blob: 5c9ba41267ac4495a8f8b0a558917bd15ed5ce60 [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 Tarreaufba74ea2018-12-22 11:19:45 +01005 version 2.0
Willy Tarreau6a06a402007-07-15 20:15:28 +02006 willy tarreau
William Lallemand4f392792020-06-12 17:31:06 +02007 2020/06/12
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
Willy Tarreauc57f0e22009-05-10 13:12:33 +020055
564. Proxies
574.1. Proxy keywords matrix
584.2. Alphabetically sorted keywords reference
59
Davor Ocelice9ed2812017-12-25 17:49:28 +0100605. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200615.1. Bind options
625.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200635.3. Server DNS resolution
645.3.1. Global overview
655.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020066
676. HTTP header manipulation
68
Willy Tarreau74ca5042013-06-11 23:12:07 +0200697. Using ACLs and fetching samples
707.1. ACL basics
717.1.1. Matching booleans
727.1.2. Matching integers
737.1.3. Matching strings
747.1.4. Matching regular expressions (regexes)
757.1.5. Matching arbitrary data blocks
767.1.6. Matching IPv4 and IPv6 addresses
777.2. Using ACLs to form conditions
787.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200797.3.1. Converters
807.3.2. Fetching samples from internal states
817.3.3. Fetching samples at Layer 4
827.3.4. Fetching samples at Layer 5
837.3.5. Fetching samples from buffer contents (Layer 6)
847.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +0200857.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020086
878. Logging
888.1. Log levels
898.2. Log formats
908.2.1. Default log format
918.2.2. TCP log format
928.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +0100938.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +0100948.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200958.3. Advanced logging options
968.3.1. Disabling logging of external tests
978.3.2. Logging before waiting for the session to terminate
988.3.3. Raising log level upon errors
998.3.4. Disabling logging of successful connections
1008.4. Timing events
1018.5. Session state at disconnection
1028.6. Non-printable characters
1038.7. Capturing HTTP cookies
1048.8. Capturing HTTP headers
1058.9. Examples of logs
106
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001079. Supported filters
1089.1. Trace
1099.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001109.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001119.4. Cache
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200112
William Lallemand86d0df02017-11-24 21:36:45 +010011310. Cache
Cyril Bonté7b888f12017-11-26 22:24:31 +010011410.1. Limitation
11510.2. Setup
11610.2.1. Cache section
11710.2.2. Proxy section
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200118
1191. Quick reminder about HTTP
120----------------------------
121
Davor Ocelice9ed2812017-12-25 17:49:28 +0100122When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200123fully analyzed and indexed, thus it becomes possible to build matching criteria
124on almost anything found in the contents.
125
126However, it is important to understand how HTTP requests and responses are
127formed, and how HAProxy decomposes them. It will then become easier to write
128correct rules and to debug existing configurations.
129
130
1311.1. The HTTP transaction model
132-------------------------------
133
134The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100135to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100136from the client to the server, a request is sent by the client through the
137connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200138will involve a new connection :
139
140 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
141
142In this mode, called the "HTTP close" mode, there are as many connection
143establishments as there are HTTP transactions. Since the connection is closed
144by the server after the response, the client does not need to know the content
145length.
146
147Due to the transactional nature of the protocol, it was possible to improve it
148to avoid closing a connection between two subsequent transactions. In this mode
149however, it is mandatory that the server indicates the content length for each
150response so that the client does not wait indefinitely. For this, a special
151header is used: "Content-length". This mode is called the "keep-alive" mode :
152
153 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
154
155Its advantages are a reduced latency between transactions, and less processing
156power required on the server side. It is generally better than the close mode,
157but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200158a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200159
Willy Tarreau95c4e142017-11-26 12:18:55 +0100160Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200161keep-alive, but the client does not wait for the first response to send the
162second request. This is useful for fetching large number of images composing a
163page :
164
165 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
166
167This can obviously have a tremendous benefit on performance because the network
168latency is eliminated between subsequent requests. Many HTTP agents do not
169correctly support pipelining since there is no way to associate a response with
170the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100171server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172
Willy Tarreau95c4e142017-11-26 12:18:55 +0100173The next improvement is the multiplexed mode, as implemented in HTTP/2. This
174time, each transaction is assigned a single stream identifier, and all streams
175are multiplexed over an existing connection. Many requests can be sent in
176parallel by the client, and responses can arrive in any order since they also
177carry the stream identifier.
178
Willy Tarreau70dffda2014-01-30 03:07:23 +0100179By default HAProxy operates in keep-alive mode with regards to persistent
180connections: for each connection it processes each request and response, and
181leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100182start of a new request. When it receives HTTP/2 connections from a client, it
183processes all the requests in parallel and leaves the connection idling,
184waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200185
Christopher Faulet315b39c2018-09-21 16:26:19 +0200186HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100187 - keep alive : all requests and responses are processed (default)
188 - tunnel : only the first request and response are processed,
Christopher Faulet6c9bbb22019-03-26 21:37:23 +0100189 everything else is forwarded with no analysis (deprecated).
Willy Tarreau70dffda2014-01-30 03:07:23 +0100190 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200191 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100192
Davor Ocelice9ed2812017-12-25 17:49:28 +0100193For HTTP/2, the connection mode resembles more the "server close" mode : given
194the independence of all streams, there is currently no place to hook the idle
Willy Tarreau95c4e142017-11-26 12:18:55 +0100195server connection after a response, so it is closed after the response. HTTP/2
196is only supported for incoming connections, not on connections going to
197servers.
198
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200199
2001.2. HTTP request
201-----------------
202
203First, let's consider this HTTP request :
204
205 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100206 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200207 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
208 2 Host: www.mydomain.com
209 3 User-agent: my small browser
210 4 Accept: image/jpeg, image/gif
211 5 Accept: image/png
212
213
2141.2.1. The Request line
215-----------------------
216
217Line 1 is the "request line". It is always composed of 3 fields :
218
219 - a METHOD : GET
220 - a URI : /serv/login.php?lang=en&profile=2
221 - a version tag : HTTP/1.1
222
223All of them are delimited by what the standard calls LWS (linear white spaces),
224which are commonly spaces, but can also be tabs or line feeds/carriage returns
225followed by spaces/tabs. The method itself cannot contain any colon (':') and
226is limited to alphabetic letters. All those various combinations make it
227desirable that HAProxy performs the splitting itself rather than leaving it to
228the user to write a complex or inaccurate regular expression.
229
230The URI itself can have several forms :
231
232 - A "relative URI" :
233
234 /serv/login.php?lang=en&profile=2
235
236 It is a complete URL without the host part. This is generally what is
237 received by servers, reverse proxies and transparent proxies.
238
239 - An "absolute URI", also called a "URL" :
240
241 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
242
243 It is composed of a "scheme" (the protocol name followed by '://'), a host
244 name or address, optionally a colon (':') followed by a port number, then
245 a relative URI beginning at the first slash ('/') after the address part.
246 This is generally what proxies receive, but a server supporting HTTP/1.1
247 must accept this form too.
248
249 - a star ('*') : this form is only accepted in association with the OPTIONS
250 method and is not relayable. It is used to inquiry a next hop's
251 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100252
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200253 - an address:port combination : 192.168.0.12:80
254 This is used with the CONNECT method, which is used to establish TCP
255 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
256 other protocols too.
257
258In a relative URI, two sub-parts are identified. The part before the question
259mark is called the "path". It is typically the relative path to static objects
260on the server. The part after the question mark is called the "query string".
261It is mostly used with GET requests sent to dynamic scripts and is very
262specific to the language, framework or application in use.
263
Willy Tarreau95c4e142017-11-26 12:18:55 +0100264HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100265assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100266However, haproxy natively processes HTTP/1.x requests and headers, so requests
267received over an HTTP/2 connection are transcoded to HTTP/1.1 before being
268processed. This explains why they still appear as "HTTP/1.1" in haproxy's logs
269as well as in server logs.
270
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200271
2721.2.2. The request headers
273--------------------------
274
275The headers start at the second line. They are composed of a name at the
276beginning of the line, immediately followed by a colon (':'). Traditionally,
277an LWS is added after the colon but that's not required. Then come the values.
278Multiple identical headers may be folded into one single line, delimiting the
279values with commas, provided that their order is respected. This is commonly
280encountered in the "Cookie:" field. A header may span over multiple lines if
281the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
282define a total of 3 values for the "Accept:" header.
283
Davor Ocelice9ed2812017-12-25 17:49:28 +0100284Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200285their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100286"Connection:" header). In HTTP/2, header names are always sent in lower case,
287as can be seen when running in debug mode.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200288
289The end of the headers is indicated by the first empty line. People often say
290that it's a double line feed, which is not exact, even if a double line feed
291is one valid form of empty line.
292
293Fortunately, HAProxy takes care of all these complex combinations when indexing
294headers, checking values and counting them, so there is no reason to worry
295about the way they could be written, but it is important not to accuse an
296application of being buggy if it does unusual, valid things.
297
298Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000299 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200300 in the middle of headers by LWS in order to join multi-line headers. This
301 is necessary for proper analysis and helps less capable HTTP parsers to work
302 correctly and not to be fooled by such complex constructs.
303
304
3051.3. HTTP response
306------------------
307
308An HTTP response looks very much like an HTTP request. Both are called HTTP
309messages. Let's consider this HTTP response :
310
311 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100312 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200313 1 HTTP/1.1 200 OK
314 2 Content-length: 350
315 3 Content-Type: text/html
316
Willy Tarreau816b9792009-09-15 21:25:21 +0200317As a special case, HTTP supports so called "Informational responses" as status
318codes 1xx. These messages are special in that they don't convey any part of the
319response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100320continue to post its request for instance. In the case of a status 100 response
321the requested information will be carried by the next non-100 response message
322following the informational one. This implies that multiple responses may be
323sent to a single request, and that this only works when keep-alive is enabled
324(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
325correctly forward and skip them, and only process the next non-100 response. As
326such, these messages are neither logged nor transformed, unless explicitly
327state otherwise. Status 101 messages indicate that the protocol is changing
328over the same connection and that haproxy must switch to tunnel mode, just as
329if a CONNECT had occurred. Then the Upgrade header would contain additional
330information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200331
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200332
Davor Ocelice9ed2812017-12-25 17:49:28 +01003331.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200334------------------------
335
336Line 1 is the "response line". It is always composed of 3 fields :
337
338 - a version tag : HTTP/1.1
339 - a status code : 200
340 - a reason : OK
341
342The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100343 - 1xx = informational message to be skipped (e.g. 100, 101)
344 - 2xx = OK, content is following (e.g. 200, 206)
345 - 3xx = OK, no content following (e.g. 302, 304)
346 - 4xx = error caused by the client (e.g. 401, 403, 404)
347 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200348
Lukas Tribus23953682017-04-28 13:24:30 +0000349Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100350"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200351found there, but it's a common practice to respect the well-established
352messages. It can be composed of one or multiple words, such as "OK", "Found",
353or "Authentication Required".
354
Davor Ocelice9ed2812017-12-25 17:49:28 +0100355HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200356
357 Code When / reason
358 200 access to stats page, and when replying to monitoring requests
359 301 when performing a redirection, depending on the configured code
360 302 when performing a redirection, depending on the configured code
361 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100362 307 when performing a redirection, depending on the configured code
363 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200364 400 for an invalid or too large request
365 401 when an authentication is required to perform the action (when
366 accessing the stats page)
367 403 when a request is forbidden by a "block" ACL or "reqdeny" filter
Florian Tham9f3bda02020-01-08 13:35:30 +0100368 404 when the requested resource could not be found
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200369 408 when the request timeout strikes before the request is complete
Florian Thamc09f7972020-01-08 10:19:05 +0100370 410 when the requested resource is no longer available and will not
371 be available again
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200372 500 when haproxy encounters an unrecoverable internal error, such as a
373 memory allocation failure, which should never happen
374 502 when the server returns an empty, invalid or incomplete response, or
375 when an "rspdeny" filter blocks the response.
376 503 when no server was available to handle the request, or in response to
377 monitoring requests which match the "monitor fail" condition
378 504 when the response timeout strikes before the server responds
379
380The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3814.2).
382
383
3841.3.2. The response headers
385---------------------------
386
387Response headers work exactly like request headers, and as such, HAProxy uses
388the same parsing function for both. Please refer to paragraph 1.2.2 for more
389details.
390
391
3922. Configuring HAProxy
393----------------------
394
3952.1. Configuration file format
396------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200397
398HAProxy's configuration process involves 3 major sources of parameters :
399
400 - the arguments from the command-line, which always take precedence
401 - the "global" section, which sets process-wide parameters
402 - the proxies sections which can take form of "defaults", "listen",
403 "frontend" and "backend".
404
Willy Tarreau0ba27502007-12-24 16:55:16 +0100405The configuration file syntax consists in lines beginning with a keyword
406referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200407delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100408
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200409
William Lallemandf9873ba2015-05-05 17:37:14 +02004102.2. Quoting and escaping
411-------------------------
412
413HAProxy's configuration introduces a quoting and escaping system similar to
414many programming languages. The configuration file supports 3 types: escaping
415with a backslash, weak quoting with double quotes, and strong quoting with
416single quotes.
417
418If spaces have to be entered in strings, then they must be escaped by preceding
419them by a backslash ('\') or by quoting them. Backslashes also have to be
420escaped by doubling or strong quoting them.
421
422Escaping is achieved by preceding a special character by a backslash ('\'):
423
424 \ to mark a space and differentiate it from a delimiter
425 \# to mark a hash and differentiate it from a comment
426 \\ to use a backslash
427 \' to use a single quote and differentiate it from strong quoting
428 \" to use a double quote and differentiate it from weak quoting
429
430Weak quoting is achieved by using double quotes (""). Weak quoting prevents
431the interpretation of:
432
433 space as a parameter separator
434 ' single quote as a strong quoting delimiter
435 # hash as a comment start
436
William Lallemandb2f07452015-05-12 14:27:13 +0200437Weak quoting permits the interpretation of variables, if you want to use a non
438-interpreted dollar within a double quoted string, you should escape it with a
439backslash ("\$"), it does not work outside weak quoting.
440
441Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200442quoting.
443
444Strong quoting is achieved by using single quotes (''). Inside single quotes,
445nothing is interpreted, it's the efficient way to quote regexes.
446
447Quoted and escaped strings are replaced in memory by their interpreted
448equivalent, it allows you to perform concatenation.
449
450 Example:
451 # those are equivalents:
452 log-format %{+Q}o\ %t\ %s\ %{-Q}r
453 log-format "%{+Q}o %t %s %{-Q}r"
454 log-format '%{+Q}o %t %s %{-Q}r'
455 log-format "%{+Q}o %t"' %s %{-Q}r'
456 log-format "%{+Q}o %t"' %s'\ %{-Q}r
457
458 # those are equivalents:
459 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
460 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
461 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
462 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
463
464
William Lallemandb2f07452015-05-12 14:27:13 +02004652.3. Environment variables
466--------------------------
467
468HAProxy's configuration supports environment variables. Those variables are
469interpreted only within double quotes. Variables are expanded during the
470configuration parsing. Variable names must be preceded by a dollar ("$") and
471optionally enclosed with braces ("{}") similarly to what is done in Bourne
472shell. Variable names can contain alphanumerical characters or the character
473underscore ("_") but should not start with a digit.
474
475 Example:
476
477 bind "fd@${FD_APP1}"
478
479 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
480
481 user "$HAPROXY_USER"
482
William Lallemand4d03e432019-06-14 15:35:37 +0200483Some variables are defined by HAProxy, they can be used in the configuration
484file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200485
William Lallemand4d03e432019-06-14 15:35:37 +0200486* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
487 name of the local peer. (See "-L" in the management guide.)
488
489* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
490 separated by semicolons. Can be useful in the case you specified a
491 directory.
492
493* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
494
John Roesler27c754c2019-07-10 15:45:51 -0500495* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200496 processes, separated by semicolons.
