-ssl-cn".
14208 - ssl-cipher: Name of the used cipher.
14209 - cert-sig : Signature algorithm of the used certificate.
14210 - cert-key : Key algorithm of the used certificate
14211 - authority : Host name value passed by the client (only SNI from a TLS
14212 connection is supported).
14213 - crc32c : Checksum of the PROXYv2 header.
14214 - unique-id : Send a unique ID generated using the frontend's
14215 "unique-id-format" within the PROXYv2 header.
14216 This unique-id is primarily meant for "mode tcp". It can
14217 lead to unexpected results in "mode http", because the
14218 generated unique ID is also used for the first HTTP request
14219 within a Keep-Alive connection.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010014220
David Safb76832014-05-08 23:42:08 -040014221send-proxy-v2-ssl
14222 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
14223 2 over any connection established to this server. The PROXY protocol informs
14224 the other end about the layer 3/4 addresses of the incoming connection, so
14225 that it can know the client's address or the public address it accessed to,
14226 whatever the upper layer protocol. In addition, the SSL information extension
14227 of the PROXY protocol is added to the PROXY protocol header. This setting
14228 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014229 See also the "no-send-proxy-v2-ssl" option of this section and the
14230 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014231
14232send-proxy-v2-ssl-cn
14233 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
14234 2 over any connection established to this server. The PROXY protocol informs
14235 the other end about the layer 3/4 addresses of the incoming connection, so
14236 that it can know the client's address or the public address it accessed to,
14237 whatever the upper layer protocol. In addition, the SSL information extension
14238 of the PROXY protocol, along along with the Common Name from the subject of
14239 the client certificate (if any), is added to the PROXY protocol header. This
14240 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014241 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
14242 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040014243
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014244slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014245 The "slowstart" parameter for a server accepts a value in milliseconds which
14246 indicates after how long a server which has just come back up will run at
14247 full speed. Just as with every other time-based parameter, it can be entered
14248 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
14249 linearly from 0 to 100% during this time. The limitation applies to two
14250 parameters :
14251
14252 - maxconn: the number of connections accepted by the server will grow from 1
14253 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
14254
14255 - weight: when the backend uses a dynamic weighted algorithm, the weight
14256 grows linearly from 1 to 100%. In this case, the weight is updated at every
14257 health-check. For this reason, it is important that the "inter" parameter
14258 is smaller than the "slowstart", in order to maximize the number of steps.
14259
14260 The slowstart never applies when haproxy starts, otherwise it would cause
14261 trouble to running servers. It only applies when a server has been previously
14262 seen as failed.
14263
Willy Tarreau732eac42015-07-09 11:40:25 +020014264sni <expression>
14265 The "sni" parameter evaluates the sample fetch expression, converts it to a
14266 string and uses the result as the host name sent in the SNI TLS extension to
14267 the server. A typical use case is to send the SNI received from the client in
14268 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020014269 expression, though alternatives such as req.hdr(host) can also make sense. If
14270 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014271 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010014272 "verify" directive for more details. If you want to set a SNI for health
14273 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020014274
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014275source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020014276source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014277source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014278 The "source" parameter sets the source address which will be used when
14279 connecting to the server. It follows the exact same parameters and principle
14280 as the backend "source" keyword, except that it only applies to the server
14281 referencing it. Please consult the "source" keyword for details.
14282
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020014283 Additionally, the "source" statement on a server line allows one to specify a
14284 source port range by indicating the lower and higher bounds delimited by a
14285 dash ('-'). Some operating systems might require a valid IP address when a
14286 source port range is specified. It is permitted to have the same IP/range for
14287 several servers. Doing so makes it possible to bypass the maximum of 64k
14288 total concurrent connections. The limit will then reach 64k connections per
14289 server.
14290
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000014291 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
14292 specifying the source address without port(s).
14293
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014294ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020014295 This option enables SSL ciphering on outgoing connections to the server. It
14296 is critical to verify server certificates using "verify" when using SSL to
14297 connect to servers, otherwise the communication is prone to trivial man in
14298 the-middle attacks rendering SSL useless. When this option is used, health
14299 checks are automatically sent in SSL too unless there is a "port" or an
14300 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010014301 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
14302 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020014303
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020014304ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
14305 This option enforces use of <version> or lower when SSL is used to communicate
14306 with the server. This option is also available on global statement
14307 "ssl-default-server-options". See also "ssl-min-ver".
14308
14309ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
14310 This option enforces use of <version> or upper when SSL is used to communicate
14311 with the server. This option is also available on global statement
14312 "ssl-default-server-options". See also "ssl-max-ver".
14313
Frédéric Lécailled2376272017-03-21 18:52:12 +010014314ssl-reuse
14315 This option may be used as "server" setting to reset any "no-ssl-reuse"
14316 setting which would have been inherited from "default-server" directive as
14317 default value.
14318 It may also be used as "default-server" setting to reset any previous
14319 "default-server" "no-ssl-reuse" setting.
14320
14321stick
14322 This option may be used as "server" setting to reset any "non-stick"
14323 setting which would have been inherited from "default-server" directive as
14324 default value.
14325 It may also be used as "default-server" setting to reset any previous
14326 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020014327
Alexander Liu2a54bb72019-05-22 19:44:48 +080014328socks4 <addr>:<port>
John Roeslerfb2fce12019-07-10 15:45:51 -050014329 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080014330 server. Using this option won't force the health check to go via socks4 by
14331 default. You will have to use the keyword "check-via-socks4" to enable it.
14332
Willy Tarreau163d4622015-10-13 16:16:41 +020014333tcp-ut <delay>
14334 Sets the TCP User Timeout for all outgoing connections to this server. This
14335 option is available on Linux since version 2.6.37. It allows haproxy to
14336 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010014337 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020014338 long-lived connections experiencing long idle periods such as remote
14339 terminals or database connection pools, where the client and server timeouts
14340 must remain high to allow a long period of idle, but where it is important to
14341 detect that the server has disappeared in order to release all resources
14342 associated with its connection (and the client's session). One typical use
14343 case is also to force dead server connections to die when health checks are
14344 too slow or during a soft reload since health checks are then disabled. The
14345 argument is a delay expressed in milliseconds by default. This only works for
14346 regular TCP connections, and is ignored for other protocols.
14347
Willy Tarreau034c88c2017-01-23 23:36:45 +010014348tfo
14349 This option enables using TCP fast open when connecting to servers, on
14350 systems that support it (currently only the Linux kernel >= 4.11).
14351 See the "tfo" bind option for more information about TCP fast open.
14352 Please note that when using tfo, you should also use the "conn-failure",
14353 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécaille1b9423d2019-07-04 14:19:06 +020014354 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010014355
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014356track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020014357 This option enables ability to set the current state of the server by tracking
14358 another one. It is possible to track a server which itself tracks another
14359 server, provided that at the end of the chain, a server has health checks
14360 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014361 used, it has to be enabled on both proxies.
14362
Frédéric Lécailled2376272017-03-21 18:52:12 +010014363tls-tickets
14364 This option may be used as "server" setting to reset any "no-tls-tickets"
14365 setting which would have been inherited from "default-server" directive as
14366 default value.
Lukas Tribusbdb386d2020-03-10 00:56:09 +010014367 The TLS ticket mechanism is only used up to TLS 1.2.
14368 Forward Secrecy is compromised with TLS tickets, unless ticket keys
14369 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010014370 It may also be used as "default-server" setting to reset any previous
Bjoern Jacke5ab7eb62020-02-13 14:16:16 +010014371 "default-server" "no-tls-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014372
Emeric Brunef42d922012-10-11 16:11:36 +020014373verify [none|required]
14374 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010014375 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014376 certificate provided by the server is verified using CAs from 'ca-file' and
14377 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010014378 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014379 the name passed using the "sni" directive, or if not provided, the static
14380 host name passed using the "verifyhost" directive. When no name is found, the
14381 certificate's names are ignored. For this reason, without SNI it's important
14382 to use "verifyhost". On verification failure the handshake is aborted. It is
14383 critically important to verify server certificates when using SSL to connect
14384 to servers, otherwise the communication is prone to trivial man-in-the-middle
14385 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
14386 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020014387
Evan Broderbe554312013-06-27 00:05:25 -070014388verifyhost <hostname>
14389 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020014390 only takes effect if 'verify required' is also specified. This directive sets
14391 a default static hostname to check the server's certificate against when no
14392 SNI was used to connect to the server. If SNI is not used, this is the only
14393 way to enable hostname verification. This static hostname, when set, will
14394 also be used for health checks (which cannot provide an SNI value). If none
14395 of the hostnames in the certificate match the specified hostname, the
14396 handshake is aborted. The hostnames in the server-provided certificate may
14397 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070014398
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010014399weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014400 The "weight" parameter is used to adjust the server's weight relative to
14401 other servers. All servers will receive a load proportional to their weight
14402 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020014403 load. The default weight is 1, and the maximal value is 256. A value of 0
14404 means the server will not participate in load-balancing but will still accept
14405 persistent connections. If this parameter is used to distribute the load
14406 according to server's capacity, it is recommended to start with values which
14407 can both grow and shrink, for instance between 10 and 100 to leave enough
14408 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014409
14410
Cyril Bonté46175dd2015-07-02 22:45:32 +0200144115.3. Server IP address resolution using DNS
14412-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014413
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014414HAProxy allows using a host name on the server line to retrieve its IP address
14415using name servers. By default, HAProxy resolves the name when parsing the
Thayne McCombscdbcca92021-01-07 21:24:41 -070014416configuration file, at startup and cache the result for the process's life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014417This is not sufficient in some cases, such as in Amazon where a server's IP
14418can change after a reboot or an ELB Virtual IP can change based on current
14419workload.
14420This chapter describes how HAProxy can be configured to process server's name
14421resolution at run time.
14422Whether run time server name resolution has been enable or not, HAProxy will
14423carry on doing the first resolution when parsing the configuration.
14424
14425
Cyril Bonté46175dd2015-07-02 22:45:32 +0200144265.3.1. Global overview
14427----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014428
14429As we've seen in introduction, name resolution in HAProxy occurs at two
14430different steps of the process life:
14431
14432 1. when starting up, HAProxy parses the server line definition and matches a
14433 host name. It uses libc functions to get the host name resolved. This
14434 resolution relies on /etc/resolv.conf file.
14435
Christopher Faulet67957bd2017-09-27 11:00:59 +020014436 2. at run time, HAProxy performs periodically name resolutions for servers
14437 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014438
14439A few other events can trigger a name resolution at run time:
14440 - when a server's health check ends up in a connection timeout: this may be
14441 because the server has a new IP address. So we need to trigger a name
14442 resolution to know this new IP.
14443
Christopher Faulet67957bd2017-09-27 11:00:59 +020014444When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010014445HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020014446SRV label is specified, then the corresponding SRV records will be retrieved
14447from the DNS server, and the provided hostnames will be used. The SRV label
14448will be checked periodically, and if any server are added or removed, haproxy
14449will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020014450
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014451A few things important to notice:
John Roeslerfb2fce12019-07-10 15:45:51 -050014452 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014453 first valid response.
14454
14455 - a resolution is considered as invalid (NX, timeout, refused), when all the
14456 servers return an error.
14457
14458
Cyril Bonté46175dd2015-07-02 22:45:32 +0200144595.3.2. The resolvers section
14460----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014461
14462This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020014463HAProxy. There can be as many as resolvers section as needed. Each section can
14464contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014465
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014466When multiple name servers are configured in a resolvers section, then HAProxy
14467uses the first valid response. In case of invalid responses, only the last one
14468is treated. Purpose is to give the chance to a slow server to deliver a valid
14469answer after a fast faulty or outdated server.
14470
14471When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020014472used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014473
Christopher Faulet67957bd2017-09-27 11:00:59 +020014474 1. HAProxy retries the same DNS query with a new query type. The A queries are
14475 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
14476 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014477
Christopher Faulet67957bd2017-09-27 11:00:59 +020014478 2. When the fallback on the query type was done (or not applicable), HAProxy
14479 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014480
Christopher Faulet67957bd2017-09-27 11:00:59 +020014481 3. HAProxy retries previous steps <resolve_retires> times. If no valid
14482 response is received after that, it stops the DNS resolution and reports
14483 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014484
Christopher Faulet67957bd2017-09-27 11:00:59 +020014485For example, with 2 name servers configured in a resolvers section, the
14486following scenarios are possible:
14487
14488 - First response is valid and is applied directly, second response is
14489 ignored
14490
14491 - First response is invalid and second one is valid, then second response is
14492 applied
14493
14494 - First response is a NX domain and second one a truncated response, then
14495 HAProxy retries the query with a new type
14496
14497 - First response is a NX domain and second one is a timeout, then HAProxy
14498 retries the query with a new type
14499
14500 - Query timed out for both name servers, then HAProxy retries it with the
14501 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014502
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014503As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
14504a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020014505<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014506
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014507
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014508resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014509 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014510
14511A resolvers section accept the following parameters:
14512
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014513accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010014514 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020014515 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020014516 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
14517 by RFC 6891)
14518
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020014519 Note: the maximum allowed value is 8192.
14520
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014521nameserver <id> <ip>:<port>
14522 DNS server description:
14523 <id> : label of the server, should be unique
14524 <ip> : IP address of the server
14525 <port> : port where the DNS service actually runs
14526
Ben Draut44e609b2018-05-29 15:40:08 -060014527parse-resolv-conf
14528 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
14529 list. Ordered as if each nameserver in /etc/resolv.conf was individually
14530 placed in the resolvers section in place of this directive.
14531
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014532hold <status> <period>
14533 Defines <period> during which the last name resolution should be kept based
14534 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014535 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014536 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014537 <period> : interval between two successive name resolution when the last
14538 answer was in <status>. It follows the HAProxy time format.
14539 <period> is in milliseconds by default.
14540
Baptiste Assmann686408b2017-08-18 10:15:42 +020014541 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014542
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014543resolve_retries <nb>
14544 Defines the number <nb> of queries to send to resolve a server name before
14545 giving up.
14546 Default value: 3
14547
Baptiste Assmann62b75b42015-09-09 01:11:36 +020014548 A retry occurs on name server timeout or when the full sequence of DNS query
14549 type failover is over and we need to start up from the default ANY query
14550 type.
14551
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014552timeout <event> <time>
14553 Defines timeouts related to name resolution
14554 <event> : the event on which the <time> timeout period applies to.
14555 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014556 - resolve : default time to trigger name resolutions when no
14557 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014558 Default value: 1s
14559 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014560 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020014561 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014562 <time> : time related to the event. It follows the HAProxy time format.
14563 <time> is expressed in milliseconds.
14564
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020014565 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014566
14567 resolvers mydns
14568 nameserver dns1 10.0.0.1:53
14569 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060014570 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014571 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020014572 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014573 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010014574 hold other 30s
14575 hold refused 30s
14576 hold nx 30s
14577 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014578 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020014579 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020014580
14581
Christopher Faulet87f1f3d2019-07-18 14:51:20 +0200145826. Cache
14583---------
14584
14585HAProxy provides a cache, which was designed to perform cache on small objects
14586(favicon, css...). This is a minimalist low-maintenance cache which runs in
14587RAM.
14588
14589The cache is based on a memory which is shared between processes and threads,
14590this memory is split in blocks of 1k.
14591
14592If an object is not used anymore, it can be deleted to store a new object
14593independently of its expiration date. The oldest objects are deleted first
14594when we try to allocate a new one.
14595
14596The cache uses a hash of the host header and the URI as the key.
14597
14598It's possible to view the status of a cache using the Unix socket command
14599"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
14600for more details.
14601
14602When an object is delivered from the cache, the server name in the log is
14603replaced by "<CACHE>".
14604
14605
146066.1. Limitation
14607----------------
14608
14609The cache won't store and won't deliver objects in these cases:
14610
14611- If the response is not a 200
Remi Tricot-Le Breton4f730832020-11-26 15:51:50 +010014612- If the response contains a Vary header and either the process-vary option is
14613 disabled, or a currently unmanaged header is specified in the Vary value (only
14614 accept-encoding and referer are managed for now)
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020014615- If the Content-Length + the headers size is greater than "max-object-size"
14616- If the response is not cacheable
Remi Tricot-Le Bretond493bc82020-11-26 15:51:29 +010014617- If the response does not have an explicit expiration time (s-maxage or max-age
14618 Cache-Control directives or Expires header) or a validator (ETag or Last-Modified
14619 headers)
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010014620- If the process-vary option is enabled and there are already max-secondary-entries
14621 entries with the same primary key as the current response
Remi Tricot-Le Breton6ca89162021-01-07 14:50:51 +010014622- If the process-vary option is enabled and the response has an unknown encoding (not
14623 mentioned in https://www.iana.org/assignments/http-parameters/http-parameters.xhtml)
14624 while varying on the accept-encoding client header
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020014625
14626- If the request is not a GET
14627- If the HTTP version of the request is smaller than 1.1
14628- If the request contains an Authorization header
14629
14630
146316.2. Setup
14632-----------
14633
14634To setup a cache, you must define a cache section and use it in a proxy with
14635the corresponding http-request and response actions.
14636
14637
146386.2.1. Cache section
14639---------------------
14640
14641cache <name>
14642 Declare a cache section, allocate a shared cache memory named <name>, the
14643 size of cache is mandatory.
14644
14645total-max-size <megabytes>
14646 Define the size in RAM of the cache in megabytes. This size is split in
14647 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
14648
14649max-object-size <bytes>
14650 Define the maximum size of the objects to be cached. Must not be greater than
14651 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
14652 All objects with sizes larger than "max-object-size" will not be cached.
14653
14654max-age <seconds>
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010014655 Define the maximum expiration duration. The expiration is set as the lowest
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020014656 value between the s-maxage or max-age (in this order) directive in the
14657 Cache-Control response header and this value. The default value is 60
14658 seconds, which means that you can't cache an object more than 60 seconds by
14659 default.
14660
Remi Tricot-Le Bretone6cc5b52020-12-23 18:13:53 +010014661process-vary <on/off>
14662 Enable or disable the processing of the Vary header. When disabled, a response
Remi Tricot-Le Breton754b2422020-11-16 15:56:10 +010014663 containing such a header will never be cached. When enabled, we need to calculate
14664 a preliminary hash for a subset of request headers on all the incoming requests
14665 (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 +010014666 for a given request (see RFC 7234#4.1). The default value is off (disabled).
Remi Tricot-Le Breton754b2422020-11-16 15:56:10 +010014667
Remi Tricot-Le Breton5853c0c2020-12-10 17:58:43 +010014668max-secondary-entries <number>
14669 Define the maximum number of simultaneous secondary entries with the same primary
14670 key in the cache. This needs the vary support to be enabled. Its default value is 10
14671 and should be passed a strictly positive integer.
14672
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020014673
146746.2.2. Proxy section
14675---------------------
14676
14677http-request cache-use <name> [ { if | unless } <condition> ]
14678 Try to deliver a cached object from the cache <name>. This directive is also
14679 mandatory to store the cache as it calculates the cache hash. If you want to
14680 use a condition for both storage and delivering that's a good idea to put it
14681 after this one.
14682
14683http-response cache-store <name> [ { if | unless } <condition> ]
14684 Store an http-response within the cache. The storage of the response headers
14685 is done at this step, which means you can use others http-response actions
14686 to modify headers before or after the storage of the response. This action
14687 is responsible for the setup of the cache storage filter.
14688
14689
14690Example:
14691
14692 backend bck1
14693 mode http
14694
14695 http-request cache-use foobar
14696 http-response cache-store foobar
14697 server srv1 127.0.0.1:80
14698
14699 cache foobar
14700 total-max-size 4
14701 max-age 240
14702
14703
Willy Tarreau74ca5042013-06-11 23:12:07 +0200147047. Using ACLs and fetching samples
14705----------------------------------
14706
Davor Ocelice9ed2812017-12-25 17:49:28 +010014707HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014708client or server information, from tables, environmental information etc...
14709The action of extracting such data is called fetching a sample. Once retrieved,
14710these samples may be used for various purposes such as a key to a stick-table,
14711but most common usages consist in matching them against predefined constant
14712data called patterns.
14713
14714
147157.1. ACL basics
14716---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014717
14718The use of Access Control Lists (ACL) provides a flexible solution to perform
14719content switching and generally to take decisions based on content extracted
14720from the request, the response or any environmental status. The principle is
14721simple :
14722
Willy Tarreau74ca5042013-06-11 23:12:07 +020014723 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014724 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020014725 - apply one or multiple pattern matching methods on this sample
14726 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014727
Willy Tarreau74ca5042013-06-11 23:12:07 +020014728The actions generally consist in blocking a request, selecting a backend, or
14729adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014730
14731In order to define a test, the "acl" keyword is used. The syntax is :
14732
Willy Tarreau74ca5042013-06-11 23:12:07 +020014733 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014734
14735This creates a new ACL <aclname> or completes an existing one with new tests.
14736Those tests apply to the portion of request/response specified in <criterion>
14737and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014738an operator which may be specified before the set of values. Optionally some
14739conversion operators may be applied to the sample, and they will be specified
14740as a comma-delimited list of keywords just after the first keyword. The values
14741are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014742
14743ACL names must be formed from upper and lower case letters, digits, '-' (dash),
14744'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
14745which means that "my_acl" and "My_Acl" are two different ACLs.
14746
14747There is no enforced limit to the number of ACLs. The unused ones do not affect
14748performance, they just consume a small amount of memory.
14749
Willy Tarreau74ca5042013-06-11 23:12:07 +020014750The criterion generally is the name of a sample fetch method, or one of its ACL
14751specific declinations. The default test method is implied by the output type of
14752this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014753methods of a same sample fetch method. The sample fetch methods are the only
14754ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014755
14756Sample fetch methods return data which can be of the following types :
14757 - boolean
14758 - integer (signed or unsigned)
14759 - IPv4 or IPv6 address
14760 - string
14761 - data block
14762
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014763Converters transform any of these data into any of these. For example, some
14764converters might convert a string to a lower-case string while other ones
14765would turn a string to an IPv4 address, or apply a netmask to an IP address.
14766The resulting sample is of the type of the last converter applied to the list,
14767which defaults to the type of the sample fetch method.
14768
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014769Each sample or converter returns data of a specific type, specified with its
14770keyword in this documentation. When an ACL is declared using a standard sample
14771fetch method, certain types automatically involved a default matching method
14772which are summarized in the table below :
14773
14774 +---------------------+-----------------+
14775 | Sample or converter | Default |
14776 | output type | matching method |
14777 +---------------------+-----------------+
14778 | boolean | bool |
14779 +---------------------+-----------------+
14780 | integer | int |
14781 +---------------------+-----------------+
14782 | ip | ip |
14783 +---------------------+-----------------+
14784 | string | str |
14785 +---------------------+-----------------+
14786 | binary | none, use "-m" |
14787 +---------------------+-----------------+
14788
14789Note that in order to match a binary samples, it is mandatory to specify a
14790matching method, see below.
14791
Willy Tarreau74ca5042013-06-11 23:12:07 +020014792The ACL engine can match these types against patterns of the following types :
14793 - boolean
14794 - integer or integer range
14795 - IP address / network
14796 - string (exact, substring, suffix, prefix, subdir, domain)
14797 - regular expression
14798 - hex block
14799
Willy Tarreauc57f0e22009-05-10 13:12:33 +020014800The following ACL flags are currently supported :
14801
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014802 -i : ignore case during matching of all subsequent patterns.
14803 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014804 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014805 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010014806 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010014807 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020014808 -- : force end of flags. Useful when a string looks like one of the flags.
14809
Willy Tarreau74ca5042013-06-11 23:12:07 +020014810The "-f" flag is followed by the name of a file from which all lines will be
14811read as individual values. It is even possible to pass multiple "-f" arguments
14812if the patterns are to be loaded from multiple files. Empty lines as well as
14813lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
14814will be stripped. If it is absolutely necessary to insert a valid pattern
14815beginning with a sharp, just prefix it with a space so that it is not taken for
14816a comment. Depending on the data type and match method, haproxy may load the
14817lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
14818exact string matching. In this case, duplicates will automatically be removed.
14819
Thierry FOURNIER9860c412014-01-29 14:23:29 +010014820The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
14821parsed as two column file. The first column contains the patterns used by the
14822ACL, and the second column contain the samples. The sample can be used later by
14823a map. This can be useful in some rare cases where an ACL would just be used to
14824check for the existence of a pattern in a map before a mapping is applied.
14825
Thierry FOURNIER3534d882014-01-20 17:01:44 +010014826The "-u" flag forces the unique id of the ACL. This unique id is used with the
14827socket interface to identify ACL and dynamically change its values. Note that a
14828file is always identified by its name even if an id is set.
14829
Willy Tarreau74ca5042013-06-11 23:12:07 +020014830Also, note that the "-i" flag applies to subsequent entries and not to entries
14831loaded from files preceding it. For instance :
14832
14833 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
14834
14835In this example, each line of "exact-ua.lst" will be exactly matched against
14836the "user-agent" header of the request. Then each line of "generic-ua" will be
14837case-insensitively matched. Then the word "test" will be insensitively matched
14838as well.
14839
14840The "-m" flag is used to select a specific pattern matching method on the input
14841sample. All ACL-specific criteria imply a pattern matching method and generally
14842do not need this flag. However, this flag is useful with generic sample fetch
14843methods to describe how they're going to be matched against the patterns. This
14844is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010014845obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020014846followed by a pattern matching method name, this method is used instead of the
14847default one for the criterion. This makes it possible to match contents in ways
14848that were not initially planned, or with sample fetch methods which return a
14849string. The matching method also affects the way the patterns are parsed.
14850
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014851The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
14852By default, if the parser cannot parse ip address it considers that the parsed
14853string is maybe a domain name and try dns resolution. The flag "-n" disable this
14854resolution. It is useful for detecting malformed ip lists. Note that if the DNS
14855server is not reachable, the haproxy configuration parsing may last many minutes
John Roeslerfb2fce12019-07-10 15:45:51 -050014856waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010014857flag "-n" disable this behavior. Note also that during the runtime, this
14858function is disabled for the dynamic acl modifications.
14859
Willy Tarreau74ca5042013-06-11 23:12:07 +020014860There are some restrictions however. Not all methods can be used with all
14861sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
14862be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020014863
14864 - "found" : only check if the requested sample could be found in the stream,
14865 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020014866 to pass any pattern to avoid confusion. This matching method is
14867 particularly useful to detect presence of certain contents such
14868 as headers, cookies, etc... even if they are empty and without
14869 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014870
14871 - "bool" : check the value as a boolean. It can only be applied to fetches
14872 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014873 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014874
14875 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020014876 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014877
14878 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020014879 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014880
Davor Ocelice9ed2812017-12-25 17:49:28 +010014881 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020014882 binary sequence. This may be used with binary or string samples.
14883
14884 - "len" : match the sample's length as an integer. This may be used with
14885 binary or string samples.
14886
Willy Tarreau74ca5042013-06-11 23:12:07 +020014887 - "str" : exact match : match the contents against a string. This may be
14888 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014889
Willy Tarreau74ca5042013-06-11 23:12:07 +020014890 - "sub" : substring match : check that the contents contain at least one of
14891 the provided string patterns. This may be used with binary or
14892 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014893
Willy Tarreau74ca5042013-06-11 23:12:07 +020014894 - "reg" : regex match : match the contents against a list of regular
14895 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014896
Willy Tarreau74ca5042013-06-11 23:12:07 +020014897 - "beg" : prefix match : check that the contents begin like the provided
14898 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014899
Willy Tarreau74ca5042013-06-11 23:12:07 +020014900 - "end" : suffix match : check that the contents end like the provided
14901 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014902
Willy Tarreau74ca5042013-06-11 23:12:07 +020014903 - "dir" : subdir match : check that a slash-delimited portion of the
14904 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014905 This may be used with binary or string samples.
14906
Willy Tarreau74ca5042013-06-11 23:12:07 +020014907 - "dom" : domain match : check that a dot-delimited portion of the contents
14908 exactly match one of the provided string patterns. This may be
14909 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020014910
14911For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
14912request, it is possible to do :
14913
14914 acl jsess_present cook(JSESSIONID) -m found
14915
14916In order to apply a regular expression on the 500 first bytes of data in the
14917buffer, one would use the following acl :
14918
14919 acl script_tag payload(0,500) -m reg -i <script>
14920
Willy Tarreaue6b11e42013-11-26 19:02:32 +010014921On systems where the regex library is much slower when using "-i", it is
14922possible to convert the sample to lowercase before matching, like this :
14923
14924 acl script_tag payload(0,500),lower -m reg <script>
14925
Willy Tarreau74ca5042013-06-11 23:12:07 +020014926All ACL-specific criteria imply a default matching method. Most often, these
14927criteria are composed by concatenating the name of the original sample fetch
14928method and the matching method. For example, "hdr_beg" applies the "beg" match
14929to samples retrieved using the "hdr" fetch method. Since all ACL-specific
14930criteria rely on a sample fetch method, it is always possible instead to use
14931the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014932
Willy Tarreau74ca5042013-06-11 23:12:07 +020014933If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014934the matching method is simply applied to the underlying sample fetch method.
14935For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014936
Willy Tarreau74ca5042013-06-11 23:12:07 +020014937 acl short_form hdr_beg(host) www.
14938 acl alternate1 hdr_beg(host) -m beg www.
14939 acl alternate2 hdr_dom(host) -m beg www.
14940 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014941
Willy Tarreau2b5285d2010-05-09 23:45:24 +020014942
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014943The table below summarizes the compatibility matrix between sample or converter
14944types and the pattern types to fetch against. It indicates for each compatible
14945combination the name of the matching method to be used, surrounded with angle
14946brackets ">" and "<" when the method is the default one and will work by
14947default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010014948
Willy Tarreau74ca5042013-06-11 23:12:07 +020014949 +-------------------------------------------------+
14950 | Input sample type |
14951 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014952 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014953 +----------------------+---------+---------+---------+---------+---------+
14954 | none (presence only) | found | found | found | found | found |
14955 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014956 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014957 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014958 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014959 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014960 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014961 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014962 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014963 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020014964 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014965 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014966 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014967 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014968 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014969 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014970 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014971 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014972 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014973 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014974 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014975 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010014976 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020014977 +----------------------+---------+---------+---------+---------+---------+
14978 | hex block | | | | bin | bin |
14979 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020014980
14981
Willy Tarreau74ca5042013-06-11 23:12:07 +0200149827.1.1. Matching booleans
14983------------------------
14984
14985In order to match a boolean, no value is needed and all values are ignored.
14986Boolean matching is used by default for all fetch methods of type "boolean".
14987When boolean matching is used, the fetched value is returned as-is, which means
14988that a boolean "true" will always match and a boolean "false" will never match.
14989
14990Boolean matching may also be enforced using "-m bool" on fetch methods which
14991return an integer value. Then, integer value 0 is converted to the boolean
14992"false" and all other values are converted to "true".
14993
Willy Tarreau6a06a402007-07-15 20:15:28 +020014994
Willy Tarreau74ca5042013-06-11 23:12:07 +0200149957.1.2. Matching integers
14996------------------------
14997
14998Integer matching applies by default to integer fetch methods. It can also be
14999enforced on boolean fetches using "-m int". In this case, "false" is converted
15000to the integer 0, and "true" is converted to the integer 1.
15001
15002Integer matching also supports integer ranges and operators. Note that integer
15003matching only applies to positive values. A range is a value expressed with a
15004lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015005
15006For instance, "1024:65535" is a valid range to represent a range of
15007unprivileged ports, and "1024:" would also work. "0:1023" is a valid
15008representation of privileged ports, and ":1023" would also work.
15009
Willy Tarreau62644772008-07-16 18:36:06 +020015010As a special case, some ACL functions support decimal numbers which are in fact
15011two integers separated by a dot. This is used with some version checks for
15012instance. All integer properties apply to those decimal numbers, including
15013ranges and operators.
15014
Willy Tarreau6a06a402007-07-15 20:15:28 +020015015For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010015016operators with ranges does not make much sense and is strongly discouraged.
15017Similarly, it does not make much sense to perform order comparisons with a set
15018of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015019
Willy Tarreau0ba27502007-12-24 16:55:16 +010015020Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020015021
15022 eq : true if the tested value equals at least one value
15023 ge : true if the tested value is greater than or equal to at least one value
15024 gt : true if the tested value is greater than at least one value
15025 le : true if the tested value is less than or equal to at least one value
15026 lt : true if the tested value is less than at least one value
15027
Willy Tarreau0ba27502007-12-24 16:55:16 +010015028For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020015029
15030 acl negative-length hdr_val(content-length) lt 0
15031
Willy Tarreau62644772008-07-16 18:36:06 +020015032This one matches SSL versions between 3.0 and 3.1 (inclusive) :
15033
15034 acl sslv3 req_ssl_ver 3:3.1
15035
Willy Tarreau6a06a402007-07-15 20:15:28 +020015036
Willy Tarreau74ca5042013-06-11 23:12:07 +0200150377.1.3. Matching strings
15038-----------------------
15039
15040String matching applies to string or binary fetch methods, and exists in 6
15041different forms :
15042
15043 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015044 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020015045
15046 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015047 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020015048
15049 - prefix match (-m beg) : the patterns are compared with the beginning of
15050 the extracted string, and the ACL matches if any of them matches.
15051
15052 - suffix match (-m end) : the patterns are compared with the end of the
15053 extracted string, and the ACL matches if any of them matches.
15054
Baptiste Assmann33db6002016-03-06 23:32:10 +010015055 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020015056 string, delimited with slashes ("/"), and the ACL matches if any of them
15057 matches.
15058
15059 - domain match (-m dom) : the patterns are looked up inside the extracted
15060 string, delimited with dots ("."), and the ACL matches if any of them
15061 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015062
15063String matching applies to verbatim strings as they are passed, with the
15064exception of the backslash ("\") which makes it possible to escape some
15065characters such as the space. If the "-i" flag is passed before the first
15066string, then the matching will be performed ignoring the case. In order
15067to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010015068before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020015069
Mathias Weiersmuellercb250fc2019-12-02 09:43:40 +010015070Do not use string matches for binary fetches which might contain null bytes
15071(0x00), as the comparison stops at the occurrence of the first null byte.
15072Instead, convert the binary fetch to a hex string with the hex converter first.
15073
15074Example:
15075 # matches if the string <tag> is present in the binary sample
15076 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
15077
Willy Tarreau6a06a402007-07-15 20:15:28 +020015078
Willy Tarreau74ca5042013-06-11 23:12:07 +0200150797.1.4. Matching regular expressions (regexes)
15080---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020015081
15082Just like with string matching, regex matching applies to verbatim strings as
15083they are passed, with the exception of the backslash ("\") which makes it
15084possible to escape some characters such as the space. If the "-i" flag is
15085passed before the first regex, then the matching will be performed ignoring
15086the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010015087the "--" flag before the first string. Same principle applies of course to
15088match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020015089
15090
Willy Tarreau74ca5042013-06-11 23:12:07 +0200150917.1.5. Matching arbitrary data blocks
15092-------------------------------------
15093
15094It is possible to match some extracted samples against a binary block which may
15095not safely be represented as a string. For this, the patterns must be passed as
15096a series of hexadecimal digits in an even number, when the match method is set
15097to binary. Each sequence of two digits will represent a byte. The hexadecimal
15098digits may be used upper or lower case.
15099
15100Example :
15101 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
15102 acl hello payload(0,6) -m bin 48656c6c6f0a
15103
15104
151057.1.6. Matching IPv4 and IPv6 addresses
15106---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020015107
15108IPv4 addresses values can be specified either as plain addresses or with a
15109netmask appended, in which case the IPv4 address matches whenever it is
15110within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010015111host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010015112difficult to read and debug configurations. If hostnames are used, you should
15113at least ensure that they are present in /etc/hosts so that the configuration
15114does not depend on any random DNS match at the moment the configuration is
15115parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015116
Daniel Schnellereba56342016-04-13 00:26:52 +020015117The dotted IPv4 address notation is supported in both regular as well as the
15118abbreviated form with all-0-octets omitted:
15119
15120 +------------------+------------------+------------------+
15121 | Example 1 | Example 2 | Example 3 |
15122 +------------------+------------------+------------------+
15123 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
15124 | 192.168.1 | 10.12 | 127.1 |
15125 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
15126 | 192.168.1/22 | 10.12/8 | 127.1/8 |
15127 +------------------+------------------+------------------+
15128
15129Notice that this is different from RFC 4632 CIDR address notation in which
15130192.168.42/24 would be equivalent to 192.168.42.0/24.
15131
Willy Tarreauceb4ac92012-04-28 00:41:46 +020015132IPv6 may be entered in their usual form, with or without a netmask appended.