497
John Roesler27c754c2019-07-10 15:45:51 -0500498* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200499 CLI, separated by semicolons.
500
501See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200502
5032.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200504----------------
505
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100506Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100507values are generally expressed in milliseconds (unless explicitly stated
508otherwise) but may be expressed in any other unit by suffixing the unit to the
509numeric value. It is important to consider this because it will not be repeated
510for every keyword. Supported units are :
511
512 - us : microseconds. 1 microsecond = 1/1000000 second
513 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
514 - s : seconds. 1s = 1000ms
515 - m : minutes. 1m = 60s = 60000ms
516 - h : hours. 1h = 60m = 3600s = 3600000ms
517 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
518
519
Lukas Tribusaa83a312017-03-21 09:25:09 +00005202.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200521-------------
522
523 # Simple configuration for an HTTP proxy listening on port 80 on all
524 # interfaces and forwarding requests to a single backend "servers" with a
525 # single server "server1" listening on 127.0.0.1:8000
526 global
527 daemon
528 maxconn 256
529
530 defaults
531 mode http
532 timeout connect 5000ms
533 timeout client 50000ms
534 timeout server 50000ms
535
536 frontend http-in
537 bind *:80
538 default_backend servers
539
540 backend servers
541 server server1 127.0.0.1:8000 maxconn 32
542
543
544 # The same configuration defined with a single listen block. Shorter but
545 # less expressive, especially in HTTP mode.
546 global
547 daemon
548 maxconn 256
549
550 defaults
551 mode http
552 timeout connect 5000ms
553 timeout client 50000ms
554 timeout server 50000ms
555
556 listen http-in
557 bind *:80
558 server server1 127.0.0.1:8000 maxconn 32
559
560
561Assuming haproxy is in $PATH, test these configurations in a shell with:
562
Willy Tarreauccb289d2010-12-11 20:19:38 +0100563 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200564
565
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005663. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200567--------------------
568
569Parameters in the "global" section are process-wide and often OS-specific. They
570are generally set once for all and do not need being changed once correct. Some
571of them have command-line equivalents.
572
573The following keywords are supported in the "global" section :
574
575 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200576 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200577 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200578 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200579 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200580 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200581 - description
582 - deviceatlas-json-file
583 - deviceatlas-log-level
584 - deviceatlas-separator
585 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900586 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200587 - gid
588 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100589 - hard-stop-after
Christopher Fauleta99ff4d2019-07-22 16:18:24 +0200590 - h1-case-adjust
591 - h1-case-adjust-file
Willy Tarreau6a06a402007-07-15 20:15:28 +0200592 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200593 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100594 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200595 - lua-load
William Lallemand27edc4b2019-05-07 17:49:33 +0200596 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200597 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200598 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200599 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200600 - pidfile
Willy Tarreau1d549722016-02-16 12:41:57 +0100601 - presetenv
602 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200603 - uid
604 - ulimit-n
605 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200606 - set-dumpable
Willy Tarreau1d549722016-02-16 12:41:57 +0100607 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200608 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200609 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200610 - ssl-default-bind-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200611 - ssl-default-bind-options
612 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200613 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200614 - ssl-default-server-options
615 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100616 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100617 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100618 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100619 - 51degrees-data-file
620 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200621 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200622 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200623 - wurfl-data-file
624 - wurfl-information-list
625 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200626 - wurfl-cache-size
Willy Tarreaud72758d2010-01-12 10:42:19 +0100627
Willy Tarreau6a06a402007-07-15 20:15:28 +0200628 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200629 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200630 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200631 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100632 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100633 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100634 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200635 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200636 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200637 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200638 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200639 - noepoll
640 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000641 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200642 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100643 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300644 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000645 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100646 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200647 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200648 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200649 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000650 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000651 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200652 - tune.buffers.limit
653 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200654 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200655 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100656 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200657 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200658 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200659 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100660 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200661 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200662 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100663 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100664 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100665 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100666 - tune.lua.session-timeout
667 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200668 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100669 - tune.maxaccept
670 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200671 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200672 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200673 - tune.pipesize
Willy Tarreaua5e11c02020-07-01 18:27:16 +0200674 - tune.pool-high-fd-ratio
675 - tune.pool-low-fd-ratio
Willy Tarreaue803de22010-01-21 17:43:04 +0100676 - tune.rcvbuf.client
677 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100678 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200679 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100680 - tune.sndbuf.client
681 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100682 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100683 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200684 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100685 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200686 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200687 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100688 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200689 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100690 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200691 - tune.vars.reqres-max-size
692 - tune.vars.sess-max-size
693 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100694 - tune.zlib.memlevel
695 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100696
Willy Tarreau6a06a402007-07-15 20:15:28 +0200697 * Debugging
698 - debug
699 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200700
701
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007023.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200703------------------------------------
704
Emeric Brunc8e8d122012-10-02 18:42:10 +0200705ca-base <dir>
706 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200707 relative path is used with "ca-file" or "crl-file" directives. Absolute
708 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200709
Willy Tarreau6a06a402007-07-15 20:15:28 +0200710chroot <jail dir>
711 Changes current directory to <jail dir> and performs a chroot() there before
712 dropping privileges. This increases the security level in case an unknown
713 vulnerability would be exploited, since it would make it very hard for the
714 attacker to exploit the system. This only works when the process is started
715 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100716 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100717
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100718cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
719 On Linux 2.6 and above, it is possible to bind a process or a thread to a
720 specific CPU set. This means that the process or the thread will never run on
721 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
722 sets. The first argument is a process set, eventually followed by a thread
723 set. These sets have the format
724
725 all | odd | even | number[-[number]]
726
727 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100728 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100729 ignored. It is possible to specify a range with two such number delimited by
730 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100731 "all", only odd numbers using "odd" or even numbers using "even", just like
732 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100733 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 +0100734 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100735 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100736 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100737 specified. Each "cpu-map" directive will replace the previous ones when they
738 overlap. A thread will be bound on the intersection of its mapping and the
739 one of the process on which it is attached. If the intersection is null, no
740 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100741
Christopher Fauletff4121f2017-11-22 16:38:49 +0100742 Ranges can be partially defined. The higher bound can be omitted. In such
743 case, it is replaced by the corresponding maximum value, 32 or 64 depending
744 on the machine's word size.
745
Christopher Faulet26028f62017-11-22 15:01:51 +0100746 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100747 automatically bind a process or a thread to a CPU by incrementing
748 process/thread and CPU sets. To be valid, both sets must have the same
749 size. No matter the declaration order of the CPU sets, it will be bound from
750 the lowest to the highest bound. Having a process and a thread range with the
751 "auto:" prefix is not supported. Only one range is supported, the other one
752 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100753
754 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100755 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
756
757 cpu-map 1/all 0-3 # bind all threads of the first process on the
758 # first 4 CPUs
759
760 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
761 # or "cpu-map 1-32 0-31" depending on the machine's
762 # word size.
763
Christopher Faulet26028f62017-11-22 15:01:51 +0100764 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100765 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100766 cpu-map auto:1-4 0-3
767 cpu-map auto:1-4 0-1 2-3
768 cpu-map auto:1-4 3 2 1 0
769
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100770 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
771 # and so on.
772 cpu-map auto:1/1-4 0-3
773 cpu-map auto:1/1-4 0-1 2-3
774 cpu-map auto:1/1-4 3 2 1 0
775
Davor Ocelice9ed2812017-12-25 17:49:28 +0100776 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100777 cpu-map auto:all 0-63
778 cpu-map auto:even 0-31
779 cpu-map auto:odd 32-63
780
781 # invalid cpu-map because process and CPU sets have different sizes.
782 cpu-map auto:1-4 0 # invalid
783 cpu-map auto:1 0-3 # invalid
784
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100785 # invalid cpu-map because automatic binding is used with a process range
786 # and a thread range.
787 cpu-map auto:all/all 0 # invalid
788 cpu-map auto:all/1-4 0 # invalid
789 cpu-map auto:1-4/all 0 # invalid
790
Emeric Brunc8e8d122012-10-02 18:42:10 +0200791crt-base <dir>
792 Assigns a default directory to fetch SSL certificates from when a relative
793 path is used with "crtfile" directives. Absolute locations specified after
794 "crtfile" prevail and ignore "crt-base".
795
Willy Tarreau6a06a402007-07-15 20:15:28 +0200796daemon
797 Makes the process fork into background. This is the recommended mode of
798 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100799 disabled by the command line "-db" argument. This option is ignored in
800 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200801
David Carlier8167f302015-06-01 13:50:06 +0200802deviceatlas-json-file <path>
803 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100804 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200805
806deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100807 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200808 optional and set to 0 by default if not set.
809
810deviceatlas-separator <char>
811 Sets the character separator for the API properties results. This directive
812 is optional and set to | by default if not set.
813
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100814deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200815 Sets the client cookie's name used for the detection if the DeviceAtlas
816 Client-side component was used during the request. This directive is optional
817 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100818
Simon Horman98637e52014-06-20 12:30:16 +0900819external-check
820 Allows the use of an external agent to perform health checks.
821 This is disabled by default as a security precaution.
822 See "option external-check".
823
Willy Tarreau6a06a402007-07-15 20:15:28 +0200824gid <number>
825 Changes the process' group ID to <number>. It is recommended that the group
826 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
827 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100828 Note that if haproxy is started from a user having supplementary groups, it
829 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200830 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100831
Willy Tarreau8b852462019-12-03 08:29:22 +0100832group <group name>
833 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
834 See also "gid" and "user".
835
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100836hard-stop-after <time>
837 Defines the maximum time allowed to perform a clean soft-stop.
838
839 Arguments :
840 <time> is the maximum time (by default in milliseconds) for which the
841 instance will remain alive when a soft-stop is received via the
842 SIGUSR1 signal.
843
844 This may be used to ensure that the instance will quit even if connections
845 remain opened during a soft-stop (for example with long timeouts for a proxy
846 in tcp mode). It applies both in TCP and HTTP mode.
847
848 Example:
849 global
850 hard-stop-after 30s
851
Christopher Fauleta99ff4d2019-07-22 16:18:24 +0200852h1-case-adjust <from> <to>
853 Defines the case adjustment to apply, when enabled, to the header name
854 <from>, to change it to <to> before sending it to HTTP/1 clients or
855 servers. <from> must be in lower case, and <from> and <to> must not differ
856 except for their case. It may be repeated if several header names need to be
Ilya Shipitsin5c836fd2020-02-29 12:34:59 +0500857 adjusted. Duplicate entries are not allowed. If a lot of header names have to
Christopher Fauleta99ff4d2019-07-22 16:18:24 +0200858 be adjusted, it might be more convenient to use "h1-case-adjust-file".
859 Please note that no transformation will be applied unless "option
860 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
861 specified in a proxy.
862
863 There is no standard case for header names because, as stated in RFC7230,
864 they are case-insensitive. So applications must handle them in a case-
865 insensitive manner. But some bogus applications violate the standards and
866 erroneously rely on the cases most commonly used by browsers. This problem
867 becomes critical with HTTP/2 because all header names must be exchanged in
868 lower case, and HAProxy follows the same convention. All header names are
869 sent in lower case to clients and servers, regardless of the HTTP version.
870
871 Applications which fail to properly process requests or responses may require
872 to temporarily use such workarounds to adjust header names sent to them for
873 the time it takes the application to be fixed. Please note that an
874 application which requires such workarounds might be vulnerable to content
875 smuggling attacks and must absolutely be fixed.
876
877 Example:
878 global
879 h1-case-adjust content-length Content-Length
880
881 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
882 "option h1-case-adjust-bogus-server".
883
884h1-case-adjust-file <hdrs-file>
885 Defines a file containing a list of key/value pairs used to adjust the case
886 of some header names before sending them to HTTP/1 clients or servers. The
887 file <hdrs-file> must contain 2 header names per line. The first one must be
888 in lower case and both must not differ except for their case. Lines which
889 start with '#' are ignored, just like empty lines. Leading and trailing tabs
890 and spaces are stripped. Duplicate entries are not allowed. Please note that
891 no transformation will be applied unless "option h1-case-adjust-bogus-client"
892 or "option h1-case-adjust-bogus-server" is specified in a proxy.
893
894 If this directive is repeated, only the last one will be processed. It is an
895 alternative to the directive "h1-case-adjust" if a lot of header names need
896 to be adjusted. Please read the risks associated with using this.
897
898 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
899 "option h1-case-adjust-bogus-server".
900
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200901log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
902 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100903 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100904 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100905 configured with "log global".
906
907 <address> can be one of:
908
Willy Tarreau2769aa02007-12-27 18:26:09 +0100909 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100910 no port is specified, 514 is used by default (the standard syslog
911 port).
912
David du Colombier24bb5f52011-03-17 10:40:23 +0100913 - An IPv6 address followed by a colon and optionally a UDP port. If
914 no port is specified, 514 is used by default (the standard syslog
915 port).
916
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100917 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100918 considerations for chroot (be sure the path is accessible inside
919 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100920 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100921
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100922 - A file descriptor number in the form "fd@<number>", which may point
923 to a pipe, terminal, or socket. In this case unbuffered logs are used
924 and one writev() call per log is performed. This is a bit expensive
925 but acceptable for most workloads. Messages sent this way will not be
926 truncated but may be dropped, in which case the DroppedLogs counter
927 will be incremented. The writev() call is atomic even on pipes for
928 messages up to PIPE_BUF size, which POSIX recommends to be at least
929 512 and which is 4096 bytes on most modern operating systems. Any
930 larger message may be interleaved with messages from other processes.
931 Exceptionally for debugging purposes the file descriptor may also be
932 directed to a file, but doing so will significantly slow haproxy down
933 as non-blocking calls will be ignored. Also there will be no way to
934 purge nor rotate this file without restarting the process. Note that
935 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +0100936 for use with a TCP syslog server. See also the "short" and "raw"
937 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100938
939 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
940 "fd@2", see above.
941
William Lallemandb2f07452015-05-12 14:27:13 +0200942 You may want to reference some environment variables in the address
943 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100944
Willy Tarreau18324f52014-06-27 18:10:07 +0200945 <length> is an optional maximum line length. Log lines larger than this value
946 will be truncated before being sent. The reason is that syslog
947 servers act differently on log line length. All servers support the
948 default value of 1024, but some servers simply drop larger lines
949 while others do log them. If a server supports long lines, it may
950 make sense to set this value here in order to avoid truncating long
951 lines. Similarly, if a server drops long lines, it is preferable to
952 truncate them before sending them. Accepted values are 80 to 65535
953 inclusive. The default value of 1024 is generally fine for all
954 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100955 JSON-formatted logs may require larger values. You may also need to
956 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200957
Dragan Dosen7ad31542015-09-28 17:16:47 +0200958 <format> is the log format used when generating syslog messages. It may be
959 one of the following :
960
961 rfc3164 The RFC3164 syslog message format. This is the default.
962 (https://tools.ietf.org/html/rfc3164)
963
964 rfc5424 The RFC5424 syslog message format.
965 (https://tools.ietf.org/html/rfc5424)
966
Willy Tarreaue8746a02018-11-12 08:45:00 +0100967 short A message containing only a level between angle brackets such as
968 '<3>', followed by the text. The PID, date, time, process name
969 and system name are omitted. This is designed to be used with a
970 local log server. This format is compatible with what the systemd
971 logger consumes.
972
Willy Tarreauc1b06452018-11-12 11:57:56 +0100973 raw A message containing only the text. The level, PID, date, time,
974 process name and system name are omitted. This is designed to be
975 used in containers or during development, where the severity only
976 depends on the file descriptor used (stdout/stderr).
977
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200978 <ranges> A list of comma-separated ranges to identify the logs to sample.
979 This is used to balance the load of the logs to send to the log
980 server. The limits of the ranges cannot be null. They are numbered
981 from 1. The size or period (in number of logs) of the sample must be
982 set with <sample_size> parameter.