15133Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
15134trouble with randomly resolved IP addresses, host names are never allowed in
15135IPv6 patterns.
15136
15137HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
15138following situations :
15139 - tested address is IPv4, pattern address is IPv4, the match applies
15140 in IPv4 using the supplied mask if any.
15141 - tested address is IPv6, pattern address is IPv6, the match applies
15142 in IPv6 using the supplied mask if any.
15143 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
15144 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
15145 ::IPV4 or ::ffff:IPV4, otherwise it fails.
15146 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
15147 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
15148 applied in IPv6 using the supplied IPv6 mask.
15149
Willy Tarreau74ca5042013-06-11 23:12:07 +020015150
151517.2. Using ACLs to form conditions
15152----------------------------------
15153
15154Some actions are only performed upon a valid condition. A condition is a
15155combination of ACLs with operators. 3 operators are supported :
15156
15157 - AND (implicit)
15158 - OR (explicit with the "or" keyword or the "||" operator)
15159 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020015160
Willy Tarreau74ca5042013-06-11 23:12:07 +020015161A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020015162
Willy Tarreau74ca5042013-06-11 23:12:07 +020015163 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020015164
Willy Tarreau74ca5042013-06-11 23:12:07 +020015165Such conditions are generally used after an "if" or "unless" statement,
15166indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020015167
Willy Tarreau74ca5042013-06-11 23:12:07 +020015168For instance, to block HTTP requests to the "*" URL with methods other than
15169"OPTIONS", as well as POST requests without content-length, and GET or HEAD
15170requests with a content-length greater than 0, and finally every request which
15171is not either GET/HEAD/POST/OPTIONS !
15172
15173 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015174 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
15175 http-request deny if METH_GET HTTP_CONTENT
15176 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020015177
15178To select a different backend for requests to static contents on the "www" site
15179and to every request on the "img", "video", "download" and "ftp" hosts :
15180
15181 acl url_static path_beg /static /images /img /css
15182 acl url_static path_end .gif .png .jpg .css .js
15183 acl host_www hdr_beg(host) -i www
15184 acl host_static hdr_beg(host) -i img. video. download. ftp.
15185
Davor Ocelice9ed2812017-12-25 17:49:28 +010015186 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020015187 # of host "www". Use backend "www" for the rest.
15188 use_backend static if host_static or host_www url_static
15189 use_backend www if host_www
15190
15191It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
15192expressions that are built on the fly without needing to be declared. They must
15193be enclosed between braces, with a space before and after each brace (because
15194the braces must be seen as independent words). Example :
15195
15196 The following rule :
15197
15198 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015199 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020015200
15201 Can also be written that way :
15202
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015203 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020015204
15205It is generally not recommended to use this construct because it's a lot easier
15206to leave errors in the configuration when written that way. However, for very
15207simple rules matching only one source IP address for instance, it can make more
15208sense to use them than to declare ACLs with random names. Another example of
15209good use is the following :
15210
15211 With named ACLs :
15212
15213 acl site_dead nbsrv(dynamic) lt 2
15214 acl site_dead nbsrv(static) lt 2
15215 monitor fail if site_dead
15216
15217 With anonymous ACLs :
15218
15219 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
15220
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030015221See section 4.2 for detailed help on the "http-request deny" and "use_backend"
15222keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015223
15224
152257.3. Fetching samples
15226---------------------
15227
15228Historically, sample fetch methods were only used to retrieve data to match
15229against patterns using ACLs. With the arrival of stick-tables, a new class of
15230sample fetch methods was created, most often sharing the same syntax as their
15231ACL counterpart. These sample fetch methods are also known as "fetches". As
15232of now, ACLs and fetches have converged. All ACL fetch methods have been made
15233available as fetch methods, and ACLs may use any sample fetch method as well.
15234
15235This section details all available sample fetch methods and their output type.
15236Some sample fetch methods have deprecated aliases that are used to maintain
15237compatibility with existing configurations. They are then explicitly marked as
15238deprecated and should not be used in new setups.
15239
15240The ACL derivatives are also indicated when available, with their respective
15241matching methods. These ones all have a well defined default pattern matching
15242method, so it is never necessary (though allowed) to pass the "-m" option to
15243indicate how the sample will be matched using ACLs.
15244
15245As indicated in the sample type versus matching compatibility matrix above,
15246when using a generic sample fetch method in an ACL, the "-m" option is
15247mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
15248the same keyword exists as an ACL keyword and as a standard fetch method, the
15249ACL engine will automatically pick the ACL-only one by default.
15250
15251Some of these keywords support one or multiple mandatory arguments, and one or
15252multiple optional arguments. These arguments are strongly typed and are checked
15253when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010015254incorrect argument (e.g. an unresolved backend name). Fetch function arguments
15255are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020015256is optional, it will be indicated below between square brackets ('[ ]'). When
15257all arguments are optional, the parenthesis may be omitted.
15258
15259Thus, the syntax of a standard sample fetch method is one of the following :
15260 - name
15261 - name(arg1)
15262 - name(arg1,arg2)
15263
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015264
152657.3.1. Converters
15266-----------------
15267
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015268Sample fetch methods may be combined with transformations to be applied on top
15269of the fetched sample (also called "converters"). These combinations form what
15270is called "sample expressions" and the result is a "sample". Initially this
15271was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010015272has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010015273unique-id-format, add-header, ...).
15274
15275These transformations are enumerated as a series of specific keywords after the
15276sample fetch method. These keywords may equally be appended immediately after
15277the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010015278support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015279
Willy Tarreau97707872015-01-27 15:12:13 +010015280A certain category of converters are bitwise and arithmetic operators which
15281support performing basic operations on integers. Some bitwise operations are
15282supported (and, or, xor, cpl) and some arithmetic operations are supported
15283(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
15284bool) which make it possible to report a match without having to write an ACL.
15285
Willy Tarreau74ca5042013-06-11 23:12:07 +020015286The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010015287
Ben Shillitof25e8e52016-12-02 14:25:37 +00001528851d.single(<prop>[,<prop>*])
15289 Returns values for the properties requested as a string, where values are
15290 separated by the delimiter specified with "51degrees-property-separator".
15291 The device is identified using the User-Agent header passed to the
15292 converter. The function can be passed up to five property names, and if a
15293 property name can't be found, the value "NoData" is returned.
15294
15295 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010015296 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
15297 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000015298 # User-Agent passed to the converter.
15299 frontend http-in
15300 bind *:8081
15301 default_backend servers
15302 http-request set-header X-51D-DeviceTypeMobileTablet \
15303 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
15304
Willy Tarreau97707872015-01-27 15:12:13 +010015305add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015306 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015307 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015308 name. The name of the variable starts with an indication about its scope. The
15309 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015310 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015311 "sess" : the variable is shared with the whole session
15312 "txn" : the variable is shared with the transaction (request and response)
15313 "req" : the variable is shared only during request processing
15314 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015315 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015316 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015317
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010015318aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
15319 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
15320 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
15321 need to be base64 encoded and the returned result is in raw byte format.
15322 If the <aead_tag> validation fails, the converter doesn't return any data.
15323 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
15324 converter requires at least OpenSSL 1.0.1.
15325
15326 Example:
15327 http-response set-header X-Decrypted-Text %[var(txn.enc),\
15328 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
15329
Willy Tarreau97707872015-01-27 15:12:13 +010015330and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015331 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015332 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010015333 numeric value or a variable name. The name of the variable starts with an
15334 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015335 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015336 "sess" : the variable is shared with the whole session
15337 "txn" : the variable is shared with the transaction (request and response)
15338 "req" : the variable is shared only during request processing
15339 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015340 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015341 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015342
Holger Just1bfc24b2017-05-06 00:56:53 +020015343b64dec
15344 Converts (decodes) a base64 encoded input string to its binary
15345 representation. It performs the inverse operation of base64().
15346
Emeric Brun53d1a982014-04-30 18:21:37 +020015347base64
15348 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010015349 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020015350 an SSL ID can be copied in a header).
15351
Willy Tarreau97707872015-01-27 15:12:13 +010015352bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015353 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010015354 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010015355 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010015356 presence of a flag).
15357
Emeric Brun54c4ac82014-11-03 15:32:43 +010015358bytes(<offset>[,<length>])
15359 Extracts some bytes from an input binary sample. The result is a binary
15360 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010015361 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010015362
Willy Tarreau280f42b2018-02-19 15:34:12 +010015363concat([<start>],[<var>],[<end>])
15364 Concatenates up to 3 fields after the current sample which is then turned to
15365 a string. The first one, <start>, is a constant string, that will be appended
15366 immediately after the existing sample. It may be omitted if not used. The
15367 second one, <var>, is a variable name. The variable will be looked up, its
15368 contents converted to a string, and it will be appended immediately after the
15369 <first> part. If the variable is not found, nothing is appended. It may be
15370 omitted as well. The third field, <end> is a constant string that will be
15371 appended after the variable. It may also be omitted. Together, these elements
15372 allow to concatenate variables with delimiters to an existing set of
15373 variables. This can be used to build new variables made of a succession of
Willy Tarreauef21fac2020-02-14 13:37:20 +010015374 other variables, such as colon-delimited values. If commas or closing
Daniel Corbett67a82712020-07-06 23:01:19 -040015375 parenthesis are needed as delimiters, they must be protected by quotes or
Willy Tarreauef21fac2020-02-14 13:37:20 +010015376 backslashes, themselves protected so that they are not stripped by the first
15377 level parser. See examples below.
Willy Tarreau280f42b2018-02-19 15:34:12 +010015378
15379 Example:
15380 tcp-request session set-var(sess.src) src
15381 tcp-request session set-var(sess.dn) ssl_c_s_dn
15382 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
Willy Tarreauef21fac2020-02-14 13:37:20 +010015383 tcp-request session set-var(txn.ipport) "str(),concat('addr=(',sess.ip),concat(',',sess.port,')')"
Willy Tarreau280f42b2018-02-19 15:34:12 +010015384 http-request set-header x-hap-sig %[var(txn.sig)]
15385
Willy Tarreau97707872015-01-27 15:12:13 +010015386cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015387 Takes the input value of type signed integer, applies a ones-complement
15388 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010015389
Willy Tarreau80599772015-01-20 19:35:24 +010015390crc32([<avalanche>])
15391 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
15392 hash function. Optionally, it is possible to apply a full avalanche hash
15393 function to the output if the optional <avalanche> argument equals 1. This
15394 converter uses the same functions as used by the various hash-based load
15395 balancing algorithms, so it will provide exactly the same results. It is
15396 provided for compatibility with other software which want a CRC32 to be
15397 computed on some input keys, so it follows the most common implementation as
15398 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
15399 but may provide a better or at least less predictable distribution. It must
15400 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015401 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
15402
15403crc32c([<avalanche>])
15404 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
15405 hash function. Optionally, it is possible to apply a full avalanche hash
15406 function to the output if the optional <avalanche> argument equals 1. This
15407 converter uses the same functions as described in RFC4960, Appendix B [8].
15408 It is provided for compatibility with other software which want a CRC32C to be
15409 computed on some input keys. It is slower than the other algorithms and it must
15410 not be used for security purposes as a 32-bit hash is trivial to break. See
15411 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010015412
Christopher Fauletea159d62020-04-01 16:21:44 +020015413cut_crlf
15414 Cuts the string representation of the input sample on the first carriage
15415 return ('\r') or newline ('\n') character found. Only the string length is
15416 updated.
15417
David Carlier29b3ca32015-09-25 14:09:21 +010015418da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020015419 Asks the DeviceAtlas converter to identify the User Agent string passed on
15420 input, and to emit a string made of the concatenation of the properties
15421 enumerated in argument, delimited by the separator defined by the global
15422 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000015423 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020015424 configuration language.
15425
15426 Example:
15427 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015428 bind *:8881
15429 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000015430 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 +020015431
Willy Tarreau0851fd52019-12-17 10:07:25 +010015432debug([<prefix][,<destination>])
15433 This converter is used as debug tool. It takes a capture of the input sample
15434 and sends it to event sink <destination>, which may designate a ring buffer
15435 such as "buf0", as well as "stdout", or "stderr". Available sinks may be
15436 checked at run time by issuing "show events" on the CLI. When not specified,
15437 the output will be "buf0", which may be consulted via the CLI's "show events"
15438 command. An optional prefix <prefix> may be passed to help distinguish
15439 outputs from multiple expressions. It will then appear before the colon in
15440 the output message. The input sample is passed as-is on the output, so that
15441 it is safe to insert the debug converter anywhere in a chain, even with non-
15442 printable sample types.
15443
15444 Example:
15445 tcp-request connection track-sc0 src,debug(track-sc)
Thierry FOURNIER9687c772015-05-07 15:46:29 +020015446
Patrick Gansterer8e366512020-04-22 16:47:57 +020015447digest(<algorithm>)
15448 Converts a binary input sample to a message digest. The result is a binary
15449 sample. The <algorithm> must be an OpenSSL message digest name (e.g. sha256).
15450
15451 Please note that this converter is only available when haproxy has been
15452 compiled with USE_OPENSSL.
15453
Willy Tarreau97707872015-01-27 15:12:13 +010015454div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015455 Divides the input value of type signed integer by <value>, and returns the
15456 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015457 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010015458 variable name. The name of the variable starts with an indication about its
15459 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015460 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015461 "sess" : the variable is shared with the whole session
15462 "txn" : the variable is shared with the transaction (request and response)
15463 "req" : the variable is shared only during request processing
15464 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015465 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015466 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015467
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015468djb2([<avalanche>])
15469 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
15470 hash function. Optionally, it is possible to apply a full avalanche hash
15471 function to the output if the optional <avalanche> argument equals 1. This
15472 converter uses the same functions as used by the various hash-based load
15473 balancing algorithms, so it will provide exactly the same results. It is
15474 mostly intended for debugging, but can be used as a stick-table entry to
15475 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015476 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
15477 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015478
Willy Tarreau97707872015-01-27 15:12:13 +010015479even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015480 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010015481 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
15482
Marcin Deranek9631a282018-04-16 14:30:46 +020015483field(<index>,<delimiters>[,<count>])
15484 Extracts the substring at the given index counting from the beginning
15485 (positive index) or from the end (negative index) considering given delimiters
15486 from an input string. Indexes start at 1 or -1 and delimiters are a string
15487 formatted list of chars. Optionally you can specify <count> of fields to
15488 extract (default: 1). Value of 0 indicates extraction of all remaining
15489 fields.
15490
15491 Example :
15492 str(f1_f2_f3__f5),field(5,_) # f5
15493 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
15494 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
15495 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
15496 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010015497
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015498fix_is_valid
15499 Parses a binary payload and performs sanity checks regarding FIX (Financial
15500 Information eXchange):
15501
15502 - checks that all tag IDs and values are not empty and the tags IDs are well
15503 numeric
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050015504 - checks the BeginString tag is the first tag with a valid FIX version
Baptiste Assmanne138dda2020-10-22 15:39:03 +020015505 - checks the BodyLength tag is the second one with the right body length
15506 - checks the MstType tag is the third tag.
15507 - checks that last tag in the message is the CheckSum tag with a valid
15508 checksum
15509
15510 Due to current HAProxy design, only the first message sent by the client and
15511 the server can be parsed.
15512
15513 This converter returns a boolean, true if the payload contains a valid FIX
15514 message, false if not.
15515
15516 See also the fix_tag_value converter.
15517
15518 Example:
15519 tcp-request inspect-delay 10s
15520 tcp-request content reject unless { req.payload(0,0),fix_is_valid }
15521
15522fix_tag_value(<tag>)
15523 Parses a FIX (Financial Information eXchange) message and extracts the value
15524 from the tag <tag>. <tag> can be a string or an integer pointing to the
15525 desired tag. Any integer value is accepted, but only the following strings
15526 are translated into their integer equivalent: BeginString, BodyLength,
15527 MsgType, SenderComID, TargetComID, CheckSum. More tag names can be easily
15528 added.
15529
15530 Due to current HAProxy design, only the first message sent by the client and
15531 the server can be parsed. No message validation is performed by this
15532 converter. It is highly recommended to validate the message first using
15533 fix_is_valid converter.
15534
15535 See also the fix_is_valid converter.
15536
15537 Example:
15538 tcp-request inspect-delay 10s
15539 tcp-request content reject unless { req.payload(0,0),fix_is_valid }
15540 # MsgType tag ID is 35, so both lines below will return the same content
15541 tcp-request content set-var(txn.foo) req.payload(0,0),fix_tag_value(35)
15542 tcp-request content set-var(txn.bar) req.payload(0,0),fix_tag_value(MsgType)
15543
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015544hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010015545 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015546 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010015547 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 +020015548 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010015549
Dragan Dosen3f957b22017-10-24 09:27:34 +020015550hex2i
15551 Converts a hex string containing two hex digits per input byte to an
John Roeslerfb2fce12019-07-10 15:45:51 -050015552 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020015553
Christopher Faulet4ccc12f2020-04-01 09:08:32 +020015554htonl
15555 Converts the input integer value to its 32-bit binary representation in the
15556 network byte order. Because sample fetches own signed 64-bit integer, when
15557 this converter is used, the input integer value is first casted to an
15558 unsigned 32-bit integer.
15559
Tim Duesterhusa3082092021-01-21 17:40:49 +010015560hmac(<algorithm>,<key>)
Patrick Gansterer8e366512020-04-22 16:47:57 +020015561 Converts a binary input sample to a message authentication code with the given
15562 key. The result is a binary sample. The <algorithm> must be one of the
15563 registered OpenSSL message digest names (e.g. sha256). The <key> parameter must
15564 be base64 encoded and can either be a string or a variable.
15565
15566 Please note that this converter is only available when haproxy has been
15567 compiled with USE_OPENSSL.
15568
Cyril Bonté6bcd1822019-11-05 23:13:59 +010015569http_date([<offset],[<unit>])
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015570 Converts an integer supposed to contain a date since epoch to a string
15571 representing this date in a format suitable for use in HTTP header fields. If
Damien Claisseae6f1252019-10-30 15:57:28 +000015572 an offset value is specified, then it is added to the date before the
15573 conversion is operated. This is particularly useful to emit Date header fields,
15574 Expires values in responses when combined with a positive offset, or
15575 Last-Modified values when the offset is negative.
15576 If a unit value is specified, then consider the timestamp as either
15577 "s" for seconds (default behavior), "ms" for milliseconds, or "us" for
15578 microseconds since epoch. Offset is assumed to have the same unit as
15579 input timestamp.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015580
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015581iif(<true>,<false>)
15582 Returns the <true> string if the input value is true. Returns the <false>
15583 string otherwise.
15584
15585 Example:
Tim Duesterhus870713b2020-09-11 17:13:12 +020015586 http-request set-header x-forwarded-proto %[ssl_fc,iif(https,http)]
Tim Duesterhus3943e4f2020-09-11 14:25:23 +020015587
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015588in_table(<table>)
15589 Uses the string representation of the input sample to perform a look up in
15590 the specified table. If the key is not found in the table, a boolean false
15591 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010015592 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020015593 or not a source IP address or an Authorization header was already seen).
15594
Tim Duesterhusa3082092021-01-21 17:40:49 +010015595ipmask(<mask4>,[<mask6>])
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015596 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015597 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010015598 table entries and as such use the same server. The mask4 can be passed in
15599 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
15600 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
15601 If no mask6 is given IPv6 addresses will fail to convert for backwards
15602 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015603
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015604json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010015605 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015606 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020015607 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015608 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
15609 of errors:
15610 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
15611 bytes, ...)
15612 - invalid range (the decoded value is within a UTF-8 prohibited range),
15613 - code overlong (the value is encoded with more bytes than necessary).
15614
15615 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
15616 character is greater than 0xffff because the JSON string escape specification
15617 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
15618 in 4 variants designated by a combination of two suffix letters : "p" for
15619 "permissive" and "s" for "silently ignore". The behaviors of the decoders
15620 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010015621 - "ascii" : never fails;
15622 - "utf8" : fails on any detected errors;
15623 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015624 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010015625 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015626 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
15627 characters corresponding to the other errors.
15628
15629 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010015630 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015631
15632 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015633 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020015634 capture request header user-agent len 150
15635 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020015636
15637 Input request from client 127.0.0.1:
15638 GET / HTTP/1.0
15639 User-Agent: Very "Ugly" UA 1/2
15640
15641 Output log:
15642 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
15643
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015644language(<value>[,<default>])
15645 Returns the value with the highest q-factor from a list as extracted from the
15646 "accept-language" header using "req.fhdr". Values with no q-factor have a
15647 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
15648 belong to the list of semi-colon delimited <values> will be considered. The
15649 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
15650 given list and a default value is provided, it is returned. Note that language
15651 names may have a variant after a dash ('-'). If this variant is present in the
15652 list, it will be matched, but if it is not, only the base language is checked.
15653 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010015654 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015655 ordering of the values in the request counts, as the first value among
15656 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020015657
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015658 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020015659
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015660 # this configuration switches to the backend matching a
15661 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020015662
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015663 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
15664 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
15665 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
15666 use_backend spanish if es
15667 use_backend french if fr
15668 use_backend english if en
15669 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020015670
Willy Tarreau60a2ee72017-12-15 07:13:48 +010015671length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010015672 Get the length of the string. This can only be placed after a string
15673 sample fetch function or after a transformation keyword returning a string
15674 type. The result is of type integer.
15675
Willy Tarreauffcb2e42014-07-10 16:29:08 +020015676lower
15677 Convert a string sample to lower case. This can only be placed after a string
15678 sample fetch function or after a transformation keyword returning a string
15679 type. The result is of type string.
15680
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015681ltime(<format>[,<offset>])
15682 Converts an integer supposed to contain a date since epoch to a string
15683 representing this date in local time using a format defined by the <format>
15684 string using strftime(3). The purpose is to allow any date format to be used
15685 in logs. An optional <offset> in seconds may be applied to the input date
15686 (positive or negative). See the strftime() man page for the format supported
15687 by your operating system. See also the utime converter.
15688
15689 Example :
15690
15691 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010015692 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020015693 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
15694
Christopher Faulet51fc9d12020-04-01 17:24:41 +020015695ltrim(<chars>)
15696 Skips any characters from <chars> from the beginning of the string
15697 representation of the input sample.
15698
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015699map(<map_file>[,<default_value>])
15700map_<match_type>(<map_file>[,<default_value>])
15701map_<match_type>_<output_type>(<map_file>[,<default_value>])
15702 Search the input value from <map_file> using the <match_type> matching method,
15703 and return the associated value converted to the type <output_type>. If the
15704 input value cannot be found in the <map_file>, the converter returns the
15705 <default_value>. If the <default_value> is not set, the converter fails and
15706 acts as if no input value could be fetched. If the <match_type> is not set, it
15707 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
15708 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
15709 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015710
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015711 It is important to avoid overlapping between the keys : IP addresses and
15712 strings are stored in trees, so the first of the finest match will be used.
15713 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015714
Tim Düsterhus4896c442016-11-29 02:15:19 +010015715 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015716 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015717
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015718 input type | match method | output type str | output type int | output type ip
15719 -----------+--------------+-----------------+-----------------+---------------
15720 str | str | map_str | map_str_int | map_str_ip
15721 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020015722 str | beg | map_beg | map_beg_int | map_end_ip
15723 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015724 str | sub | map_sub | map_sub_int | map_sub_ip
15725 -----------+--------------+-----------------+-----------------+---------------
15726 str | dir | map_dir | map_dir_int | map_dir_ip
15727 -----------+--------------+-----------------+-----------------+---------------
15728 str | dom | map_dom | map_dom_int | map_dom_ip
15729 -----------+--------------+-----------------+-----------------+---------------
15730 str | end | map_end | map_end_int | map_end_ip
15731 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080015732 str | reg | map_reg | map_reg_int | map_reg_ip
15733 -----------+--------------+-----------------+-----------------+---------------
15734 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015735 -----------+--------------+-----------------+-----------------+---------------
15736 int | int | map_int | map_int_int | map_int_ip
15737 -----------+--------------+-----------------+-----------------+---------------
15738 ip | ip | map_ip | map_ip_int | map_ip_ip
15739 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015740
Thierry Fournier8feaa662016-02-10 22:55:20 +010015741 The special map called "map_regm" expect matching zone in the regular
15742 expression and modify the output replacing back reference (like "\1") by
15743 the corresponding match text.
15744
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015745 The file contains one key + value per line. Lines which start with '#' are
15746 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
15747 is then the first "word" (series of non-space/tabs characters), and the value
15748 is what follows this series of space/tab till the end of the line excluding
15749 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015750
Thierry FOURNIER060762e2014-04-23 13:29:15 +020015751 Example :
15752
15753 # this is a comment and is ignored
15754 2.22.246.0/23 United Kingdom \n
15755 <-><-----------><--><------------><---->
15756 | | | | `- trailing spaces ignored
15757 | | | `---------- value
15758 | | `-------------------- middle spaces ignored
15759 | `---------------------------- key
15760 `------------------------------------ leading spaces ignored
15761
Willy Tarreau97707872015-01-27 15:12:13 +010015762mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015763 Divides the input value of type signed integer by <value>, and returns the
15764 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015765 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015766 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015767 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015768 "sess" : the variable is shared with the whole session
15769 "txn" : the variable is shared with the transaction (request and response)
15770 "req" : the variable is shared only during request processing
15771 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015772 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015773 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015774
Baptiste Assmanne279ca62020-10-27 18:10:06 +010015775mqtt_field_value(<packettype>,<fieldname or property ID>)
15776 Returns value of <fieldname> found in input MQTT payload of type
15777 <packettype>.
15778 <packettype> can be either a string (case insensitive matching) or a numeric
15779 value corresponding to the type of packet we're supposed to extract data
15780 from.
15781 Supported string and integers can be found here:
15782 https://docs.oasis-open.org/mqtt/mqtt/v3.1.1/os/mqtt-v3.1.1-os.html#_Toc398718021
15783 https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html#_Toc3901022
15784
15785 <fieldname> depends on <packettype> and can be any of the following below.
15786 (note that <fieldname> matching is case insensitive).
15787 <property id> can only be found in MQTT v5.0 streams. check this table:
15788 https://docs.oasis-open.org/mqtt/mqtt/v5.0/os/mqtt-v5.0-os.html#_Toc3901029
15789
15790 - CONNECT (or 1): flags, protocol_name, protocol_version, client_identifier,
15791 will_topic, will_payload, username, password, keepalive
15792 OR any property ID as a numeric value (for MQTT v5.0
15793 packets only):
15794 17: Session Expiry Interval
15795 33: Receive Maximum
15796 39: Maximum Packet Size
15797 34: Topic Alias Maximum
15798 25: Request Response Information
15799 23: Request Problem Information
15800 21: Authentication Method
15801 22: Authentication Data
15802 18: Will Delay Interval
15803 1: Payload Format Indicator
15804 2: Message Expiry Interval
15805 3: Content Type
15806 8: Response Topic
15807 9: Correlation Data
15808 Not supported yet:
15809 38: User Property
15810
15811 - CONNACK (or 2): flags, protocol_version, reason_code
15812 OR any property ID as a numeric value (for MQTT v5.0
15813 packets only):
15814 17: Session Expiry Interval
15815 33: Receive Maximum
15816 36: Maximum QoS
15817 37: Retain Available
15818 39: Maximum Packet Size
15819 18: Assigned Client Identifier
15820 34: Topic Alias Maximum
15821 31: Reason String
15822 40; Wildcard Subscription Available
15823 41: Subscription Identifiers Available
15824 42: Shared Subscription Available
15825 19: Server Keep Alive
15826 26: Response Information
15827 28: Server Reference
15828 21: Authentication Method
15829 22: Authentication Data
15830 Not supported yet:
15831 38: User Property
15832
15833 Due to current HAProxy design, only the first message sent by the client and
15834 the server can be parsed. Thus this converter can extract data only from
15835 CONNECT and CONNACK packet types. CONNECT is the first message sent by the
15836 client and CONNACK is the first response sent by the server.
15837
15838 Example:
15839
15840 acl data_in_buffer req.len ge 4
15841 tcp-request content set-var(txn.username) \
15842 req.payload(0,0),mqtt_field_value(connect,protocol_name) \
15843 if data_in_buffer
15844 # do the same as above
15845 tcp-request content set-var(txn.username) \
15846 req.payload(0,0),mqtt_field_value(1,protocol_name) \
15847 if data_in_buffer
15848
15849mqtt_is_valid
15850 Checks that the binary input is a valid MQTT packet. It returns a boolean.
15851
15852 Due to current HAProxy design, only the first message sent by the client and
15853 the server can be parsed. Thus this converter can extract data only from
15854 CONNECT and CONNACK packet types. CONNECT is the first message sent by the
15855 client and CONNACK is the first response sent by the server.
15856
15857 Example:
15858
15859 acl data_in_buffer req.len ge 4
15860 tcp-request content reject unless req.payload(0,0),mqtt_is_valid
15861
Willy Tarreau97707872015-01-27 15:12:13 +010015862mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015863 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020015864 the product as an signed integer. In case of overflow, the largest possible
15865 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015866 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010015867 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015868 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015869 "sess" : the variable is shared with the whole session
15870 "txn" : the variable is shared with the transaction (request and response)
15871 "req" : the variable is shared only during request processing
15872 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015873 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015874 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015875
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010015876nbsrv
15877 Takes an input value of type string, interprets it as a backend name and
15878 returns the number of usable servers in that backend. Can be used in places
15879 where we want to look up a backend from a dynamic name, like a result of a
15880 map lookup.
15881
Willy Tarreau97707872015-01-27 15:12:13 +010015882neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015883 Takes the input value of type signed integer, computes the opposite value,
15884 and returns the remainder as an signed integer. 0 is identity. This operator
15885 is provided for reversed subtracts : in order to subtract the input from a
15886 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010015887
15888not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015889 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010015890 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010015891 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010015892 absence of a flag).
15893
15894odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015895 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010015896 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
15897
15898or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020015899 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020015900 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010015901 numeric value or a variable name. The name of the variable starts with an
15902 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015903 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015904 "sess" : the variable is shared with the whole session
15905 "txn" : the variable is shared with the transaction (request and response)
15906 "req" : the variable is shared only during request processing
15907 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010015908 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015909 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010015910
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010015911protobuf(<field_number>,[<field_type>])
15912 This extracts the protocol buffers message field in raw mode of an input binary
15913 sample representation of a protocol buffer message with <field_number> as field
15914 number (dotted notation) if <field_type> is not present, or as an integer sample
15915 if this field is present (see also "ungrpc" below).
15916 The list of the authorized types is the following one: "int32", "int64", "uint32",
15917 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
15918 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
15919 "float" for the wire type 5. Note that "string" is considered as a length-delimited
15920 type, so it does not require any <field_type> argument to be extracted.
15921 More information may be found here about the protocol buffers message field types:
15922 https://developers.google.com/protocol-buffers/docs/encoding
15923
Willy Tarreauc4dc3502015-01-23 20:39:28 +010015924regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010015925 Applies a regex-based substitution to the input string. It does the same
15926 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
15927 default it will replace in the input string the first occurrence of the
15928 largest part matching the regular expression <regex> with the substitution
15929 string <subst>. It is possible to replace all occurrences instead by adding
15930 the flag "g" in the third argument <flags>. It is also possible to make the
15931 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
15932 string, it is made up from the concatenation of all desired flags. Thus if
15933 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
Willy Tarreauef21fac2020-02-14 13:37:20 +010015934 The first use of this converter is to replace certain characters or sequence
15935 of characters with other ones.
15936
15937 It is highly recommended to enclose the regex part using protected quotes to
15938 improve clarity and never have a closing parenthesis from the regex mixed up
15939 with the parenthesis from the function. Just like in Bourne shell, the first
15940 level of quotes is processed when delimiting word groups on the line, a
15941 second level is usable for argument. It is recommended to use single quotes
15942 outside since these ones do not try to resolve backslashes nor dollar signs.
Willy Tarreau7eda8492015-01-20 19:47:06 +010015943
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010015944 Examples:
Willy Tarreau7eda8492015-01-20 19:47:06 +010015945
15946 # de-duplicate "/" in header "x-path".
15947 # input: x-path: /////a///b/c/xzxyz/
15948 # output: x-path: /a/b/c/xzxyz/
Willy Tarreauef21fac2020-02-14 13:37:20 +010015949 http-request set-header x-path "%[hdr(x-path),regsub('/+','/','g')]"
Willy Tarreau7eda8492015-01-20 19:47:06 +010015950
Willy Tarreaucd0d2ed2020-02-14 17:33:06 +010015951 # copy query string to x-query and drop all leading '?', ';' and '&'
15952 http-request set-header x-query "%[query,regsub([?;&]*,'')]"
15953
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010015954 # capture groups and backreferences
15955 # both lines do the same.
Willy Tarreau465dc7d2020-10-08 18:05:56 +020015956 http-request redirect location %[url,'regsub("(foo|bar)([0-9]+)?","\2\1",i)']
Jerome Magnin07e1e3c2020-02-16 19:20:19 +010015957 http-request redirect location %[url,regsub(\"(foo|bar)([0-9]+)?\",\"\2\1\",i)]
15958
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015959capture-req(<id>)
15960 Capture the string entry in the request slot <id> and returns the entry as
15961 is. If the slot doesn't exist, the capture fails silently.
15962
15963 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020015964 "http-response capture", "capture.req.hdr" and
15965 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015966
15967capture-res(<id>)
15968 Capture the string entry in the response slot <id> and returns the entry as
15969 is. If the slot doesn't exist, the capture fails silently.
15970
15971 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020015972 "http-response capture", "capture.req.hdr" and
15973 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020015974
Christopher Faulet568415a2020-04-01 17:24:47 +020015975rtrim(<chars>)
15976 Skips any characters from <chars> from the end of the string representation
15977 of the input sample.
15978
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015979sdbm([<avalanche>])
15980 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
15981 hash function. Optionally, it is possible to apply a full avalanche hash
15982 function to the output if the optional <avalanche> argument equals 1. This
15983 converter uses the same functions as used by the various hash-based load
15984 balancing algorithms, so it will provide exactly the same results. It is
15985 mostly intended for debugging, but can be used as a stick-table entry to
15986 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010015987 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
15988 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015989
Tim Duesterhusf38175c2020-06-09 11:48:42 +020015990secure_memcmp(<var>)
15991 Compares the contents of <var> with the input value. Both values are treated
15992 as a binary string. Returns a boolean indicating whether both binary strings
15993 match.
15994
15995 If both binary strings have the same length then the comparison will be
15996 performed in constant time.
15997
15998 Please note that this converter is only available when haproxy has been
15999 compiled with USE_OPENSSL.
16000
16001 Example :
16002
16003 http-request set-var(txn.token) hdr(token)
16004 # Check whether the token sent by the client matches the secret token
16005 # value, without leaking the contents using a timing attack.
16006 acl token_given str(my_secret_token),secure_memcmp(txn.token)
16007
Tim Duesterhusef4e45c2021-01-21 17:40:50 +010016008set-var(<var>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010016009 Sets a variable with the input content and returns the content on the output
16010 as-is. The variable keeps the value and the associated input type. The name of
16011 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016012 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016013 "sess" : the variable is shared with the whole session
16014 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016015 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016016 "req" : the variable is shared only during request processing,
16017 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016018 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016019 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016020
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020016021sha1
Tim Duesterhusd4376302019-06-17 12:41:44 +020016022 Converts a binary input sample to a SHA-1 digest. The result is a binary
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020016023 sample with length of 20 bytes.
16024
Tim Duesterhusd4376302019-06-17 12:41:44 +020016025sha2([<bits>])
16026 Converts a binary input sample to a digest in the SHA-2 family. The result
16027 is a binary sample with length of <bits>/8 bytes.
16028
16029 Valid values for <bits> are 224, 256, 384, 512, each corresponding to
16030 SHA-<bits>. The default value is 256.
16031
16032 Please note that this converter is only available when haproxy has been
16033 compiled with USE_OPENSSL.
16034
Nenad Merdanovic177adc92019-08-27 01:58:13 +020016035srv_queue
16036 Takes an input value of type string, either a server name or <backend>/<server>
16037 format and returns the number of queued sessions on that server. Can be used
16038 in places where we want to look up queued sessions from a dynamic name, like a
16039 cookie value (e.g. req.cook(SRVID),srv_queue) and then make a decision to break
16040 persistence or direct a request elsewhere.
16041
Tim Duesterhusca097c12018-04-27 21:18:45 +020016042strcmp(<var>)
16043 Compares the contents of <var> with the input value of type string. Returns
16044 the result as a signed integer compatible with strcmp(3): 0 if both strings
16045 are identical. A value less than 0 if the left string is lexicographically
16046 smaller than the right string or if the left string is shorter. A value greater
16047 than 0 otherwise (right string greater than left string or the right string is
16048 shorter).
16049
Tim Duesterhusf38175c2020-06-09 11:48:42 +020016050 See also the secure_memcmp converter if you need to compare two binary
16051 strings in constant time.
16052
Tim Duesterhusca097c12018-04-27 21:18:45 +020016053 Example :
16054
16055 http-request set-var(txn.host) hdr(host)
16056 # Check whether the client is attempting domain fronting.
16057 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
16058
16059
Willy Tarreau97707872015-01-27 15:12:13 +010016060sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016061 Subtracts <value> from the input value of type signed integer, and returns
16062 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016063 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010016064 or a variable name. The name of the variable starts with an indication about
16065 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016066 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016067 "sess" : the variable is shared with the whole session
16068 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016069 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016070 "req" : the variable is shared only during request processing,
16071 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016072 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016073 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016074
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016075table_bytes_in_rate(<table>)
16076 Uses the string representation of the input sample to perform a look up in
16077 the specified table. If the key is not found in the table, integer value zero
16078 is returned. Otherwise the converter returns the average client-to-server
16079 bytes rate associated with the input sample in the designated table, measured
16080 in amount of bytes over the period configured in the table. See also the
16081 sc_bytes_in_rate sample fetch keyword.
16082
16083
16084table_bytes_out_rate(<table>)
16085 Uses the string representation of the input sample to perform a look up in
16086 the specified table. If the key is not found in the table, integer value zero
16087 is returned. Otherwise the converter returns the average server-to-client
16088 bytes rate associated with the input sample in the designated table, measured
16089 in amount of bytes over the period configured in the table. See also the
16090 sc_bytes_out_rate sample fetch keyword.
16091
16092table_conn_cnt(<table>)
16093 Uses the string representation of the input sample to perform a look up in
16094 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016095 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016096 connections associated with the input sample in the designated table. See
16097 also the sc_conn_cnt sample fetch keyword.
16098
16099table_conn_cur(<table>)
16100 Uses the string representation of the input sample to perform a look up in
16101 the specified table. If the key is not found in the table, integer value zero
16102 is returned. Otherwise the converter returns the current amount of concurrent
16103 tracked connections associated with the input sample in the designated table.
16104 See also the sc_conn_cur sample fetch keyword.
16105
16106table_conn_rate(<table>)
16107 Uses the string representation of the input sample to perform a look up in
16108 the specified table. If the key is not found in the table, integer value zero
16109 is returned. Otherwise the converter returns the average incoming connection
16110 rate associated with the input sample in the designated table. See also the
16111 sc_conn_rate sample fetch keyword.
16112
Thierry FOURNIER236657b2015-08-19 08:25:14 +020016113table_gpt0(<table>)
16114 Uses the string representation of the input sample to perform a look up in
16115 the specified table. If the key is not found in the table, boolean value zero
16116 is returned. Otherwise the converter returns the current value of the first
16117 general purpose tag associated with the input sample in the designated table.
16118 See also the sc_get_gpt0 sample fetch keyword.
16119
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016120table_gpc0(<table>)
16121 Uses the string representation of the input sample to perform a look up in
16122 the specified table. If the key is not found in the table, integer value zero
16123 is returned. Otherwise the converter returns the current value of the first
16124 general purpose counter associated with the input sample in the designated
16125 table. See also the sc_get_gpc0 sample fetch keyword.
16126
16127table_gpc0_rate(<table>)
16128 Uses the string representation of the input sample to perform a look up in
16129 the specified table. If the key is not found in the table, integer value zero
16130 is returned. Otherwise the converter returns the frequency which the gpc0
16131 counter was incremented over the configured period in the table, associated
16132 with the input sample in the designated table. See also the sc_get_gpc0_rate
16133 sample fetch keyword.
16134
Frédéric Lécaille6778b272018-01-29 15:22:53 +010016135table_gpc1(<table>)
16136 Uses the string representation of the input sample to perform a look up in
16137 the specified table. If the key is not found in the table, integer value zero
16138 is returned. Otherwise the converter returns the current value of the second
16139 general purpose counter associated with the input sample in the designated
16140 table. See also the sc_get_gpc1 sample fetch keyword.
16141
16142table_gpc1_rate(<table>)
16143 Uses the string representation of the input sample to perform a look up in
16144 the specified table. If the key is not found in the table, integer value zero
16145 is returned. Otherwise the converter returns the frequency which the gpc1
16146 counter was incremented over the configured period in the table, associated
16147 with the input sample in the designated table. See also the sc_get_gpc1_rate
16148 sample fetch keyword.
16149
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016150table_http_err_cnt(<table>)
16151 Uses the string representation of the input sample to perform a look up in
16152 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016153 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016154 errors associated with the input sample in the designated table. See also the
16155 sc_http_err_cnt sample fetch keyword.
16156
16157table_http_err_rate(<table>)
16158 Uses the string representation of the input sample to perform a look up in
16159 the specified table. If the key is not found in the table, integer value zero
16160 is returned. Otherwise the average rate of HTTP errors associated with the
16161 input sample in the designated table, measured in amount of errors over the
16162 period configured in the table. See also the sc_http_err_rate sample fetch
16163 keyword.
16164
16165table_http_req_cnt(<table>)
16166 Uses the string representation of the input sample to perform a look up in
16167 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016168 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016169 requests associated with the input sample in the designated table. See also
16170 the sc_http_req_cnt sample fetch keyword.
16171
16172table_http_req_rate(<table>)
16173 Uses the string representation of the input sample to perform a look up in
16174 the specified table. If the key is not found in the table, integer value zero
16175 is returned. Otherwise the average rate of HTTP requests associated with the
16176 input sample in the designated table, measured in amount of requests over the
16177 period configured in the table. See also the sc_http_req_rate sample fetch
16178 keyword.
16179
16180table_kbytes_in(<table>)
16181 Uses the string representation of the input sample to perform a look up in
16182 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016183 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016184 to-server data associated with the input sample in the designated table,
16185 measured in kilobytes. The test is currently performed on 32-bit integers,
16186 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
16187 keyword.
16188
16189table_kbytes_out(<table>)
16190 Uses the string representation of the input sample to perform a look up in
16191 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016192 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016193 to-client data associated with the input sample in the designated table,
16194 measured in kilobytes. The test is currently performed on 32-bit integers,
16195 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
16196 keyword.
16197
16198table_server_id(<table>)
16199 Uses the string representation of the input sample to perform a look up in
16200 the specified table. If the key is not found in the table, integer value zero
16201 is returned. Otherwise the converter returns the server ID associated with
16202 the input sample in the designated table. A server ID is associated to a
16203 sample by a "stick" rule when a connection to a server succeeds. A server ID
16204 zero means that no server is associated with this key.
16205
16206table_sess_cnt(<table>)
16207 Uses the string representation of the input sample to perform a look up in
16208 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010016209 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020016210 sessions associated with the input sample in the designated table. Note that
16211 a session here refers to an incoming connection being accepted by the
16212 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
16213 keyword.
16214
16215table_sess_rate(<table>)
16216 Uses the string representation of the input sample to perform a look up in
16217 the specified table. If the key is not found in the table, integer value zero
16218 is returned. Otherwise the converter returns the average incoming session
16219 rate associated with the input sample in the designated table. Note that a
16220 session here refers to an incoming connection being accepted by the
16221 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
16222 keyword.
16223
16224table_trackers(<table>)
16225 Uses the string representation of the input sample to perform a look up in
16226 the specified table. If the key is not found in the table, integer value zero
16227 is returned. Otherwise the converter returns the current amount of concurrent
16228 connections tracking the same key as the input sample in the designated
16229 table. It differs from table_conn_cur in that it does not rely on any stored
16230 information but on the table's reference count (the "use" value which is
16231 returned by "show table" on the CLI). This may sometimes be more suited for
16232 layer7 tracking. It can be used to tell a server how many concurrent
16233 connections there are from a given address for example. See also the
16234 sc_trackers sample fetch keyword.
16235
Willy Tarreauffcb2e42014-07-10 16:29:08 +020016236upper
16237 Convert a string sample to upper case. This can only be placed after a string
16238 sample fetch function or after a transformation keyword returning a string
16239 type. The result is of type string.
16240
Willy Tarreau62ba9ba2020-04-23 17:54:47 +020016241url_dec([<in_form>])
16242 Takes an url-encoded string provided as input and returns the decoded version
16243 as output. The input and the output are of type string. If the <in_form>
16244 argument is set to a non-zero integer value, the input string is assumed to
16245 be part of a form or query string and the '+' character will be turned into a
16246 space (' '). Otherwise this will only happen after a question mark indicating
16247 a query string ('?').
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020016248
William Dauchy888b0ae2021-01-06 23:39:50 +010016249url_enc([<enc_type>])
16250 Takes a string provided as input and returns the encoded version as output.
16251 The input and the output are of type string. By default the type of encoding
16252 is meant for `query` type. There is no other type supported for now but the
16253 optional argument is here for future changes.
16254
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016255ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016256 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016257 sample representation of a gRPC message with <field_number> as field number
16258 (dotted notation) if <field_type> is not present, or as an integer sample if this
16259 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016260 The list of the authorized types is the following one: "int32", "int64", "uint32",
16261 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
16262 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
16263 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010016264 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016265 More information may be found here about the protocol buffers message field types:
16266 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016267
16268 Example:
16269 // with such a protocol buffer .proto file content adapted from
16270 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
16271
16272 message Point {
16273 int32 latitude = 1;
16274 int32 longitude = 2;
16275 }
16276
16277 message PPoint {
16278 Point point = 59;
16279 }
16280
16281 message Rectangle {
16282 // One corner of the rectangle.
16283 PPoint lo = 48;
16284 // The other corner of the rectangle.
16285 PPoint hi = 49;
16286 }
16287
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016288 let's say a body request is made of a "Rectangle" object value (two PPoint
16289 protocol buffers messages), the four protocol buffers fields could be
16290 extracted with these "ungrpc" directives:
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016291
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016292 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
16293 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roeslerfb2fce12019-07-10 15:45:51 -050016294 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016295 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
16296
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016297 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016298
Frédéric Lécaille93d33162019-03-06 09:35:59 +010016299 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010016300
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016301 As a gRPC message is always made of a gRPC header followed by protocol buffers
16302 messages, in the previous example the "latitude" of "lo" first PPoint
16303 could be extracted with these equivalent directives:
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016304
16305 req.body,ungrpc(48.59),protobuf(1,int32)
16306 req.body,ungrpc(48),protobuf(59.1,int32)
16307 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
16308
Peter Gervaidf4c9d22020-06-11 18:05:11 +020016309 Note that the first convert must be "ungrpc", the remaining ones must be
16310 "protobuf" and only the last one may have or not a second argument to
16311 interpret the previous binary sample.
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010016312
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010016313
Tim Duesterhusef4e45c2021-01-21 17:40:50 +010016314unset-var(<var>)
Christopher Faulet85d79c92016-11-09 16:54:56 +010016315 Unsets a variable if the input content is defined. The name of the variable
16316 starts with an indication about its scope. The scopes allowed are:
16317 "proc" : the variable is shared with the whole process
16318 "sess" : the variable is shared with the whole session
16319 "txn" : the variable is shared with the transaction (request and
16320 response),
16321 "req" : the variable is shared only during request processing,
16322 "res" : the variable is shared only during response processing.
16323 This prefix is followed by a name. The separator is a '.'. The name may only
16324 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
16325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016326utime(<format>[,<offset>])
16327 Converts an integer supposed to contain a date since epoch to a string
16328 representing this date in UTC time using a format defined by the <format>
16329 string using strftime(3). The purpose is to allow any date format to be used
16330 in logs. An optional <offset> in seconds may be applied to the input date
16331 (positive or negative). See the strftime() man page for the format supported
16332 by your operating system. See also the ltime converter.
16333
16334 Example :
16335
16336 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010016337 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020016338 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
16339
Marcin Deranek9631a282018-04-16 14:30:46 +020016340word(<index>,<delimiters>[,<count>])
16341 Extracts the nth word counting from the beginning (positive index) or from
16342 the end (negative index) considering given delimiters from an input string.
16343 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
Jerome Magnin88209322020-01-28 13:33:44 +010016344 Delimiters at the beginning or end of the input string are ignored.
Marcin Deranek9631a282018-04-16 14:30:46 +020016345 Optionally you can specify <count> of words to extract (default: 1).
16346 Value of 0 indicates extraction of all remaining words.
16347
16348 Example :
16349 str(f1_f2_f3__f5),word(4,_) # f5
16350 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
16351 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
16352 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
16353 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Jerome Magnin88209322020-01-28 13:33:44 +010016354 str(/f1/f2/f3/f4),word(1,/) # f1
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010016355
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016356wt6([<avalanche>])
16357 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
16358 hash function. Optionally, it is possible to apply a full avalanche hash
16359 function to the output if the optional <avalanche> argument equals 1. This
16360 converter uses the same functions as used by the various hash-based load
16361 balancing algorithms, so it will provide exactly the same results. It is
16362 mostly intended for debugging, but can be used as a stick-table entry to
16363 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010016364 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
16365 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016366
Willy Tarreau97707872015-01-27 15:12:13 +010016367xor(<value>)
16368 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016369 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016370 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010016371 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016372 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016373 "sess" : the variable is shared with the whole session
16374 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016375 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016376 "req" : the variable is shared only during request processing,
16377 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020016378 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016379 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010016380
Dragan Dosen04bf0cc2020-12-22 21:44:33 +010016381xxh3([<seed>])
16382 Hashes a binary input sample into a signed 64-bit quantity using the XXH3
16383 64-bit variant of the XXhash hash function. This hash supports a seed which
16384 defaults to zero but a different value maybe passed as the <seed> argument.
16385 This hash is known to be very good and very fast so it can be used to hash
16386 URLs and/or URL parameters for use as stick-table keys to collect statistics
16387 with a low collision rate, though care must be taken as the algorithm is not
16388 considered as cryptographically secure.
16389
Thierry FOURNIER01e09742016-12-26 11:46:11 +010016390xxh32([<seed>])
16391 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
16392 variant of the XXHash hash function. This hash supports a seed which defaults
16393 to zero but a different value maybe passed as the <seed> argument. This hash
16394 is known to be very good and very fast so it can be used to hash URLs and/or
16395 URL parameters for use as stick-table keys to collect statistics with a low
16396 collision rate, though care must be taken as the algorithm is not considered
16397 as cryptographically secure.
16398
16399xxh64([<seed>])
16400 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
16401 variant of the XXHash hash function. This hash supports a seed which defaults
16402 to zero but a different value maybe passed as the <seed> argument. This hash
16403 is known to be very good and very fast so it can be used to hash URLs and/or
16404 URL parameters for use as stick-table keys to collect statistics with a low
16405 collision rate, though care must be taken as the algorithm is not considered
16406 as cryptographically secure.
16407
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010016408
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200164097.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020016410--------------------------------------------
16411
16412A first set of sample fetch methods applies to internal information which does
16413not even relate to any client information. These ones are sometimes used with
16414"monitor-fail" directives to report an internal status to external watchers.
16415The sample fetch methods described in this section are usable anywhere.
16416
16417always_false : boolean
16418 Always returns the boolean "false" value. It may be used with ACLs as a
16419 temporary replacement for another one when adjusting configurations.
16420
16421always_true : boolean
16422 Always returns the boolean "true" value. It may be used with ACLs as a
16423 temporary replacement for another one when adjusting configurations.
16424
16425avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016426 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020016427 divided by the number of active servers. The current backend is used if no
16428 backend is specified. This is very similar to "queue" except that the size of
16429 the farm is considered, in order to give a more accurate measurement of the
16430 time it may take for a new connection to be processed. The main usage is with
16431 ACL to return a sorry page to new users when it becomes certain they will get
16432 a degraded service, or to pass to the backend servers in a header so that
16433 they decide to work in degraded mode or to disable some functions to speed up
16434 the processing a bit. Note that in the event there would not be any active
16435 server anymore, twice the number of queued connections would be considered as
16436 the measured value. This is a fair estimate, as we expect one server to get
16437 back soon anyway, but we still prefer to send new traffic to another backend
16438 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
16439 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010016440
Willy Tarreau74ca5042013-06-11 23:12:07 +020016441be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020016442 Applies to the number of currently established connections on the backend,
16443 possibly including the connection being evaluated. If no backend name is
16444 specified, the current one is used. But it is also possible to check another
16445 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040016446 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
16447
16448be_conn_free([<backend>]) : integer
16449 Returns an integer value corresponding to the number of available connections
16450 across available servers in the backend. Queue slots are not included. Backup
16451 servers are also not included, unless all other servers are down. If no
16452 backend name is specified, the current one is used. But it is also possible
16453 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040016454 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
16455 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040016456
16457 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
16458 (meaning unlimited), then this fetch clearly does not make sense, in which
16459 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016460
Willy Tarreau74ca5042013-06-11 23:12:07 +020016461be_sess_rate([<backend>]) : integer
16462 Returns an integer value corresponding to the sessions creation rate on the
16463 backend, in number of new sessions per second. This is used with ACLs to
16464 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010016465 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020016466 online dictionary). It can also be useful to add this element to logs using a
16467 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016468
16469 Example :
16470 # Redirect to an error page if the dictionary is requested too often
16471 backend dynamic
16472 mode http
16473 acl being_scanned be_sess_rate gt 100
16474 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010016475
Davor Ocelice9ed2812017-12-25 17:49:28 +010016476bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016477 Returns a binary chain. The input is the hexadecimal representation
16478 of the string.
16479
16480bool(<bool>) : bool
16481 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
16482 'false' and '0' are the same. 'true' and '1' are the same.
16483
Willy Tarreau74ca5042013-06-11 23:12:07 +020016484connslots([<backend>]) : integer
16485 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016486 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020016487 connections on all servers and the maximum queue size. This is probably only
16488 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050016489
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016490 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020016491 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016492 usage; see "use_backend" keyword) can be redirected to a different backend.
16493
Willy Tarreau55165fe2009-05-10 12:02:55 +020016494 'connslots' = number of available server connection slots, + number of
16495 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016496
Willy Tarreaua36af912009-10-10 12:02:45 +020016497 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020016498 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020016499 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020016500 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010016501 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020016502 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020016503 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016504
Willy Tarreau55165fe2009-05-10 12:02:55 +020016505 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
16506 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020016507 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020016508 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080016509
Willy Tarreau70fe9442018-11-22 16:07:39 +010016510cpu_calls : integer
16511 Returns the number of calls to the task processing the stream or current
16512 request since it was allocated. This number is reset for each new request on
16513 the same connections in case of HTTP keep-alive. This value should usually be
16514 low and stable (around 2 calls for a typically simple request) but may become
16515 high if some processing (compression, caching or analysis) is performed. This
16516 is purely for performance monitoring purposes.
16517
16518cpu_ns_avg : integer
16519 Returns the average number of nanoseconds spent in each call to the task
16520 processing the stream or current request. This number is reset for each new
16521 request on the same connections in case of HTTP keep-alive. This value
16522 indicates the overall cost of processing the request or the connection for
16523 each call. There is no good nor bad value but the time spent in a call
16524 automatically causes latency for other processing (see lat_ns_avg below),
16525 and may affect other connection's apparent response time. Certain operations
16526 like compression, complex regex matching or heavy Lua operations may directly
16527 affect this value, and having it in the logs will make it easier to spot the
16528 faulty processing that needs to be fixed to recover decent performance.
16529 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
16530
16531cpu_ns_tot : integer
16532 Returns the total number of nanoseconds spent in each call to the task
16533 processing the stream or current request. This number is reset for each new
16534 request on the same connections in case of HTTP keep-alive. This value
16535 indicates the overall cost of processing the request or the connection for
16536 each call. There is no good nor bad value but the time spent in a call
16537 automatically causes latency for other processing (see lat_ns_avg below),
16538 induces CPU costs on the machine, and may affect other connection's apparent
16539 response time. Certain operations like compression, complex regex matching or
16540 heavy Lua operations may directly affect this value, and having it in the
16541 logs will make it easier to spot the faulty processing that needs to be fixed
16542 to recover decent performance. The value may be artificially high due to a
16543 high cpu_calls count, for example when processing many HTTP chunks, and for
16544 this reason it is often preferred to log cpu_ns_avg instead.
16545
Cyril Bonté6bcd1822019-11-05 23:13:59 +010016546date([<offset>],[<unit>]) : integer
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016547 Returns the current date as the epoch (number of seconds since 01/01/1970).
Damien Claisseae6f1252019-10-30 15:57:28 +000016548
16549 If an offset value is specified, then it is added to the current date before
16550 returning the value. This is particularly useful to compute relative dates,
16551 as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020016552 It is useful combined with the http_date converter.
16553
Damien Claisseae6f1252019-10-30 15:57:28 +000016554 <unit> is facultative, and can be set to "s" for seconds (default behavior),
16555 "ms" for milliseconds or "us" for microseconds.
16556 If unit is set, return value is an integer reflecting either seconds,
16557 milliseconds or microseconds since epoch, plus offset.
16558 It is useful when a time resolution of less than a second is needed.
16559
Willy Tarreau276fae92013-07-25 14:36:01 +020016560 Example :
16561
16562 # set an expires header to now+1 hour in every response
16563 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020016564
Damien Claisseae6f1252019-10-30 15:57:28 +000016565 # set an expires header to now+1 hour in every response, with
16566 # millisecond granularity
16567 http-response set-header Expires %[date(3600000,ms),http_date(0,ms)]
16568
Etienne Carrierea792a0a2018-01-17 13:43:24 +010016569date_us : integer
16570 Return the microseconds part of the date (the "second" part is returned by
16571 date sample). This sample is coherent with the date sample as it is comes
16572 from the same timeval structure.
16573
Willy Tarreaud716f9b2017-10-13 11:03:15 +020016574distcc_body(<token>[,<occ>]) : binary
16575 Parses a distcc message and returns the body associated to occurrence #<occ>
16576 of the token <token>. Occurrences start at 1, and when unspecified, any may
16577 match though in practice only the first one is checked for now. This can be
16578 used to extract file names or arguments in files built using distcc through
16579 haproxy. Please refer to distcc's protocol documentation for the complete
16580 list of supported tokens.
16581
16582distcc_param(<token>[,<occ>]) : integer
16583 Parses a distcc message and returns the parameter associated to occurrence
16584 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
16585 any may match though in practice only the first one is checked for now. This
16586 can be used to extract certain information such as the protocol version, the
16587 file size or the argument in files built using distcc through haproxy.
16588 Another use case consists in waiting for the start of the preprocessed file
16589 contents before connecting to the server to avoid keeping idle connections.
16590 Please refer to distcc's protocol documentation for the complete list of
16591 supported tokens.
16592
16593 Example :
16594 # wait up to 20s for the pre-processed file to be uploaded
16595 tcp-request inspect-delay 20s
16596 tcp-request content accept if { distcc_param(DOTI) -m found }
16597 # send large files to the big farm
16598 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
16599
Willy Tarreau595ec542013-06-12 21:34:28 +020016600env(<name>) : string
16601 Returns a string containing the value of environment variable <name>. As a
16602 reminder, environment variables are per-process and are sampled when the
16603 process starts. This can be useful to pass some information to a next hop
16604 server, or with ACLs to take specific action when the process is started a
16605 certain way.
16606
16607 Examples :
16608 # Pass the Via header to next hop with the local hostname in it
16609 http-request add-header Via 1.1\ %[env(HOSTNAME)]
16610
16611 # reject cookie-less requests when the STOP environment variable is set
16612 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
16613
Willy Tarreau74ca5042013-06-11 23:12:07 +020016614fe_conn([<frontend>]) : integer
16615 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010016616 possibly including the connection being evaluated. If no frontend name is
16617 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020016618 frontend. It can be used to return a sorry page before hard-blocking, or to
16619 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010016620 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020016621 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
16622 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020016623
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020016624fe_req_rate([<frontend>]) : integer
16625 Returns an integer value corresponding to the number of HTTP requests per
16626 second sent to a frontend. This number can differ from "fe_sess_rate" in
16627 situations where client-side keep-alive is enabled.
16628
Willy Tarreau74ca5042013-06-11 23:12:07 +020016629fe_sess_rate([<frontend>]) : integer
16630 Returns an integer value corresponding to the sessions creation rate on the
16631 frontend, in number of new sessions per second. This is used with ACLs to
16632 limit the incoming session rate to an acceptable range in order to prevent
16633 abuse of service at the earliest moment, for example when combined with other
16634 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
16635 down below the limit. It can also be useful to add this element to logs using
16636 a log-format directive. See also the "rate-limit sessions" directive for use
16637 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010016638
16639 Example :
16640 # This frontend limits incoming mails to 10/s with a max of 100
16641 # concurrent connections. We accept any connection below 10/s, and
16642 # force excess clients to wait for 100 ms. Since clients are limited to
16643 # 100 max, there cannot be more than 10 incoming mails per second.
16644 frontend mail
16645 bind :25
16646 mode tcp
16647 maxconn 100
16648 acl too_fast fe_sess_rate ge 10
16649 tcp-request inspect-delay 100ms
16650 tcp-request content accept if ! too_fast
16651 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010016652
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010016653hostname : string
16654 Returns the system hostname.
16655
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020016656int(<integer>) : signed integer
16657 Returns a signed integer.
16658
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016659ipv4(<ipv4>) : ipv4
16660 Returns an ipv4.
16661
16662ipv6(<ipv6>) : ipv6
16663 Returns an ipv6.
16664
Willy Tarreau70fe9442018-11-22 16:07:39 +010016665lat_ns_avg : integer
16666 Returns the average number of nanoseconds spent between the moment the task
16667 handling the stream is woken up and the moment it is effectively called. This
16668 number is reset for each new request on the same connections in case of HTTP
16669 keep-alive. This value indicates the overall latency inflicted to the current
16670 request by all other requests being processed in parallel, and is a direct
16671 indicator of perceived performance due to noisy neighbours. In order to keep
16672 the value low, it is possible to reduce the scheduler's run queue depth using
16673 "tune.runqueue-depth", to reduce the number of concurrent events processed at
16674 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020016675 the "nice" option on the "bind" lines or in the frontend, to enable low
16676 latency scheduling using "tune.sched.low-latency", or to look for other heavy
16677 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
16678 processing needs to be adjusted or fixed. Compression of large buffers could
16679 be a culprit, like heavy regex or long lists of regex. Note: this value is
16680 exactly lat_ns_tot divided by cpu_calls.
Willy Tarreau70fe9442018-11-22 16:07:39 +010016681
16682lat_ns_tot : integer
16683 Returns the total number of nanoseconds spent between the moment the task
16684 handling the stream is woken up and the moment it is effectively called. This
16685 number is reset for each new request on the same connections in case of HTTP
16686 keep-alive. This value indicates the overall latency inflicted to the current
16687 request by all other requests being processed in parallel, and is a direct
16688 indicator of perceived performance due to noisy neighbours. In order to keep
16689 the value low, it is possible to reduce the scheduler's run queue depth using
16690 "tune.runqueue-depth", to reduce the number of concurrent events processed at
16691 once using "tune.maxpollevents", to decrease the stream's nice value using
Willy Tarreaue7723bd2020-06-24 11:11:02 +020016692 the "nice" option on the "bind" lines or in the frontend, to enable low
16693 latency scheduling using "tune.sched.low-latency", or to look for other heavy
16694 requests in logs (those exhibiting large values of "cpu_ns_avg"), whose
16695 processing needs to be adjusted or fixed. Compression of large buffers could
16696 be a culprit, like heavy regex or long lists of regex. Note: while it
Willy Tarreau70fe9442018-11-22 16:07:39 +010016697 may intuitively seem that the total latency adds to a transfer time, it is
16698 almost never true because while a task waits for the CPU, network buffers
16699 continue to fill up and the next call will process more at once. The value
16700 may be artificially high due to a high cpu_calls count, for example when
16701 processing many HTTP chunks, and for this reason it is often preferred to log
16702 lat_ns_avg instead, which is a more relevant performance indicator.
16703
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016704meth(<method>) : method
16705 Returns a method.
16706
Willy Tarreau0f30d262014-11-24 16:02:05 +010016707nbproc : integer
16708 Returns an integer value corresponding to the number of processes that were
16709 started (it equals the global "nbproc" setting). This is useful for logging
16710 and debugging purposes.
16711
Willy Tarreau74ca5042013-06-11 23:12:07 +020016712nbsrv([<backend>]) : integer
16713 Returns an integer value corresponding to the number of usable servers of
16714 either the current backend or the named backend. This is mostly used with
16715 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010016716 switch to an alternate backend when the number of servers is too low to
16717 to handle some load. It is useful to report a failure when combined with
16718 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010016719
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040016720prio_class : integer
16721 Returns the priority class of the current session for http mode or connection
16722 for tcp mode. The value will be that set by the last call to "http-request
16723 set-priority-class" or "tcp-request content set-priority-class".
16724
16725prio_offset : integer
16726 Returns the priority offset of the current session for http mode or
16727 connection for tcp mode. The value will be that set by the last call to
16728 "http-request set-priority-offset" or "tcp-request content
16729 set-priority-offset".
16730
Willy Tarreau0f30d262014-11-24 16:02:05 +010016731proc : integer
16732 Returns an integer value corresponding to the position of the process calling
16733 the function, between 1 and global.nbproc. This is useful for logging and
16734 debugging purposes.
16735
Willy Tarreau74ca5042013-06-11 23:12:07 +020016736queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016737 Returns the total number of queued connections of the designated backend,
16738 including all the connections in server queues. If no backend name is
16739 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020016740 one. This is useful with ACLs or to pass statistics to backend servers. This
16741 can be used to take actions when queuing goes above a known level, generally
16742 indicating a surge of traffic or a massive slowdown on the servers. One
16743 possible action could be to reject new users but still accept old ones. See
16744 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
16745
Willy Tarreau84310e22014-02-14 11:59:04 +010016746rand([<range>]) : integer
16747 Returns a random integer value within a range of <range> possible values,
16748 starting at zero. If the range is not specified, it defaults to 2^32, which
16749 gives numbers between 0 and 4294967295. It can be useful to pass some values
16750 needed to take some routing decisions for example, or just for debugging
16751 purposes. This random must not be used for security purposes.
16752
Luca Schimweg8a694b82019-09-10 15:42:52 +020016753uuid([<version>]) : string
16754 Returns a UUID following the RFC4122 standard. If the version is not
16755 specified, a UUID version 4 (fully random) is returned.
16756 Currently, only version 4 is supported.
16757
Willy Tarreau74ca5042013-06-11 23:12:07 +020016758srv_conn([<backend>/]<server>) : integer
16759 Returns an integer value corresponding to the number of currently established
16760 connections on the designated server, possibly including the connection being
16761 evaluated. If <backend> is omitted, then the server is looked up in the
16762 current backend. It can be used to use a specific farm when one server is
16763 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040016764 connections with it. See also the "fe_conn", "be_conn", "queue", and
16765 "srv_conn_free" fetch methods.
16766
16767srv_conn_free([<backend>/]<server>) : integer
16768 Returns an integer value corresponding to the number of available connections
16769 on the designated server, possibly including the connection being evaluated.
16770 The value does not include queue slots. If <backend> is omitted, then the
16771 server is looked up in the current backend. It can be used to use a specific
16772 farm when one server is full, or to inform the server about our view of the
16773 number of active connections with it. See also the "be_conn_free" and
16774 "srv_conn" fetch methods.
16775
16776 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
16777 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016778
16779srv_is_up([<backend>/]<server>) : boolean
16780 Returns true when the designated server is UP, and false when it is either
16781 DOWN or in maintenance mode. If <backend> is omitted, then the server is
16782 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010016783 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020016784 availability). Another possible use which is more of a hack consists in
16785 using dummy servers as boolean variables that can be enabled or disabled from
16786 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
16787
Willy Tarreauff2b7af2017-10-13 11:46:26 +020016788srv_queue([<backend>/]<server>) : integer
16789 Returns an integer value corresponding to the number of connections currently
16790 pending in the designated server's queue. If <backend> is omitted, then the
16791 server is looked up in the current backend. It can sometimes be used together
16792 with the "use-server" directive to force to use a known faster server when it
16793 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
16794 fetch methods.
16795
Willy Tarreau74ca5042013-06-11 23:12:07 +020016796srv_sess_rate([<backend>/]<server>) : integer
16797 Returns an integer corresponding to the sessions creation rate on the
16798 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016799 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020016800 used with ACLs but can make sense with logs too. This is used to switch to an
16801 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010016802 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020016803 overloading servers).
16804
16805 Example :
16806 # Redirect to a separate back
16807 acl srv1_full srv_sess_rate(be1/srv1) gt 50
16808 acl srv2_full srv_sess_rate(be1/srv2) gt 50
16809 use_backend be2 if srv1_full or srv2_full
16810
Christopher Faulet1bea8652020-07-10 16:03:45 +020016811srv_iweight([<backend>/]<server>): integer
16812 Returns an integer corresponding to the server's initial weight. If <backend>
16813 is omitted, then the server is looked up in the current backend. See also
16814 "srv_weight" and "srv_uweight".
16815
16816srv_uweight([<backend>/]<server>): integer
16817 Returns an integer corresponding to the user visible server's weight. If
16818 <backend> is omitted, then the server is looked up in the current
16819 backend. See also "srv_weight" and "srv_iweight".
16820
16821srv_weight([<backend>/]<server>): integer
16822 Returns an integer corresponding to the current (or effective) server's
16823 weight. If <backend> is omitted, then the server is looked up in the current
16824 backend. See also "srv_iweight" and "srv_uweight".
16825
Willy Tarreau0f30d262014-11-24 16:02:05 +010016826stopping : boolean
16827 Returns TRUE if the process calling the function is currently stopping. This
16828 can be useful for logging, or for relaxing certain checks or helping close
16829 certain connections upon graceful shutdown.
16830
Thierry FOURNIERcc103292015-06-06 19:30:17 +020016831str(<string>) : string
16832 Returns a string.
16833
Willy Tarreau74ca5042013-06-11 23:12:07 +020016834table_avl([<table>]) : integer
16835 Returns the total number of available entries in the current proxy's
16836 stick-table or in the designated stick-table. See also table_cnt.
16837
16838table_cnt([<table>]) : integer
16839 Returns the total number of entries currently in use in the current proxy's
16840 stick-table or in the designated stick-table. See also src_conn_cnt and
16841 table_avl for other entry counting methods.
16842
Christopher Faulet34adb2a2017-11-21 21:45:38 +010016843thread : integer
16844 Returns an integer value corresponding to the position of the thread calling
16845 the function, between 0 and (global.nbthread-1). This is useful for logging
16846 and debugging purposes.
16847
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016848var(<var-name>) : undefined
16849 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010016850 sample fetch fails. The name of the variable starts with an indication
16851 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010016852 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010016853 "sess" : the variable is shared with the whole session
16854 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016855 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010016856 "req" : the variable is shared only during request processing,
16857 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016858 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010016859 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020016860
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200168617.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020016862----------------------------------
16863
16864The layer 4 usually describes just the transport layer which in haproxy is
16865closest to the connection, where no content is yet made available. The fetch
16866methods described here are usable as low as the "tcp-request connection" rule
16867sets unless they require some future information. Those generally include
16868TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020016869the incoming connection. For retrieving a value from a sticky counters, the
16870counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020016871"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
16872used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
16873can be specified as the first integer argument when using the "sc_" prefix.
16874Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
16875table may be specified with the "sc*" form, in which case the currently
16876tracked key will be looked up into this alternate table instead of the table
16877currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016878
Jérôme Magnin35e53a62019-01-16 14:38:37 +010016879bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010016880 Returns the backend connection's HTTP major version encoding, which may be 1
16881 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
16882 encoding and not the version present in the request header.
16883
Willy Tarreau74ca5042013-06-11 23:12:07 +020016884be_id : integer
16885 Returns an integer containing the current backend's id. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020016886 frontends with responses to check which backend processed the request. It can
16887 also be used in a tcp-check or an http-check ruleset.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016888
Marcin Deranekd2471c22016-12-12 14:08:05 +010016889be_name : string
16890 Returns a string containing the current backend's name. It can be used in
Christopher Fauletd1b44642020-04-30 09:51:15 +020016891 frontends with responses to check which backend processed the request. It can
16892 also be used in a tcp-check or an http-check ruleset.
Marcin Deranekd2471c22016-12-12 14:08:05 +010016893
Amaury Denoyelled91d7792020-12-10 13:43:56 +010016894be_server_timeout : integer
16895 Returns the configuration value in millisecond for the server timeout of the
16896 current backend. This timeout can be overwritten by a "set-timeout" rule. See
16897 also the "cur_server_timeout".