983
984 <sample_size>
985 The size of the sample in number of logs to consider when balancing
986 their logging loads. It is used to balance the load of the logs to
987 send to the syslog server. This size must be greater or equal to the
988 maximum of the high limits of the ranges.
989 (see also <ranges> parameter).
990
Robert Tsai81ae1952007-12-05 10:47:29 +0100991 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200992
Willy Tarreaue8746a02018-11-12 08:45:00 +0100993 kern user mail daemon auth syslog lpr news
994 uucp cron auth2 ftp ntp audit alert cron2
995 local0 local1 local2 local3 local4 local5 local6 local7
996
Willy Tarreauc1b06452018-11-12 11:57:56 +0100997 Note that the facility is ignored for the "short" and "raw"
998 formats, but still required as a positional field. It is
999 recommended to use "daemon" in this case to make it clear that
1000 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001001
1002 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +02001003 all messages are sent. If a maximum level is specified, only messages with a
1004 severity at least as important as this level will be sent. An optional minimum
1005 level can be specified. If it is set, logs emitted with a more severe level
1006 than this one will be capped to this level. This is used to avoid sending
1007 "emerg" messages on all terminals on some default syslog configurations.
1008 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001009
Cyril Bontédc4d9032012-04-08 21:57:39 +02001010 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001011
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001012log-send-hostname [<string>]
1013 Sets the hostname field in the syslog header. If optional "string" parameter
1014 is set the header is set to the string contents, otherwise uses the hostname
1015 of the system. Generally used if one is not relaying logs through an
1016 intermediate syslog server or for simply customizing the hostname printed in
1017 the logs.
1018
Kevinm48936af2010-12-22 16:08:21 +00001019log-tag <string>
1020 Sets the tag field in the syslog header to this string. It defaults to the
1021 program name as launched from the command line, which usually is "haproxy".
1022 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001023 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001024
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001025lua-load <file>
1026 This global directive loads and executes a Lua file. This directive can be
1027 used multiple times.
1028
William Lallemand4cfede82017-11-24 22:02:34 +01001029master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001030 Master-worker mode. It is equivalent to the command line "-W" argument.
1031 This mode will launch a "master" which will monitor the "workers". Using
1032 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001033 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001034 or daemon mode. It is recommended to use this mode with multiprocess and
1035 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001036 By default, if a worker exits with a bad return code, in the case of a
1037 segfault for example, all workers will be killed, and the master will leave.
1038 It is convenient to combine this behavior with Restart=on-failure in a
1039 systemd unit file in order to relaunch the whole process. If you don't want
1040 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001041
William Lallemand4cfede82017-11-24 22:02:34 +01001042 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001043
William Lallemand27edc4b2019-05-07 17:49:33 +02001044mworker-max-reloads <number>
1045 In master-worker mode, this option limits the number of time a worker can
John Roesler27c754c2019-07-10 15:45:51 -05001046 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001047 number of reloads is greater than this number, the worker will receive a
1048 SIGTERM. This option helps to keep under control the number of workers.
1049 See also "show proc" in the Management Guide.
1050
Willy Tarreau6a06a402007-07-15 20:15:28 +02001051nbproc <number>
1052 Creates <number> processes when going daemon. This requires the "daemon"
1053 mode. By default, only one process is created, which is the recommended mode
1054 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001055 process, it may be needed to fork multiple daemons. When set to a value
1056 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreau1f672a82019-01-26 14:20:55 +01001057 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon" and
1058 "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001059
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001060nbthread <number>
1061 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001062 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1063 "nbproc" historically used to be the only way to use multiple processors, it
1064 also involved a number of shortcomings related to the lack of synchronization
1065 between processes (health-checks, peers, stick-tables, stats, ...) which do
1066 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001067 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1068 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1069 when nbproc is not used, the default "nbthread" value is automatically set to
1070 the number of CPUs the process is bound to upon startup. This means that the
1071 thread count can easily be adjusted from the calling process using commands
1072 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1073 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001074
Willy Tarreau6a06a402007-07-15 20:15:28 +02001075pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001076 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +02001077 the "-p" command line argument. The file must be accessible to the user
1078 starting the process. See also "daemon".
1079
Willy Tarreau1d549722016-02-16 12:41:57 +01001080presetenv <name> <value>
1081 Sets environment variable <name> to value <value>. If the variable exists, it
1082 is NOT overwritten. The changes immediately take effect so that the next line
1083 in the configuration file sees the new value. See also "setenv", "resetenv",
1084 and "unsetenv".
1085
1086resetenv [<name> ...]
1087 Removes all environment variables except the ones specified in argument. It
1088 allows to use a clean controlled environment before setting new values with
1089 setenv or unsetenv. Please note that some internal functions may make use of
1090 some environment variables, such as time manipulation functions, but also
1091 OpenSSL or even external checks. This must be used with extreme care and only
1092 after complete validation. The changes immediately take effect so that the
1093 next line in the configuration file sees the new environment. See also
1094 "setenv", "presetenv", and "unsetenv".
1095
Christopher Fauletff4121f2017-11-22 16:38:49 +01001096stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001097 Limits the stats socket to a certain set of processes numbers. By default the
1098 stats socket is bound to all processes, causing a warning to be emitted when
1099 nbproc is greater than 1 because there is no way to select the target process
1100 when connecting. However, by using this setting, it becomes possible to pin
1101 the stats socket to a specific set of processes, typically the first one. The
1102 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001103 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001104 word size (32 or 64). Ranges can be partially defined. The higher bound can
1105 be omitted. In such case, it is replaced by the corresponding maximum
1106 value. A better option consists in using the "process" setting of the "stats
1107 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001108
Baptiste Assmann5626f482015-08-23 10:00:10 +02001109server-state-base <directory>
1110 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001111 file names which do not start with a '/'. See also "server-state-file",
1112 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001113
1114server-state-file <file>
1115 Specifies the path to the file containing state of servers. If the path starts
1116 with a slash ('/'), it is considered absolute, otherwise it is considered
1117 relative to the directory specified using "server-state-base" (if set) or to
1118 the current directory. Before reloading HAProxy, it is possible to save the
1119 servers' current state using the stats command "show servers state". The
1120 output of this command must be written in the file pointed by <file>. When
1121 starting up, before handling traffic, HAProxy will read, load and apply state
1122 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001123 configuration. See also "server-state-base" and "show servers state",
1124 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001125
Willy Tarreau1d549722016-02-16 12:41:57 +01001126setenv <name> <value>
1127 Sets environment variable <name> to value <value>. If the variable exists, it
1128 is overwritten. The changes immediately take effect so that the next line in
1129 the configuration file sees the new value. See also "presetenv", "resetenv",
1130 and "unsetenv".
1131
Willy Tarreau636848a2019-04-15 19:38:50 +02001132set-dumpable
1133 This option is better left disabled by default and enabled only upon a
1134 developer's request. It has no impact on performance nor stability but will
1135 try hard to re-enable core dumps that were possibly disabled by file size
1136 limitations (ulimit -f), core size limitations (ulimit -c), or "dumpability"
1137 of a process after changing its UID/GID (such as /proc/sys/fs/suid_dumpable
1138 on Linux). Core dumps might still be limited by the current directory's
1139 permissions (check what directory the file is started from), the chroot
1140 directory's permission (it may be needed to temporarily disable the chroot
1141 directive or to move it to a dedicated writable location), or any other
1142 system-specific constraint. For example, some Linux flavours are notorious
1143 for replacing the default core file with a path to an executable not even
1144 installed on the system (check /proc/sys/kernel/core_pattern). Often, simply
1145 writing "core", "core.%p" or "/var/log/core/core.%p" addresses the issue.
1146 When trying to enable this option waiting for a rare issue to re-appear, it's
1147 often a good idea to first try to obtain such a dump by issuing, for example,
1148 "kill -11" to the haproxy process and verify that it leaves a core where
1149 expected when dying.
1150
Willy Tarreau610f04b2014-02-13 11:36:41 +01001151ssl-default-bind-ciphers <ciphers>
1152 This setting is only available when support for OpenSSL was built in. It sets
1153 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001154 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001155 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001156 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1157 information and recommendations see e.g.
1158 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1159 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1160 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1161 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001162
1163ssl-default-bind-ciphersuites <ciphersuites>
1164 This setting is only available when support for OpenSSL was built in and
1165 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1166 describing the list of cipher algorithms ("cipher suite") that are negotiated
1167 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1168 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001169 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1170 cipher configuration for TLSv1.2 and earlier, please check the
1171 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001172 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001173
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001174ssl-default-bind-options [<option>]...
1175 This setting is only available when support for OpenSSL was built in. It sets
1176 default ssl-options to force on all "bind" lines. Please check the "bind"
1177 keyword to see available options.
1178
1179 Example:
1180 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001181 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001182
Willy Tarreau610f04b2014-02-13 11:36:41 +01001183ssl-default-server-ciphers <ciphers>
1184 This setting is only available when support for OpenSSL was built in. It
1185 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001186 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001187 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001188 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1189 information and recommendations see e.g.
1190 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1191 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1192 For TLSv1.3 cipher configuration, please check the
1193 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1194 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001195
1196ssl-default-server-ciphersuites <ciphersuites>
1197 This setting is only available when support for OpenSSL was built in and
1198 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1199 string describing the list of cipher algorithms that are negotiated during
1200 the TLSv1.3 handshake with the server, for all "server" lines which do not
1201 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001202 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1203 cipher configuration for TLSv1.2 and earlier, please check the
1204 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1205 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001206
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001207ssl-default-server-options [<option>]...
1208 This setting is only available when support for OpenSSL was built in. It sets
1209 default ssl-options to force on all "server" lines. Please check the "server"
1210 keyword to see available options.
1211
Remi Gacogne47783ef2015-05-29 15:53:22 +02001212ssl-dh-param-file <file>
1213 This setting is only available when support for OpenSSL was built in. It sets
1214 the default DH parameters that are used during the SSL/TLS handshake when
1215 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001216 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001217 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001218 are not specified either by using ssl-dh-param-file or by setting them
1219 directly in the certificate file, pre-generated DH parameters of the size
1220 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1221 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001222 Custom DH parameters may be generated by using the OpenSSL command
1223 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1224 parameters should not be considered secure anymore.
1225
Emeric Brun850efd52014-01-29 12:24:34 +01001226ssl-server-verify [none|required]
1227 The default behavior for SSL verify on servers side. If specified to 'none',
1228 servers certificates are not verified. The default is 'required' except if
1229 forced using cmdline option '-dV'.
1230
Willy Tarreauabb175f2012-09-24 12:43:26 +02001231stats socket [<address:port>|<path>] [param*]
1232 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1233 Connections to this socket will return various statistics outputs and even
1234 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001235 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001236 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001237
Willy Tarreauabb175f2012-09-24 12:43:26 +02001238 All parameters supported by "bind" lines are supported, for instance to
1239 restrict access to some users or their access rights. Please consult
1240 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001241
1242stats timeout <timeout, in milliseconds>
1243 The default timeout on the stats socket is set to 10 seconds. It is possible
1244 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001245 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001246
1247stats maxconn <connections>
1248 By default, the stats socket is limited to 10 concurrent connections. It is
1249 possible to change this value with "stats maxconn".
1250
Willy Tarreau6a06a402007-07-15 20:15:28 +02001251uid <number>
1252 Changes the process' user ID to <number>. It is recommended that the user ID
1253 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1254 be started with superuser privileges in order to be able to switch to another
1255 one. See also "gid" and "user".
1256
1257ulimit-n <number>
1258 Sets the maximum number of per-process file-descriptors to <number>. By
1259 default, it is automatically computed, so it is recommended not to use this
1260 option.
1261
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001262unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1263 [ group <group> ] [ gid <gid> ]
1264
1265 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1266 This is mainly used to simplify declaration of those UNIX sockets and reduce
1267 the risk of errors, since those settings are most commonly required but are
1268 also process-specific. The <prefix> setting can be used to force all socket
1269 path to be relative to that directory. This might be needed to access another
1270 component's chroot. Note that those paths are resolved before haproxy chroots
1271 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1272 all have the same meaning as their homonyms used by the "bind" statement. If
1273 both are specified, the "bind" statement has priority, meaning that the
1274 "unix-bind" settings may be seen as process-wide default settings.
1275
Willy Tarreau1d549722016-02-16 12:41:57 +01001276unsetenv [<name> ...]
1277 Removes environment variables specified in arguments. This can be useful to
1278 hide some sensitive information that are occasionally inherited from the
1279 user's environment during some operations. Variables which did not exist are
1280 silently ignored so that after the operation, it is certain that none of
1281 these variables remain. The changes immediately take effect so that the next
1282 line in the configuration file will not see these variables. See also
1283 "setenv", "presetenv", and "resetenv".
1284
Willy Tarreau6a06a402007-07-15 20:15:28 +02001285user <user name>
1286 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1287 See also "uid" and "group".
1288
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001289node <name>
1290 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1291
1292 This statement is useful in HA configurations where two or more processes or
1293 servers share the same IP address. By setting a different node-name on all
1294 nodes, it becomes easy to immediately spot what server is handling the
1295 traffic.
1296
1297description <text>
1298 Add a text that describes the instance.
1299
1300 Please note that it is required to escape certain characters (# for example)
1301 and this text is inserted into a html page so you should avoid using
1302 "<" and ">" characters.
1303
Thomas Holmesdb04f192015-05-18 13:21:39 +0100130451degrees-data-file <file path>
1305 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001306 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001307
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001308 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001309 compiled with USE_51DEGREES.
1310
Ben Shillitof25e8e52016-12-02 14:25:37 +0000131151degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001312 A list of 51Degrees property names to be load from the dataset. A full list
1313 of names is available on the 51Degrees website:
1314 https://51degrees.com/resources/property-dictionary
1315
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001316 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001317 compiled with USE_51DEGREES.
1318
Dragan Dosen93b38d92015-06-29 16:43:25 +0200131951degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001320 A char that will be appended to every property value in a response header
1321 containing 51Degrees results. If not set that will be set as ','.
1322
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001323 Please note that this option is only available when haproxy has been
1324 compiled with USE_51DEGREES.
1325
132651degrees-cache-size <number>
1327 Sets the size of the 51Degrees converter cache to <number> entries. This
1328 is an LRU cache which reminds previous device detections and their results.
1329 By default, this cache is disabled.
1330
1331 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001332 compiled with USE_51DEGREES.
1333
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001334wurfl-data-file <file path>
1335 The path of the WURFL data file to provide device detection services. The
1336 file should be accessible by HAProxy with relevant permissions.
1337
1338 Please note that this option is only available when haproxy has been compiled
1339 with USE_WURFL=1.
1340
1341wurfl-information-list [<capability>]*
1342 A space-delimited list of WURFL capabilities, virtual capabilities, property
1343 names we plan to use in injected headers. A full list of capability and
1344 virtual capability names is available on the Scientiamobile website :
1345
1346 https://www.scientiamobile.com/wurflCapability
1347
1348 Valid WURFL properties are:
1349 - wurfl_id Contains the device ID of the matched device.
1350
1351 - wurfl_root_id Contains the device root ID of the matched
1352 device.
1353
1354 - wurfl_isdevroot Tells if the matched device is a root device.
1355 Possible values are "TRUE" or "FALSE".
1356
1357 - wurfl_useragent The original useragent coming with this
1358 particular web request.
1359
1360 - wurfl_api_version Contains a string representing the currently
1361 used Libwurfl API version.
1362
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001363 - wurfl_info A string containing information on the parsed
1364 wurfl.xml and its full path.
1365
1366 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1367 WURFL has been loaded successfully.
1368
1369 - wurfl_normalized_useragent The normalized useragent.
1370
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001371 Please note that this option is only available when haproxy has been compiled
1372 with USE_WURFL=1.
1373
1374wurfl-information-list-separator <char>
1375 A char that will be used to separate values in a response header containing
1376 WURFL results. If not set that a comma (',') will be used by default.
1377
1378 Please note that this option is only available when haproxy has been compiled
1379 with USE_WURFL=1.