16898
16899be_tunnel_timeout : integer
16900 Returns the configuration value in millisecond for the tunnel timeout of the
16901 current backend. This timeout can be overwritten by a "set-timeout" rule. See
16902 also the "cur_tunnel_timeout".
16903
Amaury Denoyellef7719a22020-12-10 13:43:58 +010016904cur_server_timeout : integer
16905 Returns the currently applied server timeout in millisecond for the stream.
16906 In the default case, this will be equal to be_server_timeout unless a
16907 "set-timeout" rule has been applied. See also "be_server_timeout".
16908
16909cur_tunnel_timeout : integer
16910 Returns the currently applied tunnel timeout in millisecond for the stream.
16911 In the default case, this will be equal to be_tunnel_timeout unless a
16912 "set-timeout" rule has been applied. See also "be_tunnel_timeout".
16913
Willy Tarreau74ca5042013-06-11 23:12:07 +020016914dst : ip
16915 This is the destination IPv4 address of the connection on the client side,
16916 which is the address the client connected to. It can be useful when running
16917 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
16918 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010016919 RFC 4291. When the incoming connection passed through address translation or
16920 redirection involving connection tracking, the original destination address
16921 before the redirection will be reported. On Linux systems, the source and
16922 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
16923 is set, because a late response may reopen a timed out connection and switch
16924 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016925
16926dst_conn : integer
16927 Returns an integer value corresponding to the number of currently established
16928 connections on the same socket including the one being evaluated. It is
16929 normally used with ACLs but can as well be used to pass the information to
16930 servers in an HTTP header or in logs. It can be used to either return a sorry
16931 page before hard-blocking, or to use a specific backend to drain new requests
16932 when the socket is considered saturated. This offers the ability to assign
16933 different limits to different listening ports or addresses. See also the
16934 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016935
Willy Tarreau16e01562016-08-09 16:46:18 +020016936dst_is_local : boolean
16937 Returns true if the destination address of the incoming connection is local
16938 to the system, or false if the address doesn't exist on the system, meaning
16939 that it was intercepted in transparent mode. It can be useful to apply
16940 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010016941 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020016942 be delivered only on this address, or SSH access could be locally redirected.
16943 Please note that the check involves a few system calls, so it's better to do
16944 it only once per connection.
16945
Willy Tarreau74ca5042013-06-11 23:12:07 +020016946dst_port : integer
16947 Returns an integer value corresponding to the destination TCP port of the
16948 connection on the client side, which is the port the client connected to.
16949 This might be used when running in transparent mode, when assigning dynamic
16950 ports to some clients for a whole application session, to stick all users to
16951 a same server, or to pass the destination port information to a server using
16952 an HTTP header.
16953
Willy Tarreau60ca10a2017-08-18 15:26:54 +020016954fc_http_major : integer
16955 Reports the front connection's HTTP major version encoding, which may be 1
16956 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
16957 encoding and not on the version present in the request header.
16958
Geoff Simmons7185b782019-08-27 18:31:16 +020016959fc_pp_authority : string
16960 Returns the authority TLV sent by the client in the PROXY protocol header,
16961 if any.
16962
Tim Duesterhusd1b15b62020-03-13 12:34:23 +010016963fc_pp_unique_id : string
16964 Returns the unique ID TLV sent by the client in the PROXY protocol header,
16965 if any.
16966
Emeric Brun4f603012017-01-05 15:11:44 +010016967fc_rcvd_proxy : boolean
16968 Returns true if the client initiated the connection with a PROXY protocol
16969 header.
16970
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020016971fc_rtt(<unit>) : integer
16972 Returns the Round Trip Time (RTT) measured by the kernel for the client
16973 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
16974 can be set to "ms" for milliseconds or "us" for microseconds. If the server
16975 connection is not established, if the connection is not TCP or if the
16976 operating system does not support TCP_INFO, for example Linux kernels before
16977 2.4, the sample fetch fails.
16978
16979fc_rttvar(<unit>) : integer
16980 Returns the Round Trip Time (RTT) variance measured by the kernel for the
16981 client connection. <unit> is facultative, by default the unit is milliseconds.
16982 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
16983 server connection is not established, if the connection is not TCP or if the
16984 operating system does not support TCP_INFO, for example Linux kernels before
16985 2.4, the sample fetch fails.
16986
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016987fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016988 Returns the unacked counter measured by the kernel for the client connection.
16989 If the server connection is not established, if the connection is not TCP or
16990 if the operating system does not support TCP_INFO, for example Linux kernels
16991 before 2.4, the sample fetch fails.
16992
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016993fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070016994 Returns the sacked counter measured by the kernel for the client connection.
16995 If the server connection is not established, if the connection is not TCP or
16996 if the operating system does not support TCP_INFO, for example Linux kernels
16997 before 2.4, the sample fetch fails.
16998
Christopher Fauletba0c53e2019-10-17 14:40:48 +020016999fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017000 Returns the retransmits counter measured by the kernel for the client
17001 connection. If the server connection is not established, if the connection is
17002 not TCP or if the operating system does not support TCP_INFO, for example
17003 Linux kernels before 2.4, the sample fetch fails.
17004
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017005fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017006 Returns the fack counter measured by the kernel for the client
17007 connection. If the server connection is not established, if the connection is
17008 not TCP or if the operating system does not support TCP_INFO, for example
17009 Linux kernels before 2.4, the sample fetch fails.
17010
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017011fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017012 Returns the lost counter measured by the kernel for the client
17013 connection. If the server connection is not established, if the connection is
17014 not TCP or if the operating system does not support TCP_INFO, for example
17015 Linux kernels before 2.4, the sample fetch fails.
17016
Christopher Fauletba0c53e2019-10-17 14:40:48 +020017017fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070017018 Returns the reordering counter measured by the kernel for the client
17019 connection. If the server connection is not established, if the connection is
17020 not TCP or if the operating system does not support TCP_INFO, for example
17021 Linux kernels before 2.4, the sample fetch fails.
17022
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020017023fe_defbe : string
17024 Returns a string containing the frontend's default backend name. It can be
17025 used in frontends to check which backend will handle requests by default.
17026
Willy Tarreau74ca5042013-06-11 23:12:07 +020017027fe_id : integer
17028 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010017029 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020017030 coming via a same frontend to the same server.
17031
Marcin Deranekd2471c22016-12-12 14:08:05 +010017032fe_name : string
17033 Returns a string containing the current frontend's name. It can be used in
17034 backends to check from which frontend it was called, or to stick all users
17035 coming via a same frontend to the same server.
17036
Amaury Denoyelleda184d52020-12-10 13:43:55 +010017037fe_client_timeout : integer
17038 Returns the configuration value in millisecond for the client timeout of the
17039 current frontend.
17040
Cyril Bonté62ba8702014-04-22 23:52:25 +020017041sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017042sc0_bytes_in_rate([<table>]) : integer
17043sc1_bytes_in_rate([<table>]) : integer
17044sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017045 Returns the average client-to-server bytes rate from the currently tracked
17046 counters, measured in amount of bytes over the period configured in the
17047 table. See also src_bytes_in_rate.
17048
Cyril Bonté62ba8702014-04-22 23:52:25 +020017049sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017050sc0_bytes_out_rate([<table>]) : integer
17051sc1_bytes_out_rate([<table>]) : integer
17052sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017053 Returns the average server-to-client bytes rate from the currently tracked
17054 counters, measured in amount of bytes over the period configured in the
17055 table. See also src_bytes_out_rate.
17056
Cyril Bonté62ba8702014-04-22 23:52:25 +020017057sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017058sc0_clr_gpc0([<table>]) : integer
17059sc1_clr_gpc0([<table>]) : integer
17060sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020017061 Clears the first General Purpose Counter associated to the currently tracked
17062 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010017063 stored value is zero, so first invocation will always return zero. This is
17064 typically used as a second ACL in an expression in order to mark a connection
17065 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020017066
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017067 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020017068 # block if 5 consecutive requests continue to come faster than 10 sess
17069 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017070 acl abuse sc0_http_req_rate gt 10
17071 acl kill sc0_inc_gpc0 gt 5
17072 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020017073 tcp-request connection accept if !abuse save
17074 tcp-request connection reject if abuse kill
17075
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017076sc_clr_gpc1(<ctr>[,<table>]) : integer
17077sc0_clr_gpc1([<table>]) : integer
17078sc1_clr_gpc1([<table>]) : integer
17079sc2_clr_gpc1([<table>]) : integer
17080 Clears the second General Purpose Counter associated to the currently tracked
17081 counters, and returns its previous value. Before the first invocation, the
17082 stored value is zero, so first invocation will always return zero. This is
17083 typically used as a second ACL in an expression in order to mark a connection
17084 when a first ACL was verified.
17085
Cyril Bonté62ba8702014-04-22 23:52:25 +020017086sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017087sc0_conn_cnt([<table>]) : integer
17088sc1_conn_cnt([<table>]) : integer
17089sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017090 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017091 counters. See also src_conn_cnt.
17092
Cyril Bonté62ba8702014-04-22 23:52:25 +020017093sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017094sc0_conn_cur([<table>]) : integer
17095sc1_conn_cur([<table>]) : integer
17096sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017097 Returns the current amount of concurrent connections tracking the same
17098 tracked counters. This number is automatically incremented when tracking
17099 begins and decremented when tracking stops. See also src_conn_cur.
17100
Cyril Bonté62ba8702014-04-22 23:52:25 +020017101sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017102sc0_conn_rate([<table>]) : integer
17103sc1_conn_rate([<table>]) : integer
17104sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017105 Returns the average connection rate from the currently tracked counters,
17106 measured in amount of connections over the period configured in the table.
17107 See also src_conn_rate.
17108
Cyril Bonté62ba8702014-04-22 23:52:25 +020017109sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017110sc0_get_gpc0([<table>]) : integer
17111sc1_get_gpc0([<table>]) : integer
17112sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017113 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017114 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017115
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017116sc_get_gpc1(<ctr>[,<table>]) : integer
17117sc0_get_gpc1([<table>]) : integer
17118sc1_get_gpc1([<table>]) : integer
17119sc2_get_gpc1([<table>]) : integer
17120 Returns the value of the second General Purpose Counter associated to the
17121 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
17122
Thierry FOURNIER236657b2015-08-19 08:25:14 +020017123sc_get_gpt0(<ctr>[,<table>]) : integer
17124sc0_get_gpt0([<table>]) : integer
17125sc1_get_gpt0([<table>]) : integer
17126sc2_get_gpt0([<table>]) : integer
17127 Returns the value of the first General Purpose Tag associated to the
17128 currently tracked counters. See also src_get_gpt0.
17129
Cyril Bonté62ba8702014-04-22 23:52:25 +020017130sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017131sc0_gpc0_rate([<table>]) : integer
17132sc1_gpc0_rate([<table>]) : integer
17133sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017134 Returns the average increment rate of the first General Purpose Counter
17135 associated to the currently tracked counters. It reports the frequency
17136 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017137 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
17138 that the "gpc0_rate" counter must be stored in the stick-table for a value to
17139 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020017140
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017141sc_gpc1_rate(<ctr>[,<table>]) : integer
17142sc0_gpc1_rate([<table>]) : integer
17143sc1_gpc1_rate([<table>]) : integer
17144sc2_gpc1_rate([<table>]) : integer
17145 Returns the average increment rate of the second General Purpose Counter
17146 associated to the currently tracked counters. It reports the frequency
17147 which the gpc1 counter was incremented over the configured period. See also
17148 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
17149 that the "gpc1_rate" counter must be stored in the stick-table for a value to
17150 be returned, as "gpc1" only holds the event count.
17151
Cyril Bonté62ba8702014-04-22 23:52:25 +020017152sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017153sc0_http_err_cnt([<table>]) : integer
17154sc1_http_err_cnt([<table>]) : integer
17155sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017156 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017157 counters. This includes the both request errors and 4xx error responses.
17158 See also src_http_err_cnt.
17159
Cyril Bonté62ba8702014-04-22 23:52:25 +020017160sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017161sc0_http_err_rate([<table>]) : integer
17162sc1_http_err_rate([<table>]) : integer
17163sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017164 Returns the average rate of HTTP errors from the currently tracked counters,
17165 measured in amount of errors over the period configured in the table. This
17166 includes the both request errors and 4xx error responses. See also
17167 src_http_err_rate.
17168
Cyril Bonté62ba8702014-04-22 23:52:25 +020017169sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017170sc0_http_req_cnt([<table>]) : integer
17171sc1_http_req_cnt([<table>]) : integer
17172sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017173 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020017174 counters. This includes every started request, valid or not. See also
17175 src_http_req_cnt.
17176
Cyril Bonté62ba8702014-04-22 23:52:25 +020017177sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017178sc0_http_req_rate([<table>]) : integer
17179sc1_http_req_rate([<table>]) : integer
17180sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017181 Returns the average rate of HTTP requests from the currently tracked
17182 counters, measured in amount of requests over the period configured in
17183 the table. This includes every started request, valid or not. See also
17184 src_http_req_rate.
17185
Cyril Bonté62ba8702014-04-22 23:52:25 +020017186sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017187sc0_inc_gpc0([<table>]) : integer
17188sc1_inc_gpc0([<table>]) : integer
17189sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017190 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010017191 tracked counters, and returns its new value. Before the first invocation,
17192 the stored value is zero, so first invocation will increase it to 1 and will
17193 return 1. This is typically used as a second ACL in an expression in order
17194 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020017195
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017196 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017197 acl abuse sc0_http_req_rate gt 10
17198 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020017199 tcp-request connection reject if abuse kill
17200
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017201sc_inc_gpc1(<ctr>[,<table>]) : integer
17202sc0_inc_gpc1([<table>]) : integer
17203sc1_inc_gpc1([<table>]) : integer
17204sc2_inc_gpc1([<table>]) : integer
17205 Increments the second General Purpose Counter associated to the currently
17206 tracked counters, and returns its new value. Before the first invocation,
17207 the stored value is zero, so first invocation will increase it to 1 and will
17208 return 1. This is typically used as a second ACL in an expression in order
17209 to mark a connection when a first ACL was verified.
17210
Cyril Bonté62ba8702014-04-22 23:52:25 +020017211sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017212sc0_kbytes_in([<table>]) : integer
17213sc1_kbytes_in([<table>]) : integer
17214sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017215 Returns the total amount of client-to-server data from the currently tracked
17216 counters, measured in kilobytes. The test is currently performed on 32-bit
17217 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020017218
Cyril Bonté62ba8702014-04-22 23:52:25 +020017219sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017220sc0_kbytes_out([<table>]) : integer
17221sc1_kbytes_out([<table>]) : integer
17222sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017223 Returns the total amount of server-to-client data from the currently tracked
17224 counters, measured in kilobytes. The test is currently performed on 32-bit
17225 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020017226
Cyril Bonté62ba8702014-04-22 23:52:25 +020017227sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017228sc0_sess_cnt([<table>]) : integer
17229sc1_sess_cnt([<table>]) : integer
17230sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017231 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020017232 into sessions, which means that they were accepted by a "tcp-request
17233 connection" rule, from the currently tracked counters. A backend may count
17234 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017235 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020017236 with the client. See also src_sess_cnt.
17237
Cyril Bonté62ba8702014-04-22 23:52:25 +020017238sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017239sc0_sess_rate([<table>]) : integer
17240sc1_sess_rate([<table>]) : integer
17241sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020017242 Returns the average session rate from the currently tracked counters,
17243 measured in amount of sessions over the period configured in the table. A
17244 session is a connection that got past the early "tcp-request connection"
17245 rules. A backend may count more sessions than connections because each
17246 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017247 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020017248
Cyril Bonté62ba8702014-04-22 23:52:25 +020017249sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020017250sc0_tracked([<table>]) : boolean
17251sc1_tracked([<table>]) : boolean
17252sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020017253 Returns true if the designated session counter is currently being tracked by
17254 the current session. This can be useful when deciding whether or not we want
17255 to set some values in a header passed to the server.
17256
Cyril Bonté62ba8702014-04-22 23:52:25 +020017257sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020017258sc0_trackers([<table>]) : integer
17259sc1_trackers([<table>]) : integer
17260sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010017261 Returns the current amount of concurrent connections tracking the same
17262 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017263 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010017264 that it does not rely on any stored information but on the table's reference
17265 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020017266 may sometimes be more suited for layer7 tracking. It can be used to tell a
17267 server how many concurrent connections there are from a given address for
17268 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010017269
Willy Tarreau74ca5042013-06-11 23:12:07 +020017270so_id : integer
17271 Returns an integer containing the current listening socket's id. It is useful
17272 in frontends involving many "bind" lines, or to stick all users coming via a
17273 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017274
Jerome Magnineb421b22020-03-27 22:08:40 +010017275so_name : string
17276 Returns a string containing the current listening socket's name, as defined
17277 with name on a "bind" line. It can serve the same purposes as so_id but with
17278 strings instead of integers.
17279
Willy Tarreau74ca5042013-06-11 23:12:07 +020017280src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010017281 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020017282 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
17283 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
17284 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017285 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
17286 directive is used, it can be the address of a client behind another
17287 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010017288 "tcp-request connection" which sees the real address. When the incoming
17289 connection passed through address translation or redirection involving
17290 connection tracking, the original destination address before the redirection
17291 will be reported. On Linux systems, the source and destination may seldom
17292 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
17293 response may reopen a timed out connection and switch what is believed to be
17294 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010017295
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010017296 Example:
17297 # add an HTTP header in requests with the originating address' country
17298 http-request set-header X-Country %[src,map_ip(geoip.lst)]
17299
Willy Tarreau74ca5042013-06-11 23:12:07 +020017300src_bytes_in_rate([<table>]) : integer
17301 Returns the average bytes rate from the incoming connection's source address
17302 in the current proxy's stick-table or in the designated stick-table, measured
17303 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017304 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017305
Willy Tarreau74ca5042013-06-11 23:12:07 +020017306src_bytes_out_rate([<table>]) : integer
17307 Returns the average bytes rate to the incoming connection's source address in
17308 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017309 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017310 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017311
Willy Tarreau74ca5042013-06-11 23:12:07 +020017312src_clr_gpc0([<table>]) : integer
17313 Clears the first General Purpose Counter associated to the incoming
17314 connection's source address in the current proxy's stick-table or in the
17315 designated stick-table, and returns its previous value. If the address is not
17316 found, an entry is created and 0 is returned. This is typically used as a
17317 second ACL in an expression in order to mark a connection when a first ACL
17318 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020017319
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017320 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020017321 # block if 5 consecutive requests continue to come faster than 10 sess
17322 # per second, and reset the counter as soon as the traffic slows down.
17323 acl abuse src_http_req_rate gt 10
17324 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010017325 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020017326 tcp-request connection accept if !abuse save
17327 tcp-request connection reject if abuse kill
17328
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017329src_clr_gpc1([<table>]) : integer
17330 Clears the second General Purpose Counter associated to the incoming
17331 connection's source address in the current proxy's stick-table or in the
17332 designated stick-table, and returns its previous value. If the address is not
17333 found, an entry is created and 0 is returned. This is typically used as a
17334 second ACL in an expression in order to mark a connection when a first ACL
17335 was verified.
17336
Willy Tarreau74ca5042013-06-11 23:12:07 +020017337src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017338 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020017339 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017340 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017341 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017342
Willy Tarreau74ca5042013-06-11 23:12:07 +020017343src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020017344 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017345 current incoming connection's source address in the current proxy's
17346 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017347 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017348
Willy Tarreau74ca5042013-06-11 23:12:07 +020017349src_conn_rate([<table>]) : integer
17350 Returns the average connection rate from the incoming connection's source
17351 address in the current proxy's stick-table or in the designated stick-table,
17352 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017353 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017354
Willy Tarreau74ca5042013-06-11 23:12:07 +020017355src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020017356 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020017357 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020017358 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017359 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017360
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017361src_get_gpc1([<table>]) : integer
17362 Returns the value of the second General Purpose Counter associated to the
17363 incoming connection's source address in the current proxy's stick-table or in
17364 the designated stick-table. If the address is not found, zero is returned.
17365 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
17366
Thierry FOURNIER236657b2015-08-19 08:25:14 +020017367src_get_gpt0([<table>]) : integer
17368 Returns the value of the first General Purpose Tag associated to the
17369 incoming connection's source address in the current proxy's stick-table or in
17370 the designated stick-table. If the address is not found, zero is returned.
17371 See also sc/sc0/sc1/sc2_get_gpt0.
17372
Willy Tarreau74ca5042013-06-11 23:12:07 +020017373src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017374 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020017375 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017376 stick-table or in the designated stick-table. It reports the frequency
17377 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017378 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
17379 that the "gpc0_rate" counter must be stored in the stick-table for a value to
17380 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020017381
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017382src_gpc1_rate([<table>]) : integer
17383 Returns the average increment rate of the second General Purpose Counter
17384 associated to the incoming connection's source address in the current proxy's
17385 stick-table or in the designated stick-table. It reports the frequency
17386 which the gpc1 counter was incremented over the configured period. See also
17387 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
17388 that the "gpc1_rate" counter must be stored in the stick-table for a value to
17389 be returned, as "gpc1" only holds the event count.
17390
Willy Tarreau74ca5042013-06-11 23:12:07 +020017391src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017392 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017393 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020017394 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017395 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020017396 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017397
Willy Tarreau74ca5042013-06-11 23:12:07 +020017398src_http_err_rate([<table>]) : integer
17399 Returns the average rate of HTTP errors from the incoming connection's source
17400 address in the current proxy's stick-table or in the designated stick-table,
17401 measured in amount of errors over the period configured in the table. This
17402 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017403 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017404
Willy Tarreau74ca5042013-06-11 23:12:07 +020017405src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017406 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020017407 source address in the current proxy's stick-table or in the designated stick-
17408 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017409 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017410
Willy Tarreau74ca5042013-06-11 23:12:07 +020017411src_http_req_rate([<table>]) : integer
17412 Returns the average rate of HTTP requests from the incoming connection's
17413 source address in the current proxy's stick-table or in the designated stick-
17414 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020017415 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017416 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017417
Willy Tarreau74ca5042013-06-11 23:12:07 +020017418src_inc_gpc0([<table>]) : integer
17419 Increments the first General Purpose Counter associated to the incoming
17420 connection's source address in the current proxy's stick-table or in the
17421 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020017422 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020017423 This is typically used as a second ACL in an expression in order to mark a
17424 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020017425
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017426 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020017427 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010017428 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020017429 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020017430
Frédéric Lécaille6778b272018-01-29 15:22:53 +010017431src_inc_gpc1([<table>]) : integer
17432 Increments the second General Purpose Counter associated to the incoming
17433 connection's source address in the current proxy's stick-table or in the
17434 designated stick-table, and returns its new value. If the address is not
17435 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
17436 This is typically used as a second ACL in an expression in order to mark a
17437 connection when a first ACL was verified.
17438
Willy Tarreau16e01562016-08-09 16:46:18 +020017439src_is_local : boolean
17440 Returns true if the source address of the incoming connection is local to the
17441 system, or false if the address doesn't exist on the system, meaning that it
17442 comes from a remote machine. Note that UNIX addresses are considered local.
17443 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017444 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020017445 note that the check involves a few system calls, so it's better to do it only
17446 once per connection.
17447
Willy Tarreau74ca5042013-06-11 23:12:07 +020017448src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017449 Returns the total amount of data received from the incoming connection's
17450 source address in the current proxy's stick-table or in the designated
17451 stick-table, measured in kilobytes. If the address is not found, zero is
17452 returned. The test is currently performed on 32-bit integers, which limits
17453 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017454
Willy Tarreau74ca5042013-06-11 23:12:07 +020017455src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020017456 Returns the total amount of data sent to the incoming connection's source
17457 address in the current proxy's stick-table or in the designated stick-table,
17458 measured in kilobytes. If the address is not found, zero is returned. The
17459 test is currently performed on 32-bit integers, which limits values to 4
17460 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017461
Willy Tarreau74ca5042013-06-11 23:12:07 +020017462src_port : integer
17463 Returns an integer value corresponding to the TCP source port of the
17464 connection on the client side, which is the port the client connected from.
17465 Usage of this function is very limited as modern protocols do not care much
17466 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010017467
Willy Tarreau74ca5042013-06-11 23:12:07 +020017468src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010017469 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020017470 connection's source IPv4 address in the current proxy's stick-table or in the
17471 designated stick-table, that were transformed into sessions, which means that
17472 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017473 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017474
Willy Tarreau74ca5042013-06-11 23:12:07 +020017475src_sess_rate([<table>]) : integer
17476 Returns the average session rate from the incoming connection's source
17477 address in the current proxy's stick-table or in the designated stick-table,
17478 measured in amount of sessions over the period configured in the table. A
17479 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020017480 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020017481
Willy Tarreau74ca5042013-06-11 23:12:07 +020017482src_updt_conn_cnt([<table>]) : integer
17483 Creates or updates the entry associated to the incoming connection's source
17484 address in the current proxy's stick-table or in the designated stick-table.
17485 This table must be configured to store the "conn_cnt" data type, otherwise
17486 the match will be ignored. The current count is incremented by one, and the
17487 expiration timer refreshed. The updated count is returned, so this match
17488 can't return zero. This was used to reject service abusers based on their
17489 source address. Note: it is recommended to use the more complete "track-sc*"
17490 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017491
17492 Example :
17493 # This frontend limits incoming SSH connections to 3 per 10 second for
17494 # each source address, and rejects excess connections until a 10 second
17495 # silence is observed. At most 20 addresses are tracked.
17496 listen ssh
17497 bind :22
17498 mode tcp
17499 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020017500 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020017501 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020017502 server local 127.0.0.1:22
17503
Willy Tarreau74ca5042013-06-11 23:12:07 +020017504srv_id : integer
17505 Returns an integer containing the server's id when processing the response.
17506 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020017507 debugging. It can also be used in a tcp-check or an http-check ruleset.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020017508
vkill1dfd1652019-10-30 16:58:14 +080017509srv_name : string
17510 Returns a string containing the server's name when processing the response.
17511 While it's almost only used with ACLs, it may be used for logging or
Christopher Fauletd1b44642020-04-30 09:51:15 +020017512 debugging. It can also be used in a tcp-check or an http-check ruleset.
vkill1dfd1652019-10-30 16:58:14 +080017513
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200175147.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020017515----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020017516
Willy Tarreau74ca5042013-06-11 23:12:07 +020017517The layer 5 usually describes just the session layer which in haproxy is
17518closest to the session once all the connection handshakes are finished, but
17519when no content is yet made available. The fetch methods described here are
17520usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017521future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020017522
Ben Shillitof25e8e52016-12-02 14:25:37 +00001752351d.all(<prop>[,<prop>*]) : string
17524 Returns values for the properties requested as a string, where values are
17525 separated by the delimiter specified with "51degrees-property-separator".
17526 The device is identified using all the important HTTP headers from the
17527 request. The function can be passed up to five property names, and if a
17528 property name can't be found, the value "NoData" is returned.
17529
17530 Example :
17531 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
17532 # containing the three properties requested using all relevant headers from
17533 # the request.
17534 frontend http-in
17535 bind *:8081
17536 default_backend servers
17537 http-request set-header X-51D-DeviceTypeMobileTablet \
17538 %[51d.all(DeviceType,IsMobile,IsTablet)]
17539
Emeric Brun645ae792014-04-30 14:21:06 +020017540ssl_bc : boolean
17541 Returns true when the back connection was made via an SSL/TLS transport
17542 layer and is locally deciphered. This means the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017543 other a server with the "ssl" option. It can be used in a tcp-check or an
17544 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017545
17546ssl_bc_alg_keysize : integer
17547 Returns the symmetric cipher key size supported in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017548 connection was made over an SSL/TLS transport layer. It can be used in a
17549 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017550
Olivier Houchard6b77f492018-11-22 18:18:29 +010017551ssl_bc_alpn : string
17552 This extracts the Application Layer Protocol Negotiation field from an
17553 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020017554 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010017555 server. The SSL library must have been built with support for TLS
17556 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
17557 not advertised unless the "alpn" keyword on the "server" line specifies a
17558 protocol list. Also, nothing forces the server to pick a protocol from this
17559 list, any other one may be requested. The TLS ALPN extension is meant to
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017560 replace the TLS NPN extension. See also "ssl_bc_npn". It can be used in a
17561 tcp-check or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010017562
Emeric Brun645ae792014-04-30 14:21:06 +020017563ssl_bc_cipher : string
17564 Returns the name of the used cipher when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017565 over an SSL/TLS transport layer. It can be used in a tcp-check or an
17566 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017567
Patrick Hemmer65674662019-06-04 08:13:03 -040017568ssl_bc_client_random : binary
17569 Returns the client random of the back connection when the incoming connection
17570 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17571 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017572 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040017573
Emeric Brun74f7ffa2018-02-19 16:14:12 +010017574ssl_bc_is_resumed : boolean
17575 Returns true when the back connection was made over an SSL/TLS transport
17576 layer and the newly created SSL session was resumed using a cached
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017577 session or a TLS ticket. It can be used in a tcp-check or an http-check
17578 ruleset.
Emeric Brun74f7ffa2018-02-19 16:14:12 +010017579
Olivier Houchard6b77f492018-11-22 18:18:29 +010017580ssl_bc_npn : string
17581 This extracts the Next Protocol Negotiation field from an outgoing connection
17582 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020017583 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010017584 built with support for TLS extensions enabled (check haproxy -vv). Note that
17585 the TLS NPN extension is not advertised unless the "npn" keyword on the
17586 "server" line specifies a protocol list. Also, nothing forces the server to
17587 pick a protocol from this list, any other one may be used. Please note that
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017588 the TLS NPN extension was replaced with ALPN. It can be used in a tcp-check
17589 or an http-check ruleset.
Olivier Houchard6b77f492018-11-22 18:18:29 +010017590
Emeric Brun645ae792014-04-30 14:21:06 +020017591ssl_bc_protocol : string
17592 Returns the name of the used protocol when the outgoing connection was made
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017593 over an SSL/TLS transport layer. It can be used in a tcp-check or an
17594 http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017595
Emeric Brunb73a9b02014-04-30 18:49:19 +020017596ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020017597 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020017598 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017599 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64". It
17600 can be used in a tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017601
Patrick Hemmer65674662019-06-04 08:13:03 -040017602ssl_bc_server_random : binary
17603 Returns the server random of the back connection when the incoming connection
17604 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17605 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017606 It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmer65674662019-06-04 08:13:03 -040017607
Emeric Brun645ae792014-04-30 14:21:06 +020017608ssl_bc_session_id : binary
17609 Returns the SSL ID of the back connection when the outgoing connection was
17610 made over an SSL/TLS transport layer. It is useful to log if we want to know
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017611 if session was reused or not. It can be used in a tcp-check or an http-check
17612 ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017613
Patrick Hemmere0275472018-04-28 19:15:51 -040017614ssl_bc_session_key : binary
17615 Returns the SSL session master key of the back connection when the outgoing
17616 connection was made over an SSL/TLS transport layer. It is useful to decrypt
17617 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017618 BoringSSL. It can be used in a tcp-check or an http-check ruleset.
Patrick Hemmere0275472018-04-28 19:15:51 -040017619
Emeric Brun645ae792014-04-30 14:21:06 +020017620ssl_bc_use_keysize : integer
17621 Returns the symmetric cipher key size used in bits when the outgoing
Christopher Fauletd92ea7f2020-04-30 10:03:55 +020017622 connection was made over an SSL/TLS transport layer. It can be used in a
17623 tcp-check or an http-check ruleset.
Emeric Brun645ae792014-04-30 14:21:06 +020017624
Willy Tarreau74ca5042013-06-11 23:12:07 +020017625ssl_c_ca_err : integer
17626 When the incoming connection was made over an SSL/TLS transport layer,
17627 returns the ID of the first error detected during verification of the client
17628 certificate at depth > 0, or 0 if no error was encountered during this
17629 verification process. Please refer to your SSL library's documentation to
17630 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020017631
Willy Tarreau74ca5042013-06-11 23:12:07 +020017632ssl_c_ca_err_depth : integer
17633 When the incoming connection was made over an SSL/TLS transport layer,
17634 returns the depth in the CA chain of the first error detected during the
17635 verification of the client certificate. If no error is encountered, 0 is
17636 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010017637
Christopher Faulet70d10d12020-11-06 12:10:33 +010017638ssl_c_chain_der : binary
William Dauchya598b502020-08-06 18:11:38 +020017639 Returns the DER formatted chain certificate presented by the client when the
17640 incoming connection was made over an SSL/TLS transport layer. When used for
17641 an ACL, the value(s) to match against can be passed in hexadecimal form. One
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050017642 can parse the result with any lib accepting ASN.1 DER data. It currently
William Dauchya598b502020-08-06 18:11:38 +020017643 does not support resumed sessions.
17644
Christopher Faulet70d10d12020-11-06 12:10:33 +010017645ssl_c_der : binary
17646 Returns the DER formatted certificate presented by the client when the
17647 incoming connection was made over an SSL/TLS transport layer. When used for
17648 an ACL, the value(s) to match against can be passed in hexadecimal form.
17649
Willy Tarreau74ca5042013-06-11 23:12:07 +020017650ssl_c_err : integer
17651 When the incoming connection was made over an SSL/TLS transport layer,
17652 returns the ID of the first error detected during verification at depth 0, or
17653 0 if no error was encountered during this verification process. Please refer
17654 to your SSL library's documentation to find the exhaustive list of error
17655 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020017656
Elliot Otchet71f82972020-01-15 08:12:14 -050017657ssl_c_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017658 When the incoming connection was made over an SSL/TLS transport layer,
17659 returns the full distinguished name of the issuer of the certificate
17660 presented by the client when no <entry> is specified, or the value of the
17661 first given entry found from the beginning of the DN. If a positive/negative
17662 occurrence number is specified as the optional second argument, it returns
17663 the value of the nth given entry value from the beginning/end of the DN.
17664 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
17665 "ssl_c_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017666 The <format> parameter allows you to receive the DN suitable for
17667 consumption by different protocols. Currently supported is rfc2253 for
17668 LDAP v3.
17669 If you'd like to modify the format only you can specify an empty string
17670 and zero for the first two parameters. Example: ssl_c_i_dn(,0,rfc2253)
Willy Tarreau62644772008-07-16 18:36:06 +020017671
Willy Tarreau74ca5042013-06-11 23:12:07 +020017672ssl_c_key_alg : string
17673 Returns the name of the algorithm used to generate the key of the certificate
17674 presented by the client when the incoming connection was made over an SSL/TLS
17675 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020017676
Willy Tarreau74ca5042013-06-11 23:12:07 +020017677ssl_c_notafter : string
17678 Returns the end date presented by the client as a formatted string
17679 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17680 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020017681
Willy Tarreau74ca5042013-06-11 23:12:07 +020017682ssl_c_notbefore : string
17683 Returns the start date presented by the client as a formatted string
17684 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17685 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010017686
Elliot Otchet71f82972020-01-15 08:12:14 -050017687ssl_c_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017688 When the incoming connection was made over an SSL/TLS transport layer,
17689 returns the full distinguished name of the subject of the certificate
17690 presented by the client when no <entry> is specified, or the value of the
17691 first given entry found from the beginning of the DN. If a positive/negative
17692 occurrence number is specified as the optional second argument, it returns
17693 the value of the nth given entry value from the beginning/end of the DN.
17694 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
17695 "ssl_c_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017696 The <format> parameter allows you to receive the DN suitable for
17697 consumption by different protocols. Currently supported is rfc2253 for
17698 LDAP v3.
17699 If you'd like to modify the format only you can specify an empty string
17700 and zero for the first two parameters. Example: ssl_c_s_dn(,0,rfc2253)
Willy Tarreaub6672b52011-12-12 17:23:41 +010017701
Willy Tarreau74ca5042013-06-11 23:12:07 +020017702ssl_c_serial : binary
17703 Returns the serial of the certificate presented by the client when the
17704 incoming connection was made over an SSL/TLS transport layer. When used for
17705 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017706
Willy Tarreau74ca5042013-06-11 23:12:07 +020017707ssl_c_sha1 : binary
17708 Returns the SHA-1 fingerprint of the certificate presented by the client when
17709 the incoming connection was made over an SSL/TLS transport layer. This can be
17710 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020017711 Note that the output is binary, so if you want to pass that signature to the
17712 server, you need to encode it in hex or base64, such as in the example below:
17713
Jarno Huuskonen676f6222017-03-30 09:19:45 +030017714 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020017715 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020017716
Willy Tarreau74ca5042013-06-11 23:12:07 +020017717ssl_c_sig_alg : string
17718 Returns the name of the algorithm used to sign the certificate presented by
17719 the client when the incoming connection was made over an SSL/TLS transport
17720 layer.
Emeric Brun87855892012-10-17 17:39:35 +020017721
Willy Tarreau74ca5042013-06-11 23:12:07 +020017722ssl_c_used : boolean
17723 Returns true if current SSL session uses a client certificate even if current
17724 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020017725
Willy Tarreau74ca5042013-06-11 23:12:07 +020017726ssl_c_verify : integer
17727 Returns the verify result error ID when the incoming connection was made over
17728 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
17729 refer to your SSL library's documentation for an exhaustive list of error
17730 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020017731
Willy Tarreau74ca5042013-06-11 23:12:07 +020017732ssl_c_version : integer
17733 Returns the version of the certificate presented by the client when the
17734 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020017735
Emeric Brun43e79582014-10-29 19:03:26 +010017736ssl_f_der : binary
17737 Returns the DER formatted certificate presented by the frontend when the
17738 incoming connection was made over an SSL/TLS transport layer. When used for
17739 an ACL, the value(s) to match against can be passed in hexadecimal form.
17740
Elliot Otchet71f82972020-01-15 08:12:14 -050017741ssl_f_i_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017742 When the incoming connection was made over an SSL/TLS transport layer,
17743 returns the full distinguished name of the issuer of the certificate
17744 presented by the frontend when no <entry> is specified, or the value of the
17745 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020017746 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020017747 the value of the nth given entry value from the beginning/end of the DN.
17748 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
17749 "ssl_f_i_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017750 The <format> parameter allows you to receive the DN suitable for
17751 consumption by different protocols. Currently supported is rfc2253 for
17752 LDAP v3.
17753 If you'd like to modify the format only you can specify an empty string
17754 and zero for the first two parameters. Example: ssl_f_i_dn(,0,rfc2253)
Emeric Brun87855892012-10-17 17:39:35 +020017755
Willy Tarreau74ca5042013-06-11 23:12:07 +020017756ssl_f_key_alg : string
17757 Returns the name of the algorithm used to generate the key of the certificate
17758 presented by the frontend when the incoming connection was made over an
17759 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020017760
Willy Tarreau74ca5042013-06-11 23:12:07 +020017761ssl_f_notafter : string
17762 Returns the end date presented by the frontend as a formatted string
17763 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17764 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017765
Willy Tarreau74ca5042013-06-11 23:12:07 +020017766ssl_f_notbefore : string
17767 Returns the start date presented by the frontend as a formatted string
17768 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
17769 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020017770
Elliot Otchet71f82972020-01-15 08:12:14 -050017771ssl_f_s_dn([<entry>[,<occ>[,<format>]]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020017772 When the incoming connection was made over an SSL/TLS transport layer,
17773 returns the full distinguished name of the subject of the certificate
17774 presented by the frontend when no <entry> is specified, or the value of the
17775 first given entry found from the beginning of the DN. If a positive/negative
17776 occurrence number is specified as the optional second argument, it returns
17777 the value of the nth given entry value from the beginning/end of the DN.
17778 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
17779 "ssl_f_s_dn(CN)" retrieves the common name.
Elliot Otchet71f82972020-01-15 08:12:14 -050017780 The <format> parameter allows you to receive the DN suitable for
17781 consumption by different protocols. Currently supported is rfc2253 for
17782 LDAP v3.
17783 If you'd like to modify the format only you can specify an empty string
17784 and zero for the first two parameters. Example: ssl_f_s_dn(,0,rfc2253)
Emeric Brunce5ad802012-10-22 14:11:22 +020017785
Willy Tarreau74ca5042013-06-11 23:12:07 +020017786ssl_f_serial : binary
17787 Returns the serial of the certificate presented by the frontend when the
17788 incoming connection was made over an SSL/TLS transport layer. When used for
17789 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020017790
Emeric Brun55f4fa82014-04-30 17:11:25 +020017791ssl_f_sha1 : binary
17792 Returns the SHA-1 fingerprint of the certificate presented by the frontend
17793 when the incoming connection was made over an SSL/TLS transport layer. This
17794 can be used to know which certificate was chosen using SNI.
17795
Willy Tarreau74ca5042013-06-11 23:12:07 +020017796ssl_f_sig_alg : string
17797 Returns the name of the algorithm used to sign the certificate presented by
17798 the frontend when the incoming connection was made over an SSL/TLS transport
17799 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020017800
Willy Tarreau74ca5042013-06-11 23:12:07 +020017801ssl_f_version : integer
17802 Returns the version of the certificate presented by the frontend when the
17803 incoming connection was made over an SSL/TLS transport layer.
17804
17805ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020017806 Returns true when the front connection was made via an SSL/TLS transport
17807 layer and is locally deciphered. This means it has matched a socket declared
17808 with a "bind" line having the "ssl" option.
17809
Willy Tarreau74ca5042013-06-11 23:12:07 +020017810 Example :
17811 # This passes "X-Proto: https" to servers when client connects over SSL
17812 listen http-https
17813 bind :80
17814 bind :443 ssl crt /etc/haproxy.pem
17815 http-request add-header X-Proto https if { ssl_fc }
17816
17817ssl_fc_alg_keysize : integer
17818 Returns the symmetric cipher key size supported in bits when the incoming
17819 connection was made over an SSL/TLS transport layer.
17820
17821ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017822 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020017823 incoming connection made via a TLS transport layer and locally deciphered by
17824 haproxy. The result is a string containing the protocol name advertised by
17825 the client. The SSL library must have been built with support for TLS
17826 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
17827 not advertised unless the "alpn" keyword on the "bind" line specifies a
17828 protocol list. Also, nothing forces the client to pick a protocol from this
17829 list, any other one may be requested. The TLS ALPN extension is meant to
17830 replace the TLS NPN extension. See also "ssl_fc_npn".
17831
Willy Tarreau74ca5042013-06-11 23:12:07 +020017832ssl_fc_cipher : string
17833 Returns the name of the used cipher when the incoming connection was made
17834 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020017835
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017836ssl_fc_cipherlist_bin : binary
17837 Returns the binary form of the client hello cipher list. The maximum returned
17838 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017839 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017840
17841ssl_fc_cipherlist_hex : string
17842 Returns the binary form of the client hello cipher list encoded as
17843 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017844 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017845
17846ssl_fc_cipherlist_str : string
17847 Returns the decoded text form of the client hello cipher list. The maximum
17848 number of ciphers returned is according with the value of
17849 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010017850 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020017851 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017852
17853ssl_fc_cipherlist_xxh : integer
17854 Returns a xxh64 of the cipher list. This hash can be return only is the value
17855 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010017856 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010017857
Patrick Hemmer65674662019-06-04 08:13:03 -040017858ssl_fc_client_random : binary
17859 Returns the client random of the front connection when the incoming connection
17860 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17861 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
17862
William Lallemand7d42ef52020-07-06 11:41:30 +020017863ssl_fc_client_early_traffic_secret : string
17864 Return the CLIENT_EARLY_TRAFFIC_SECRET as an hexadecimal string for the
17865 front connection when the incoming connection was made over a TLS 1.3
17866 transport layer.
17867 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17868 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17869 activated with "tune.ssl.keylog on" in the global section. See also
17870 "tune.ssl.keylog"
17871
17872ssl_fc_client_handshake_traffic_secret : string
17873 Return the CLIENT_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
17874 front connection when the incoming connection was made over a TLS 1.3
17875 transport layer.
17876 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17877 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17878 activated with "tune.ssl.keylog on" in the global section. See also
17879 "tune.ssl.keylog"
17880
17881ssl_fc_client_traffic_secret_0 : string
17882 Return the CLIENT_TRAFFIC_SECRET_0 as an hexadecimal string for the
17883 front connection when the incoming connection was made over a TLS 1.3
17884 transport layer.
17885 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17886 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17887 activated with "tune.ssl.keylog on" in the global section. See also
17888 "tune.ssl.keylog"
17889
17890ssl_fc_exporter_secret : string
17891 Return the EXPORTER_SECRET as an hexadecimal string for the
17892 front connection when the incoming connection was made over a TLS 1.3
17893 transport layer.
17894 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17895 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17896 activated with "tune.ssl.keylog on" in the global section. See also
17897 "tune.ssl.keylog"
17898
17899ssl_fc_early_exporter_secret : string
17900 Return the EARLY_EXPORTER_SECRET as an hexadecimal string for the
17901 front connection when the incoming connection was made over an TLS 1.3
17902 transport layer.
17903 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17904 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17905 activated with "tune.ssl.keylog on" in the global section. See also
17906 "tune.ssl.keylog"
17907
Willy Tarreau74ca5042013-06-11 23:12:07 +020017908ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020017909 Returns true if a client certificate is present in an incoming connection over
17910 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010017911 Note: on SSL session resumption with Session ID or TLS ticket, client
17912 certificate is not present in the current connection but may be retrieved
17913 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
17914 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020017915
Olivier Houchardccaa7de2017-10-02 11:51:03 +020017916ssl_fc_has_early : boolean
17917 Returns true if early data were sent, and the handshake didn't happen yet. As
17918 it has security implications, it is useful to be able to refuse those, or
17919 wait until the handshake happened.
17920
Willy Tarreau74ca5042013-06-11 23:12:07 +020017921ssl_fc_has_sni : boolean
17922 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020017923 in an incoming connection was made over an SSL/TLS transport layer. Returns
17924 true when the incoming connection presents a TLS SNI field. This requires
John Roeslerfb2fce12019-07-10 15:45:51 -050017925 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020017926 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020017927
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020017928ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020017929 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010017930 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
17931 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020017932
Willy Tarreau74ca5042013-06-11 23:12:07 +020017933ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017934 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020017935 made via a TLS transport layer and locally deciphered by haproxy. The result
17936 is a string containing the protocol name advertised by the client. The SSL
17937 library must have been built with support for TLS extensions enabled (check
17938 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
17939 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
17940 forces the client to pick a protocol from this list, any other one may be
17941 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020017942
Willy Tarreau74ca5042013-06-11 23:12:07 +020017943ssl_fc_protocol : string
17944 Returns the name of the used protocol when the incoming connection was made
17945 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020017946
Emeric Brunb73a9b02014-04-30 18:49:19 +020017947ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040017948 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020017949 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
17950 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040017951
William Lallemand7d42ef52020-07-06 11:41:30 +020017952ssl_fc_server_handshake_traffic_secret : string
17953 Return the SERVER_HANDSHAKE_TRAFFIC_SECRET as an hexadecimal string for the
17954 front connection when the incoming connection was made over a TLS 1.3
17955 transport layer.
17956 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17957 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17958 activated with "tune.ssl.keylog on" in the global section. See also
17959 "tune.ssl.keylog"
17960
17961ssl_fc_server_traffic_secret_0 : string
17962 Return the SERVER_TRAFFIC_SECRET_0 as an hexadecimal string for the
17963 front connection when the incoming connection was made over an TLS 1.3
17964 transport layer.
17965 Require OpenSSL >= 1.1.1. This is one of the keys dumped by the OpenSSL
17966 keylog callback to generate the SSLKEYLOGFILE. The SSL Key logging must be
17967 activated with "tune.ssl.keylog on" in the global section. See also
17968 "tune.ssl.keylog"
17969
Patrick Hemmer65674662019-06-04 08:13:03 -040017970ssl_fc_server_random : binary
17971 Returns the server random of the front connection when the incoming connection
17972 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
17973 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
17974
Willy Tarreau74ca5042013-06-11 23:12:07 +020017975ssl_fc_session_id : binary
17976 Returns the SSL ID of the front connection when the incoming connection was
17977 made over an SSL/TLS transport layer. It is useful to stick a given client to
17978 a server. It is important to note that some browsers refresh their session ID
17979 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020017980
Patrick Hemmere0275472018-04-28 19:15:51 -040017981ssl_fc_session_key : binary
17982 Returns the SSL session master key of the front connection when the incoming
17983 connection was made over an SSL/TLS transport layer. It is useful to decrypt
17984 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
17985 BoringSSL.
17986
17987
Willy Tarreau74ca5042013-06-11 23:12:07 +020017988ssl_fc_sni : string
17989 This extracts the Server Name Indication TLS extension (SNI) field from an
17990 incoming connection made via an SSL/TLS transport layer and locally
17991 deciphered by haproxy. The result (when present) typically is a string
17992 matching the HTTPS host name (253 chars or less). The SSL library must have
17993 been built with support for TLS extensions enabled (check haproxy -vv).
17994
17995 This fetch is different from "req_ssl_sni" above in that it applies to the
17996 connection being deciphered by haproxy and not to SSL contents being blindly
17997 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roeslerfb2fce12019-07-10 15:45:51 -050017998 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020017999 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020018000
Willy Tarreau74ca5042013-06-11 23:12:07 +020018001 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020018002 ssl_fc_sni_end : suffix match
18003 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020018004
Willy Tarreau74ca5042013-06-11 23:12:07 +020018005ssl_fc_use_keysize : integer
18006 Returns the symmetric cipher key size used in bits when the incoming
18007 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018008
William Lallemandbfa3e812020-06-25 20:07:18 +020018009ssl_s_der : binary
18010 Returns the DER formatted certificate presented by the server when the
18011 outgoing connection was made over an SSL/TLS transport layer. When used for
18012 an ACL, the value(s) to match against can be passed in hexadecimal form.
18013
William Dauchya598b502020-08-06 18:11:38 +020018014ssl_s_chain_der : binary
18015 Returns the DER formatted chain certificate presented by the server when the
18016 outgoing connection was made over an SSL/TLS transport layer. When used for
18017 an ACL, the value(s) to match against can be passed in hexadecimal form. One
Ilya Shipitsin2272d8a2020-12-21 01:22:40 +050018018 can parse the result with any lib accepting ASN.1 DER data. It currently
William Dauchya598b502020-08-06 18:11:38 +020018019 does not support resumed sessions.
18020
William Lallemandbfa3e812020-06-25 20:07:18 +020018021ssl_s_key_alg : string
18022 Returns the name of the algorithm used to generate the key of the certificate
18023 presented by the server when the outgoing connection was made over an
18024 SSL/TLS transport layer.
18025
18026ssl_s_notafter : string
18027 Returns the end date presented by the server as a formatted string
18028 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
18029 transport layer.
18030
18031ssl_s_notbefore : string
18032 Returns the start date presented by the server as a formatted string
18033 YYMMDDhhmmss[Z] when the outgoing connection was made over an SSL/TLS
18034 transport layer.
18035
18036ssl_s_i_dn([<entry>[,<occ>[,<format>]]]) : string
18037 When the outgoing connection was made over an SSL/TLS transport layer,
18038 returns the full distinguished name of the issuer of the certificate
18039 presented by the server when no <entry> is specified, or the value of the
18040 first given entry found from the beginning of the DN. If a positive/negative
18041 occurrence number is specified as the optional second argument, it returns
18042 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020018043 For instance, "ssl_s_i_dn(OU,2)" the second organization unit, and
18044 "ssl_s_i_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020018045 The <format> parameter allows you to receive the DN suitable for
18046 consumption by different protocols. Currently supported is rfc2253 for
18047 LDAP v3.
18048 If you'd like to modify the format only you can specify an empty string
18049 and zero for the first two parameters. Example: ssl_s_i_dn(,0,rfc2253)
18050
18051ssl_s_s_dn([<entry>[,<occ>[,<format>]]]) : string
18052 When the outgoing connection was made over an SSL/TLS transport layer,
18053 returns the full distinguished name of the subject of the certificate
18054 presented by the server when no <entry> is specified, or the value of the
18055 first given entry found from the beginning of the DN. If a positive/negative
18056 occurrence number is specified as the optional second argument, it returns
18057 the value of the nth given entry value from the beginning/end of the DN.
William Lallemand8f600c82020-06-26 09:55:06 +020018058 For instance, "ssl_s_s_dn(OU,2)" the second organization unit, and
18059 "ssl_s_s_dn(CN)" retrieves the common name.
William Lallemandbfa3e812020-06-25 20:07:18 +020018060 The <format> parameter allows you to receive the DN suitable for
18061 consumption by different protocols. Currently supported is rfc2253 for
18062 LDAP v3.
18063 If you'd like to modify the format only you can specify an empty string
18064 and zero for the first two parameters. Example: ssl_s_s_dn(,0,rfc2253)
18065
18066ssl_s_serial : binary
18067 Returns the serial of the certificate presented by the server when the
18068 outgoing connection was made over an SSL/TLS transport layer. When used for
18069 an ACL, the value(s) to match against can be passed in hexadecimal form.
18070
18071ssl_s_sha1 : binary
18072 Returns the SHA-1 fingerprint of the certificate presented by the server
18073 when the outgoing connection was made over an SSL/TLS transport layer. This
18074 can be used to know which certificate was chosen using SNI.
18075
18076ssl_s_sig_alg : string
18077 Returns the name of the algorithm used to sign the certificate presented by
18078 the server when the outgoing connection was made over an SSL/TLS transport
18079 layer.
18080
18081ssl_s_version : integer
18082 Returns the version of the certificate presented by the server when the
18083 outgoing connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018084
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200180857.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020018086------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020018087
Willy Tarreau74ca5042013-06-11 23:12:07 +020018088Fetching samples from buffer contents is a bit different from the previous
18089sample fetches above because the sampled data are ephemeral. These data can
18090only be used when they're available and will be lost when they're forwarded.
18091For this reason, samples fetched from buffer contents during a request cannot
18092be used in a response for example. Even while the data are being fetched, they
18093can change. Sometimes it is necessary to set some delays or combine multiple
18094sample fetch methods to ensure that the expected data are complete and usable,
18095for example through TCP request content inspection. Please see the "tcp-request
18096content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020018097
Willy Tarreau74ca5042013-06-11 23:12:07 +020018098payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010018099 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018100 "stick on", "stick match"), and for "res.payload" when used in the context of
18101 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010018102
Willy Tarreau74ca5042013-06-11 23:12:07 +020018103payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
18104 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010018105 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020018106 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010018107
Willy Tarreau74ca5042013-06-11 23:12:07 +020018108req.len : integer
18109req_len : integer (deprecated)
18110 Returns an integer value corresponding to the number of bytes present in the
18111 request buffer. This is mostly used in ACL. It is important to understand
18112 that this test does not return false as long as the buffer is changing. This
18113 means that a check with equality to zero will almost always immediately match
18114 at the beginning of the session, while a test for more data will wait for
18115 that data to come in and return false only when haproxy is certain that no
18116 more data will come in. This test was designed to be used with TCP request
18117 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018118
Willy Tarreau74ca5042013-06-11 23:12:07 +020018119req.payload(<offset>,<length>) : binary
18120 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020018121 in the request buffer. As a special case, if the <length> argument is zero,
18122 the the whole buffer from <offset> to the end is extracted. This can be used
18123 with ACLs in order to check for the presence of some content in a buffer at
18124 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018125
Willy Tarreau74ca5042013-06-11 23:12:07 +020018126 ACL alternatives :
18127 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018128
Willy Tarreau74ca5042013-06-11 23:12:07 +020018129req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
18130 This extracts a binary block whose size is specified at <offset1> for <length>
18131 bytes, and which starts at <offset2> if specified or just after the length in
18132 the request buffer. The <offset2> parameter also supports relative offsets if
18133 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018134
Willy Tarreau74ca5042013-06-11 23:12:07 +020018135 ACL alternatives :
18136 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018137
Willy Tarreau74ca5042013-06-11 23:12:07 +020018138 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018139
Willy Tarreau74ca5042013-06-11 23:12:07 +020018140req.proto_http : boolean
18141req_proto_http : boolean (deprecated)
18142 Returns true when data in the request buffer look like HTTP and correctly
18143 parses as such. It is the same parser as the common HTTP request parser which
18144 is used so there should be no surprises. The test does not match until the
18145 request is complete, failed or timed out. This test may be used to report the
18146 protocol in TCP logs, but the biggest use is to block TCP request analysis
18147 until a complete HTTP request is present in the buffer, for example to track
18148 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018149
Willy Tarreau74ca5042013-06-11 23:12:07 +020018150 Example:
18151 # track request counts per "base" (concatenation of Host+URL)
18152 tcp-request inspect-delay 10s
18153 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020018154 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020018155
Willy Tarreau74ca5042013-06-11 23:12:07 +020018156req.rdp_cookie([<name>]) : string
18157rdp_cookie([<name>]) : string (deprecated)
18158 When the request buffer looks like the RDP protocol, extracts the RDP cookie
18159 <name>, or any cookie if unspecified. The parser only checks for the first
18160 cookie, as illustrated in the RDP protocol specification. The cookie name is
18161 case insensitive. Generally the "MSTS" cookie name will be used, as it can
18162 contain the user name of the client connecting to the server if properly
18163 configured on the client. The "MSTSHASH" cookie is often used as well for
18164 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018165
Willy Tarreau74ca5042013-06-11 23:12:07 +020018166 This differs from "balance rdp-cookie" in that any balancing algorithm may be
18167 used and thus the distribution of clients to backend servers is not linked to
18168 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
18169 such as "balance roundrobin" or "balance leastconn" will lead to a more even
18170 distribution of clients to backend servers than the hash used by "balance
18171 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018172
Willy Tarreau74ca5042013-06-11 23:12:07 +020018173 ACL derivatives :
18174 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018175
Willy Tarreau74ca5042013-06-11 23:12:07 +020018176 Example :
18177 listen tse-farm
18178 bind 0.0.0.0:3389
18179 # wait up to 5s for an RDP cookie in the request
18180 tcp-request inspect-delay 5s
18181 tcp-request content accept if RDP_COOKIE
18182 # apply RDP cookie persistence
18183 persist rdp-cookie
18184 # Persist based on the mstshash cookie
18185 # This is only useful makes sense if
18186 # balance rdp-cookie is not used
18187 stick-table type string size 204800
18188 stick on req.rdp_cookie(mstshash)
18189 server srv1 1.1.1.1:3389
18190 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018191
Willy Tarreau74ca5042013-06-11 23:12:07 +020018192 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
18193 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018194
Willy Tarreau74ca5042013-06-11 23:12:07 +020018195req.rdp_cookie_cnt([name]) : integer
18196rdp_cookie_cnt([name]) : integer (deprecated)
18197 Tries to parse the request buffer as RDP protocol, then returns an integer
18198 corresponding to the number of RDP cookies found. If an optional cookie name
18199 is passed, only cookies matching this name are considered. This is mostly
18200 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018201
Willy Tarreau74ca5042013-06-11 23:12:07 +020018202 ACL derivatives :
18203 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018204
Alex Zorin4afdd132018-12-30 13:56:28 +110018205req.ssl_alpn : string
18206 Returns a string containing the values of the Application-Layer Protocol
18207 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
18208 ClientHello message. Note that this only applies to raw contents found in the
18209 request buffer and not to the contents deciphered via an SSL data layer, so
18210 this will not work with "bind" lines having the "ssl" option. This is useful
18211 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020018212 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110018213
18214 Examples :
18215 # Wait for a client hello for at most 5 seconds
18216 tcp-request inspect-delay 5s
18217 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020018218 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110018219 default_backend bk_default
18220
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020018221req.ssl_ec_ext : boolean
18222 Returns a boolean identifying if client sent the Supported Elliptic Curves
18223 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020018224 message. This can be used to present ECC compatible clients with EC
18225 certificate and to use RSA for all others, on the same IP address. Note that
18226 this only applies to raw contents found in the request buffer and not to
18227 contents deciphered via an SSL data layer, so this will not work with "bind"
18228 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020018229
Willy Tarreau74ca5042013-06-11 23:12:07 +020018230req.ssl_hello_type : integer
18231req_ssl_hello_type : integer (deprecated)
18232 Returns an integer value containing the type of the SSL hello message found
18233 in the request buffer if the buffer contains data that parse as a complete
18234 SSL (v3 or superior) client hello message. Note that this only applies to raw
18235 contents found in the request buffer and not to contents deciphered via an
18236 SSL data layer, so this will not work with "bind" lines having the "ssl"
18237 option. This is mostly used in ACL to detect presence of an SSL hello message
18238 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018239
Willy Tarreau74ca5042013-06-11 23:12:07 +020018240req.ssl_sni : string
18241req_ssl_sni : string (deprecated)
18242 Returns a string containing the value of the Server Name TLS extension sent
18243 by a client in a TLS stream passing through the request buffer if the buffer
18244 contains data that parse as a complete SSL (v3 or superior) client hello
18245 message. Note that this only applies to raw contents found in the request
18246 buffer and not to contents deciphered via an SSL data layer, so this will not
Lukas Tribusa267b5d2020-07-19 00:25:06 +020018247 work with "bind" lines having the "ssl" option. This will only work for actual
18248 implicit TLS based protocols like HTTPS (443), IMAPS (993), SMTPS (465),
18249 however it will not work for explicit TLS based protocols, like SMTP (25/587)
18250 or IMAP (143). SNI normally contains the name of the host the client tries to
18251 connect to (for recent browsers). SNI is useful for allowing or denying access
18252 to certain hosts when SSL/TLS is used by the client. This test was designed to
18253 be used with TCP request content inspection. If content switching is needed,
18254 it is recommended to first wait for a complete client hello (type 1), like in
18255 the example below. See also "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018256
Willy Tarreau74ca5042013-06-11 23:12:07 +020018257 ACL derivatives :
18258 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018259
Willy Tarreau74ca5042013-06-11 23:12:07 +020018260 Examples :
18261 # Wait for a client hello for at most 5 seconds
18262 tcp-request inspect-delay 5s
18263 tcp-request content accept if { req_ssl_hello_type 1 }
18264 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
18265 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020018266
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053018267req.ssl_st_ext : integer
18268 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
18269 Returns 1 if the client sent SessionTicket TLS Extension
18270 Returns 2 if the client also sent non-zero length TLS SessionTicket
18271 Note that this only applies to raw contents found in the request buffer and
18272 not to contents deciphered via an SSL data layer, so this will not work with
18273 "bind" lines having the "ssl" option. This can for example be used to detect
18274 whether the client sent a SessionTicket or not and stick it accordingly, if
18275 no SessionTicket then stick on SessionID or don't stick as there's no server
18276 side state is there when SessionTickets are in use.
18277
Willy Tarreau74ca5042013-06-11 23:12:07 +020018278req.ssl_ver : integer
18279req_ssl_ver : integer (deprecated)
18280 Returns an integer value containing the version of the SSL/TLS protocol of a
18281 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
18282 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
18283 composed of the major version multiplied by 65536, added to the minor
18284 version. Note that this only applies to raw contents found in the request
18285 buffer and not to contents deciphered via an SSL data layer, so this will not
18286 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010018287 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020018288 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018289
Willy Tarreau74ca5042013-06-11 23:12:07 +020018290 ACL derivatives :
18291 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010018292
Willy Tarreau47e8eba2013-09-11 23:28:46 +020018293res.len : integer
18294 Returns an integer value corresponding to the number of bytes present in the
18295 response buffer. This is mostly used in ACL. It is important to understand
18296 that this test does not return false as long as the buffer is changing. This
18297 means that a check with equality to zero will almost always immediately match
18298 at the beginning of the session, while a test for more data will wait for
18299 that data to come in and return false only when haproxy is certain that no
18300 more data will come in. This test was designed to be used with TCP response
Christopher Faulete596d182020-05-05 17:46:34 +020018301 content inspection. But it may also be used in tcp-check based expect rules.
Willy Tarreau47e8eba2013-09-11 23:28:46 +020018302
Willy Tarreau74ca5042013-06-11 23:12:07 +020018303res.payload(<offset>,<length>) : binary
18304 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020018305 in the response buffer. As a special case, if the <length> argument is zero,
Christopher Faulete596d182020-05-05 17:46:34 +020018306 the whole buffer from <offset> to the end is extracted. This can be used
Willy Tarreau00f00842013-08-02 11:07:32 +020018307 with ACLs in order to check for the presence of some content in a buffer at
Christopher Faulete596d182020-05-05 17:46:34 +020018308 any location. It may also be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018309
Willy Tarreau74ca5042013-06-11 23:12:07 +020018310res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
18311 This extracts a binary block whose size is specified at <offset1> for <length>
18312 bytes, and which starts at <offset2> if specified or just after the length in
18313 the response buffer. The <offset2> parameter also supports relative offsets
Christopher Faulete596d182020-05-05 17:46:34 +020018314 if prepended with a '+' or '-' sign. It may also be used in tcp-check based
18315 expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018316
Willy Tarreau74ca5042013-06-11 23:12:07 +020018317 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018318
Willy Tarreau971f7b62015-09-29 14:06:59 +020018319res.ssl_hello_type : integer
18320rep_ssl_hello_type : integer (deprecated)
18321 Returns an integer value containing the type of the SSL hello message found
18322 in the response buffer if the buffer contains data that parses as a complete
18323 SSL (v3 or superior) hello message. Note that this only applies to raw
18324 contents found in the response buffer and not to contents deciphered via an
18325 SSL data layer, so this will not work with "server" lines having the "ssl"
18326 option. This is mostly used in ACL to detect presence of an SSL hello message
18327 that is supposed to contain an SSL session ID usable for stickiness.
18328
Willy Tarreau74ca5042013-06-11 23:12:07 +020018329wait_end : boolean
18330 This fetch either returns true when the inspection period is over, or does
18331 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018332 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020018333 actions, such as a delayed reject for some special addresses. Since it either
18334 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018335 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020018336 "WAIT_END" is always usable without prior declaration. This test was designed
18337 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018338
Willy Tarreau74ca5042013-06-11 23:12:07 +020018339 Examples :
18340 # delay every incoming request by 2 seconds
18341 tcp-request inspect-delay 2s
18342 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010018343
Willy Tarreau74ca5042013-06-11 23:12:07 +020018344 # don't immediately tell bad guys they are rejected
18345 tcp-request inspect-delay 10s
18346 acl goodguys src 10.0.0.0/24
18347 acl badguys src 10.0.1.0/24
18348 tcp-request content accept if goodguys
18349 tcp-request content reject if badguys WAIT_END
18350 tcp-request content reject
18351
18352
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200183537.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020018354--------------------------------------
18355
18356It is possible to fetch samples from HTTP contents, requests and responses.
18357This application layer is also called layer 7. It is only possible to fetch the
18358data in this section when a full HTTP request or response has been parsed from
18359its respective request or response buffer. This is always the case with all
18360HTTP specific rules and for sections running with "mode http". When using TCP
18361content inspection, it may be necessary to support an inspection delay in order
18362to let the request or response come in first. These fetches may require a bit
18363more CPU resources than the layer 4 ones, but not much since the request and
18364response are indexed.
18365
18366base : string
18367 This returns the concatenation of the first Host header and the path part of
18368 the request, which starts at the first slash and ends before the question
18369 mark. It can be useful in virtual hosted environments to detect URL abuses as
18370 well as to improve shared caches efficiency. Using this with a limited size
18371 stick table also allows one to collect statistics about most commonly
18372 requested objects by host/path. With ACLs it can allow simple content
18373 switching rules involving the host and the path at the same time, such as
18374 "www.example.com/favicon.ico". See also "path" and "uri".
18375
18376 ACL derivatives :
18377 base : exact string match
18378 base_beg : prefix match
18379 base_dir : subdir match
18380 base_dom : domain match
18381 base_end : suffix match
18382 base_len : length match
18383 base_reg : regex match
18384 base_sub : substring match
18385
18386base32 : integer
18387 This returns a 32-bit hash of the value returned by the "base" fetch method
18388 above. This is useful to track per-URL activity on high traffic sites without
18389 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020018390 memory. The output type is an unsigned integer. The hash function used is
18391 SDBM with full avalanche on the output. Technically, base32 is exactly equal
18392 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020018393
18394base32+src : binary
18395 This returns the concatenation of the base32 fetch above and the src fetch
18396 below. The resulting type is of type binary, with a size of 8 or 20 bytes
18397 depending on the source address family. This can be used to track per-IP,
18398 per-URL counters.
18399
William Lallemand65ad6e12014-01-31 15:08:02 +010018400capture.req.hdr(<idx>) : string
18401 This extracts the content of the header captured by the "capture request
18402 header", idx is the position of the capture keyword in the configuration.
18403 The first entry is an index of 0. See also: "capture request header".
18404
18405capture.req.method : string
18406 This extracts the METHOD of an HTTP request. It can be used in both request
18407 and response. Unlike "method", it can be used in both request and response
18408 because it's allocated.
18409
18410capture.req.uri : string
18411 This extracts the request's URI, which starts at the first slash and ends
18412 before the first space in the request (without the host part). Unlike "path"
18413 and "url", it can be used in both request and response because it's
18414 allocated.
18415
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020018416capture.req.ver : string
18417 This extracts the request's HTTP version and returns either "HTTP/1.0" or
18418 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
18419 logs because it relies on a persistent flag.
18420
William Lallemand65ad6e12014-01-31 15:08:02 +010018421capture.res.hdr(<idx>) : string
18422 This extracts the content of the header captured by the "capture response
18423 header", idx is the position of the capture keyword in the configuration.
18424 The first entry is an index of 0.
18425 See also: "capture response header"
18426
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020018427capture.res.ver : string
18428 This extracts the response's HTTP version and returns either "HTTP/1.0" or
18429 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
18430 persistent flag.
18431
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018432req.body : binary
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018433 This returns the HTTP request's available body as a block of data. It is
18434 recommended to use "option http-buffer-request" to be sure to wait, as much
18435 as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018436
Thierry FOURNIER9826c772015-05-20 15:50:54 +020018437req.body_param([<name>) : string
18438 This fetch assumes that the body of the POST request is url-encoded. The user
18439 can check if the "content-type" contains the value
18440 "application/x-www-form-urlencoded". This extracts the first occurrence of the
18441 parameter <name> in the body, which ends before '&'. The parameter name is
18442 case-sensitive. If no name is given, any parameter will match, and the first
18443 one will be returned. The result is a string corresponding to the value of the
18444 parameter <name> as presented in the request body (no URL decoding is
18445 performed). Note that the ACL version of this fetch iterates over multiple
18446 parameters and will iteratively report all parameters values if no name is
18447 given.
18448
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018449req.body_len : integer
18450 This returns the length of the HTTP request's available body in bytes. It may
18451 be lower than the advertised length if the body is larger than the buffer. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018452 is recommended to use "option http-buffer-request" to be sure to wait, as
18453 much as possible, for the request's body.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018454
18455req.body_size : integer
18456 This returns the advertised length of the HTTP request's body in bytes. It
Christopher Fauletaf4dc4c2020-05-05 17:33:25 +020018457 will represent the advertised Content-Length header, or the size of the
18458 available data in case of chunked encoding.
Willy Tarreaua5910cc2015-05-02 00:46:08 +020018459
Willy Tarreau74ca5042013-06-11 23:12:07 +020018460req.cook([<name>]) : string
18461cook([<name>]) : string (deprecated)
18462 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18463 header line from the request, and returns its value as string. If no name is
18464 specified, the first cookie value is returned. When used with ACLs, all
18465 matching cookies are evaluated. Spaces around the name and the value are
18466 ignored as requested by the Cookie header specification (RFC6265). The cookie
18467 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
18468 well return an empty value if it is present. Use the "found" match to detect
18469 presence. Use the res.cook() variant for response cookies sent by the server.
18470
18471 ACL derivatives :
18472 cook([<name>]) : exact string match
18473 cook_beg([<name>]) : prefix match
18474 cook_dir([<name>]) : subdir match
18475 cook_dom([<name>]) : domain match
18476 cook_end([<name>]) : suffix match
18477 cook_len([<name>]) : length match
18478 cook_reg([<name>]) : regex match
18479 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010018480
Willy Tarreau74ca5042013-06-11 23:12:07 +020018481req.cook_cnt([<name>]) : integer
18482cook_cnt([<name>]) : integer (deprecated)
18483 Returns an integer value representing the number of occurrences of the cookie
18484 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018485
Willy Tarreau74ca5042013-06-11 23:12:07 +020018486req.cook_val([<name>]) : integer
18487cook_val([<name>]) : integer (deprecated)
18488 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18489 header line from the request, and converts its value to an integer which is
18490 returned. If no name is specified, the first cookie value is returned. When
18491 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020018492
Willy Tarreau74ca5042013-06-11 23:12:07 +020018493cookie([<name>]) : string (deprecated)
18494 This extracts the last occurrence of the cookie name <name> on a "Cookie"
18495 header line from the request, or a "Set-Cookie" header from the response, and
18496 returns its value as a string. A typical use is to get multiple clients
18497 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020018498 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020018499 multi-peer synchronization and state keeping across restarts. If no name is
18500 specified, the first cookie value is returned. This fetch should not be used
18501 anymore and should be replaced by req.cook() or res.cook() instead as it
18502 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018503
Willy Tarreau74ca5042013-06-11 23:12:07 +020018504hdr([<name>[,<occ>]]) : string
18505 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
18506 used on responses. Please refer to these respective fetches for more details.