1380
1381wurfl-patch-file [<file path>]
1382 A list of WURFL patch file paths. Note that patches are loaded during startup
1383 thus before the chroot.
1384
1385 Please note that this option is only available when haproxy has been compiled
1386 with USE_WURFL=1.
1387
paulborilebad132c2019-04-18 11:57:04 +02001388wurfl-cache-size <size>
1389 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1390 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001391 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001392 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001393
1394 Please note that this option is only available when haproxy has been compiled
1395 with USE_WURFL=1.
1396
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013973.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001398-----------------------
1399
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001400busy-polling
1401 In some situations, especially when dealing with low latency on processors
1402 supporting a variable frequency or when running inside virtual machines, each
1403 time the process waits for an I/O using the poller, the processor goes back
1404 to sleep or is offered to another VM for a long time, and it causes
1405 excessively high latencies. This option provides a solution preventing the
1406 processor from sleeping by always using a null timeout on the pollers. This
1407 results in a significant latency reduction (30 to 100 microseconds observed)
1408 at the expense of a risk to overheat the processor. It may even be used with
1409 threads, in which case improperly bound threads may heavily conflict,
1410 resulting in a worse performance and high values for the CPU stolen fields
1411 in "show info" output, indicating which threads are misconfigured. It is
1412 important not to let the process run on the same processor as the network
1413 interrupts when this option is used. It is also better to avoid using it on
1414 multiple CPU threads sharing the same core. This option is disabled by
1415 default. If it has been enabled, it may still be forcibly disabled by
1416 prefixing it with the "no" keyword. It is ignored by the "select" and
1417 "poll" pollers.
1418
William Dauchy857b9432019-12-28 15:36:02 +01001419 This option is automatically disabled on old processes in the context of
1420 seamless reload; it avoids too much cpu conflicts when multiple processes
1421 stay around for some time waiting for the end of their current connections.
1422
Willy Tarreau1746eec2014-04-25 10:46:47 +02001423max-spread-checks <delay in milliseconds>
1424 By default, haproxy tries to spread the start of health checks across the
1425 smallest health check interval of all the servers in a farm. The principle is
1426 to avoid hammering services running on the same server. But when using large
1427 check intervals (10 seconds or more), the last servers in the farm take some
1428 time before starting to be tested, which can be a problem. This parameter is
1429 used to enforce an upper bound on delay between the first and the last check,
1430 even if the servers' check intervals are larger. When servers run with
1431 shorter intervals, their intervals will be respected though.
1432
Willy Tarreau6a06a402007-07-15 20:15:28 +02001433maxconn <number>
1434 Sets the maximum per-process number of concurrent connections to <number>. It
1435 is equivalent to the command-line argument "-n". Proxies will stop accepting
1436 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001437 automatically adjusted according to this value. See also "ulimit-n". Note:
1438 the "select" poller cannot reliably use more than 1024 file descriptors on
1439 some platforms. If your platform only supports select and reports "select
1440 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001441 below 500 in general). If this value is not set, it will automatically be
1442 calculated based on the current file descriptors limit reported by the
1443 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1444 is enforced, based on the buffer size, memory allocated to compression, SSL
1445 cache size, and use or not of SSL and the associated maxsslconn (which can
1446 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001447
Willy Tarreau81c25d02011-09-07 15:17:21 +02001448maxconnrate <number>
1449 Sets the maximum per-process number of connections per second to <number>.
1450 Proxies will stop accepting connections when this limit is reached. It can be
1451 used to limit the global capacity regardless of each frontend capacity. It is
1452 important to note that this can only be used as a service protection measure,
1453 as there will not necessarily be a fair share between frontends when the
1454 limit is reached, so it's a good idea to also limit each frontend to some
1455 value close to its expected share. Also, lowering tune.maxaccept can improve
1456 fairness.
1457
William Lallemandd85f9172012-11-09 17:05:39 +01001458maxcomprate <number>
1459 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001460 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001461 level will be decreased during the session. If the maximum is reached at the
1462 beginning of a session, the session will not compress at all. If the maximum
1463 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001464 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001465 default value.
1466
William Lallemand072a2bf2012-11-20 17:01:01 +01001467maxcompcpuusage <number>
1468 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1469 for new requests or decreasing the compression level of current requests.
1470 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1471 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1472 case of multiple processes (nbproc > 1), each process manages its individual
1473 usage. A value of 100 disable the limit. The default value is 100. Setting
1474 a lower value will prevent the compression work from slowing the whole
1475 process down and from introducing high latencies.
1476
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001477maxpipes <number>
1478 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1479 are only used by kernel-based tcp splicing. Since a pipe contains two file
1480 descriptors, the "ulimit-n" value will be increased accordingly. The default
1481 value is maxconn/4, which seems to be more than enough for most heavy usages.
1482 The splice code dynamically allocates and releases pipes, and can fall back
1483 to standard copy, so setting this value too low may only impact performance.
1484
Willy Tarreau93e7c002013-10-07 18:51:07 +02001485maxsessrate <number>
1486 Sets the maximum per-process number of sessions per second to <number>.
1487 Proxies will stop accepting connections when this limit is reached. It can be
1488 used to limit the global capacity regardless of each frontend capacity. It is
1489 important to note that this can only be used as a service protection measure,
1490 as there will not necessarily be a fair share between frontends when the
1491 limit is reached, so it's a good idea to also limit each frontend to some
1492 value close to its expected share. Also, lowering tune.maxaccept can improve
1493 fairness.
1494
Willy Tarreau403edff2012-09-06 11:58:37 +02001495maxsslconn <number>
1496 Sets the maximum per-process number of concurrent SSL connections to
1497 <number>. By default there is no SSL-specific limit, which means that the
1498 global maxconn setting will apply to all connections. Setting this limit
1499 avoids having openssl use too much memory and crash when malloc returns NULL
1500 (since it unfortunately does not reliably check for such conditions). Note
1501 that the limit applies both to incoming and outgoing connections, so one
1502 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001503 If this value is not set, but a memory limit is enforced, this value will be
1504 automatically computed based on the memory limit, maxconn, the buffer size,
1505 memory allocated to compression, SSL cache size, and use of SSL in either
1506 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1507 when there is a memory limit, haproxy will automatically adjust these values
1508 so that 100% of the connections can be made over SSL with no risk, and will
1509 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001510
Willy Tarreaue43d5322013-10-07 20:01:52 +02001511maxsslrate <number>
1512 Sets the maximum per-process number of SSL sessions per second to <number>.
1513 SSL listeners will stop accepting connections when this limit is reached. It
1514 can be used to limit the global SSL CPU usage regardless of each frontend
1515 capacity. It is important to note that this can only be used as a service
1516 protection measure, as there will not necessarily be a fair share between
1517 frontends when the limit is reached, so it's a good idea to also limit each
1518 frontend to some value close to its expected share. It is also important to
1519 note that the sessions are accounted before they enter the SSL stack and not
1520 after, which also protects the stack against bad handshakes. Also, lowering
1521 tune.maxaccept can improve fairness.
1522
William Lallemand9d5f5482012-11-07 16:12:57 +01001523maxzlibmem <number>
1524 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1525 When the maximum amount is reached, future sessions will not compress as long
1526 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001527 The default value is 0. The value is available in bytes on the UNIX socket
1528 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1529 "ZlibMemUsage" in bytes.
1530
Willy Tarreau6a06a402007-07-15 20:15:28 +02001531noepoll
1532 Disables the use of the "epoll" event polling system on Linux. It is
1533 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001534 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001535
1536nokqueue
1537 Disables the use of the "kqueue" event polling system on BSD. It is
1538 equivalent to the command-line argument "-dk". The next polling system
1539 used will generally be "poll". See also "nopoll".
1540
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001541noevports
1542 Disables the use of the event ports event polling system on SunOS systems
1543 derived from Solaris 10 and later. It is equivalent to the command-line
1544 argument "-dv". The next polling system used will generally be "poll". See
1545 also "nopoll".
1546
Willy Tarreau6a06a402007-07-15 20:15:28 +02001547nopoll
1548 Disables the use of the "poll" event polling system. It is equivalent to the
1549 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001550 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001551 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
1552 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001553
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001554nosplice
1555 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001556 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001557 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001558 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001559 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1560 be used. This option makes it easier to globally disable kernel splicing in
1561 case of doubt. See also "option splice-auto", "option splice-request" and
1562 "option splice-response".
1563
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001564nogetaddrinfo
1565 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1566 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1567
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001568noreuseport
1569 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1570 command line argument "-dR".
1571
Willy Tarreaud2d33482019-04-25 17:09:07 +02001572profiling.tasks { auto | on | off }
1573 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
1574 the profiling automatically turns on a thread when it starts to suffer from
1575 an average latency of 1000 microseconds or higher as reported in the
1576 "avg_loop_us" activity field, and automatically turns off when the latency
John Roesler27c754c2019-07-10 15:45:51 -05001577 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02001578 loops so it does not vary quickly and tends to significantly smooth short
1579 spikes). It may also spontaneously trigger from time to time on overloaded
1580 systems, containers, or virtual machines, or when the system swaps (which
1581 must absolutely never happen on a load balancer).
1582
1583 CPU profiling per task can be very convenient to report where the time is
1584 spent and which requests have what effect on which other request. Enabling
1585 it will typically affect the overall's performance by less than 1%, thus it
1586 is recommended to leave it to the default 'auto' value so that it only
1587 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01001588 supporting the clock_gettime(2) syscall with clock identifiers
1589 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1590 be zero. This option may be changed at run time using "set profiling" on the
1591 CLI.
1592
Willy Tarreaufe255b72007-10-14 23:09:26 +02001593spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001594 Sometimes it is desirable to avoid sending agent and health checks to
1595 servers at exact intervals, for instance when many logical servers are
1596 located on the same physical server. With the help of this parameter, it
1597 becomes possible to add some randomness in the check interval between 0
1598 and +/- 50%. A value between 2 and 5 seems to show good results. The
1599 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001600
Davor Ocelice9ed2812017-12-25 17:49:28 +01001601ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001602 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001603 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001604 multiple times, it will simply enable multiple crypto engines. Referencing an
1605 unsupported engine will prevent haproxy from starting. Note that many engines
1606 will lead to lower HTTPS performance than pure software with recent
1607 processors. The optional command "algo" sets the default algorithms an ENGINE
1608 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001609 of "ALL" uses the engine for all cryptographic operations. If no list of
1610 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001611 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1612 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1613 openssl configuration file uses:
1614 https://www.openssl.org/docs/man1.0.2/apps/config.html
1615
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001616ssl-mode-async
1617 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001618 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001619 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1620 doesn't support moving read/write buffers and is not compliant with
1621 haproxy's buffer management. So the asynchronous mode is disabled on
John Roesler27c754c2019-07-10 15:45:51 -05001622 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00001623 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001624
Willy Tarreau33cb0652014-12-23 22:52:37 +01001625tune.buffers.limit <number>
1626 Sets a hard limit on the number of buffers which may be allocated per process.
1627 The default value is zero which means unlimited. The minimum non-zero value
1628 will always be greater than "tune.buffers.reserve" and should ideally always
1629 be about twice as large. Forcing this value can be particularly useful to
1630 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001631 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001632 another one to be released by another session. Since buffers are dynamically
1633 allocated and released, the waiting time is very short and not perceptible
1634 provided that limits remain reasonable. In fact sometimes reducing the limit
1635 may even increase performance by increasing the CPU cache's efficiency. Tests
1636 have shown good results on average HTTP traffic with a limit to 1/10 of the
1637 expected global maxconn setting, which also significantly reduces memory
1638 usage. The memory savings come from the fact that a number of connections
1639 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1640 advised to do so by an haproxy core developer.
1641
Willy Tarreau1058ae72014-12-23 22:40:40 +01001642tune.buffers.reserve <number>
1643 Sets the number of buffers which are pre-allocated and reserved for use only
1644 during memory shortage conditions resulting in failed memory allocations. The
1645 minimum value is 2 and is also the default. There is no reason a user would
1646 want to change this value, it's mostly aimed at haproxy core developers.
1647
Willy Tarreau27a674e2009-08-17 07:23:33 +02001648tune.bufsize <number>
1649 Sets the buffer size to this size (in bytes). Lower values allow more
1650 sessions to coexist in the same amount of RAM, and higher values allow some
1651 applications with very large cookies to work. The default value is 16384 and
1652 can be changed at build time. It is strongly recommended not to change this
1653 from the default value, as very low values will break some services such as
1654 statistics, and values larger than default size will increase memory usage,
1655 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001656 parameter should be decreased by the same factor as this one is increased. In
1657 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1658 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001659 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001660 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1661 value set using this parameter will automatically be rounded up to the next
1662 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001663
Willy Tarreau43961d52010-10-04 20:39:20 +02001664tune.chksize <number>
1665 Sets the check buffer size to this size (in bytes). Higher values may help
1666 find string or regex patterns in very large pages, though doing so may imply
1667 more memory and CPU usage. The default value is 16384 and can be changed at
1668 build time. It is not recommended to change this value, but to use better
1669 checks whenever possible.
1670
William Lallemandf3747832012-11-09 12:33:10 +01001671tune.comp.maxlevel <number>
1672 Sets the maximum compression level. The compression level affects CPU
1673 usage during compression. This value affects CPU usage during compression.
1674 Each session using compression initializes the compression algorithm with
1675 this value. The default value is 1.
1676
Willy Tarreauc299e1e2019-02-27 11:35:12 +01001677tune.fail-alloc
1678 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
1679 allocation attempt fails. Must be between 0 (no failure) and 100 (no
1680 success). This is useful to debug and make sure memory failures are handled
1681 gracefully.
1682
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001683tune.h2.header-table-size <number>
1684 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1685 cannot be larger than 65536 bytes. A larger value may help certain clients
1686 send more compact requests, depending on their capabilities. This amount of
1687 memory is consumed for each HTTP/2 connection. It is recommended not to
1688 change it.
1689
Willy Tarreaue6baec02017-07-27 11:45:11 +02001690tune.h2.initial-window-size <number>
1691 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001692 can upload before waiting for an acknowledgment from haproxy. This setting
1693 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001694 The default value is 65535, which roughly allows up to 5 Mbps of upload
1695 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1696 over a 1-ms local network. It can make sense to increase this value to allow
1697 faster uploads, or to reduce it to increase fairness when dealing with many
1698 clients. It doesn't affect resource usage.
1699
Willy Tarreau5242ef82017-07-27 11:47:28 +02001700tune.h2.max-concurrent-streams <number>
1701 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1702 number of outstanding requests on a single connection). The default value is
1703 100. A larger one may slightly improve page load time for complex sites when
1704 visited over high latency networks, but increases the amount of resources a
1705 single client may allocate. A value of zero disables the limit so a single
1706 client may create as many streams as allocatable by haproxy. It is highly
1707 recommended not to change this value.
1708
Willy Tarreaua24b35c2019-02-21 13:24:36 +01001709tune.h2.max-frame-size <number>
1710 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
1711 receive to its peers. The default value is the largest between 16384 and the
1712 buffer size (tune.bufsize). In any case, haproxy will not announce support
1713 for frame sizes larger than buffers. The main purpose of this setting is to
1714 allow to limit the maximum frame size setting when using large buffers. Too
1715 large frame sizes might have performance impact or cause some peers to
1716 misbehave. It is highly recommended not to change this value.
1717
Willy Tarreau193b8c62012-11-22 00:17:38 +01001718tune.http.cookielen <number>
1719 Sets the maximum length of captured cookies. This is the maximum value that
1720 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1721 will automatically be truncated to this one. It is important not to set too
1722 high a value because all cookie captures still allocate this size whatever
1723 their configured value (they share a same pool). This value is per request
1724 per response, so the memory allocated is twice this value per connection.
1725 When not specified, the limit is set to 63 characters. It is recommended not
1726 to change this value.