18507 In case of doubt about the fetch direction, please use the explicit ones.
18508 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018509 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018510
Willy Tarreau74ca5042013-06-11 23:12:07 +020018511req.fhdr(<name>[,<occ>]) : string
18512 This extracts the last occurrence of header <name> in an HTTP request. When
18513 used from an ACL, all occurrences are iterated over until a match is found.
18514 Optionally, a specific occurrence might be specified as a position number.
18515 Positive values indicate a position from the first occurrence, with 1 being
18516 the first one. Negative values indicate positions relative to the last one,
18517 with -1 being the last one. It differs from req.hdr() in that any commas
18518 present in the value are returned and are not used as delimiters. This is
18519 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018520
Willy Tarreau74ca5042013-06-11 23:12:07 +020018521req.fhdr_cnt([<name>]) : integer
18522 Returns an integer value representing the number of occurrences of request
18523 header field name <name>, or the total number of header fields if <name> is
18524 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
18525 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018526
Willy Tarreau74ca5042013-06-11 23:12:07 +020018527req.hdr([<name>[,<occ>]]) : string
18528 This extracts the last occurrence of header <name> in an HTTP request. When
18529 used from an ACL, all occurrences are iterated over until a match is found.
18530 Optionally, a specific occurrence might be specified as a position number.
18531 Positive values indicate a position from the first occurrence, with 1 being
18532 the first one. Negative values indicate positions relative to the last one,
18533 with -1 being the last one. A typical use is with the X-Forwarded-For header
18534 once converted to IP, associated with an IP stick-table. The function
18535 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000018536 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020018537 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010018538 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010018539
Willy Tarreau74ca5042013-06-11 23:12:07 +020018540 ACL derivatives :
18541 hdr([<name>[,<occ>]]) : exact string match
18542 hdr_beg([<name>[,<occ>]]) : prefix match
18543 hdr_dir([<name>[,<occ>]]) : subdir match
18544 hdr_dom([<name>[,<occ>]]) : domain match
18545 hdr_end([<name>[,<occ>]]) : suffix match
18546 hdr_len([<name>[,<occ>]]) : length match
18547 hdr_reg([<name>[,<occ>]]) : regex match
18548 hdr_sub([<name>[,<occ>]]) : substring match
18549
18550req.hdr_cnt([<name>]) : integer
18551hdr_cnt([<header>]) : integer (deprecated)
18552 Returns an integer value representing the number of occurrences of request
18553 header field name <name>, or the total number of header field values if
18554 <name> is not specified. It is important to remember that one header line may
18555 count as several headers if it has several values. The function considers any
18556 comma as a delimiter for distinct values. If full-line headers are desired
18557 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
18558 detect presence, absence or abuse of a specific header, as well as to block
18559 request smuggling attacks by rejecting requests which contain more than one
18560 of certain headers. See "req.hdr" for more information on header matching.
18561
18562req.hdr_ip([<name>[,<occ>]]) : ip
18563hdr_ip([<name>[,<occ>]]) : ip (deprecated)
18564 This extracts the last occurrence of header <name> in an HTTP request,
18565 converts it to an IPv4 or IPv6 address and returns this address. When used
18566 with ACLs, all occurrences are checked, and if <name> is omitted, every value
18567 of every header is checked. Optionally, a specific occurrence might be
18568 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010018569 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020018570 positions relative to the last one, with -1 being the last one. A typical use
18571 is with the X-Forwarded-For and X-Client-IP headers.
18572
18573req.hdr_val([<name>[,<occ>]]) : integer
18574hdr_val([<name>[,<occ>]]) : integer (deprecated)
18575 This extracts the last occurrence of header <name> in an HTTP request, and
18576 converts it to an integer value. When used with ACLs, all occurrences are
18577 checked, and if <name> is omitted, every value of every header is checked.
18578 Optionally, a specific occurrence might be specified as a position number.
18579 Positive values indicate a position from the first occurrence, with 1 being
18580 the first one. Negative values indicate positions relative to the last one,
18581 with -1 being the last one. A typical use is with the X-Forwarded-For header.
18582
Christopher Faulet687a68e2020-11-24 17:13:24 +010018583req.hdrs : string
18584 Returns the current request headers as string including the last empty line
18585 separating headers from the request body. The last empty line can be used to
18586 detect a truncated header block. This sample fetch is useful for some SPOE
18587 headers analyzers and for advanced logging.
18588
18589req.hdrs_bin : binary
18590 Returns the current request headers contained in preparsed binary form. This
18591 is useful for offloading some processing with SPOE. Each string is described
18592 by a length followed by the number of bytes indicated in the length. The
18593 length is represented using the variable integer encoding detailed in the
18594 SPOE documentation. The end of the list is marked by a couple of empty header
18595 names and values (length of 0 for both).
18596
18597 *(<str:header-name><str:header-value>)<empty string><empty string>
Frédéric Lécailleec891192019-02-26 15:02:35 +010018598
Christopher Faulet687a68e2020-11-24 17:13:24 +010018599 int: refer to the SPOE documentation for the encoding
18600 str: <int:length><bytes>
Frédéric Lécailleec891192019-02-26 15:02:35 +010018601
Willy Tarreau74ca5042013-06-11 23:12:07 +020018602http_auth(<userlist>) : boolean
18603 Returns a boolean indicating whether the authentication data received from
18604 the client match a username & password stored in the specified userlist. This
18605 fetch function is not really useful outside of ACLs. Currently only http
18606 basic auth is supported.
18607
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010018608http_auth_group(<userlist>) : string
18609 Returns a string corresponding to the user name found in the authentication
18610 data received from the client if both the user name and password are valid
18611 according to the specified userlist. The main purpose is to use it in ACLs
18612 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018613 This fetch function is not really useful outside of ACLs. Currently only http
18614 basic auth is supported.
18615
18616 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010018617 http_auth_group(<userlist>) : group ...
18618 Returns true when the user extracted from the request and whose password is
18619 valid according to the specified userlist belongs to at least one of the
18620 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020018621
Christopher Fauleta4063562019-08-02 11:51:37 +020018622http_auth_pass : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018623 Returns the user's password found in the authentication data received from
18624 the client, as supplied in the Authorization header. Not checks are
18625 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018626
18627http_auth_type : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018628 Returns the authentication method found in the authentication data received from
18629 the client, as supplied in the Authorization header. Not checks are
18630 performed by this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018631
18632http_auth_user : string
Willy Tarreauc9c6cdb2020-03-05 16:03:58 +010018633 Returns the user name found in the authentication data received from the
18634 client, as supplied in the Authorization header. Not checks are performed by
18635 this sample fetch. Only Basic authentication is supported.
Christopher Fauleta4063562019-08-02 11:51:37 +020018636
Willy Tarreau74ca5042013-06-11 23:12:07 +020018637http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020018638 Returns true when the request being processed is the first one of the
18639 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020018640 from some requests when a request is not the first one, or to help grouping
18641 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020018642
Willy Tarreau74ca5042013-06-11 23:12:07 +020018643method : integer + string
18644 Returns an integer value corresponding to the method in the HTTP request. For
18645 example, "GET" equals 1 (check sources to establish the matching). Value 9
18646 means "other method" and may be converted to a string extracted from the
18647 stream. This should not be used directly as a sample, this is only meant to
18648 be used from ACLs, which transparently convert methods from patterns to these
18649 integer + string values. Some predefined ACL already check for most common
18650 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018651
Willy Tarreau74ca5042013-06-11 23:12:07 +020018652 ACL derivatives :
18653 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020018654
Willy Tarreau74ca5042013-06-11 23:12:07 +020018655 Example :
18656 # only accept GET and HEAD requests
18657 acl valid_method method GET HEAD
18658 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020018659
Willy Tarreau74ca5042013-06-11 23:12:07 +020018660path : string
18661 This extracts the request's URL path, which starts at the first slash and
18662 ends before the question mark (without the host part). A typical use is with
18663 prefetch-capable caches, and with portals which need to aggregate multiple
18664 information from databases and keep them in caches. Note that with outgoing
18665 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010018666 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020018667 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018668
Willy Tarreau74ca5042013-06-11 23:12:07 +020018669 ACL derivatives :
18670 path : exact string match
18671 path_beg : prefix match
18672 path_dir : subdir match
18673 path_dom : domain match
18674 path_end : suffix match
18675 path_len : length match
18676 path_reg : regex match
18677 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020018678
Christopher Faulete720c322020-09-02 17:25:18 +020018679pathq : string
18680 This extracts the request's URL path with the query-string, which starts at
18681 the first slash. This sample fetch is pretty handy to always retrieve a
18682 relative URI, excluding the scheme and the authority part, if any. Indeed,
18683 while it is the common representation for an HTTP/1.1 request target, in
18684 HTTP/2, an absolute URI is often used. This sample fetch will return the same
18685 result in both cases.
18686
Willy Tarreau49ad95c2015-01-19 15:06:26 +010018687query : string
18688 This extracts the request's query string, which starts after the first
18689 question mark. If no question mark is present, this fetch returns nothing. If
18690 a question mark is present but nothing follows, it returns an empty string.
18691 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010018692 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010018693 which stops before the question mark.
18694
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018695req.hdr_names([<delim>]) : string
18696 This builds a string made from the concatenation of all header names as they
18697 appear in the request when the rule is evaluated. The default delimiter is
18698 the comma (',') but it may be overridden as an optional argument <delim>. In
18699 this case, only the first character of <delim> is considered.
18700
Willy Tarreau74ca5042013-06-11 23:12:07 +020018701req.ver : string
18702req_ver : string (deprecated)
18703 Returns the version string from the HTTP request, for example "1.1". This can
18704 be useful for logs, but is mostly there for ACL. Some predefined ACL already
18705 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018706
Willy Tarreau74ca5042013-06-11 23:12:07 +020018707 ACL derivatives :
18708 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020018709
Christopher Faulete596d182020-05-05 17:46:34 +020018710res.body : binary
18711 This returns the HTTP response's available body as a block of data. Unlike
18712 the request side, there is no directive to wait for the response's body. This
18713 sample fetch is really useful (and usable) in the health-check context. It
18714 may be used in tcp-check based expect rules.
18715
18716res.body_len : integer
18717 This returns the length of the HTTP response available body in bytes. Unlike
18718 the request side, there is no directive to wait for the response's body. This
18719 sample fetch is really useful (and usable) in the health-check context. It
18720 may be used in tcp-check based expect rules.
18721
18722res.body_size : integer
18723 This returns the advertised length of the HTTP response body in bytes. It
18724 will represent the advertised Content-Length header, or the size of the
18725 available data in case of chunked encoding. Unlike the request side, there is
18726 no directive to wait for the response body. This sample fetch is really
18727 useful (and usable) in the health-check context. It may be used in tcp-check
18728 based expect rules.
18729
Remi Tricot-Le Bretonbf971212020-10-27 11:55:57 +010018730res.cache_hit : boolean
18731 Returns the boolean "true" value if the response has been built out of an
18732 HTTP cache entry, otherwise returns boolean "false".
18733
18734res.cache_name : string
18735 Returns a string containing the name of the HTTP cache that was used to
18736 build the HTTP response if res.cache_hit is true, otherwise returns an
18737 empty string.
18738
Willy Tarreau74ca5042013-06-11 23:12:07 +020018739res.comp : boolean
18740 Returns the boolean "true" value if the response has been compressed by
18741 HAProxy, otherwise returns boolean "false". This may be used to add
18742 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018743
Willy Tarreau74ca5042013-06-11 23:12:07 +020018744res.comp_algo : string
18745 Returns a string containing the name of the algorithm used if the response
18746 was compressed by HAProxy, for example : "deflate". This may be used to add
18747 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018748
Willy Tarreau74ca5042013-06-11 23:12:07 +020018749res.cook([<name>]) : string
18750scook([<name>]) : string (deprecated)
18751 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18752 header line from the response, and returns its value as string. If no name is
Christopher Faulete596d182020-05-05 17:46:34 +020018753 specified, the first cookie value is returned. It may be used in tcp-check
18754 based expect rules.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020018755
Willy Tarreau74ca5042013-06-11 23:12:07 +020018756 ACL derivatives :
18757 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020018758
Willy Tarreau74ca5042013-06-11 23:12:07 +020018759res.cook_cnt([<name>]) : integer
18760scook_cnt([<name>]) : integer (deprecated)
18761 Returns an integer value representing the number of occurrences of the cookie
18762 <name> in the response, or all cookies if <name> is not specified. This is
Christopher Faulete596d182020-05-05 17:46:34 +020018763 mostly useful when combined with ACLs to detect suspicious responses. It may
18764 be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018765
Willy Tarreau74ca5042013-06-11 23:12:07 +020018766res.cook_val([<name>]) : integer
18767scook_val([<name>]) : integer (deprecated)
18768 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18769 header line from the response, and converts its value to an integer which is
Christopher Faulete596d182020-05-05 17:46:34 +020018770 returned. If no name is specified, the first cookie value is returned. It may
18771 be used in tcp-check based expect rules.
Willy Tarreaud63335a2010-02-26 12:56:52 +010018772
Willy Tarreau74ca5042013-06-11 23:12:07 +020018773res.fhdr([<name>[,<occ>]]) : string
18774 This extracts the last occurrence of header <name> in an HTTP response, or of
18775 the last header if no <name> is specified. Optionally, a specific occurrence
18776 might be specified as a position number. Positive values indicate a position
18777 from the first occurrence, with 1 being the first one. Negative values
18778 indicate positions relative to the last one, with -1 being the last one. It
18779 differs from res.hdr() in that any commas present in the value are returned
18780 and are not used as delimiters. If this is not desired, the res.hdr() fetch
18781 should be used instead. This is sometimes useful with headers such as Date or
Christopher Faulete596d182020-05-05 17:46:34 +020018782 Expires. It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018783
Willy Tarreau74ca5042013-06-11 23:12:07 +020018784res.fhdr_cnt([<name>]) : integer
18785 Returns an integer value representing the number of occurrences of response
18786 header field name <name>, or the total number of header fields if <name> is
18787 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
18788 the number of full line headers and does not stop on commas. If this is not
Christopher Faulete596d182020-05-05 17:46:34 +020018789 desired, the res.hdr_cnt() fetch should be used instead. It may be used in
18790 tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018791
Willy Tarreau74ca5042013-06-11 23:12:07 +020018792res.hdr([<name>[,<occ>]]) : string
18793shdr([<name>[,<occ>]]) : string (deprecated)
18794 This extracts the last occurrence of header <name> in an HTTP response, or of
18795 the last header if no <name> is specified. Optionally, a specific occurrence
18796 might be specified as a position number. Positive values indicate a position
18797 from the first occurrence, with 1 being the first one. Negative values
18798 indicate positions relative to the last one, with -1 being the last one. This
18799 can be useful to learn some data into a stick-table. The function considers
18800 any comma as a delimiter for distinct values. If this is not desired, the
Christopher Faulete596d182020-05-05 17:46:34 +020018801 res.fhdr() fetch should be used instead. It may be used in tcp-check based
18802 expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018803
Willy Tarreau74ca5042013-06-11 23:12:07 +020018804 ACL derivatives :
18805 shdr([<name>[,<occ>]]) : exact string match
18806 shdr_beg([<name>[,<occ>]]) : prefix match
18807 shdr_dir([<name>[,<occ>]]) : subdir match
18808 shdr_dom([<name>[,<occ>]]) : domain match
18809 shdr_end([<name>[,<occ>]]) : suffix match
18810 shdr_len([<name>[,<occ>]]) : length match
18811 shdr_reg([<name>[,<occ>]]) : regex match
18812 shdr_sub([<name>[,<occ>]]) : substring match
18813
18814res.hdr_cnt([<name>]) : integer
18815shdr_cnt([<name>]) : integer (deprecated)
18816 Returns an integer value representing the number of occurrences of response
18817 header field name <name>, or the total number of header fields if <name> is
18818 not specified. The function considers any comma as a delimiter for distinct
18819 values. If this is not desired, the res.fhdr_cnt() fetch should be used
Christopher Faulete596d182020-05-05 17:46:34 +020018820 instead. It may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018821
Willy Tarreau74ca5042013-06-11 23:12:07 +020018822res.hdr_ip([<name>[,<occ>]]) : ip
18823shdr_ip([<name>[,<occ>]]) : ip (deprecated)
18824 This extracts the last occurrence of header <name> in an HTTP response,
18825 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
18826 specific occurrence might be specified as a position number. Positive values
18827 indicate a position from the first occurrence, with 1 being the first one.
18828 Negative values indicate positions relative to the last one, with -1 being
Christopher Faulete596d182020-05-05 17:46:34 +020018829 the last one. This can be useful to learn some data into a stick table. It
18830 may be used in tcp-check based expect rules.
Willy Tarreau6a06a402007-07-15 20:15:28 +020018831
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018832res.hdr_names([<delim>]) : string
18833 This builds a string made from the concatenation of all header names as they
18834 appear in the response when the rule is evaluated. The default delimiter is
18835 the comma (',') but it may be overridden as an optional argument <delim>. In
Christopher Faulete596d182020-05-05 17:46:34 +020018836 this case, only the first character of <delim> is considered. It may be used
18837 in tcp-check based expect rules.
Willy Tarreaueb27ec72015-02-20 13:55:29 +010018838
Willy Tarreau74ca5042013-06-11 23:12:07 +020018839res.hdr_val([<name>[,<occ>]]) : integer
18840shdr_val([<name>[,<occ>]]) : integer (deprecated)
18841 This extracts the last occurrence of header <name> in an HTTP response, and
18842 converts it to an integer value. Optionally, a specific occurrence might be
18843 specified as a position number. Positive values indicate a position from the
18844 first occurrence, with 1 being the first one. Negative values indicate
18845 positions relative to the last one, with -1 being the last one. This can be
Christopher Faulete596d182020-05-05 17:46:34 +020018846 useful to learn some data into a stick table. It may be used in tcp-check
18847 based expect rules.
18848
18849res.hdrs : string
18850 Returns the current response headers as string including the last empty line
18851 separating headers from the request body. The last empty line can be used to
18852 detect a truncated header block. This sample fetch is useful for some SPOE
18853 headers analyzers and for advanced logging. It may also be used in tcp-check
18854 based expect rules.
18855
18856res.hdrs_bin : binary
18857 Returns the current response headers contained in preparsed binary form. This
18858 is useful for offloading some processing with SPOE. It may be used in
18859 tcp-check based expect rules. Each string is described by a length followed
18860 by the number of bytes indicated in the length. The length is represented
18861 using the variable integer encoding detailed in the SPOE documentation. The
18862 end of the list is marked by a couple of empty header names and values
18863 (length of 0 for both).
18864
18865 *(<str:header-name><str:header-value>)<empty string><empty string>
18866
18867 int: refer to the SPOE documentation for the encoding
18868 str: <int:length><bytes>
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010018869
Willy Tarreau74ca5042013-06-11 23:12:07 +020018870res.ver : string
18871resp_ver : string (deprecated)
18872 Returns the version string from the HTTP response, for example "1.1". This
Christopher Faulete596d182020-05-05 17:46:34 +020018873 can be useful for logs, but is mostly there for ACL. It may be used in
18874 tcp-check based expect rules.
Willy Tarreau0e698542011-09-16 08:32:32 +020018875
Willy Tarreau74ca5042013-06-11 23:12:07 +020018876 ACL derivatives :
18877 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010018878
Willy Tarreau74ca5042013-06-11 23:12:07 +020018879set-cookie([<name>]) : string (deprecated)
18880 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
18881 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020018882 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020018883 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018884
Willy Tarreau74ca5042013-06-11 23:12:07 +020018885 This fetch function is deprecated and has been superseded by the "res.cook"
18886 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018887
Willy Tarreau74ca5042013-06-11 23:12:07 +020018888status : integer
18889 Returns an integer containing the HTTP status code in the HTTP response, for
18890 example, 302. It is mostly used within ACLs and integer ranges, for example,
Christopher Faulete596d182020-05-05 17:46:34 +020018891 to remove any Location header if the response is not a 3xx. It may be used in
18892 tcp-check based expect rules.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018893
Thierry Fournier0e00dca2016-04-07 15:47:40 +020018894unique-id : string
18895 Returns the unique-id attached to the request. The directive
18896 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
18897 fails. Note that the unique-id is usually used with HTTP requests, however this
18898 sample fetch can be used with other protocols. Obviously, if it is used with
18899 other protocols than HTTP, the unique-id-format directive must not contain
18900 HTTP parts. See: unique-id-format and unique-id-header
18901
Willy Tarreau74ca5042013-06-11 23:12:07 +020018902url : string
18903 This extracts the request's URL as presented in the request. A typical use is
18904 with prefetch-capable caches, and with portals which need to aggregate
18905 multiple information from databases and keep them in caches. With ACLs, using
18906 "path" is preferred over using "url", because clients may send a full URL as
18907 is normally done with proxies. The only real use is to match "*" which does
18908 not match in "path", and for which there is already a predefined ACL. See
18909 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018910
Willy Tarreau74ca5042013-06-11 23:12:07 +020018911 ACL derivatives :
18912 url : exact string match
18913 url_beg : prefix match
18914 url_dir : subdir match
18915 url_dom : domain match
18916 url_end : suffix match
18917 url_len : length match
18918 url_reg : regex match
18919 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018920
Willy Tarreau74ca5042013-06-11 23:12:07 +020018921url_ip : ip
18922 This extracts the IP address from the request's URL when the host part is
18923 presented as an IP address. Its use is very limited. For instance, a
18924 monitoring system might use this field as an alternative for the source IP in
18925 order to test what path a given source address would follow, or to force an
18926 entry in a table for a given source address. With ACLs it can be used to
18927 restrict access to certain systems through a proxy, for example when combined
18928 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018929
Willy Tarreau74ca5042013-06-11 23:12:07 +020018930url_port : integer
18931 This extracts the port part from the request's URL. Note that if the port is
18932 not specified in the request, port 80 is assumed. With ACLs it can be used to
18933 restrict access to certain systems through a proxy, for example when combined
18934 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018935
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018936urlp([<name>[,<delim>]]) : string
18937url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020018938 This extracts the first occurrence of the parameter <name> in the query
18939 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018940 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
18941 any parameter will match, and the first one will be returned. The result is
18942 a string corresponding to the value of the parameter <name> as presented in
18943 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020018944 stickiness based on a client ID, to extract an application cookie passed as a
18945 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020018946 this fetch iterates over multiple parameters and will iteratively report all
18947 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018948
Willy Tarreau74ca5042013-06-11 23:12:07 +020018949 ACL derivatives :
18950 urlp(<name>[,<delim>]) : exact string match
18951 urlp_beg(<name>[,<delim>]) : prefix match
18952 urlp_dir(<name>[,<delim>]) : subdir match
18953 urlp_dom(<name>[,<delim>]) : domain match
18954 urlp_end(<name>[,<delim>]) : suffix match
18955 urlp_len(<name>[,<delim>]) : length match
18956 urlp_reg(<name>[,<delim>]) : regex match
18957 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018958
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018959
Willy Tarreau74ca5042013-06-11 23:12:07 +020018960 Example :
18961 # match http://example.com/foo?PHPSESSIONID=some_id
18962 stick on urlp(PHPSESSIONID)
18963 # match http://example.com/foo;JSESSIONID=some_id
18964 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020018965
Jarno Huuskonen676f6222017-03-30 09:19:45 +030018966urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020018967 See "urlp" above. This one extracts the URL parameter <name> in the request
18968 and converts it to an integer value. This can be used for session stickiness
18969 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020018970
Dragan Dosen0070cd52016-06-16 12:19:49 +020018971url32 : integer
18972 This returns a 32-bit hash of the value obtained by concatenating the first
18973 Host header and the whole URL including parameters (not only the path part of
18974 the request, as in the "base32" fetch above). This is useful to track per-URL
18975 activity. A shorter hash is stored, saving a lot of memory. The output type
18976 is an unsigned integer.
18977
18978url32+src : binary
18979 This returns the concatenation of the "url32" fetch and the "src" fetch. The
18980 resulting type is of type binary, with a size of 8 or 20 bytes depending on
18981 the source address family. This can be used to track per-IP, per-URL counters.
18982
Christopher Faulet16032ab2020-04-30 11:30:00 +020018983
Christopher Faulete596d182020-05-05 17:46:34 +0200189847.3.7. Fetching samples for developers
Christopher Fauletd47941d2020-01-08 14:40:19 +010018985---------------------------------------
18986
18987This set of sample fetch methods is reserved to developers and must never be
18988used on a production environment, except on developer demand, for debugging
18989purposes. Moreover, no special care will be taken on backwards compatibility.
18990There is no warranty the following sample fetches will never change, be renamed
18991or simply removed. So be really careful if you should use one of them. To avoid
18992any ambiguity, these sample fetches are placed in the dedicated scope "internal",
18993for instance "internal.strm.is_htx".
18994
18995internal.htx.data : integer
18996 Returns the size in bytes used by data in the HTX message associated to a
18997 channel. The channel is chosen depending on the sample direction.
18998
18999internal.htx.free : integer
19000 Returns the free space (size - used) in bytes in the HTX message associated
19001 to a channel. The channel is chosen depending on the sample direction.
19002
19003internal.htx.free_data : integer
19004 Returns the free space for the data in bytes in the HTX message associated to
19005 a channel. The channel is chosen depending on the sample direction.
19006
19007internal.htx.has_eom : boolean
19008 Returns true if the HTX message associated to a channel contains an
19009 end-of-message block (EOM). Otherwise, it returns false. The channel is
19010 chosen depending on the sample direction.
19011
19012internal.htx.nbblks : integer
19013 Returns the number of blocks present in the HTX message associated to a
19014 channel. The channel is chosen depending on the sample direction.
19015
19016internal.htx.size : integer
19017 Returns the total size in bytes of the HTX message associated to a
19018 channel. The channel is chosen depending on the sample direction.
19019
19020internal.htx.used : integer
19021 Returns the total size used in bytes (data + metadata) in the HTX message
19022 associated to a channel. The channel is chosen depending on the sample
19023 direction.
19024
19025internal.htx_blk.size(<idx>) : integer
19026 Returns the size of the block at the position <idx> in the HTX message
19027 associated to a channel or 0 if it does not exist. The channel is chosen
19028 depending on the sample direction. <idx> may be any positive integer or one
19029 of the special value :
19030 * head : The oldest inserted block
19031 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019032 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019033
19034internal.htx_blk.type(<idx>) : string
19035 Returns the type of the block at the position <idx> in the HTX message
19036 associated to a channel or "HTX_BLK_UNUSED" if it does not exist. The channel
19037 is chosen depending on the sample direction. <idx> may be any positive
19038 integer or one of the special value :
19039 * head : The oldest inserted block
19040 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019041 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019042
19043internal.htx_blk.data(<idx>) : binary
19044 Returns the value of the DATA block at the position <idx> in the HTX message
19045 associated to a channel or an empty string if it does not exist or if it is
19046 not a DATA block. The channel is chosen depending on the sample direction.
19047 <idx> may be any positive integer or one of the special value :
19048
19049 * head : The oldest inserted block
19050 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019051 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019052
19053internal.htx_blk.hdrname(<idx>) : string
19054 Returns the header name of the HEADER block at the position <idx> in the HTX
19055 message associated to a channel or an empty string if it does not exist or if
19056 it is not an HEADER block. The channel is chosen depending on the sample
19057 direction. <idx> may be any positive integer or one of the special value :
19058
19059 * head : The oldest inserted block
19060 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019061 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019062
19063internal.htx_blk.hdrval(<idx>) : string
19064 Returns the header value of the HEADER block at the position <idx> in the HTX
19065 message associated to a channel or an empty string if it does not exist or if
19066 it is not an HEADER block. The channel is chosen depending on the sample
19067 direction. <idx> may be any positive integer or one of the special value :
19068
19069 * head : The oldest inserted block
19070 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019071 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019072
19073internal.htx_blk.start_line(<idx>) : string
19074 Returns the value of the REQ_SL or RES_SL block at the position <idx> in the
19075 HTX message associated to a channel or an empty string if it does not exist
19076 or if it is not a SL block. The channel is chosen depending on the sample
19077 direction. <idx> may be any positive integer or one of the special value :
19078
19079 * head : The oldest inserted block
19080 * tail : The newest inserted block
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050019081 * first : The first block where to (re)start the analysis
Christopher Fauletd47941d2020-01-08 14:40:19 +010019082
19083internal.strm.is_htx : boolean
19084 Returns true if the current stream is an HTX stream. It means the data in the
19085 channels buffers are stored using the internal HTX representation. Otherwise,
19086 it returns false.
19087
19088
Willy Tarreau74ca5042013-06-11 23:12:07 +0200190897.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019090---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010019091
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019092Some predefined ACLs are hard-coded so that they do not have to be declared in
19093every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020019094order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010019095
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019096ACL name Equivalent to Usage
19097---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019098FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020019099HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019100HTTP_1.0 req_ver 1.0 match HTTP version 1.0
19101HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010019102HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
19103HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
19104HTTP_URL_SLASH url_beg / match URL beginning with "/"
19105HTTP_URL_STAR url * match URL equal to "*"
19106LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019107METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020019108METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019109METH_GET method GET HEAD match HTTP GET or HEAD method
19110METH_HEAD method HEAD match HTTP HEAD method
19111METH_OPTIONS method OPTIONS match HTTP OPTIONS method
19112METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020019113METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019114METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020019115RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019116REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010019117TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019118WAIT_END wait_end wait for end of content analysis
19119---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010019120
Willy Tarreaub937b7e2010-01-12 15:27:54 +010019121
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200191228. Logging
19123----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010019124
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019125One of HAProxy's strong points certainly lies is its precise logs. It probably
19126provides the finest level of information available for such a product, which is
19127very important for troubleshooting complex environments. Standard information
19128provided in logs include client ports, TCP/HTTP state timers, precise session
19129state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010019130to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019131headers.
19132
19133In order to improve administrators reactivity, it offers a great transparency
19134about encountered problems, both internal and external, and it is possible to
19135send logs to different sources at the same time with different level filters :
19136
19137 - global process-level logs (system errors, start/stop, etc..)
19138 - per-instance system and internal errors (lack of resource, bugs, ...)
19139 - per-instance external troubles (servers up/down, max connections)
19140 - per-instance activity (client connections), either at the establishment or
19141 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010019142 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060019143 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019144
19145The ability to distribute different levels of logs to different log servers
19146allow several production teams to interact and to fix their problems as soon
19147as possible. For example, the system team might monitor system-wide errors,
19148while the application team might be monitoring the up/down for their servers in
19149real time, and the security team might analyze the activity logs with one hour
19150delay.
19151
19152
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200191538.1. Log levels
19154---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019155
Simon Hormandf791f52011-05-29 15:01:10 +090019156TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019157source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090019158HTTP request, HTTP return code, number of bytes transmitted, conditions
19159in which the session ended, and even exchanged cookies values. For example
19160track a particular user's problems. All messages may be sent to up to two
19161syslog servers. Check the "log" keyword in section 4.2 for more information
19162about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019163
19164
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200191658.2. Log formats
19166----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019167
William Lallemand48940402012-01-30 16:47:22 +010019168HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090019169and will be detailed in the following sections. A few of them may vary
19170slightly with the configuration, due to indicators specific to certain
19171options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019172
19173 - the default format, which is very basic and very rarely used. It only
19174 provides very basic information about the incoming connection at the moment
19175 it is accepted : source IP:port, destination IP:port, and frontend-name.
19176 This mode will eventually disappear so it will not be described to great
19177 extents.
19178
19179 - the TCP format, which is more advanced. This format is enabled when "option
19180 tcplog" is set on the frontend. HAProxy will then usually wait for the
19181 connection to terminate before logging. This format provides much richer
19182 information, such as timers, connection counts, queue size, etc... This
19183 format is recommended for pure TCP proxies.
19184
19185 - the HTTP format, which is the most advanced for HTTP proxying. This format
19186 is enabled when "option httplog" is set on the frontend. It provides the
19187 same information as the TCP format with some HTTP-specific fields such as
19188 the request, the status code, and captures of headers and cookies. This
19189 format is recommended for HTTP proxies.
19190
Emeric Brun3a058f32009-06-30 18:26:00 +020019191 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
19192 fields arranged in the same order as the CLF format. In this mode, all
19193 timers, captures, flags, etc... appear one per field after the end of the
19194 common fields, in the same order they appear in the standard HTTP format.
19195
William Lallemand48940402012-01-30 16:47:22 +010019196 - the custom log format, allows you to make your own log line.
19197
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019198Next sections will go deeper into details for each of these formats. Format
19199specification will be performed on a "field" basis. Unless stated otherwise, a
19200field is a portion of text delimited by any number of spaces. Since syslog
19201servers are susceptible of inserting fields at the beginning of a line, it is
19202always assumed that the first field is the one containing the process name and
19203identifier.
19204
19205Note : Since log lines may be quite long, the log examples in sections below
19206 might be broken into multiple lines. The example log lines will be
19207 prefixed with 3 closing angle brackets ('>>>') and each time a log is
19208 broken into multiple lines, each non-final line will end with a
19209 backslash ('\') and the next line will start indented by two characters.
19210
19211
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200192128.2.1. Default log format
19213-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019214
19215This format is used when no specific option is set. The log is emitted as soon
19216as the connection is accepted. One should note that this currently is the only
19217format which logs the request's destination IP and ports.
19218
19219 Example :
19220 listen www
19221 mode http
19222 log global
19223 server srv1 127.0.0.1:8000
19224
19225 >>> Feb 6 12:12:09 localhost \
19226 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
19227 (www/HTTP)
19228
19229 Field Format Extract from the example above
19230 1 process_name '[' pid ']:' haproxy[14385]:
19231 2 'Connect from' Connect from
19232 3 source_ip ':' source_port 10.0.1.2:33312
19233 4 'to' to
19234 5 destination_ip ':' destination_port 10.0.3.31:8012
19235 6 '(' frontend_name '/' mode ')' (www/HTTP)
19236
19237Detailed fields description :
19238 - "source_ip" is the IP address of the client which initiated the connection.
19239 - "source_port" is the TCP port of the client which initiated the connection.
19240 - "destination_ip" is the IP address the client connected to.
19241 - "destination_port" is the TCP port the client connected to.
19242 - "frontend_name" is the name of the frontend (or listener) which received
19243 and processed the connection.
19244 - "mode is the mode the frontend is operating (TCP or HTTP).
19245
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019246In case of a UNIX socket, the source and destination addresses are marked as
19247"unix:" and the ports reflect the internal ID of the socket which accepted the
19248connection (the same ID as reported in the stats).
19249
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019250It is advised not to use this deprecated format for newer installations as it
19251will eventually disappear.
19252
19253
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200192548.2.2. TCP log format
19255---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019256
19257The TCP format is used when "option tcplog" is specified in the frontend, and
19258is the recommended format for pure TCP proxies. It provides a lot of precious
19259information for troubleshooting. Since this format includes timers and byte
19260counts, the log is normally emitted at the end of the session. It can be
19261emitted earlier if "option logasap" is specified, which makes sense in most
19262environments with long sessions such as remote terminals. Sessions which match
19263the "monitor" rules are never logged. It is also possible not to emit logs for
19264sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019265specifying "option dontlognull" in the frontend. Successful connections will
19266not be logged if "option dontlog-normal" is specified in the frontend. A few
19267fields may slightly vary depending on some configuration options, those are
19268marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019269
19270 Example :
19271 frontend fnt
19272 mode tcp
19273 option tcplog
19274 log global
19275 default_backend bck
19276
19277 backend bck
19278 server srv1 127.0.0.1:8000
19279
19280 >>> Feb 6 12:12:56 localhost \
19281 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
19282 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
19283
19284 Field Format Extract from the example above
19285 1 process_name '[' pid ']:' haproxy[14387]:
19286 2 client_ip ':' client_port 10.0.1.2:33313
19287 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
19288 4 frontend_name fnt
19289 5 backend_name '/' server_name bck/srv1
19290 6 Tw '/' Tc '/' Tt* 0/0/5007
19291 7 bytes_read* 212
19292 8 termination_state --
19293 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
19294 10 srv_queue '/' backend_queue 0/0
19295
19296Detailed fields description :
19297 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019298 connection to haproxy. If the connection was accepted on a UNIX socket
19299 instead, the IP address would be replaced with the word "unix". Note that
19300 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019301 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010019302 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019303 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019304
19305 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019306 If the connection was accepted on a UNIX socket instead, the port would be
19307 replaced with the ID of the accepting socket, which is also reported in the
19308 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019309
19310 - "accept_date" is the exact date when the connection was received by haproxy
19311 (which might be very slightly different from the date observed on the
19312 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020019313 the same date which may appear in any upstream firewall's log. When used in
19314 HTTP mode, the accept_date field will be reset to the first moment the
19315 connection is ready to receive a new request (end of previous response for
19316 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019317
19318 - "frontend_name" is the name of the frontend (or listener) which received
19319 and processed the connection.
19320
19321 - "backend_name" is the name of the backend (or listener) which was selected
19322 to manage the connection to the server. This will be the same as the
19323 frontend if no switching rule has been applied, which is common for TCP
19324 applications.
19325
19326 - "server_name" is the name of the last server to which the connection was
19327 sent, which might differ from the first one if there were connection errors
19328 and a redispatch occurred. Note that this server belongs to the backend
19329 which processed the request. If the connection was aborted before reaching
19330 a server, "<NOSRV>" is indicated instead of a server name.
19331
19332 - "Tw" is the total time in milliseconds spent waiting in the various queues.
19333 It can be "-1" if the connection was aborted before reaching the queue.
19334 See "Timers" below for more details.
19335
19336 - "Tc" is the total time in milliseconds spent waiting for the connection to
19337 establish to the final server, including retries. It can be "-1" if the
19338 connection was aborted before a connection could be established. See
19339 "Timers" below for more details.
19340
19341 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030019342 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019343 "option logasap" was specified, then the time counting stops at the moment
19344 the log is emitted. In this case, a '+' sign is prepended before the value,
19345 indicating that the final one will be larger. See "Timers" below for more
19346 details.
19347
19348 - "bytes_read" is the total number of bytes transmitted from the server to
19349 the client when the log is emitted. If "option logasap" is specified, the
19350 this value will be prefixed with a '+' sign indicating that the final one
19351 may be larger. Please note that this value is a 64-bit counter, so log
19352 analysis tools must be able to handle it without overflowing.
19353
19354 - "termination_state" is the condition the session was in when the session
19355 ended. This indicates the session state, which side caused the end of
19356 session to happen, and for what reason (timeout, error, ...). The normal
19357 flags should be "--", indicating the session was closed by either end with
19358 no data remaining in buffers. See below "Session state at disconnection"
19359 for more details.
19360
19361 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019362 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019363 limits have been reached. For instance, if actconn is close to 512 when
19364 multiple connection errors occur, chances are high that the system limits
19365 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019366 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019367
19368 - "feconn" is the total number of concurrent connections on the frontend when
19369 the session was logged. It is useful to estimate the amount of resource
19370 required to sustain high loads, and to detect when the frontend's "maxconn"
19371 has been reached. Most often when this value increases by huge jumps, it is
19372 because there is congestion on the backend servers, but sometimes it can be
19373 caused by a denial of service attack.
19374
19375 - "beconn" is the total number of concurrent connections handled by the
19376 backend when the session was logged. It includes the total number of
19377 concurrent connections active on servers as well as the number of
19378 connections pending in queues. It is useful to estimate the amount of
19379 additional servers needed to support high loads for a given application.
19380 Most often when this value increases by huge jumps, it is because there is
19381 congestion on the backend servers, but sometimes it can be caused by a
19382 denial of service attack.
19383
19384 - "srv_conn" is the total number of concurrent connections still active on
19385 the server when the session was logged. It can never exceed the server's
19386 configured "maxconn" parameter. If this value is very often close or equal
19387 to the server's "maxconn", it means that traffic regulation is involved a
19388 lot, meaning that either the server's maxconn value is too low, or that
19389 there aren't enough servers to process the load with an optimal response
19390 time. When only one of the server's "srv_conn" is high, it usually means
19391 that this server has some trouble causing the connections to take longer to
19392 be processed than on other servers.
19393
19394 - "retries" is the number of connection retries experienced by this session
19395 when trying to connect to the server. It must normally be zero, unless a
19396 server is being stopped at the same moment the connection was attempted.
19397 Frequent retries generally indicate either a network problem between
19398 haproxy and the server, or a misconfigured system backlog on the server
19399 preventing new connections from being queued. This field may optionally be
19400 prefixed with a '+' sign, indicating that the session has experienced a
19401 redispatch after the maximal retry count has been reached on the initial
19402 server. In this case, the server name appearing in the log is the one the
19403 connection was redispatched to, and not the first one, though both may
19404 sometimes be the same in case of hashing for instance. So as a general rule
19405 of thumb, when a '+' is present in front of the retry count, this count
19406 should not be attributed to the logged server.
19407
19408 - "srv_queue" is the total number of requests which were processed before
19409 this one in the server queue. It is zero when the request has not gone
19410 through the server queue. It makes it possible to estimate the approximate
19411 server's response time by dividing the time spent in queue by the number of
19412 requests in the queue. It is worth noting that if a session experiences a
19413 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010019414 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019415 backend queue unless a redispatch occurs.
19416
19417 - "backend_queue" is the total number of requests which were processed before
19418 this one in the backend's global queue. It is zero when the request has not
19419 gone through the global queue. It makes it possible to estimate the average
19420 queue length, which easily translates into a number of missing servers when
19421 divided by a server's "maxconn" parameter. It is worth noting that if a
19422 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010019423 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019424 through both the server queue and the backend queue unless a redispatch
19425 occurs.
19426
19427
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200194288.2.3. HTTP log format
19429----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019430
19431The HTTP format is the most complete and the best suited for HTTP proxies. It
19432is enabled by when "option httplog" is specified in the frontend. It provides
19433the same level of information as the TCP format with additional features which
19434are specific to the HTTP protocol. Just like the TCP format, the log is usually
19435emitted at the end of the session, unless "option logasap" is specified, which
19436generally only makes sense for download sites. A session which matches the
19437"monitor" rules will never logged. It is also possible not to log sessions for
19438which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019439frontend. Successful connections will not be logged if "option dontlog-normal"
19440is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019441
19442Most fields are shared with the TCP log, some being different. A few fields may
19443slightly vary depending on some configuration options. Those ones are marked
19444with a star ('*') after the field name below.
19445
19446 Example :
19447 frontend http-in
19448 mode http
19449 option httplog
19450 log global
19451 default_backend bck
19452
19453 backend static
19454 server srv1 127.0.0.1:8000
19455
19456 >>> Feb 6 12:14:14 localhost \
19457 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
19458 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 +010019459 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019460
19461 Field Format Extract from the example above
19462 1 process_name '[' pid ']:' haproxy[14389]:
19463 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019464 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019465 4 frontend_name http-in
19466 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019467 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019468 7 status_code 200
19469 8 bytes_read* 2750
19470 9 captured_request_cookie -
19471 10 captured_response_cookie -
19472 11 termination_state ----
19473 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
19474 13 srv_queue '/' backend_queue 0/0
19475 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
19476 15 '{' captured_response_headers* '}' {}
19477 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010019478
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019479Detailed fields description :
19480 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019481 connection to haproxy. If the connection was accepted on a UNIX socket
19482 instead, the IP address would be replaced with the word "unix". Note that
19483 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019484 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010019485 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010019486 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019487
19488 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010019489 If the connection was accepted on a UNIX socket instead, the port would be
19490 replaced with the ID of the accepting socket, which is also reported in the
19491 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019492
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019493 - "request_date" is the exact date when the first byte of the HTTP request
19494 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019495
19496 - "frontend_name" is the name of the frontend (or listener) which received
19497 and processed the connection.
19498
19499 - "backend_name" is the name of the backend (or listener) which was selected
19500 to manage the connection to the server. This will be the same as the
19501 frontend if no switching rule has been applied.
19502
19503 - "server_name" is the name of the last server to which the connection was
19504 sent, which might differ from the first one if there were connection errors
19505 and a redispatch occurred. Note that this server belongs to the backend
19506 which processed the request. If the request was aborted before reaching a
19507 server, "<NOSRV>" is indicated instead of a server name. If the request was
19508 intercepted by the stats subsystem, "<STATS>" is indicated instead.
19509
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019510 - "TR" is the total time in milliseconds spent waiting for a full HTTP
19511 request from the client (not counting body) after the first byte was
19512 received. It can be "-1" if the connection was aborted before a complete
John Roeslerfb2fce12019-07-10 15:45:51 -050019513 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019514 always be very small because a request generally fits in one single packet.
19515 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020019516 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
19517 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019518
19519 - "Tw" is the total time in milliseconds spent waiting in the various queues.
19520 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020019521 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019522
19523 - "Tc" is the total time in milliseconds spent waiting for the connection to
19524 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020019525 request was aborted before a connection could be established. See section
19526 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019527
19528 - "Tr" is the total time in milliseconds spent waiting for the server to send
19529 a full HTTP response, not counting data. It can be "-1" if the request was
19530 aborted before a complete response could be received. It generally matches
19531 the server's processing time for the request, though it may be altered by
19532 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020019533 "GET" requests generally indicate an overloaded server. See section 8.4
19534 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019535
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019536 - "Ta" is the time the request remained active in haproxy, which is the total
19537 time in milliseconds elapsed between the first byte of the request was
19538 received and the last byte of response was sent. It covers all possible
19539 processing except the handshake (see Th) and idle time (see Ti). There is
19540 one exception, if "option logasap" was specified, then the time counting
19541 stops at the moment the log is emitted. In this case, a '+' sign is
19542 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020019543 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019544
19545 - "status_code" is the HTTP status code returned to the client. This status
19546 is generally set by the server, but it might also be set by haproxy when
19547 the server cannot be reached or when its response is blocked by haproxy.
19548
19549 - "bytes_read" is the total number of bytes transmitted to the client when
19550 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roeslerfb2fce12019-07-10 15:45:51 -050019551 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019552 the final one may be larger. Please note that this value is a 64-bit
19553 counter, so log analysis tools must be able to handle it without
19554 overflowing.
19555
19556 - "captured_request_cookie" is an optional "name=value" entry indicating that
19557 the client had this cookie in the request. The cookie name and its maximum
19558 length are defined by the "capture cookie" statement in the frontend
19559 configuration. The field is a single dash ('-') when the option is not
19560 set. Only one cookie may be captured, it is generally used to track session
19561 ID exchanges between a client and a server to detect session crossing
19562 between clients due to application bugs. For more details, please consult
19563 the section "Capturing HTTP headers and cookies" below.
19564
19565 - "captured_response_cookie" is an optional "name=value" entry indicating
19566 that the server has returned a cookie with its response. The cookie name
19567 and its maximum length are defined by the "capture cookie" statement in the
19568 frontend configuration. The field is a single dash ('-') when the option is
19569 not set. Only one cookie may be captured, it is generally used to track
19570 session ID exchanges between a client and a server to detect session
19571 crossing between clients due to application bugs. For more details, please
19572 consult the section "Capturing HTTP headers and cookies" below.
19573
19574 - "termination_state" is the condition the session was in when the session
19575 ended. This indicates the session state, which side caused the end of
19576 session to happen, for what reason (timeout, error, ...), just like in TCP
19577 logs, and information about persistence operations on cookies in the last
19578 two characters. The normal flags should begin with "--", indicating the
19579 session was closed by either end with no data remaining in buffers. See
19580 below "Session state at disconnection" for more details.
19581
19582 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019583 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019584 limits have been reached. For instance, if actconn is close to 512 or 1024
19585 when multiple connection errors occur, chances are high that the system
19586 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019587 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019588 system.
19589
19590 - "feconn" is the total number of concurrent connections on the frontend when
19591 the session was logged. It is useful to estimate the amount of resource
19592 required to sustain high loads, and to detect when the frontend's "maxconn"
19593 has been reached. Most often when this value increases by huge jumps, it is
19594 because there is congestion on the backend servers, but sometimes it can be
19595 caused by a denial of service attack.
19596
19597 - "beconn" is the total number of concurrent connections handled by the
19598 backend when the session was logged. It includes the total number of
19599 concurrent connections active on servers as well as the number of
19600 connections pending in queues. It is useful to estimate the amount of
19601 additional servers needed to support high loads for a given application.
19602 Most often when this value increases by huge jumps, it is because there is
19603 congestion on the backend servers, but sometimes it can be caused by a
19604 denial of service attack.
19605
19606 - "srv_conn" is the total number of concurrent connections still active on
19607 the server when the session was logged. It can never exceed the server's
19608 configured "maxconn" parameter. If this value is very often close or equal
19609 to the server's "maxconn", it means that traffic regulation is involved a
19610 lot, meaning that either the server's maxconn value is too low, or that
19611 there aren't enough servers to process the load with an optimal response
19612 time. When only one of the server's "srv_conn" is high, it usually means
19613 that this server has some trouble causing the requests to take longer to be
19614 processed than on other servers.
19615
19616 - "retries" is the number of connection retries experienced by this session
19617 when trying to connect to the server. It must normally be zero, unless a
19618 server is being stopped at the same moment the connection was attempted.
19619 Frequent retries generally indicate either a network problem between
19620 haproxy and the server, or a misconfigured system backlog on the server
19621 preventing new connections from being queued. This field may optionally be
19622 prefixed with a '+' sign, indicating that the session has experienced a
19623 redispatch after the maximal retry count has been reached on the initial
19624 server. In this case, the server name appearing in the log is the one the
19625 connection was redispatched to, and not the first one, though both may
19626 sometimes be the same in case of hashing for instance. So as a general rule
19627 of thumb, when a '+' is present in front of the retry count, this count
19628 should not be attributed to the logged server.
19629
19630 - "srv_queue" is the total number of requests which were processed before
19631 this one in the server queue. It is zero when the request has not gone
19632 through the server queue. It makes it possible to estimate the approximate
19633 server's response time by dividing the time spent in queue by the number of
19634 requests in the queue. It is worth noting that if a session experiences a
19635 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010019636 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019637 backend queue unless a redispatch occurs.
19638
19639 - "backend_queue" is the total number of requests which were processed before
19640 this one in the backend's global queue. It is zero when the request has not
19641 gone through the global queue. It makes it possible to estimate the average
19642 queue length, which easily translates into a number of missing servers when
19643 divided by a server's "maxconn" parameter. It is worth noting that if a
19644 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010019645 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019646 through both the server queue and the backend queue unless a redispatch
19647 occurs.
19648
19649 - "captured_request_headers" is a list of headers captured in the request due
19650 to the presence of the "capture request header" statement in the frontend.
19651 Multiple headers can be captured, they will be delimited by a vertical bar
19652 ('|'). When no capture is enabled, the braces do not appear, causing a
19653 shift of remaining fields. It is important to note that this field may
19654 contain spaces, and that using it requires a smarter log parser than when
19655 it's not used. Please consult the section "Capturing HTTP headers and
19656 cookies" below for more details.
19657
19658 - "captured_response_headers" is a list of headers captured in the response
19659 due to the presence of the "capture response header" statement in the
19660 frontend. Multiple headers can be captured, they will be delimited by a
19661 vertical bar ('|'). When no capture is enabled, the braces do not appear,
19662 causing a shift of remaining fields. It is important to note that this
19663 field may contain spaces, and that using it requires a smarter log parser
19664 than when it's not used. Please consult the section "Capturing HTTP headers
19665 and cookies" below for more details.
19666
19667 - "http_request" is the complete HTTP request line, including the method,
19668 request and HTTP version string. Non-printable characters are encoded (see
19669 below the section "Non-printable characters"). This is always the last
19670 field, and it is always delimited by quotes and is the only one which can
19671 contain quotes. If new fields are added to the log format, they will be
19672 added before this field. This field might be truncated if the request is
19673 huge and does not fit in the standard syslog buffer (1024 characters). This
19674 is the reason why this field must always remain the last one.
19675
19676
Cyril Bontédc4d9032012-04-08 21:57:39 +0200196778.2.4. Custom log format
19678------------------------
William Lallemand48940402012-01-30 16:47:22 +010019679
Willy Tarreau2beef582012-12-20 17:22:52 +010019680The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010019681mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010019682
Davor Ocelice9ed2812017-12-25 17:49:28 +010019683HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010019684Variables can take arguments using braces ('{}'), and multiple arguments are
19685separated by commas within the braces. Flags may be added or removed by
19686prefixing them with a '+' or '-' sign.
19687
19688Special variable "%o" may be used to propagate its flags to all other
19689variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010019690("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010019691
Willy Tarreauc8368452012-12-21 00:09:23 +010019692If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020019693as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010019694less common information such as the client's SSL certificate's DN, or to log
19695the key that would be used to store an entry into a stick table.
19696
Dragan Dosen1e3b16f2020-06-23 18:16:44 +020019697Note: spaces must be escaped. In configuration directives "log-format",
19698"log-format-sd" and "unique-id-format", spaces are considered as
19699delimiters and are merged. In order to emit a verbatim '%', it must be
19700preceded by another '%' resulting in '%%'.
William Lallemand48940402012-01-30 16:47:22 +010019701
Dragan Dosen835b9212016-02-12 13:23:03 +010019702Note: when using the RFC5424 syslog message format, the characters '"',
19703'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
19704https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
19705such cases, the use of the flag "E" should be considered.
19706
William Lallemand48940402012-01-30 16:47:22 +010019707Flags are :
19708 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040019709 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010019710 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
19711 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010019712
19713 Example:
19714
19715 log-format %T\ %t\ Some\ Text
19716 log-format %{+Q}o\ %t\ %s\ %{-Q}r
19717
Dragan Dosen835b9212016-02-12 13:23:03 +010019718 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
19719
William Lallemand48940402012-01-30 16:47:22 +010019720At the moment, the default HTTP format is defined this way :
19721
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019722 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
19723 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010019724
William Lallemandbddd4fd2012-02-27 11:23:10 +010019725the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010019726
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019727 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
19728 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
19729 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010019730
William Lallemandbddd4fd2012-02-27 11:23:10 +010019731and the default TCP format is defined this way :
19732
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019733 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
19734 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010019735
William Lallemand48940402012-01-30 16:47:22 +010019736Please refer to the table below for currently defined variables :
19737
William Lallemandbddd4fd2012-02-27 11:23:10 +010019738 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019739 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019740 +---+------+-----------------------------------------------+-------------+
19741 | | %o | special variable, apply flags on all next var | |
19742 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010019743 | | %B | bytes_read (from server to client) | numeric |
19744 | H | %CC | captured_request_cookie | string |
19745 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020019746 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000019747 | H | %HM | HTTP method (ex: POST) | string |
Maciej Zdeb21acc332020-11-26 10:45:52 +000019748 | H | %HP | HTTP request URI without query string | string |
Maciej Zdebfcdfd852020-11-30 18:27:47 +000019749 | H | %HPO | HTTP path only (without host nor query string)| string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000019750 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000019751 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
19752 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010019753 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020019754 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019755 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019756 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019757 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020019758 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080019759 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019760 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
19761 | H | %Ti | idle time before the HTTP request | numeric |
19762 | H | %Tq | Th + Ti + TR | numeric |
19763 | H | %TR | time to receive the full request from 1st byte| numeric |
19764 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019765 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019766 | | %Tt | Tt | numeric |
Damien Claisse57c8eb92020-04-28 12:09:19 +000019767 | | %Tu | Tu | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019768 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010019769 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019770 | | %ac | actconn | numeric |
19771 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019772 | | %bc | beconn (backend concurrent connections) | numeric |
19773 | | %bi | backend_source_ip (connecting address) | IP |
19774 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019775 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010019776 | | %ci | client_ip (accepted address) | IP |
19777 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019778 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019779 | | %fc | feconn (frontend concurrent connections) | numeric |
19780 | | %fi | frontend_ip (accepting address) | IP |
19781 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020019782 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020019783 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020019784 | | %hr | captured_request_headers default style | string |
19785 | | %hrl | captured_request_headers CLF style | string list |
19786 | | %hs | captured_response_headers default style | string |
19787 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020019788 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020019789 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019790 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019791 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010019792 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019793 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019794 | | %sc | srv_conn (server concurrent connections) | numeric |
19795 | | %si | server_IP (target address) | IP |
19796 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019797 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019798 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
19799 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010019800 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019801 | H | %tr | date_time of HTTP request | date |
19802 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020019803 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019804 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019805 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010019806 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010019807
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020019808 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010019809
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010019810
198118.2.5. Error log format
19812-----------------------
19813
19814When an incoming connection fails due to an SSL handshake or an invalid PROXY
19815protocol header, haproxy will log the event using a shorter, fixed line format.
19816By default, logs are emitted at the LOG_INFO level, unless the option
19817"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010019818will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010019819logged if the "dontlognull" option is set.
19820
19821The format looks like this :
19822
19823 >>> Dec 3 18:27:14 localhost \
19824 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
19825 Connection error during SSL handshake
19826
19827 Field Format Extract from the example above
19828 1 process_name '[' pid ']:' haproxy[6103]:
19829 2 client_ip ':' client_port 127.0.0.1:56059
19830 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
19831 4 frontend_name "/" bind_name ":" frt/f1:
19832 5 message Connection error during SSL handshake
19833
19834These fields just provide minimal information to help debugging connection
19835failures.
19836
19837
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200198388.3. Advanced logging options
19839-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019840
19841Some advanced logging options are often looked for but are not easy to find out
19842just by looking at the various options. Here is an entry point for the few
19843options which can enable better logging. Please refer to the keywords reference
19844for more information about their usage.
19845
19846
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200198478.3.1. Disabling logging of external tests
19848------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019849
19850It is quite common to have some monitoring tools perform health checks on
19851haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
19852commercial load-balancer, and sometimes it will simply be a more complete
19853monitoring system such as Nagios. When the tests are very frequent, users often
19854ask how to disable logging for those checks. There are three possibilities :
19855
19856 - if connections come from everywhere and are just TCP probes, it is often
19857 desired to simply disable logging of connections without data exchange, by
19858 setting "option dontlognull" in the frontend. It also disables logging of
19859 port scans, which may or may not be desired.
19860
Willy Tarreau9e9919d2020-10-14 15:55:23 +020019861 - it is possible to use the "http-request set-log-level silent" action using
19862 a variety of conditions (source networks, paths, user-agents, etc).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019863
19864 - if the tests are performed on a known URI, use "monitor-uri" to declare
19865 this URI as dedicated to monitoring. Any host sending this request will
19866 only get the result of a health-check, and the request will not be logged.
19867
19868
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200198698.3.2. Logging before waiting for the session to terminate
19870----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019871
19872The problem with logging at end of connection is that you have no clue about
19873what is happening during very long sessions, such as remote terminal sessions
19874or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010019875"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019876just before data transfer begins. This means that in case of TCP, it will still
19877log the connection status to the server, and in case of HTTP, it will log just
19878after processing the server headers. In this case, the number of bytes reported
19879is the number of header bytes sent to the client. In order to avoid confusion
19880with normal logs, the total time field and the number of bytes are prefixed
19881with a '+' sign which means that real numbers are certainly larger.
19882
19883
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200198848.3.3. Raising log level upon errors
19885------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019886
19887Sometimes it is more convenient to separate normal traffic from errors logs,
19888for instance in order to ease error monitoring from log files. When the option
19889"log-separate-errors" is used, connections which experience errors, timeouts,
19890retries, redispatches or HTTP status codes 5xx will see their syslog level
19891raised from "info" to "err". This will help a syslog daemon store the log in
19892a separate file. It is very important to keep the errors in the normal traffic
19893file too, so that log ordering is not altered. You should also be careful if
19894you already have configured your syslog daemon to store all logs higher than
19895"notice" in an "admin" file, because the "err" level is higher than "notice".
19896
19897
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200198988.3.4. Disabling logging of successful connections
19899--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020019900
19901Although this may sound strange at first, some large sites have to deal with
19902multiple thousands of logs per second and are experiencing difficulties keeping
19903them intact for a long time or detecting errors within them. If the option
19904"dontlog-normal" is set on the frontend, all normal connections will not be
19905logged. In this regard, a normal connection is defined as one without any
19906error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
19907and a response with a status 5xx is not considered normal and will be logged
19908too. Of course, doing is is really discouraged as it will remove most of the
19909useful information from the logs. Do this only if you have no other
19910alternative.
19911
19912
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200199138.4. Timing events
19914------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019915
19916Timers provide a great help in troubleshooting network problems. All values are
19917reported in milliseconds (ms). These timers should be used in conjunction with
19918the session termination flags. In TCP mode with "option tcplog" set on the
19919frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019920mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
19921addition, three other measures are provided, "Th", "Ti", and "Tq".
19922
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010019923Timings events in HTTP mode:
19924
19925 first request 2nd request
19926 |<-------------------------------->|<-------------- ...
19927 t tr t tr ...
19928 ---|----|----|----|----|----|----|----|----|--
19929 : Th Ti TR Tw Tc Tr Td : Ti ...
19930 :<---- Tq ---->: :
19931 :<-------------- Tt -------------->:
Damien Claisse57c8eb92020-04-28 12:09:19 +000019932 :<-- -----Tu--------------->:
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010019933 :<--------- Ta --------->:
19934
19935Timings events in TCP mode:
19936
19937 TCP session
19938 |<----------------->|
19939 t t
19940 ---|----|----|----|----|---
19941 | Th Tw Tc Td |
19942 |<------ Tt ------->|
19943
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019944 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010019945 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019946 only happen once during the whole connection's lifetime. A large time here
19947 may indicate that the client only pre-established the connection without
19948 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010019949 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020019950 SSL handshake was very expensive to compute. Please note that this time is
19951 reported only before the first request, so it is safe to average it over
19952 all request to calculate the amortized value. The second and subsequent
19953 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019954
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019955 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
19956 counts between the end of the handshakes and the first byte of the HTTP
19957 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020019958 to count after the end of the transmission the previous response. When a
19959 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
19960 after the previous request. Some browsers pre-establish connections to a
19961 server in order to reduce the latency of a future request, and keep them
19962 pending until they need it. This delay will be reported as the idle time. A
19963 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020019964
19965 - TR: total time to get the client request (HTTP mode only). It's the time
19966 elapsed between the first bytes received and the moment the proxy received
19967 the empty line marking the end of the HTTP headers. The value "-1"
19968 indicates that the end of headers has never been seen. This happens when
19969 the client closes prematurely or times out. This time is usually very short
19970 since most requests fit in a single packet. A large time may indicate a
19971 request typed by hand during a test.
19972
19973 - Tq: total time to get the client request from the accept date or since the
19974 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010019975 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 +020019976 returns -1 as well. This timer used to be very useful before the arrival of
19977 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
19978 it in favor of TR nowadays, as the idle time adds a lot of noise to the
19979 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010019980
19981 - Tw: total time spent in the queues waiting for a connection slot. It
19982 accounts for backend queue as well as the server queues, and depends on the
19983 queue size, and the time needed for the server to complete previous
19984 requests. The value "-1" means that the request was killed before reaching
19985 the queue, which is generally what happens with invalid or denied requests.
19986
19987 - Tc: total time to establish the TCP connection to the server. It's the time
19988 elapsed between the moment the proxy sent the connection request, and the
19989 moment it was acknowledged by the server, or between the TCP SYN packet and
19990 the matching SYN/ACK packet in return. The value "-1" means that the
19991 connection never established.
19992
19993 - Tr: server response time (HTTP mode only). It's the time elapsed between
19994 the moment the TCP connection was established to the server and the moment
19995 the server sent its complete response headers. It purely shows its request
19996 processing time, without the network overhead due to the data transmission.
19997 It is worth noting that when the client has data to send to the server, for
19998 instance during a POST request, the time already runs, and this can distort
19999 apparent response time. For this reason, it's generally wise not to trust
20000 too much this field for POST requests initiated from clients behind an
20001 untrusted network. A value of "-1" here means that the last the response
20002 header (empty line) was never seen, most likely because the server timeout
20003 stroke before the server managed to process the request.
20004
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020005 - Ta: total active time for the HTTP request, between the moment the proxy
20006 received the first byte of the request header and the emission of the last
20007 byte of the response body. The exception is when the "logasap" option is
20008 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
20009 a '+' sign. From this field, we can deduce "Td", the data transmission time,
20010 by subtracting other timers when valid :
20011
20012 Td = Ta - (TR + Tw + Tc + Tr)
20013
20014 Timers with "-1" values have to be excluded from this equation. Note that
20015 "Ta" can never be negative.
20016
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020017 - Tt: total session duration time, between the moment the proxy accepted it
20018 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020019 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
20020 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020021 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020022
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020023 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020024
20025 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020026 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
20027 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020028
Damien Claisse57c8eb92020-04-28 12:09:19 +000020029 - Tu: total estimated time as seen from client, between the moment the proxy
20030 accepted it and the moment both ends were closed, without idle time.
20031 This is useful to roughly measure end-to-end time as a user would see it,
20032 without idle time pollution from keep-alive time between requests. This
20033 timer in only an estimation of time seen by user as it assumes network
20034 latency is the same in both directions. The exception is when the "logasap"
20035 option is specified. In this case, it only equals (Th+TR+Tw+Tc+Tr), and is
20036 prefixed with a '+' sign.
20037
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020038These timers provide precious indications on trouble causes. Since the TCP
20039protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
20040that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020041due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
20042"Tt" is close to a timeout value specified in the configuration, it often means
20043that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020044
20045Most common cases :
20046
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020047 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
20048 the client and the proxy. This is very rare on local networks but might
20049 happen when clients are on far remote networks and send large requests. It
20050 may happen that values larger than usual appear here without any network
20051 cause. Sometimes, during an attack or just after a resource starvation has
20052 ended, haproxy may accept thousands of connections in a few milliseconds.
20053 The time spent accepting these connections will inevitably slightly delay
20054 processing of other connections, and it can happen that request times in the
20055 order of a few tens of milliseconds are measured after a few thousands of
20056 new connections have been accepted at once. Using one of the keep-alive
20057 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020020058 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020059
20060 - If "Tc" is close to 3000, a packet has probably been lost between the
20061 server and the proxy during the server connection phase. This value should
20062 always be very low, such as 1 ms on local networks and less than a few tens
20063 of ms on remote networks.
20064
Willy Tarreau55165fe2009-05-10 12:02:55 +020020065 - If "Tr" is nearly always lower than 3000 except some rare values which seem
20066 to be the average majored by 3000, there are probably some packets lost
20067 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020068
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020069 - If "Ta" is large even for small byte counts, it generally is because
20070 neither the client nor the server decides to close the connection while
20071 haproxy is running in tunnel mode and both have agreed on a keep-alive
20072 connection mode. In order to solve this issue, it will be needed to specify
20073 one of the HTTP options to manipulate keep-alive or close options on either
20074 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
20075 important when connection regulation is used with the "maxconn" option on
20076 the servers, since no new connection will be sent to the server until
20077 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020078
20079Other noticeable HTTP log cases ('xx' means any value to be ignored) :
20080
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020081 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020082 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020083 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020084
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020085 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020086 or it aborted too early. Check the session termination flags
20087 then "timeout http-request" and "timeout client" settings.
20088
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020089 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020090 servers were out of order, because the request was invalid
20091 or forbidden by ACL rules. Check the session termination
20092 flags.
20093
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020094 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
20095 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020096 Check the session termination flags, then check the
20097 "timeout connect" setting. Note that the tarpit action might
20098 return similar-looking patterns, with "Tw" equal to the time
20099 the client connection was maintained open.
20100
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020101 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020102 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020020103 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020104 termination flags, then check the "timeout server" setting.
20105
20106
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200201078.5. Session state at disconnection
20108-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020109
20110TCP and HTTP logs provide a session termination indicator in the
20111"termination_state" field, just before the number of active connections. It is
201122-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
20113each of which has a special meaning :
20114
20115 - On the first character, a code reporting the first event which caused the
20116 session to terminate :
20117
20118 C : the TCP session was unexpectedly aborted by the client.
20119
20120 S : the TCP session was unexpectedly aborted by the server, or the
20121 server explicitly refused it.
20122
20123 P : the session was prematurely aborted by the proxy, because of a
20124 connection limit enforcement, because a DENY filter was matched,
20125 because of a security check which detected and blocked a dangerous
20126 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010020127 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020020128
20129 L : the session was locally processed by haproxy and was not passed to
20130 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020131
20132 R : a resource on the proxy has been exhausted (memory, sockets, source
20133 ports, ...). Usually, this appears during the connection phase, and
20134 system logs should contain a copy of the precise error. If this
20135 happens, it must be considered as a very serious anomaly which
20136 should be fixed as soon as possible by any means.
20137
20138 I : an internal error was identified by the proxy during a self-check.
20139 This should NEVER happen, and you are encouraged to report any log
20140 containing this, because this would almost certainly be a bug. It
20141 would be wise to preventively restart the process after such an
20142 event too, in case it would be caused by memory corruption.
20143
Simon Horman752dc4a2011-06-21 14:34:59 +090020144 D : the session was killed by haproxy because the server was detected
20145 as down and was configured to kill all connections when going down.
20146
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070020147 U : the session was killed by haproxy on this backup server because an
20148 active server was detected as up and was configured to kill all
20149 backup connections when going up.
20150
Willy Tarreaua2a64e92011-09-07 23:01:56 +020020151 K : the session was actively killed by an admin operating on haproxy.
20152
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020153 c : the client-side timeout expired while waiting for the client to
20154 send or receive data.
20155
20156 s : the server-side timeout expired while waiting for the server to
20157 send or receive data.
20158
20159 - : normal session completion, both the client and the server closed
20160 with nothing left in the buffers.
20161
20162 - on the second character, the TCP or HTTP session state when it was closed :
20163
Willy Tarreauf7b30a92010-12-06 22:59:17 +010020164 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020165 (HTTP mode only). Nothing was sent to any server.
20166
20167 Q : the proxy was waiting in the QUEUE for a connection slot. This can
20168 only happen when servers have a 'maxconn' parameter set. It can
20169 also happen in the global queue after a redispatch consecutive to
20170 a failed attempt to connect to a dying server. If no redispatch is
20171 reported, then no connection attempt was made to any server.
20172
20173 C : the proxy was waiting for the CONNECTION to establish on the
20174 server. The server might at most have noticed a connection attempt.
20175
20176 H : the proxy was waiting for complete, valid response HEADERS from the
20177 server (HTTP only).
20178
20179 D : the session was in the DATA phase.
20180
20181 L : the proxy was still transmitting LAST data to the client while the
20182 server had already finished. This one is very rare as it can only
20183 happen when the client dies while receiving the last packets.
20184
20185 T : the request was tarpitted. It has been held open with the client
20186 during the whole "timeout tarpit" duration or until the client
20187 closed, both of which will be reported in the "Tw" timer.
20188
20189 - : normal session completion after end of data transfer.
20190
20191 - the third character tells whether the persistence cookie was provided by
20192 the client (only in HTTP mode) :
20193
20194 N : the client provided NO cookie. This is usually the case for new
20195 visitors, so counting the number of occurrences of this flag in the
20196 logs generally indicate a valid trend for the site frequentation.
20197
20198 I : the client provided an INVALID cookie matching no known server.
20199 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020020200 cookies between HTTP/HTTPS sites, persistence conditionally
20201 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020202
20203 D : the client provided a cookie designating a server which was DOWN,
20204 so either "option persist" was used and the client was sent to
20205 this server, or it was not set and the client was redispatched to
20206 another server.
20207
Willy Tarreau996a92c2010-10-13 19:30:47 +020020208 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020209 server.
20210
Willy Tarreau996a92c2010-10-13 19:30:47 +020020211 E : the client provided a valid cookie, but with a last date which was
20212 older than what is allowed by the "maxidle" cookie parameter, so
20213 the cookie is consider EXPIRED and is ignored. The request will be
20214 redispatched just as if there was no cookie.
20215
20216 O : the client provided a valid cookie, but with a first date which was
20217 older than what is allowed by the "maxlife" cookie parameter, so
20218 the cookie is consider too OLD and is ignored. The request will be
20219 redispatched just as if there was no cookie.
20220
Willy Tarreauc89ccb62012-04-05 21:18:22 +020020221 U : a cookie was present but was not used to select the server because
20222 some other server selection mechanism was used instead (typically a
20223 "use-server" rule).
20224
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020225 - : does not apply (no cookie set in configuration).
20226
20227 - the last character reports what operations were performed on the persistence
20228 cookie returned by the server (only in HTTP mode) :
20229
20230 N : NO cookie was provided by the server, and none was inserted either.
20231
20232 I : no cookie was provided by the server, and the proxy INSERTED one.
20233 Note that in "cookie insert" mode, if the server provides a cookie,
20234 it will still be overwritten and reported as "I" here.
20235
Willy Tarreau996a92c2010-10-13 19:30:47 +020020236 U : the proxy UPDATED the last date in the cookie that was presented by
20237 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030020238 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020020239 date indicated in the cookie. If any other change happens, such as
20240 a redispatch, then the cookie will be marked as inserted instead.