1727
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001728tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001729 Sets the maximum length of request URI in logs. This prevents truncating long
1730 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001731 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001732 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001733 configuration directives too.
1734 The default value is 1024.
1735
Willy Tarreauac1932d2011-10-24 19:14:41 +02001736tune.http.maxhdr <number>
1737 Sets the maximum number of headers in a request. When a request comes with a
1738 number of headers greater than this value (including the first line), it is
1739 rejected with a "400 Bad Request" status code. Similarly, too large responses
1740 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1741 for all usages, considering that the widely deployed Apache server uses the
1742 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001743 a buggy application to work by the time it gets fixed. The accepted range is
1744 1..32767. Keep in mind that each new header consumes 32bits of memory for
1745 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001746
Willy Tarreau7e312732014-02-12 16:35:14 +01001747tune.idletimer <timeout>
1748 Sets the duration after which haproxy will consider that an empty buffer is
1749 probably associated with an idle stream. This is used to optimally adjust
1750 some packet sizes while forwarding large and small data alternatively. The
1751 decision to use splice() or to send large buffers in SSL is modulated by this
1752 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1753 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001754 which seems to correctly detect end user pauses (e.g. read a page before
John Roesler27c754c2019-07-10 15:45:51 -05001755 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01001756 tune.ssl.maxrecord below.
1757
Willy Tarreau7ac908b2019-02-27 12:02:18 +01001758tune.listener.multi-queue { on | off }
1759 Enables ('on') or disables ('off') the listener's multi-queue accept which
1760 spreads the incoming traffic to all threads a "bind" line is allowed to run
1761 on instead of taking them for itself. This provides a smoother traffic
1762 distribution and scales much better, especially in environments where threads
1763 may be unevenly loaded due to external activity (network interrupts colliding
1764 with one thread for example). This option is enabled by default, but it may
1765 be forcefully disabled for troubleshooting or for situations where it is
1766 estimated that the operating system already provides a good enough
1767 distribution and connections are extremely short-lived.
1768
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001769tune.lua.forced-yield <number>
1770 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001771 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001772 HAProxy scheduler to process other tasks like accepting connections or
1773 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001774 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001775 lowered. If the Lua code is quite long and its result is absolutely required
1776 to process the data, the <number> can be increased.
1777
Willy Tarreau32f61e22015-03-18 17:54:59 +01001778tune.lua.maxmem
1779 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1780 default it is zero which means unlimited. It is important to set a limit to
1781 ensure that a bug in a script will not result in the system running out of
1782 memory.
1783
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001784tune.lua.session-timeout <timeout>
1785 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001786 preventing infinite loops or spending too much time in Lua. This timeout
1787 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001788 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001789
1790tune.lua.task-timeout <timeout>
1791 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1792 dedicated to the tasks. By default, this timeout isn't set because a task may
1793 remain alive during of the lifetime of HAProxy. For example, a task used to
1794 check servers.
1795
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001796tune.lua.service-timeout <timeout>
1797 This is the execution timeout for the Lua services. This is useful for
1798 preventing infinite loops or spending too much time in Lua. This timeout
1799 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001800 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001801
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001802tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001803 Sets the maximum number of consecutive connections a process may accept in a
1804 row before switching to other work. In single process mode, higher numbers
1805 give better performance at high connection rates. However in multi-process
1806 modes, keeping a bit of fairness between processes generally is better to
1807 increase performance. This value applies individually to each listener, so
1808 that the number of processes a listener is bound to is taken into account.
1809 This value defaults to 64. In multi-process mode, it is divided by twice
1810 the number of processes the listener is bound to. Setting this value to -1
1811 completely disables the limitation. It should normally not be needed to tweak
1812 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001813
1814tune.maxpollevents <number>
1815 Sets the maximum amount of events that can be processed at once in a call to
1816 the polling system. The default value is adapted to the operating system. It
1817 has been noticed that reducing it below 200 tends to slightly decrease
1818 latency at the expense of network bandwidth, and increasing it above 200
1819 tends to trade latency for slightly increased bandwidth.
1820
Willy Tarreau27a674e2009-08-17 07:23:33 +02001821tune.maxrewrite <number>
1822 Sets the reserved buffer space to this size in bytes. The reserved space is
1823 used for header rewriting or appending. The first reads on sockets will never
1824 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1825 bufsize, though that does not make much sense since there are rarely large
1826 numbers of headers to add. Setting it too high prevents processing of large
1827 requests or responses. Setting it too low prevents addition of new headers
1828 to already large requests or to POST requests. It is generally wise to set it
1829 to about 1024. It is automatically readjusted to half of bufsize if it is
1830 larger than that. This means you don't have to worry about it when changing
1831 bufsize.
1832
Willy Tarreauf3045d22015-04-29 16:24:50 +02001833tune.pattern.cache-size <number>
1834 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1835 cache which reminds previous lookups and their results. It is used by ACLs
1836 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1837 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1838 strings. It applies to pattern expressions which means that it will be able
1839 to memorize the result of a lookup among all the patterns specified on a
1840 configuration line (including all those loaded from files). It automatically
1841 invalidates entries which are updated using HTTP actions or on the CLI. The
1842 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau7fdd81c2019-10-23 06:59:31 +02001843 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
1844 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02001845 risk of collision in this cache, which is in the order of the size of the
1846 cache divided by 2^64. Typically, at 10000 requests per second with the
1847 default cache size of 10000 entries, there's 1% chance that a brute force
1848 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1849 This is considered much lower than the risk of a memory corruption caused by
1850 aging components. If this is not acceptable, the cache can be disabled by
1851 setting this parameter to 0.
1852
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001853tune.pipesize <number>
1854 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1855 are the default size for the system. But sometimes when using TCP splicing,
1856 it can improve performance to increase pipe sizes, especially if it is
1857 suspected that pipes are not filled and that many calls to splice() are
1858 performed. This has an impact on the kernel's memory footprint, so this must
1859 not be changed if impacts are not understood.
1860
Olivier Houchard88698d92019-04-16 19:07:22 +02001861tune.pool-high-fd-ratio <number>
1862 This setting sets the max number of file descriptors (in percentage) used by
1863 haproxy globally against the maximum number of file descriptors haproxy can
1864 use before we start killing idle connections when we can't reuse a connection
1865 and we have to create a new one. The default is 25 (one quarter of the file
1866 descriptor will mean that roughly half of the maximum front connections can
1867 keep an idle connection behind, anything beyond this probably doesn't make
John Roesler27c754c2019-07-10 15:45:51 -05001868 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02001869
Willy Tarreauc55e3e12020-07-01 18:30:16 +02001870tune.pool-low-fd-ratio <number>
1871 This setting sets the max number of file descriptors (in percentage) used by
1872 haproxy globally against the maximum number of file descriptors haproxy can
1873 use before we stop putting connection into the idle pool for reuse. The
1874 default is 20.
1875
Willy Tarreaue803de22010-01-21 17:43:04 +01001876tune.rcvbuf.client <number>
1877tune.rcvbuf.server <number>
1878 Forces the kernel socket receive buffer size on the client or the server side
1879 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1880 and backends. It should normally never be set, and the default size (0) lets
John Roesler27c754c2019-07-10 15:45:51 -05001881 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001882 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001883 order to save kernel memory by preventing it from buffering too large amounts
1884 of received data. Lower values will significantly increase CPU usage though.
1885
Willy Tarreaub22fc302015-12-14 12:04:35 +01001886tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001887 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001888 socket buffers. One of them is that a read returns more than <recv_enough>
1889 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1890 may be changed by this setting to better deal with workloads involving lots
1891 of short messages such as telnet or SSH sessions.
1892
Olivier Houchard1599b802018-05-24 18:59:04 +02001893tune.runqueue-depth <number>
John Roesler27c754c2019-07-10 15:45:51 -05001894 Sets the maximum amount of task that can be processed at once when running
Olivier Houchard1599b802018-05-24 18:59:04 +02001895 tasks. The default value is 200. Increasing it may incur latency when
1896 dealing with I/Os, making it too small can incur extra overhead.
1897
Willy Tarreaue803de22010-01-21 17:43:04 +01001898tune.sndbuf.client <number>
1899tune.sndbuf.server <number>
1900 Forces the kernel socket send buffer size on the client or the server side to
1901 the specified value in bytes. This value applies to all TCP/HTTP frontends
1902 and backends. It should normally never be set, and the default size (0) lets
John Roesler27c754c2019-07-10 15:45:51 -05001903 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001904 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001905 order to save kernel memory by preventing it from buffering too large amounts
1906 of received data. Lower values will significantly increase CPU usage though.
1907 Another use case is to prevent write timeouts with extremely slow clients due
1908 to the kernel waiting for a large part of the buffer to be read before
1909 notifying haproxy again.
1910
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001911tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001912 Sets the size of the global SSL session cache, in a number of blocks. A block
1913 is large enough to contain an encoded session without peer certificate.
1914 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001915 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001916 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001917 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001918 and reassigned. Higher values reduce the occurrence of such a purge, hence
1919 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1920 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001921 and are shared between all processes if "nbproc" is greater than 1. Setting
1922 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001923
Emeric Brun8dc60392014-05-09 13:52:00 +02001924tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001925 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001926 should normally not be used since it will force many renegotiations due to
1927 clients hitting a random process. But it may be required on some operating
1928 systems where none of the SSL cache synchronization method may be used. In
1929 this case, adding a first layer of hash-based load balancing before the SSL
1930 layer might limit the impact of the lack of session sharing.
1931
Emeric Brun4f65bff2012-11-16 15:11:00 +01001932tune.ssl.lifetime <timeout>
1933 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001934 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001935 does not guarantee that sessions will last that long, because if the cache is
1936 full, the longest idle sessions will be purged despite their configured
1937 lifetime. The real usefulness of this setting is to prevent sessions from
1938 being used for too long.
1939
Willy Tarreaubfd59462013-02-21 07:46:09 +01001940tune.ssl.maxrecord <number>
1941 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1942 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1943 data only once it has received a full record. With large records, it means
1944 that clients might have to download up to 16kB of data before starting to
1945 process them. Limiting the value can improve page load times on browsers
1946 located over high latency or low bandwidth networks. It is suggested to find
1947 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1948 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1949 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1950 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001951 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001952 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001953
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001954tune.ssl.default-dh-param <number>
1955 Sets the maximum size of the Diffie-Hellman parameters used for generating
1956 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1957 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1958 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1959 this maximum value. Default value if 1024. Only 1024 or higher values are
1960 allowed. Higher values will increase the CPU load, and values greater than
1961 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001962 used if static Diffie-Hellman parameters are supplied either directly
1963 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001964
Christopher Faulet31af49d2015-06-09 17:29:50 +02001965tune.ssl.ssl-ctx-cache-size <number>
1966 Sets the size of the cache used to store generated certificates to <number>
1967 entries. This is a LRU cache. Because generating a SSL certificate
1968 dynamically is expensive, they are cached. The default cache size is set to
1969 1000 entries.
1970
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001971tune.ssl.capture-cipherlist-size <number>
1972 Sets the maximum size of the buffer used for capturing client-hello cipher
1973 list. If the value is 0 (default value) the capture is disabled, otherwise
1974 a buffer is allocated for each SSL/TLS connection.
1975
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001976tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001977tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001978tune.vars.reqres-max-size <size>
1979tune.vars.sess-max-size <size>
1980tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001981 These five tunes help to manage the maximum amount of memory used by the
1982 variables system. "global" limits the overall amount of memory available for
1983 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1984 memory for the session scope, "txn" for the transaction scope, and "reqres"
1985 limits the memory for each request or response processing.
1986 Memory accounting is hierarchical, meaning more coarse grained limits include
1987 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1988 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001989
Daniel Schneller0b547052016-03-21 20:46:57 +01001990 For example, when "tune.vars.sess-max-size" is limited to 100,
1991 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1992 100 either. If we create a variable "txn.var" that contains 100 bytes,
1993 all available space is consumed.
1994 Notice that exceeding the limits at runtime will not result in an error
1995 message, but values might be cut off or corrupted. So make sure to accurately
1996 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001997
William Lallemanda509e4c2012-11-07 16:54:34 +01001998tune.zlib.memlevel <number>
1999 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002000 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01002001 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01002002 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01002003 between 1 and 9. The default value is 8.
2004
2005tune.zlib.windowsize <number>
2006 Sets the window size (the size of the history buffer) as a parameter of the
2007 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01002008 in better compression at the expense of memory usage. Can be a value between
2009 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002010
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020113.3. Debugging
2012--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002013
2014debug
2015 Enables debug mode which dumps to stdout all exchanges, and disables forking
2016 into background. It is the equivalent of the command-line argument "-d". It
2017 should never be used in a production configuration since it may prevent full
2018 system startup.
2019
2020quiet
2021 Do not display any message during startup. It is equivalent to the command-
2022 line argument "-q".
2023
Emeric Brunf099e792010-09-27 12:05:28 +02002024
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010020253.4. Userlists
2026--------------
2027It is possible to control access to frontend/backend/listen sections or to
2028http stats by allowing only authenticated and authorized users. To do this,
2029it is required to create at least one userlist and to define users.
2030
2031userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002032 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002033 used to store authentication & authorization data for independent customers.
2034
2035group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002036 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002037 attach users to this group by using a comma separated list of names
2038 proceeded by "users" keyword.
2039
Cyril Bontéf0c60612010-02-06 14:44:47 +01002040user <username> [password|insecure-password <password>]
2041 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002042 Adds user <username> to the current userlist. Both secure (encrypted) and
2043 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002044 evaluated using the crypt(3) function, so depending on the system's
2045 capabilities, different algorithms are supported. For example, modern Glibc
2046 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2047 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002048
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002049 Attention: Be aware that using encrypted passwords might cause significantly
2050 increased CPU usage, depending on the number of requests, and the algorithm
2051 used. For any of the hashed variants, the password for each request must
2052 be processed through the chosen algorithm, before it can be compared to the
2053 value specified in the config file. Most current algorithms are deliberately
2054 designed to be expensive to compute to achieve resistance against brute
2055 force attacks. They do not simply salt/hash the clear text password once,
2056 but thousands of times. This can quickly become a major factor in haproxy's
2057 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002058
2059 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002060 userlist L1
2061 group G1 users tiger,scott
2062 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002063
Cyril Bontéf0c60612010-02-06 14:44:47 +01002064 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2065 user scott insecure-password elgato
2066 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002067
Cyril Bontéf0c60612010-02-06 14:44:47 +01002068 userlist L2
2069 group G1
2070 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002071
Cyril Bontéf0c60612010-02-06 14:44:47 +01002072 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2073 user scott insecure-password elgato groups G1,G2
2074 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002075
2076 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002077
Emeric Brunf099e792010-09-27 12:05:28 +02002078
20793.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002080----------
Emeric Brun94900952015-06-11 18:25:54 +02002081It is possible to propagate entries of any data-types in stick-tables between
2082several haproxy instances over TCP connections in a multi-master fashion. Each
2083instance pushes its local updates and insertions to remote peers. The pushed
2084values overwrite remote ones without aggregation. Interrupted exchanges are
2085automatically detected and recovered from the last known point.
2086In addition, during a soft restart, the old process connects to the new one
2087using such a TCP connection to push all its entries before the new process
2088tries to connect to other peers. That ensures very fast replication during a
2089reload, it typically takes a fraction of a second even for large tables.
2090Note that Server IDs are used to identify servers remotely, so it is important
2091that configurations look similar or at least that the same IDs are forced on
2092each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002093
2094peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002095 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002096 which is referenced by one or more stick-tables.
2097
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002098bind [<address>]:<port_range> [, ...] [param*]
2099 Defines the binding parameters of the local peer of this "peers" section.
2100 Such lines are not supported with "peer" line in the same "peers" section.
2101
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002102disabled
2103 Disables a peers section. It disables both listening and any synchronization
2104 related to this section. This is provided to disable synchronization of stick
2105 tables without having to comment out all "peers" references.
2106
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002107default-bind [param*]
2108 Defines the binding parameters for the local peer, excepted its address.
2109
2110default-server [param*]
2111 Change default options for a server in a "peers" section.
2112
2113 Arguments:
2114 <param*> is a list of parameters for this server. The "default-server"
2115 keyword accepts an important number of options and has a complete
2116 section dedicated to it. Please refer to section 5 for more
2117 details.
2118
2119
2120 See also: "server" and section 5 about server options
2121
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002122enable
2123 This re-enables a disabled peers section which was previously disabled.
2124
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002125peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002126 Defines a peer inside a peers section.
2127 If <peername> is set to the local peer name (by default hostname, or forced
2128 using "-L" command line option), haproxy will listen for incoming remote peer
2129 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
2130 to join the remote peer, and <peername> is used at the protocol level to
2131 identify and validate the remote peer on the server side.
2132
2133 During a soft restart, local peer <ip>:<port> is used by the old instance to
2134 connect the new one and initiate a complete replication (teaching process).
2135
2136 It is strongly recommended to have the exact same peers declaration on all
2137 peers and to only rely on the "-L" command line argument to change the local
2138 peer name. This makes it easier to maintain coherent configuration files
2139 across all peers.
2140
William Lallemandb2f07452015-05-12 14:27:13 +02002141 You may want to reference some environment variables in the address
2142 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002143
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002144 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2145 "server" keyword explanation below).
2146
2147server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002148 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002149 with a support for all "server" parameters found in 5.2 paragraph.
2150 If the underlying peer is local, <ip>:<port> parameters must not be present.
2151 These parameters must be provided on a "bind" line (see "bind" keyword
2152 of this "peers" section).
2153 Some of these parameters are irrelevant for "peers" sections.
2154
2155
Cyril Bontédc4d9032012-04-08 21:57:39 +02002156 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002157 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002158 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002159 peer haproxy1 192.168.0.1:1024
2160 peer haproxy2 192.168.0.2:1024
2161 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002162
2163 backend mybackend
2164 mode tcp
2165 balance roundrobin
2166 stick-table type ip size 20k peers mypeers
2167 stick on src
2168
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002169 server srv1 192.168.0.30:80
2170 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002171
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002172 Example:
2173 peers mypeers
2174 bind 127.0.0.11:10001 ssl crt mycerts/pem
2175 default-server ssl verify none
2176 server hostA 127.0.0.10:10000
2177 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002178
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002179
2180table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2181 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2182
2183 Configure a stickiness table for the current section. This line is parsed
2184 exactly the same way as the "stick-table" keyword in others section, except
John Roesler27c754c2019-07-10 15:45:51 -05002185 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002186 mandatory first parameter to designate the stick-table. Contrary to others
2187 sections, there may be several "table" lines in "peers" sections (see also
2188 "stick-table" keyword).
2189
2190 Also be aware of the fact that "peers" sections have their own stick-table
2191 namespaces to avoid collisions between stick-table names identical in
2192 different "peers" section. This is internally handled prepending the "peers"
2193 sections names to the name of the stick-tables followed by a '/' character.
2194 If somewhere else in the configuration file you have to refer to such
2195 stick-tables declared in "peers" sections you must use the prefixed version
2196 of the stick-table name as follows:
2197
2198 peers mypeers
2199 peer A ...
2200 peer B ...
2201 table t1 ...
2202
2203 frontend fe1
2204 tcp-request content track-sc0 src table mypeers/t1
2205
2206 This is also this prefixed version of the stick-table names which must be
2207 used to refer to stick-tables through the CLI.
2208
2209 About "peers" protocol, as only "peers" belonging to the same section may
2210 communicate with each others, there is no need to do such a distinction.
2211 Several "peers" sections may declare stick-tables with the same name.
2212 This is shorter version of the stick-table name which is sent over the network.
2213 There is only a '/' character as prefix to avoid stick-table name collisions between
2214 stick-tables declared as backends and stick-table declared in "peers" sections
2215 as follows in this weird but supported configuration:
2216
2217 peers mypeers
2218 peer A ...
2219 peer B ...
2220 table t1 type string size 10m store gpc0
2221
2222 backend t1
2223 stick-table type string size 10m store gpc0 peers mypeers
2224
2225 Here "t1" table declared in "mypeeers" section has "mypeers/t1" as global name.
2226 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2227 level the former table is named "/t1", the latter is again named "t1".
2228
Simon Horman51a1cf62015-02-03 13:00:44 +090022293.6. Mailers
2230------------
2231It is possible to send email alerts when the state of servers changes.
2232If configured email alerts are sent to each mailer that is configured
2233in a mailers section. Email is sent to mailers using SMTP.
2234
Pieter Baauw386a1272015-08-16 15:26:24 +02002235mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002236 Creates a new mailer list with the name <mailersect>. It is an
2237 independent section which is referenced by one or more proxies.
2238
2239mailer <mailername> <ip>:<port>
2240 Defines a mailer inside a mailers section.
2241
2242 Example:
2243 mailers mymailers
2244 mailer smtp1 192.168.0.1:587
2245 mailer smtp2 192.168.0.2:587
2246
2247 backend mybackend
2248 mode tcp
2249 balance roundrobin
2250
2251 email-alert mailers mymailers
2252 email-alert from test1@horms.org
2253 email-alert to test2@horms.org
2254
2255 server srv1 192.168.0.30:80
2256 server srv2 192.168.0.31:80
2257
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002258timeout mail <time>
2259 Defines the time available for a mail/connection to be made and send to
2260 the mail-server. If not defined the default value is 10 seconds. To allow
2261 for at least two SYN-ACK packets to be send during initial TCP handshake it
2262 is advised to keep this value above 4 seconds.
2263
2264 Example:
2265 mailers mymailers
2266 timeout mail 20s
2267 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002268
William Lallemandc9515522019-06-12 16:32:11 +020022693.7. Programs
2270-------------
2271In master-worker mode, it is possible to launch external binaries with the
2272master, these processes are called programs. These programs are launched and
2273managed the same way as the workers.
2274
2275During a reload of HAProxy, those processes are dealing with the same
2276sequence as a worker:
2277
2278 - the master is re-executed
2279 - the master sends a SIGUSR1 signal to the program
2280 - if "option start-on-reload" is not disabled, the master launches a new
2281 instance of the program
2282
2283During a stop, or restart, a SIGTERM is sent to the programs.
2284
2285program <name>
2286 This is a new program section, this section will create an instance <name>
2287 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2288 the management guide).
2289
2290command <command> [arguments*]
2291 Define the command to start with optional arguments. The command is looked
2292 up in the current PATH if it does not include an absolute path. This is a
2293 mandatory option of the program section. Arguments containing spaces must
2294 be enclosed in quotes or double quotes or be prefixed by a backslash.
2295
2296option start-on-reload
2297no option start-on-reload
2298 Start (or not) a new instance of the program upon a reload of the master.
2299 The default is to start a new instance. This option may only be used in a
2300 program section.
2301
2302
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023034. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002304----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002305
Willy Tarreau6a06a402007-07-15 20:15:28 +02002306Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002307 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002308 - frontend <name>
2309 - backend <name>
2310 - listen <name>
2311
2312A "defaults" section sets default parameters for all other sections following
2313its declaration. Those default parameters are reset by the next "defaults"
2314section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002315section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002316
2317A "frontend" section describes a set of listening sockets accepting client
2318connections.
2319
2320A "backend" section describes a set of servers to which the proxy will connect
2321to forward incoming connections.
2322
2323A "listen" section defines a complete proxy with its frontend and backend
2324parts combined in one section. It is generally useful for TCP-only traffic.
2325
Willy Tarreau0ba27502007-12-24 16:55:16 +01002326All proxy names must be formed from upper and lower case letters, digits,
2327'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2328case-sensitive, which means that "www" and "WWW" are two different proxies.
2329
2330Historically, all proxy names could overlap, it just caused troubles in the
2331logs. Since the introduction of content switching, it is mandatory that two
2332proxies with overlapping capabilities (frontend/backend) have different names.
2333However, it is still permitted that a frontend and a backend share the same
2334name, as this configuration seems to be commonly encountered.
2335
2336Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2337and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002338bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002339protocol, and can interact with it by allowing, blocking, switching, adding,
2340modifying, or removing arbitrary contents in requests or responses, based on
2341arbitrary criteria.
2342
Willy Tarreau70dffda2014-01-30 03:07:23 +01002343In HTTP mode, the processing applied to requests and responses flowing over
2344a connection depends in the combination of the frontend's HTTP options and
Julien Pivotto599788e2019-12-10 13:11:17 +01002345the backend's. HAProxy supports 3 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002346
2347 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2348 requests and responses are processed, and connections remain open but idle
2349 between responses and new requests.
2350
2351 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
2352 1.0 to 1.5-dev21 : only the first request and response are processed, and
2353 everything else is forwarded with no analysis at all. This mode should not
Christopher Faulet6c9bbb22019-03-26 21:37:23 +01002354 be used as it creates lots of trouble with logging and HTTP processing.
2355 And because it cannot work in HTTP/2, this option is deprecated and it is
2356 only supported on legacy HTTP frontends. In HTX, it is ignored and a
2357 warning is emitted during HAProxy startup.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002358
Willy Tarreau70dffda2014-01-30 03:07:23 +01002359 - SCL: server close ("option http-server-close") : the server-facing
2360 connection is closed after the end of the response is received, but the
2361 client-facing connection remains open.
2362
Christopher Faulet315b39c2018-09-21 16:26:19 +02002363 - CLO: close ("option httpclose"): the connection is closed after the end of
2364 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002365
2366The effective mode that will be applied to a connection passing through a
2367frontend and a backend can be determined by both proxy modes according to the
2368following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002369weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002370
Christopher Faulet315b39c2018-09-21 16:26:19 +02002371 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002372
Christopher Faulet315b39c2018-09-21 16:26:19 +02002373 | KAL | SCL | CLO
2374 ----+-----+-----+----
2375 KAL | KAL | SCL | CLO
2376 ----+-----+-----+----
2377 TUN | TUN | SCL | CLO
2378 Frontend ----+-----+-----+----
2379 mode SCL | SCL | SCL | CLO
2380 ----+-----+-----+----
2381 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002382
Willy Tarreau0ba27502007-12-24 16:55:16 +01002383
Willy Tarreau70dffda2014-01-30 03:07:23 +01002384
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023854.1. Proxy keywords matrix
2386--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002387
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002388The following list of keywords is supported. Most of them may only be used in a
2389limited set of section types. Some of them are marked as "deprecated" because
2390they are inherited from an old syntax which may be confusing or functionally
2391limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002392marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002393option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002394and must be disabled for a specific instance. Such options may also be prefixed
2395with "default" in order to restore default settings regardless of what has been
2396specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002397
Willy Tarreau6a06a402007-07-15 20:15:28 +02002398
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002399 keyword defaults frontend listen backend
2400------------------------------------+----------+----------+---------+---------
2401acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002402backlog X X X -
2403balance X - X X
2404bind - X X -
2405bind-process X X X X
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002406block (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002407capture cookie - X X -
2408capture request header - X X -
2409capture response header - X X -
2410clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002411compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002412contimeout (deprecated) X - X X
2413cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002414declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002415default-server X - X X
2416default_backend X X X -
2417description - X X X
2418disabled X X X X
2419dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002420email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002421email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002422email-alert mailers X X X X
2423email-alert myhostname X X X X
2424email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002425enabled X X X X
2426errorfile X X X X
2427errorloc X X X X
2428errorloc302 X X X X
2429-- keyword -------------------------- defaults - frontend - listen -- backend -
2430errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002431force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002432filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002433fullconn X - X X
2434grace X X X X
2435hash-type X - X X
2436http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002437http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002438http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002439http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002440http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002441http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002442http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002443id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002444ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002445load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002446log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002447log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002448log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002449log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002450max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002451maxconn X X X -
2452mode X X X X
2453monitor fail - X X -
2454monitor-net X X X -
2455monitor-uri X X X -
2456option abortonclose (*) X - X X
2457option accept-invalid-http-request (*) X X X -
2458option accept-invalid-http-response (*) X - X X
2459option allbackups (*) X - X X
2460option checkcache (*) X - X X
2461option clitcpka (*) X X X -
2462option contstats (*) X X X -
2463option dontlog-normal (*) X X X -
2464option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002465-- keyword -------------------------- defaults - frontend - listen -- backend -
2466option forwardfor X X X X
Christopher Fauleta99ff4d2019-07-22 16:18:24 +02002467option h1-case-adjust-bogus-client (*) X X X -
2468option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002469option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002470option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002471option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002472option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002473option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002474option http-server-close (*) X X X X
Christopher Faulet6c9bbb22019-03-26 21:37:23 +01002475option http-tunnel (deprecated) (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002476option http-use-proxy-header (*) X X X -
Willy Tarreau68ad3a42018-10-22 11:49:15 +02002477option http-use-htx (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002478option httpchk X - X X
2479option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01002480option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002481option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002482option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002483option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002484option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002485option log-health-checks (*) X - X X
2486option log-separate-errors (*) X X X -
2487option logasap (*) X X X -
2488option mysql-check X - X X
2489option nolinger (*) X X X X
2490option originalto X X X X
2491option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002492option pgsql-check X - X X
2493option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002494option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002495option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002496option smtpchk X - X X
2497option socket-stats (*) X X X -
2498option splice-auto (*) X X X X
2499option splice-request (*) X X X X
2500option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002501option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002502option srvtcpka (*) X - X X
2503option ssl-hello-chk X - X X
2504-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002505option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002506option tcp-smart-accept (*) X X X -
2507option tcp-smart-connect (*) X - X X
2508option tcpka X X X X
2509option tcplog X X X X
2510option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002511external-check command X - X X
2512external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002513persist rdp-cookie X - X X
2514rate-limit sessions X X X -
2515redirect - X X X
2516redisp (deprecated) X - X X
2517redispatch (deprecated) X - X X
Willy Tarreau96d51952019-05-22 20:34:35 +02002518reqadd (deprecated) - X X X
2519reqallow (deprecated) - X X X
2520reqdel (deprecated) - X X X
2521reqdeny (deprecated) - X X X
2522reqiallow (deprecated) - X X X
2523reqidel (deprecated) - X X X
2524reqideny (deprecated) - X X X
2525reqipass (deprecated) - X X X
2526reqirep (deprecated) - X X X
2527reqitarpit (deprecated) - X X X
2528reqpass (deprecated) - X X X
2529reqrep (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002530-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau96d51952019-05-22 20:34:35 +02002531reqtarpit (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002532retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02002533retry-on X - X X
Willy Tarreau96d51952019-05-22 20:34:35 +02002534rspadd (deprecated) - X X X
2535rspdel (deprecated) - X X X
2536rspdeny (deprecated) - X X X
2537rspidel (deprecated) - X X X
2538rspideny (deprecated) - X X X
2539rspirep (deprecated) - X X X
2540rsprep (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002541server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002542server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002543server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002544source X - X X
2545srvtimeout (deprecated) X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002546stats admin - X X X
2547stats auth X X X X
2548stats enable X X X X
2549stats hide-version X X X X
2550stats http-request - X X X
2551stats realm X X X X
2552stats refresh X X X X
2553stats scope X X X X
2554stats show-desc X X X X
2555stats show-legends X X X X
2556stats show-node X X X X
2557stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002558-- keyword -------------------------- defaults - frontend - listen -- backend -
2559stick match - - X X
2560stick on - - X X
2561stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002562stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002563stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002564tcp-check connect - - X X
2565tcp-check expect - - X X
2566tcp-check send - - X X
2567tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002568tcp-request connection - X X -
2569tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002570tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002571tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002572tcp-response content - - X X
2573tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002574timeout check X - X X
2575timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002576timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002577timeout clitimeout (deprecated) X X X -
2578timeout connect X - X X
2579timeout contimeout (deprecated) X - X X
2580timeout http-keep-alive X X X X
2581timeout http-request X X X X
2582timeout queue X - X X
2583timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002584timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002585timeout srvtimeout (deprecated) X - X X
2586timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002587timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002588transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002589unique-id-format X X X -
2590unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002591use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002592use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002593------------------------------------+----------+----------+---------+---------
2594 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002595
Willy Tarreau0ba27502007-12-24 16:55:16 +01002596
Willy Tarreauc57f0e22009-05-10 13:12:33 +020025974.2. Alphabetically sorted keywords reference
2598---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002599
2600This section provides a description of each keyword and its usage.