20241
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020242 P : a cookie was PROVIDED by the server and transmitted as-is.
20243
20244 R : the cookie provided by the server was REWRITTEN by the proxy, which
20245 happens in "cookie rewrite" or "cookie prefix" modes.
20246
20247 D : the cookie provided by the server was DELETED by the proxy.
20248
20249 - : does not apply (no cookie set in configuration).
20250
Willy Tarreau996a92c2010-10-13 19:30:47 +020020251The combination of the two first flags gives a lot of information about what
20252was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020253helpful to detect server saturation, network troubles, local system resource
20254starvation, attacks, etc...
20255
20256The most common termination flags combinations are indicated below. They are
20257alphabetically sorted, with the lowercase set just after the upper case for
20258easier finding and understanding.
20259
20260 Flags Reason
20261
20262 -- Normal termination.
20263
20264 CC The client aborted before the connection could be established to the
20265 server. This can happen when haproxy tries to connect to a recently
20266 dead (or unchecked) server, and the client aborts while haproxy is
20267 waiting for the server to respond or for "timeout connect" to expire.
20268
20269 CD The client unexpectedly aborted during data transfer. This can be
20270 caused by a browser crash, by an intermediate equipment between the
20271 client and haproxy which decided to actively break the connection,
20272 by network routing issues between the client and haproxy, or by a
20273 keep-alive session between the server and the client terminated first
20274 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010020275
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020276 cD The client did not send nor acknowledge any data for as long as the
20277 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020020278 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020279
20280 CH The client aborted while waiting for the server to start responding.
20281 It might be the server taking too long to respond or the client
20282 clicking the 'Stop' button too fast.
20283
20284 cH The "timeout client" stroke while waiting for client data during a
20285 POST request. This is sometimes caused by too large TCP MSS values
20286 for PPPoE networks which cannot transport full-sized packets. It can
20287 also happen when client timeout is smaller than server timeout and
20288 the server takes too long to respond.
20289
20290 CQ The client aborted while its session was queued, waiting for a server
20291 with enough empty slots to accept it. It might be that either all the
20292 servers were saturated or that the assigned server was taking too
20293 long a time to respond.
20294
20295 CR The client aborted before sending a full HTTP request. Most likely
20296 the request was typed by hand using a telnet client, and aborted
20297 too early. The HTTP status code is likely a 400 here. Sometimes this
20298 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020020299 and the client. "option http-ignore-probes" can be used to ignore
20300 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020301
20302 cR The "timeout http-request" stroke before the client sent a full HTTP
20303 request. This is sometimes caused by too large TCP MSS values on the
20304 client side for PPPoE networks which cannot transport full-sized
20305 packets, or by clients sending requests by hand and not typing fast
20306 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020020307 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020020308 some browsers started to implement a "pre-connect" feature consisting
20309 in speculatively connecting to some recently visited web sites just
20310 in case the user would like to visit them. This results in many
20311 connections being established to web sites, which end up in 408
20312 Request Timeout if the timeout strikes first, or 400 Bad Request when
20313 the browser decides to close them first. These ones pollute the log
20314 and feed the error counters. Some versions of some browsers have even
20315 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010020316 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020020317 http-ignore-probes" in the frontend, resulting in connections with
20318 zero data transfer to be totally ignored. This will definitely hide
20319 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020320
20321 CT The client aborted while its session was tarpitted. It is important to
20322 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020020323 wrong tarpit rules have been written. If a lot of them happen, it
20324 might make sense to lower the "timeout tarpit" value to something
20325 closer to the average reported "Tw" timer, in order not to consume
20326 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020327
Willy Tarreau570f2212013-06-10 16:42:09 +020020328 LR The request was intercepted and locally handled by haproxy. Generally
20329 it means that this was a redirect or a stats request.
20330
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020331 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020332 the TCP connection (the proxy received a TCP RST or an ICMP message
20333 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010020334 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020335 or no ARP response on local network). When this happens in HTTP mode,
20336 the status code is likely a 502 or 503 here.
20337
20338 sC The "timeout connect" stroke before a connection to the server could
20339 complete. When this happens in HTTP mode, the status code is likely a
20340 503 or 504 here.
20341
20342 SD The connection to the server died with an error during the data
20343 transfer. This usually means that haproxy has received an RST from
20344 the server or an ICMP message from an intermediate equipment while
20345 exchanging data with the server. This can be caused by a server crash
20346 or by a network issue on an intermediate equipment.
20347
20348 sD The server did not send nor acknowledge any data for as long as the
20349 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010020350 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020351 load-balancers, ...), as well as keep-alive sessions maintained
20352 between the client and the server expiring first on haproxy.
20353
20354 SH The server aborted before sending its full HTTP response headers, or
20355 it crashed while processing the request. Since a server aborting at
20356 this moment is very rare, it would be wise to inspect its logs to
20357 control whether it crashed and why. The logged request may indicate a
20358 small set of faulty requests, demonstrating bugs in the application.
20359 Sometimes this might also be caused by an IDS killing the connection
20360 between haproxy and the server.
20361
20362 sH The "timeout server" stroke before the server could return its
20363 response headers. This is the most common anomaly, indicating too
20364 long transactions, probably caused by server or database saturation.
20365 The immediate workaround consists in increasing the "timeout server"
20366 setting, but it is important to keep in mind that the user experience
20367 will suffer from these long response times. The only long term
20368 solution is to fix the application.
20369
20370 sQ The session spent too much time in queue and has been expired. See
20371 the "timeout queue" and "timeout connect" settings to find out how to
20372 fix this if it happens too often. If it often happens massively in
20373 short periods, it may indicate general problems on the affected
20374 servers due to I/O or database congestion, or saturation caused by
20375 external attacks.
20376
20377 PC The proxy refused to establish a connection to the server because the
Thayne McCombscdbcca92021-01-07 21:24:41 -070020378 process's socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020020379 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020380 so that it does not happen anymore. This status is very rare and
20381 might happen when the global "ulimit-n" parameter is forced by hand.
20382
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020383 PD The proxy blocked an incorrectly formatted chunked encoded message in
20384 a request or a response, after the server has emitted its headers. In
20385 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010020386 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020020387 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020388
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020389 PH The proxy blocked the server's response, because it was invalid,
20390 incomplete, dangerous (cache control), or matched a security filter.
20391 In any case, an HTTP 502 error is sent to the client. One possible
20392 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010020393 containing unauthorized characters. It is also possible but quite
20394 rare, that the proxy blocked a chunked-encoding request from the
20395 client due to an invalid syntax, before the server responded. In this
20396 case, an HTTP 400 error is sent to the client and reported in the
20397 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020398
20399 PR The proxy blocked the client's HTTP request, either because of an
20400 invalid HTTP syntax, in which case it returned an HTTP 400 error to
20401 the client, or because a deny filter matched, in which case it
20402 returned an HTTP 403 error.
20403
20404 PT The proxy blocked the client's request and has tarpitted its
20405 connection before returning it a 500 server error. Nothing was sent
20406 to the server. The connection was maintained open for as long as
20407 reported by the "Tw" timer field.
20408
20409 RC A local resource has been exhausted (memory, sockets, source ports)
20410 preventing the connection to the server from establishing. The error
20411 logs will tell precisely what was missing. This is very rare and can
20412 only be solved by proper system tuning.
20413
Willy Tarreau996a92c2010-10-13 19:30:47 +020020414The combination of the two last flags gives a lot of information about how
20415persistence was handled by the client, the server and by haproxy. This is very
20416important to troubleshoot disconnections, when users complain they have to
20417re-authenticate. The commonly encountered flags are :
20418
20419 -- Persistence cookie is not enabled.
20420
20421 NN No cookie was provided by the client, none was inserted in the
20422 response. For instance, this can be in insert mode with "postonly"
20423 set on a GET request.
20424
20425 II A cookie designating an invalid server was provided by the client,
20426 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040020427 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020020428 value can be presented by a client when no other server knows it.
20429
20430 NI No cookie was provided by the client, one was inserted in the
20431 response. This typically happens for first requests from every user
20432 in "insert" mode, which makes it an easy way to count real users.
20433
20434 VN A cookie was provided by the client, none was inserted in the
20435 response. This happens for most responses for which the client has
20436 already got a cookie.
20437
20438 VU A cookie was provided by the client, with a last visit date which is
20439 not completely up-to-date, so an updated cookie was provided in
20440 response. This can also happen if there was no date at all, or if
20441 there was a date but the "maxidle" parameter was not set, so that the
20442 cookie can be switched to unlimited time.
20443
20444 EI A cookie was provided by the client, with a last visit date which is
20445 too old for the "maxidle" parameter, so the cookie was ignored and a
20446 new cookie was inserted in the response.
20447
20448 OI A cookie was provided by the client, with a first visit date which is
20449 too old for the "maxlife" parameter, so the cookie was ignored and a
20450 new cookie was inserted in the response.
20451
20452 DI The server designated by the cookie was down, a new server was
20453 selected and a new cookie was emitted in the response.
20454
20455 VI The server designated by the cookie was not marked dead but could not
20456 be reached. A redispatch happened and selected another one, which was
20457 then advertised in the response.
20458
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020459
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200204608.6. Non-printable characters
20461-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020462
20463In order not to cause trouble to log analysis tools or terminals during log
20464consulting, non-printable characters are not sent as-is into log files, but are
20465converted to the two-digits hexadecimal representation of their ASCII code,
20466prefixed by the character '#'. The only characters that can be logged without
20467being escaped are comprised between 32 and 126 (inclusive). Obviously, the
20468escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
20469is the same for the character '"' which becomes "#22", as well as '{', '|' and
20470'}' when logging headers.
20471
20472Note that the space character (' ') is not encoded in headers, which can cause
20473issues for tools relying on space count to locate fields. A typical header
20474containing spaces is "User-Agent".
20475
20476Last, it has been observed that some syslog daemons such as syslog-ng escape
20477the quote ('"') with a backslash ('\'). The reverse operation can safely be
20478performed since no quote may appear anywhere else in the logs.
20479
20480
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200204818.7. Capturing HTTP cookies
20482---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020483
20484Cookie capture simplifies the tracking a complete user session. This can be
20485achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020486section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020487cookie will simultaneously be checked in the request ("Cookie:" header) and in
20488the response ("Set-Cookie:" header). The respective values will be reported in
20489the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020490locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020491not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
20492user switches to a new session for example, because the server will reassign it
20493a new cookie. It is also possible to detect if a server unexpectedly sets a
20494wrong cookie to a client, leading to session crossing.
20495
20496 Examples :
20497 # capture the first cookie whose name starts with "ASPSESSION"
20498 capture cookie ASPSESSION len 32
20499
20500 # capture the first cookie whose name is exactly "vgnvisitor"
20501 capture cookie vgnvisitor= len 32
20502
20503
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200205048.8. Capturing HTTP headers
20505---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020506
20507Header captures are useful to track unique request identifiers set by an upper
20508proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
20509the response, one can search for information about the response length, how the
20510server asked the cache to behave, or an object location during a redirection.
20511
20512Header captures are performed using the "capture request header" and "capture
20513response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020514section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020515
20516It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020517time. Non-existent headers are logged as empty strings, and if one header
20518appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020519are grouped within braces '{' and '}' in the same order as they were declared,
20520and delimited with a vertical bar '|' without any space. Response headers
20521follow the same representation, but are displayed after a space following the
20522request headers block. These blocks are displayed just before the HTTP request
20523in the logs.
20524
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020020525As a special case, it is possible to specify an HTTP header capture in a TCP
20526frontend. The purpose is to enable logging of headers which will be parsed in
20527an HTTP backend if the request is then switched to this HTTP backend.
20528
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020529 Example :
20530 # This instance chains to the outgoing proxy
20531 listen proxy-out
20532 mode http
20533 option httplog
20534 option logasap
20535 log global
20536 server cache1 192.168.1.1:3128
20537
20538 # log the name of the virtual server
20539 capture request header Host len 20
20540
20541 # log the amount of data uploaded during a POST
20542 capture request header Content-Length len 10
20543
20544 # log the beginning of the referrer
20545 capture request header Referer len 20
20546
20547 # server name (useful for outgoing proxies only)
20548 capture response header Server len 20
20549
20550 # logging the content-length is useful with "option logasap"
20551 capture response header Content-Length len 10
20552
Davor Ocelice9ed2812017-12-25 17:49:28 +010020553 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020554 capture response header Cache-Control len 8
20555
20556 # the Via header will report the next proxy's name
20557 capture response header Via len 20
20558
20559 # log the URL location during a redirection
20560 capture response header Location len 20
20561
20562 >>> Aug 9 20:26:09 localhost \
20563 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
20564 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
20565 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
20566 "GET http://fr.adserver.yahoo.com/"
20567
20568 >>> Aug 9 20:30:46 localhost \
20569 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
20570 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
20571 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020572 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020573
20574 >>> Aug 9 20:30:46 localhost \
20575 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
20576 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
20577 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
20578 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020579 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020580
20581
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200205828.9. Examples of logs
20583---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020584
20585These are real-world examples of logs accompanied with an explanation. Some of
20586them have been made up by hand. The syslog part has been removed for better
20587reading. Their sole purpose is to explain how to decipher them.
20588
20589 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
20590 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
20591 "HEAD / HTTP/1.0"
20592
20593 => long request (6.5s) entered by hand through 'telnet'. The server replied
20594 in 147 ms, and the session ended normally ('----')
20595
20596 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
20597 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
20598 0/9 "HEAD / HTTP/1.0"
20599
20600 => Idem, but the request was queued in the global queue behind 9 other
20601 requests, and waited there for 1230 ms.
20602
20603 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
20604 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
20605 "GET /image.iso HTTP/1.0"
20606
20607 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010020608 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020609 14 ms, 243 bytes of headers were sent to the client, and total time from
20610 accept to first data byte is 30 ms.
20611
20612 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
20613 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
20614 "GET /cgi-bin/bug.cgi? HTTP/1.0"
20615
Christopher Faulet87f1f3d2019-07-18 14:51:20 +020020616 => the proxy blocked a server response either because of an "http-response
20617 deny" rule, or because the response was improperly formatted and not
20618 HTTP-compliant, or because it blocked sensitive information which risked
20619 being cached. In this case, the response is replaced with a "502 bad
20620 gateway". The flags ("PH--") tell us that it was haproxy who decided to
20621 return the 502 and not the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020622
20623 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020624 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 +010020625
20626 => the client never completed its request and aborted itself ("C---") after
20627 8.5s, while the proxy was waiting for the request headers ("-R--").
20628 Nothing was sent to any server.
20629
20630 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
20631 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
20632
20633 => The client never completed its request, which was aborted by the
20634 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010020635 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020636 send a 408 return code to the client.
20637
20638 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
20639 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
20640
20641 => This log was produced with "option tcplog". The client timed out after
20642 5 seconds ("c----").
20643
20644 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
20645 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010020646 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020647
20648 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020649 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010020650 (config says 'retries 3'), and no redispatch (otherwise we would have
20651 seen "/+3"). Status code 503 was returned to the client. There were 115
20652 connections on this server, 202 connections on this proxy, and 205 on
20653 the global process. It is possible that the server refused the
20654 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010020655
Willy Tarreau52b2d222011-09-07 23:48:48 +020020656
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200206579. Supported filters
20658--------------------
20659
20660Here are listed officially supported filters with the list of parameters they
20661accept. Depending on compile options, some of these filters might be
20662unavailable. The list of available filters is reported in haproxy -vv.
20663
20664See also : "filter"
20665
206669.1. Trace
20667----------
20668
Christopher Fauletc41d8bd2020-11-17 10:43:26 +010020669filter trace [name <name>] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020670
20671 Arguments:
20672 <name> is an arbitrary name that will be reported in
20673 messages. If no name is provided, "TRACE" is used.
20674
Christopher Faulet96a577a2020-11-17 10:45:05 +010020675 <quiet> inhibits trace messages.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020676
Davor Ocelice9ed2812017-12-25 17:49:28 +010020677 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020678 default, this filter forwards all previously parsed
20679 data. With this parameter, it only forwards a random
20680 amount of the parsed data.
20681
Davor Ocelice9ed2812017-12-25 17:49:28 +010020682 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010020683
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020684This filter can be used as a base to develop new filters. It defines all
20685callbacks and print a message on the standard error stream (stderr) with useful
20686information for all of them. It may be useful to debug the activity of other
20687filters or, quite simply, HAProxy's activity.
20688
20689Using <random-parsing> and/or <random-forwarding> parameters is a good way to
20690tests the behavior of a filter that parses data exchanged between a client and
20691a server by adding some latencies in the processing.
20692
20693
206949.2. HTTP compression
20695---------------------
20696
20697filter compression
20698
20699The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
20700keyword must still be used to enable and configure the HTTP compression. And
Christopher Fauletb30b3102019-09-12 23:03:09 +020020701when no other filter is used, it is enough. When used with the cache or the
20702fcgi-app enabled, it is also enough. In this case, the compression is always
20703done after the response is stored in the cache. But it is mandatory to
20704explicitly use a filter line to enable the HTTP compression when at least one
20705filter other than the cache or the fcgi-app is used for the same
20706listener/frontend/backend. This is important to know the filters evaluation
20707order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020708
Christopher Fauletb30b3102019-09-12 23:03:09 +020020709See also : "compression", section 9.4 about the cache filter and section 9.5
20710 about the fcgi-app filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020020711
20712
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200207139.3. Stream Processing Offload Engine (SPOE)
20714--------------------------------------------
20715
20716filter spoe [engine <name>] config <file>
20717
20718 Arguments :
20719
20720 <name> is the engine name that will be used to find the right scope in
20721 the configuration file. If not provided, all the file will be
20722 parsed.
20723
20724 <file> is the path of the engine configuration file. This file can
20725 contain configuration of several engines. In this case, each
20726 part must be placed in its own scope.
20727
20728The Stream Processing Offload Engine (SPOE) is a filter communicating with
20729external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010020730streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020020731exchanged with them are configured in dedicated files, for the main part. It
20732also requires dedicated backends, defined in HAProxy configuration.
20733
20734SPOE communicates with external components using an in-house binary protocol,
20735the Stream Processing Offload Protocol (SPOP).
20736
Tim Düsterhus4896c442016-11-29 02:15:19 +010020737For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020020738"doc/SPOE.txt".
20739
Christopher Faulet99a17a22018-12-11 09:18:27 +0100207409.4. Cache
20741----------
20742
20743filter cache <name>
20744
20745 Arguments :
20746
20747 <name> is name of the cache section this filter will use.
20748
20749The cache uses a filter to store cacheable responses. The HTTP rules
20750"cache-store" and "cache-use" must be used to define how and when to use a
John Roeslerfb2fce12019-07-10 15:45:51 -050020751cache. By default the corresponding filter is implicitly defined. And when no
Christopher Fauletb30b3102019-09-12 23:03:09 +020020752other filters than fcgi-app or compression are used, it is enough. In such
20753case, the compression filter is always evaluated after the cache filter. But it
20754is mandatory to explicitly use a filter line to use a cache when at least one
20755filter other than the compression or the fcgi-app is used for the same
Christopher Faulet27d93c32018-12-15 22:32:02 +010020756listener/frontend/backend. This is important to know the filters evaluation
20757order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010020758
Christopher Fauletb30b3102019-09-12 23:03:09 +020020759See also : section 9.2 about the compression filter, section 9.5 about the
20760 fcgi-app filter and section 6 about cache.
20761
20762
207639.5. Fcgi-app
20764-------------
20765
Daniel Corbett67a82712020-07-06 23:01:19 -040020766filter fcgi-app <name>
Christopher Fauletb30b3102019-09-12 23:03:09 +020020767
20768 Arguments :
20769
20770 <name> is name of the fcgi-app section this filter will use.
20771
20772The FastCGI application uses a filter to evaluate all custom parameters on the
20773request path, and to process the headers on the response path. the <name> must
20774reference an existing fcgi-app section. The directive "use-fcgi-app" should be
20775used to define the application to use. By default the corresponding filter is
20776implicitly defined. And when no other filters than cache or compression are
20777used, it is enough. But it is mandatory to explicitly use a filter line to a
20778fcgi-app when at least one filter other than the compression or the cache is
20779used for the same backend. This is important to know the filters evaluation
20780order.
20781
20782See also: "use-fcgi-app", section 9.2 about the compression filter, section 9.4
20783 about the cache filter and section 10 about FastCGI application.
20784
20785
Miroslav Zagoracdc32cd92020-12-13 18:32:57 +0100207869.6. OpenTracing
20787----------------
20788
20789The OpenTracing filter adds native support for using distributed tracing in
20790HAProxy. This is enabled by sending an OpenTracing compliant request to one
20791of the supported tracers such as Datadog, Jaeger, Lightstep and Zipkin tracers.
20792Please note: tracers are not listed by any preference, but alphabetically.
20793
20794This feature is only enabled when haproxy was built with USE_OT=1.
20795
20796The OpenTracing filter activation is done explicitly by specifying it in the
20797HAProxy configuration. If this is not done, the OpenTracing filter in no way
20798participates in the work of HAProxy.
20799
20800filter opentracing [id <id>] config <file>
20801
20802 Arguments :
20803
20804 <id> is the OpenTracing filter id that will be used to find the
20805 right scope in the configuration file. If no filter id is
20806 specified, 'ot-filter' is used as default. If scope is not
20807 specified in the configuration file, it applies to all defined
20808 OpenTracing filters.
20809
20810 <file> is the path of the OpenTracing configuration file. The same
20811 file can contain configurations for multiple OpenTracing
20812 filters simultaneously. In that case we do not need to define
20813 scope so the same configuration applies to all filters or each
20814 filter must have its own scope defined.
20815
20816More detailed documentation related to the operation, configuration and use
20817of the filter can be found in the contrib/opentracing directory.
20818
20819
Christopher Fauletb30b3102019-09-12 23:03:09 +02002082010. FastCGI applications
20821-------------------------
20822
20823HAProxy is able to send HTTP requests to Responder FastCGI applications. This
20824feature was added in HAProxy 2.1. To do so, servers must be configured to use
20825the FastCGI protocol (using the keyword "proto fcgi" on the server line) and a
20826FastCGI application must be configured and used by the backend managing these
20827servers (using the keyword "use-fcgi-app" into the proxy section). Several
20828FastCGI applications may be defined, but only one can be used at a time by a
20829backend.
20830
20831HAProxy implements all features of the FastCGI specification for Responder
20832application. Especially it is able to multiplex several requests on a simple
20833connection.
20834
2083510.1. Setup
20836-----------
20837
2083810.1.1. Fcgi-app section
20839--------------------------
20840
20841fcgi-app <name>
20842 Declare a FastCGI application named <name>. To be valid, at least the
20843 document root must be defined.
20844
20845acl <aclname> <criterion> [flags] [operator] <value> ...
20846 Declare or complete an access list.
20847
20848 See "acl" keyword in section 4.2 and section 7 about ACL usage for
20849 details. ACLs defined for a FastCGI application are private. They cannot be
20850 used by any other application or by any proxy. In the same way, ACLs defined
20851 in any other section are not usable by a FastCGI application. However,
20852 Pre-defined ACLs are available.
20853
20854docroot <path>
20855 Define the document root on the remote host. <path> will be used to build
20856 the default value of FastCGI parameters SCRIPT_FILENAME and
20857 PATH_TRANSLATED. It is a mandatory setting.
20858
20859index <script-name>
20860 Define the script name that will be appended after an URI that ends with a
20861 slash ("/") to set the default value of the FastCGI parameter SCRIPT_NAME. It
20862 is an optional setting.
20863
20864 Example :
20865 index index.php
20866
20867log-stderr global
20868log-stderr <address> [len <length>] [format <format>]
Jan Wagner3e678602020-12-17 22:22:32 +010020869 [sample <ranges>:<sample_size>] <facility> [<level> [<minlevel>]]
Christopher Fauletb30b3102019-09-12 23:03:09 +020020870 Enable logging of STDERR messages reported by the FastCGI application.
20871
20872 See "log" keyword in section 4.2 for details. It is an optional setting. By
20873 default STDERR messages are ignored.
20874
20875pass-header <name> [ { if | unless } <condition> ]
20876 Specify the name of a request header which will be passed to the FastCGI
20877 application. It may optionally be followed by an ACL-based condition, in
20878 which case it will only be evaluated if the condition is true.
20879
20880 Most request headers are already available to the FastCGI application,
20881 prefixed with "HTTP_". Thus, this directive is only required to pass headers
20882 that are purposefully omitted. Currently, the headers "Authorization",
20883 "Proxy-Authorization" and hop-by-hop headers are omitted.
20884
20885 Note that the headers "Content-type" and "Content-length" are never passed to
20886 the FastCGI application because they are already converted into parameters.
20887
20888path-info <regex>
Christopher Faulet28cb3662020-02-14 14:47:37 +010020889 Define a regular expression to extract the script-name and the path-info from
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010020890 the URL-decoded path. Thus, <regex> may have two captures: the first one to
20891 capture the script name and the second one to capture the path-info. The
20892 first one is mandatory, the second one is optional. This way, it is possible
20893 to extract the script-name from the path ignoring the path-info. It is an
20894 optional setting. If it is not defined, no matching is performed on the
20895 path. and the FastCGI parameters PATH_INFO and PATH_TRANSLATED are not
20896 filled.
Christopher Faulet28cb3662020-02-14 14:47:37 +010020897
20898 For security reason, when this regular expression is defined, the newline and
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050020899 the null characters are forbidden from the path, once URL-decoded. The reason
Christopher Faulet28cb3662020-02-14 14:47:37 +010020900 to such limitation is because otherwise the matching always fails (due to a
20901 limitation one the way regular expression are executed in HAProxy). So if one
20902 of these two characters is found in the URL-decoded path, an error is
20903 returned to the client. The principle of least astonishment is applied here.
Christopher Fauletb30b3102019-09-12 23:03:09 +020020904
20905 Example :
Christopher Faulet6c57f2d2020-02-14 16:55:52 +010020906 path-info ^(/.+\.php)(/.*)?$ # both script-name and path-info may be set
20907 path-info ^(/.+\.php) # the path-info is ignored
Christopher Fauletb30b3102019-09-12 23:03:09 +020020908
20909option get-values
20910no option get-values
20911 Enable or disable the retrieve of variables about connection management.
20912
Daniel Corbett67a82712020-07-06 23:01:19 -040020913 HAProxy is able to send the record FCGI_GET_VALUES on connection
Christopher Fauletb30b3102019-09-12 23:03:09 +020020914 establishment to retrieve the value for following variables:
20915
20916 * FCGI_MAX_REQS The maximum number of concurrent requests this
20917 application will accept.
20918
William Lallemand93e548e2019-09-30 13:54:02 +020020919 * FCGI_MPXS_CONNS "0" if this application does not multiplex connections,
20920 "1" otherwise.
Christopher Fauletb30b3102019-09-12 23:03:09 +020020921
20922 Some FastCGI applications does not support this feature. Some others close
Ilya Shipitsin11057a32020-06-21 21:18:27 +050020923 the connection immediately after sending their response. So, by default, this
Christopher Fauletb30b3102019-09-12 23:03:09 +020020924 option is disabled.
20925
20926 Note that the maximum number of concurrent requests accepted by a FastCGI
20927 application is a connection variable. It only limits the number of streams
20928 per connection. If the global load must be limited on the application, the
20929 server parameters "maxconn" and "pool-max-conn" must be set. In addition, if
20930 an application does not support connection multiplexing, the maximum number
20931 of concurrent requests is automatically set to 1.
20932
20933option keep-conn
20934no option keep-conn
20935 Instruct the FastCGI application to keep the connection open or not after
20936 sending a response.
20937
20938 If disabled, the FastCGI application closes the connection after responding
20939 to this request. By default, this option is enabled.
20940
20941option max-reqs <reqs>
20942 Define the maximum number of concurrent requests this application will
20943 accept.
20944
20945 This option may be overwritten if the variable FCGI_MAX_REQS is retrieved
20946 during connection establishment. Furthermore, if the application does not
20947 support connection multiplexing, this option will be ignored. By default set
20948 to 1.
20949
20950option mpxs-conns
20951no option mpxs-conns
20952 Enable or disable the support of connection multiplexing.
20953
20954 This option may be overwritten if the variable FCGI_MPXS_CONNS is retrieved
20955 during connection establishment. It is disabled by default.
20956
20957set-param <name> <fmt> [ { if | unless } <condition> ]
20958 Set a FastCGI parameter that should be passed to this application. Its
20959 value, defined by <fmt> must follows the log-format rules (see section 8.2.4
20960 "Custom Log format"). It may optionally be followed by an ACL-based
20961 condition, in which case it will only be evaluated if the condition is true.
20962
20963 With this directive, it is possible to overwrite the value of default FastCGI
20964 parameters. If the value is evaluated to an empty string, the rule is
20965 ignored. These directives are evaluated in their declaration order.
20966
20967 Example :
20968 # PHP only, required if PHP was built with --enable-force-cgi-redirect
20969 set-param REDIRECT_STATUS 200
20970
20971 set-param PHP_AUTH_DIGEST %[req.hdr(Authorization)]
20972
20973
2097410.1.2. Proxy section
20975---------------------
20976
20977use-fcgi-app <name>
20978 Define the FastCGI application to use for the backend.
20979
20980 Arguments :
20981 <name> is the name of the FastCGI application to use.
20982
20983 This keyword is only available for HTTP proxies with the backend capability
20984 and with at least one FastCGI server. However, FastCGI servers can be mixed
20985 with HTTP servers. But except there is a good reason to do so, it is not
20986 recommended (see section 10.3 about the limitations for details). Only one
20987 application may be defined at a time per backend.
20988
20989 Note that, once a FastCGI application is referenced for a backend, depending
20990 on the configuration some processing may be done even if the request is not
20991 sent to a FastCGI server. Rules to set parameters or pass headers to an
20992 application are evaluated.
20993
20994
2099510.1.3. Example
20996---------------
20997
20998 frontend front-http
20999 mode http
21000 bind *:80
21001 bind *:
21002
21003 use_backend back-dynamic if { path_reg ^/.+\.php(/.*)?$ }
21004 default_backend back-static
21005
21006 backend back-static
21007 mode http
21008 server www A.B.C.D:80
21009
21010 backend back-dynamic
21011 mode http
21012 use-fcgi-app php-fpm
21013 server php-fpm A.B.C.D:9000 proto fcgi
21014
21015 fcgi-app php-fpm
21016 log-stderr global
21017 option keep-conn
21018
21019 docroot /var/www/my-app
21020 index index.php
21021 path-info ^(/.+\.php)(/.*)?$
21022
21023
2102410.2. Default parameters
21025------------------------
21026
21027A Responder FastCGI application has the same purpose as a CGI/1.1 program. In
21028the CGI/1.1 specification (RFC3875), several variables must be passed to the
Ilya Shipitsin8525fd92020-02-29 12:34:59 +050021029script. So HAProxy set them and some others commonly used by FastCGI
Christopher Fauletb30b3102019-09-12 23:03:09 +020021030applications. All these variables may be overwritten, with caution though.
21031
21032 +-------------------+-----------------------------------------------------+
21033 | AUTH_TYPE | Identifies the mechanism, if any, used by HAProxy |
21034 | | to authenticate the user. Concretely, only the |
21035 | | BASIC authentication mechanism is supported. |
21036 | | |
21037 +-------------------+-----------------------------------------------------+
21038 | CONTENT_LENGTH | Contains the size of the message-body attached to |
21039 | | the request. It means only requests with a known |
21040 | | size are considered as valid and sent to the |
21041 | | application. |
21042 | | |
21043 +-------------------+-----------------------------------------------------+
21044 | CONTENT_TYPE | Contains the type of the message-body attached to |
21045 | | the request. It may not be set. |
21046 | | |
21047 +-------------------+-----------------------------------------------------+
21048 | DOCUMENT_ROOT | Contains the document root on the remote host under |
21049 | | which the script should be executed, as defined in |
21050 | | the application's configuration. |
21051 | | |
21052 +-------------------+-----------------------------------------------------+
21053 | GATEWAY_INTERFACE | Contains the dialect of CGI being used by HAProxy |
21054 | | to communicate with the FastCGI application. |
21055 | | Concretely, it is set to "CGI/1.1". |
21056 | | |
21057 +-------------------+-----------------------------------------------------+
21058 | PATH_INFO | Contains the portion of the URI path hierarchy |
21059 | | following the part that identifies the script |
21060 | | itself. To be set, the directive "path-info" must |
21061 | | be defined. |
21062 | | |
21063 +-------------------+-----------------------------------------------------+
21064 | PATH_TRANSLATED | If PATH_INFO is set, it is its translated version. |
21065 | | It is the concatenation of DOCUMENT_ROOT and |
21066 | | PATH_INFO. If PATH_INFO is not set, this parameters |
21067 | | is not set too. |
21068 | | |
21069 +-------------------+-----------------------------------------------------+
21070 | QUERY_STRING | Contains the request's query string. It may not be |
21071 | | set. |
21072 | | |
21073 +-------------------+-----------------------------------------------------+
21074 | REMOTE_ADDR | Contains the network address of the client sending |
21075 | | the request. |
21076 | | |
21077 +-------------------+-----------------------------------------------------+
21078 | REMOTE_USER | Contains the user identification string supplied by |
21079 | | client as part of user authentication. |
21080 | | |
21081 +-------------------+-----------------------------------------------------+
21082 | REQUEST_METHOD | Contains the method which should be used by the |
21083 | | script to process the request. |
21084 | | |
21085 +-------------------+-----------------------------------------------------+
21086 | REQUEST_URI | Contains the request's URI. |
21087 | | |
21088 +-------------------+-----------------------------------------------------+
21089 | SCRIPT_FILENAME | Contains the absolute pathname of the script. it is |
21090 | | the concatenation of DOCUMENT_ROOT and SCRIPT_NAME. |
21091 | | |
21092 +-------------------+-----------------------------------------------------+
21093 | SCRIPT_NAME | Contains the name of the script. If the directive |
21094 | | "path-info" is defined, it is the first part of the |
21095 | | URI path hierarchy, ending with the script name. |
21096 | | Otherwise, it is the entire URI path. |
21097 | | |
21098 +-------------------+-----------------------------------------------------+
21099 | SERVER_NAME | Contains the name of the server host to which the |
21100 | | client request is directed. It is the value of the |
21101 | | header "Host", if defined. Otherwise, the |
21102 | | destination address of the connection on the client |
21103 | | side. |
21104 | | |
21105 +-------------------+-----------------------------------------------------+
21106 | SERVER_PORT | Contains the destination TCP port of the connection |
21107 | | on the client side, which is the port the client |
21108 | | connected to. |
21109 | | |
21110 +-------------------+-----------------------------------------------------+
21111 | SERVER_PROTOCOL | Contains the request's protocol. |
21112 | | |
21113 +-------------------+-----------------------------------------------------+
21114 | HTTPS | Set to a non-empty value ("on") if the script was |
21115 | | queried through the HTTPS protocol. |
21116 | | |
21117 +-------------------+-----------------------------------------------------+
21118
21119
2112010.3. Limitations
21121------------------
21122
21123The current implementation have some limitations. The first one is about the
21124way some request headers are hidden to the FastCGI applications. This happens
21125during the headers analysis, on the backend side, before the connection
21126establishment. At this stage, HAProxy know the backend is using a FastCGI
21127application but it don't know if the request will be routed to a FastCGI server
21128or not. But to hide request headers, it simply removes them from the HTX
21129message. So, if the request is finally routed to an HTTP server, it never see
21130these headers. For this reason, it is not recommended to mix FastCGI servers
21131and HTTP servers under the same backend.
21132
21133Similarly, the rules "set-param" and "pass-header" are evaluated during the
21134request headers analysis. So the evaluation is always performed, even if the
21135requests is finally forwarded to an HTTP server.
21136
21137About the rules "set-param", when a rule is applied, a pseudo header is added
21138into the HTX message. So, the same way than for HTTP header rewrites, it may
21139fail if the buffer is full. The rules "set-param" will compete with
21140"http-request" ones.
21141
21142Finally, all FastCGI params and HTTP headers are sent into a unique record
21143FCGI_PARAM. Encoding of this record must be done in one pass, otherwise a
21144processing error is returned. It means the record FCGI_PARAM, once encoded,
21145must not exceeds the size of a buffer. However, there is no reserve to respect
21146here.
William Lallemand86d0df02017-11-24 21:36:45 +010021147
Willy Tarreau0ba27502007-12-24 16:55:16 +010021148/*
21149 * Local variables:
21150 * fill-column: 79
21151 * End:
21152 */