2601
2602
2603acl <aclname> <criterion> [flags] [operator] <value> ...
2604 Declare or complete an access list.
2605 May be used in sections : defaults | frontend | listen | backend
2606 no | yes | yes | yes
2607 Example:
2608 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2609 acl invalid_src src_port 0:1023
2610 acl local_dst hdr(host) -i localhost
2611
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002612 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002613
2614
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002615backlog <conns>
2616 Give hints to the system about the approximate listen backlog desired size
2617 May be used in sections : defaults | frontend | listen | backend
2618 yes | yes | yes | no
2619 Arguments :
2620 <conns> is the number of pending connections. Depending on the operating
2621 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002622 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002623
2624 In order to protect against SYN flood attacks, one solution is to increase
2625 the system's SYN backlog size. Depending on the system, sometimes it is just
2626 tunable via a system parameter, sometimes it is not adjustable at all, and
2627 sometimes the system relies on hints given by the application at the time of
2628 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2629 to the listen() syscall. On systems which can make use of this value, it can
2630 sometimes be useful to be able to specify a different value, hence this
2631 backlog parameter.
2632
2633 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2634 used as a hint and the system accepts up to the smallest greater power of
2635 two, and never more than some limits (usually 32768).
2636
2637 See also : "maxconn" and the target operating system's tuning guide.
2638
2639
Willy Tarreau0ba27502007-12-24 16:55:16 +01002640balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002641balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002642 Define the load balancing algorithm to be used in a backend.
2643 May be used in sections : defaults | frontend | listen | backend
2644 yes | no | yes | yes
2645 Arguments :
2646 <algorithm> is the algorithm used to select a server when doing load
2647 balancing. This only applies when no persistence information
2648 is available, or when a connection is redispatched to another
2649 server. <algorithm> may be one of the following :
2650
2651 roundrobin Each server is used in turns, according to their weights.
2652 This is the smoothest and fairest algorithm when the server's
2653 processing time remains equally distributed. This algorithm
2654 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002655 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002656 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002657 large farms, when a server becomes up after having been down
2658 for a very short time, it may sometimes take a few hundreds
2659 requests for it to be re-integrated into the farm and start
2660 receiving traffic. This is normal, though very rare. It is
2661 indicated here in case you would have the chance to observe
2662 it, so that you don't worry.
2663
2664 static-rr Each server is used in turns, according to their weights.
2665 This algorithm is as similar to roundrobin except that it is
2666 static, which means that changing a server's weight on the
2667 fly will have no effect. On the other hand, it has no design
2668 limitation on the number of servers, and when a server goes
2669 up, it is always immediately reintroduced into the farm, once
2670 the full map is recomputed. It also uses slightly less CPU to
2671 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002672
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002673 leastconn The server with the lowest number of connections receives the
2674 connection. Round-robin is performed within groups of servers
2675 of the same load to ensure that all servers will be used. Use
2676 of this algorithm is recommended where very long sessions are
2677 expected, such as LDAP, SQL, TSE, etc... but is not very well
2678 suited for protocols using short sessions such as HTTP. This
2679 algorithm is dynamic, which means that server weights may be
2680 adjusted on the fly for slow starts for instance.
2681
Willy Tarreauf09c6602012-02-13 17:12:08 +01002682 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002683 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002684 identifier to the highest (see server parameter "id"), which
2685 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002686 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002687 not make sense to use this algorithm without setting maxconn.
2688 The purpose of this algorithm is to always use the smallest
2689 number of servers so that extra servers can be powered off
2690 during non-intensive hours. This algorithm ignores the server
2691 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002692 or IMAP than HTTP, though it can be useful there too. In
2693 order to use this algorithm efficiently, it is recommended
2694 that a cloud controller regularly checks server usage to turn
2695 them off when unused, and regularly checks backend queue to
2696 turn new servers on when the queue inflates. Alternatively,
2697 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002698
Willy Tarreau0ba27502007-12-24 16:55:16 +01002699 source The source IP address is hashed and divided by the total
2700 weight of the running servers to designate which server will
2701 receive the request. This ensures that the same client IP
2702 address will always reach the same server as long as no
2703 server goes down or up. If the hash result changes due to the
2704 number of running servers changing, many clients will be
2705 directed to a different server. This algorithm is generally
2706 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002707 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002708 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002709 static by default, which means that changing a server's
2710 weight on the fly will have no effect, but this can be
2711 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002712
Oskar Stolc8dc41842012-05-19 10:19:54 +01002713 uri This algorithm hashes either the left part of the URI (before
2714 the question mark) or the whole URI (if the "whole" parameter
2715 is present) and divides the hash value by the total weight of
2716 the running servers. The result designates which server will
2717 receive the request. This ensures that the same URI will
2718 always be directed to the same server as long as no server
2719 goes up or down. This is used with proxy caches and
2720 anti-virus proxies in order to maximize the cache hit rate.
2721 Note that this algorithm may only be used in an HTTP backend.
2722 This algorithm is static by default, which means that
2723 changing a server's weight on the fly will have no effect,
2724 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002725
Oskar Stolc8dc41842012-05-19 10:19:54 +01002726 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002727 "depth", both followed by a positive integer number. These
2728 options may be helpful when it is needed to balance servers
2729 based on the beginning of the URI only. The "len" parameter
2730 indicates that the algorithm should only consider that many
2731 characters at the beginning of the URI to compute the hash.
2732 Note that having "len" set to 1 rarely makes sense since most
2733 URIs start with a leading "/".
2734
2735 The "depth" parameter indicates the maximum directory depth
2736 to be used to compute the hash. One level is counted for each
2737 slash in the request. If both parameters are specified, the
2738 evaluation stops when either is reached.
2739
Willy Tarreau0ba27502007-12-24 16:55:16 +01002740 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002741 the query string of each HTTP GET request.
2742
2743 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002744 request entity will be searched for the parameter argument,
2745 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002746 ('?') in the URL. The message body will only start to be
2747 analyzed once either the advertised amount of data has been
2748 received or the request buffer is full. In the unlikely event
2749 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002750 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002751 be randomly balanced if at all. This keyword used to support
2752 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002753
2754 If the parameter is found followed by an equal sign ('=') and
2755 a value, then the value is hashed and divided by the total
2756 weight of the running servers. The result designates which
2757 server will receive the request.
2758
2759 This is used to track user identifiers in requests and ensure
2760 that a same user ID will always be sent to the same server as
2761 long as no server goes up or down. If no value is found or if
2762 the parameter is not found, then a round robin algorithm is
2763 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002764 backend. This algorithm is static by default, which means
2765 that changing a server's weight on the fly will have no
2766 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002767
Cyril Bontédc4d9032012-04-08 21:57:39 +02002768 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2769 request. Just as with the equivalent ACL 'hdr()' function,
2770 the header name in parenthesis is not case sensitive. If the
2771 header is absent or if it does not contain any value, the
2772 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002773
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002774 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002775 reducing the hash algorithm to the main domain part with some
2776 specific headers such as 'Host'. For instance, in the Host
2777 value "haproxy.1wt.eu", only "1wt" will be considered.
2778
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002779 This algorithm is static by default, which means that
2780 changing a server's weight on the fly will have no effect,
2781 but this can be changed using "hash-type".
2782
Willy Tarreau21c741a2019-01-14 18:14:27 +01002783 random
2784 random(<draws>)
2785 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02002786 hashing function. This means that the servers' weights are
2787 respected, dynamic weight changes immediately take effect, as
2788 well as new server additions. Random load balancing can be
2789 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01002790 or removed as it may avoid the hammering effect that could
2791 result from roundrobin or leastconn in this situation. The
2792 hash-balance-factor directive can be used to further improve
2793 fairness of the load balancing, especially in situations
2794 where servers show highly variable response times. When an
2795 argument <draws> is present, it must be an integer value one
2796 or greater, indicating the number of draws before selecting
2797 the least loaded of these servers. It was indeed demonstrated
2798 that picking the least loaded of two servers is enough to
2799 significantly improve the fairness of the algorithm, by
2800 always avoiding to pick the most loaded server within a farm
2801 and getting rid of any bias that could be induced by the
2802 unfair distribution of the consistent list. Higher values N
2803 will take away N-1 of the highest loaded servers at the
2804 expense of performance. With very high values, the algorithm
2805 will converge towards the leastconn's result but much slower.
2806 The default value is 2, which generally shows very good
2807 distribution and performance. This algorithm is also known as
2808 the Power of Two Random Choices and is described here :
2809 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02002810
Emeric Brun736aa232009-06-30 17:56:00 +02002811 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002812 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002813 The RDP cookie <name> (or "mstshash" if omitted) will be
2814 looked up and hashed for each incoming TCP request. Just as
2815 with the equivalent ACL 'req_rdp_cookie()' function, the name
2816 is not case-sensitive. This mechanism is useful as a degraded
2817 persistence mode, as it makes it possible to always send the
2818 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002819 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002820 used instead.
2821
2822 Note that for this to work, the frontend must ensure that an
2823 RDP cookie is already present in the request buffer. For this
2824 you must use 'tcp-request content accept' rule combined with
2825 a 'req_rdp_cookie_cnt' ACL.
2826
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002827 This algorithm is static by default, which means that
2828 changing a server's weight on the fly will have no effect,
2829 but this can be changed using "hash-type".
2830
Cyril Bontédc4d9032012-04-08 21:57:39 +02002831 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002832
Willy Tarreau0ba27502007-12-24 16:55:16 +01002833 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002834 algorithms. Right now, only "url_param" and "uri" support an
2835 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002836
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002837 The load balancing algorithm of a backend is set to roundrobin when no other
2838 algorithm, mode nor option have been set. The algorithm may only be set once
2839 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002840
Lukas Tribus80512b12018-10-27 20:07:40 +02002841 With authentication schemes that require the same connection like NTLM, URI
John Roesler27c754c2019-07-10 15:45:51 -05002842 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02002843 to be routed to different backend servers, breaking the invalid assumptions
2844 NTLM relies on.
2845
Willy Tarreau0ba27502007-12-24 16:55:16 +01002846 Examples :
2847 balance roundrobin
2848 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002849 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002850 balance hdr(User-Agent)
2851 balance hdr(host)
2852 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002853
2854 Note: the following caveats and limitations on using the "check_post"
2855 extension with "url_param" must be considered :
2856
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002857 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002858 to determine if the parameters will be found in the body or entity which
2859 may contain binary data. Therefore another method may be required to
2860 restrict consideration of POST requests that have no URL parameters in
2861 the body. (see acl reqideny http_end)
2862
2863 - using a <max_wait> value larger than the request buffer size does not
2864 make sense and is useless. The buffer size is set at build time, and
2865 defaults to 16 kB.
2866
2867 - Content-Encoding is not supported, the parameter search will probably
2868 fail; and load balancing will fall back to Round Robin.
2869
2870 - Expect: 100-continue is not supported, load balancing will fall back to
2871 Round Robin.
2872
Lukas Tribus23953682017-04-28 13:24:30 +00002873 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002874 If the entire parameter value is not present in the first chunk, the
2875 selection of server is undefined (actually, defined by how little
2876 actually appeared in the first chunk).
2877
2878 - This feature does not support generation of a 100, 411 or 501 response.
2879
2880 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002881 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002882 white space or control characters are found, indicating the end of what
2883 might be a URL parameter list. This is probably not a concern with SGML
2884 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002885
Willy Tarreau294d0f02015-08-10 19:40:12 +02002886 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002887
2888
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002889bind [<address>]:<port_range> [, ...] [param*]
2890bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002891 Define one or several listening addresses and/or ports in a frontend.
2892 May be used in sections : defaults | frontend | listen | backend
2893 no | yes | yes | no
2894 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002895 <address> is optional and can be a host name, an IPv4 address, an IPv6
2896 address, or '*'. It designates the address the frontend will
2897 listen on. If unset, all IPv4 addresses of the system will be
2898 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002899 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002900 Optionally, an address family prefix may be used before the
2901 address to force the family regardless of the address format,
2902 which can be useful to specify a path to a unix socket with
2903 no slash ('/'). Currently supported prefixes are :
2904 - 'ipv4@' -> address is always IPv4
2905 - 'ipv6@' -> address is always IPv6
2906 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002907 - 'abns@' -> address is in abstract namespace (Linux only).
2908 Note: since abstract sockets are not "rebindable", they
2909 do not cope well with multi-process mode during
2910 soft-restart, so it is better to avoid them if
2911 nbproc is greater than 1. The effect is that if the
2912 new process fails to start, only one of the old ones
2913 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002914 - 'fd@<n>' -> use file descriptor <n> inherited from the
2915 parent. The fd must be bound and may or may not already
2916 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002917 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2918 connected unix socket or of a socketpair. The bind waits
2919 to receive a FD over the unix socket and uses it as if it
2920 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02002921 You may want to reference some environment variables in the
2922 address parameter, see section 2.3 about environment
2923 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002924
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002925 <port_range> is either a unique TCP port, or a port range for which the
2926 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002927 above. The port is mandatory for TCP listeners. Note that in
2928 the case of an IPv6 address, the port is always the number
2929 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002930 - a numerical port (ex: '80')
2931 - a dash-delimited ports range explicitly stating the lower
2932 and upper bounds (ex: '2000-2100') which are included in
2933 the range.
2934
2935 Particular care must be taken against port ranges, because
2936 every <address:port> couple consumes one socket (= a file
2937 descriptor), so it's easy to consume lots of descriptors
2938 with a simple range, and to run out of sockets. Also, each
2939 <address:port> couple must be used only once among all
2940 instances running on a same system. Please note that binding
2941 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002942 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002943 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002944
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002945 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002946 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002947 receive UNIX connections on the socket located at this place.
2948 The path must begin with a slash and by default is absolute.
2949 It can be relative to the prefix defined by "unix-bind" in
2950 the global section. Note that the total length of the prefix
2951 followed by the socket path cannot exceed some system limits
2952 for UNIX sockets, which commonly are set to 107 characters.
2953
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002954 <param*> is a list of parameters common to all sockets declared on the
2955 same line. These numerous parameters depend on OS and build
2956 options and have a complete section dedicated to them. Please
2957 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002958
Willy Tarreau0ba27502007-12-24 16:55:16 +01002959 It is possible to specify a list of address:port combinations delimited by
2960 commas. The frontend will then listen on all of these addresses. There is no
2961 fixed limit to the number of addresses and ports which can be listened on in
2962 a frontend, as well as there is no limit to the number of "bind" statements
2963 in a frontend.
2964
2965 Example :
2966 listen http_proxy
2967 bind :80,:443
2968 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002969 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002970
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002971 listen http_https_proxy
2972 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002973 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002974
Willy Tarreau24709282013-03-10 21:32:12 +01002975 listen http_https_proxy_explicit
2976 bind ipv6@:80
2977 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2978 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2979
Willy Tarreaudad36a32013-03-11 01:20:04 +01002980 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002981 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002982
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002983 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2984 sun_path length is used for the address length. Some other programs
2985 such as socat use the string length only by default. Pass the option
2986 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2987 make it compatible with HAProxy's.
2988
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002989 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002990 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002991
2992
Christopher Fauletff4121f2017-11-22 16:38:49 +01002993bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002994 Limit visibility of an instance to a certain set of processes numbers.
2995 May be used in sections : defaults | frontend | listen | backend
2996 yes | yes | yes | yes
2997 Arguments :
2998 all All process will see this instance. This is the default. It
2999 may be used to override a default value.
3000
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003001 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003002 option may be combined with other numbers.
3003
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003004 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003005 option may be combined with other numbers. Do not use it
3006 with less than 2 processes otherwise some instances might be
3007 missing from all processes.
3008
Christopher Fauletff4121f2017-11-22 16:38:49 +01003009 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003010 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01003011 the machine's word size. Ranges can be partially defined. The
3012 higher bound can be omitted. In such case, it is replaced by
3013 the corresponding maximum value. If a proxy is bound to
3014 process numbers greater than the configured global.nbproc, it
3015 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003016 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003017
3018 This keyword limits binding of certain instances to certain processes. This
3019 is useful in order not to have too many processes listening to the same
3020 ports. For instance, on a dual-core machine, it might make sense to set
3021 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3022 and 'even' instances.
3023
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003024 At the moment, it is not possible to reference more than 32 or 64 processes
3025 using this keyword, but this should be more than enough for most setups.
3026 Please note that 'all' really means all processes regardless of the machine's
3027 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003028
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003029 Each "bind" line may further be limited to a subset of the proxy's processes,
3030 please consult the "process" bind keyword in section 5.1.
3031
Willy Tarreaub369a042014-09-16 13:21:03 +02003032 When a frontend has no explicit "bind-process" line, it tries to bind to all
3033 the processes referenced by its "bind" lines. That means that frontends can
3034 easily adapt to their listeners' processes.
3035
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003036 If some backends are referenced by frontends bound to other processes, the
3037 backend automatically inherits the frontend's processes.
3038
3039 Example :
3040 listen app_ip1
3041 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003042 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003043
3044 listen app_ip2
3045 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003046 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003047
3048 listen management
3049 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003050 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003051
Willy Tarreau110ecc12012-11-15 17:50:01 +01003052 listen management
3053 bind 10.0.0.4:80
3054 bind-process 1-4
3055
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003056 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003057
3058
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02003059block { if | unless } <condition> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003060 Block a layer 7 request if/unless a condition is matched
3061 May be used in sections : defaults | frontend | listen | backend
3062 no | yes | yes | yes
3063
3064 The HTTP request will be blocked very early in the layer 7 processing
3065 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003066 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02003067 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01003068 conditions are met or not met. There is no fixed limit to the number of
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03003069 "block" statements per instance. To block connections at layer 4 (without
3070 sending a 403 error) see "tcp-request connection reject" and
3071 "tcp-request content reject" rules.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003072
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02003073 This form is deprecated, do not use it in any new configuration, use the new
3074 "http-request deny" instead.
3075
Willy Tarreau0ba27502007-12-24 16:55:16 +01003076 Example:
3077 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
3078 acl invalid_src src_port 0:1023
3079 acl local_dst hdr(host) -i localhost
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +03003080 # block is deprecated. Use http-request deny instead:
3081 #block if invalid_src || local_dst
3082 http-request deny if invalid_src || local_dst
Willy Tarreau0ba27502007-12-24 16:55:16 +01003083
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03003084 See also : section 7 about ACL usage, "http-request deny",
3085 "http-response deny", "tcp-request connection reject" and
3086 "tcp-request content reject".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003087
3088capture cookie <name> len <length>
3089 Capture and log a cookie in the request and in the response.
3090 May be used in sections : defaults | frontend | listen | backend
3091 no | yes | yes | no
3092 Arguments :
3093 <name> is the beginning of the name of the cookie to capture. In order
3094 to match the exact name, simply suffix the name with an equal
3095 sign ('='). The full name will appear in the logs, which is
3096 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003097 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003098
3099 <length> is the maximum number of characters to report in the logs, which
3100 include the cookie name, the equal sign and the value, all in the
3101 standard "name=value" form. The string will be truncated on the
3102 right if it exceeds <length>.
3103
3104 Only the first cookie is captured. Both the "cookie" request headers and the
3105 "set-cookie" response headers are monitored. This is particularly useful to
3106 check for application bugs causing session crossing or stealing between
3107 users, because generally the user's cookies can only change on a login page.
3108
3109 When the cookie was not presented by the client, the associated log column
3110 will report "-". When a request does not cause a cookie to be assigned by the
3111 server, a "-" is reported in the response column.
3112
3113 The capture is performed in the frontend only because it is necessary that
3114 the log format does not change for a given frontend depending on the
3115 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003116 "capture cookie" statement in a frontend. The maximum capture length is set
3117 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3118 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003119
3120 Example:
3121 capture cookie ASPSESSION len 32
3122
3123 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003124 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003125
3126
3127capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003128 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003129 May be used in sections : defaults | frontend | listen | backend
3130 no | yes | yes | no
3131 Arguments :
3132 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003133 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003134 appear in the requests, with the first letter of each word in
3135 upper case. The header name will not appear in the logs, only the
3136 value is reported, but the position in the logs is respected.
3137
3138 <length> is the maximum number of characters to extract from the value and
3139 report in the logs. The string will be truncated on the right if
3140 it exceeds <length>.
3141
Willy Tarreau4460d032012-11-21 23:37:37 +01003142 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003143 value will be added to the logs between braces ('{}'). If multiple headers
3144 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003145 in the same order they were declared in the configuration. Non-existent
3146 headers will be logged just as an empty string. Common uses for request
3147 header captures include the "Host" field in virtual hosting environments, the
3148 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003149 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003150 environments to find where the request came from.
3151
3152 Note that when capturing headers such as "User-agent", some spaces may be
3153 logged, making the log analysis more difficult. Thus be careful about what
3154 you log if you know your log parser is not smart enough to rely on the
3155 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003156
Willy Tarreau0900abb2012-11-22 00:21:46 +01003157 There is no limit to the number of captured request headers nor to their
3158 length, though it is wise to keep them low to limit memory usage per session.
3159 In order to keep log format consistent for a same frontend, header captures
3160 can only be declared in a frontend. It is not possible to specify a capture
3161 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003162
3163 Example:
3164 capture request header Host len 15
3165 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003166 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003167
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003168 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003169 about logging.
3170
3171
3172capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003173 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003174 May be used in sections : defaults | frontend | listen | backend
3175 no | yes | yes | no
3176 Arguments :
3177 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003178 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003179 appear in the response, with the first letter of each word in
3180 upper case. The header name will not appear in the logs, only the
3181 value is reported, but the position in the logs is respected.
3182
3183 <length> is the maximum number of characters to extract from the value and
3184 report in the logs. The string will be truncated on the right if
3185 it exceeds <length>.
3186
Willy Tarreau4460d032012-11-21 23:37:37 +01003187 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003188 result will be added to the logs between braces ('{}') after the captured
3189 request headers. If multiple headers are captured, they will be delimited by
3190 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003191 the configuration. Non-existent headers will be logged just as an empty
3192 string. Common uses for response header captures include the "Content-length"
3193 header which indicates how many bytes are expected to be returned, the
3194 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003195
Willy Tarreau0900abb2012-11-22 00:21:46 +01003196 There is no limit to the number of captured response headers nor to their
3197 length, though it is wise to keep them low to limit memory usage per session.
3198 In order to keep log format consistent for a same frontend, header captures
3199 can only be declared in a frontend. It is not possible to specify a capture
3200 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003201
3202 Example:
3203 capture response header Content-length len 9
3204 capture response header Location len 15
3205
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003206 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003207 about logging.
3208
3209
Cyril Bontéf0c60612010-02-06 14:44:47 +01003210clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003211 Set the maximum inactivity time on the client side.
3212 May be used in sections : defaults | frontend | listen | backend
3213 yes | yes | yes | no
3214 Arguments :
3215 <timeout> is the timeout value is specified in milliseconds by default, but
3216 can be in any other unit if the number is suffixed by the unit,
3217 as explained at the top of this document.
3218
3219 The inactivity timeout applies when the client is expected to acknowledge or
3220 send data. In HTTP mode, this timeout is particularly important to consider
3221 during the first phase, when the client sends the request, and during the
3222 response while it is reading data sent by the server. The value is specified
3223 in milliseconds by default, but can be in any other unit if the number is
3224 suffixed by the unit, as specified at the top of this document. In TCP mode
3225 (and to a lesser extent, in HTTP mode), it is highly recommended that the
3226 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003227 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01003228 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +01003229 (e.g. 4 or 5 seconds).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003230
3231 This parameter is specific to frontends, but can be specified once for all in
3232 "defaults" sections. This is in fact one of the easiest solutions not to
3233 forget about it. An unspecified timeout results in an infinite timeout, which
3234 is not recommended. Such a usage is accepted and works but reports a warning
3235 during startup because it may results in accumulation of expired sessions in
3236 the system if the system's timeouts are not configured either.
3237
3238 This parameter is provided for compatibility but is currently deprecated.
3239 Please use "timeout client" instead.
3240
Willy Tarreau036fae02008-01-06 13:24:40 +01003241 See also : "timeout client", "timeout http-request", "timeout server", and
3242 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003243
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003244compression algo <algorithm> ...
3245compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02003246compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02003247 Enable HTTP compression.
3248 May be used in sections : defaults | frontend | listen | backend
3249 yes | yes | yes | yes
3250 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003251 algo is followed by the list of supported compression algorithms.
3252 type is followed by the list of MIME types that will be compressed.
3253 offload makes haproxy work as a compression offloader only (see notes).
3254
3255 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01003256 identity this is mostly for debugging, and it was useful for developing
3257 the compression feature. Identity does not apply any change on
3258 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003259
Willy Tarreauc91840a2015-03-28 17:00:39 +01003260 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003261 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003262
3263 deflate same as "gzip", but with deflate algorithm and zlib format.
3264 Note that this algorithm has ambiguous support on many
3265 browsers and no support at all from recent ones. It is
3266 strongly recommended not to use it for anything else than
3267 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003268 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003269
Willy Tarreauc91840a2015-03-28 17:00:39 +01003270 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3271 alternative when the browser wants "deflate". All major
3272 browsers understand it and despite violating the standards,
3273 it is known to work better than "deflate", at least on MSIE
3274 and some versions of Safari. Do not use it in conjunction
3275 with "deflate", use either one or the other since both react
3276 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003277 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003278
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003279 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003280 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003281 If backend servers support HTTP compression, these directives
3282 will be no-op: haproxy will see the compressed response and will not
3283 compress again. If backend servers do not support HTTP compression and
3284 there is Accept-Encoding header in request, haproxy will compress the
3285 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003286
3287 The "offload" setting makes haproxy remove the Accept-Encoding header to
3288 prevent backend servers from compressing responses. It is strongly
3289 recommended not to do this because this means that all the compression work
3290 will be done on the single point where haproxy is located. However in some
3291 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003292 with broken HTTP compression implementation which can't be turned off.
3293 In that case haproxy can be used to prevent that gateway from emitting
3294 invalid payloads. In this case, simply removing the header in the
3295 configuration does not work because it applies before the header is parsed,
3296 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003297 then be used for such scenarios. Note: for now, the "offload" setting is
3298 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003299
William Lallemand05097442012-11-20 12:14:28 +01003300 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003301 * the request does not advertise a supported compression algorithm in the
3302 "Accept-Encoding" header
3303 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003304 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003305 * response contain neither a "Content-Length" header nor a
3306 "Transfer-Encoding" whose last value is "chunked"
3307 * response contains a "Content-Type" header whose first value starts with
3308 "multipart"
3309 * the response contains the "no-transform" value in the "Cache-control"
3310 header
3311 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3312 and later
3313 * The response contains a "Content-Encoding" header, indicating that the
3314 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003315 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01003316
Tim Duesterhusb229f012019-01-29 16:38:56 +01003317 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003318
William Lallemand82fe75c2012-10-23 10:25:10 +02003319 Examples :
3320 compression algo gzip
3321 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003322
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003323
Cyril Bontéf0c60612010-02-06 14:44:47 +01003324contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003325 Set the maximum time to wait for a connection attempt to a server to succeed.
3326 May be used in sections : defaults | frontend | listen | backend
3327 yes | no | yes | yes
3328 Arguments :
3329 <timeout> is the timeout value is specified in milliseconds by default, but
3330 can be in any other unit if the number is suffixed by the unit,
3331 as explained at the top of this document.
3332
3333 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003334 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01003335 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +01003336 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003337 connect timeout also presets the queue timeout to the same value if this one
3338 has not been specified. Historically, the contimeout was also used to set the
3339 tarpit timeout in a listen section, which is not possible in a pure frontend.
3340
3341 This parameter is specific to backends, but can be specified once for all in
3342 "defaults" sections. This is in fact one of the easiest solutions not to
3343 forget about it. An unspecified timeout results in an infinite timeout, which
3344 is not recommended. Such a usage is accepted and works but reports a warning
3345 during startup because it may results in accumulation of failed sessions in
3346 the system if the system's timeouts are not configured either.
3347
3348 This parameter is provided for backwards compatibility but is currently
3349 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
3350 instead.
3351
3352 See also : "timeout connect", "timeout queue", "timeout tarpit",
3353 "timeout server", "contimeout".
3354
3355
Willy Tarreau55165fe2009-05-10 12:02:55 +02003356cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003357 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3358 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Christopher Fauletdb2cdbb2020-01-21 11:06:48 +01003359 [ dynamic ] [ attr <value> ]*
Willy Tarreau0ba27502007-12-24 16:55:16 +01003360 Enable cookie-based persistence in a backend.
3361 May be used in sections : defaults | frontend | listen | backend
3362 yes | no | yes | yes
3363 Arguments :
3364 <name> is the name of the cookie which will be monitored, modified or
3365 inserted in order to bring persistence. This cookie is sent to
3366 the client via a "Set-Cookie" header in the response, and is
3367 brought back by the client in a "Cookie" header in all requests.
3368 Special care should be taken to choose a name which does not
3369 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003370 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003371 HTTP/HTTPS), care should be taken to use different cookie names
3372 between all backends if persistence between them is not desired.
3373
3374 rewrite This keyword indicates that the cookie will be provided by the
3375 server and that haproxy will have to modify its value to set the
3376 server's identifier in it. This mode is handy when the management
3377 of complex combinations of "Set-cookie" and "Cache-control"
3378 headers is left to the application. The application can then
3379 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003380 cookie. Since all responses should be monitored, this mode
3381 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003382 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003383 start with this mode for new deployments. This keyword is
3384 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003385
3386 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003387 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003388
Willy Tarreaua79094d2010-08-31 22:54:15 +02003389 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003390 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02003391 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003392 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003393 existing configurations running in the "rewrite" mode. The cookie
3394 will only be a session cookie and will not be stored on the
3395 client's disk. By default, unless the "indirect" option is added,
3396 the server will see the cookies emitted by the client. Due to
3397 caching effects, it is generally wise to add the "nocache" or
3398 "postonly" keywords (see below). The "insert" keyword is not
3399 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003400
3401 prefix This keyword indicates that instead of relying on a dedicated
3402 cookie for the persistence, an existing one will be completed.
3403 This may be needed in some specific environments where the client
3404 does not support more than one single cookie and the application
3405 already needs it. In this case, whenever the server sets a cookie
3406 named <name>, it will be prefixed with the server's identifier
3407 and a delimiter. The prefix will be removed from all client
3408 requests so that the server still finds the cookie it emitted.
3409 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003410 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003411 not compatible with "rewrite" and "insert". Note: it is highly
3412 recommended not to use "indirect" with "prefix", otherwise server
3413 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003414
Willy Tarreaua79094d2010-08-31 22:54:15 +02003415 indirect When this option is specified, no cookie will be emitted to a
3416 client which already has a valid one for the server which has
3417 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003418 it will be removed, unless the "preserve" option is also set. In
3419 "insert" mode, this will additionally remove cookies from the
3420 requests transmitted to the server, making the persistence
3421 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003422 Note: it is highly recommended not to use "indirect" with
3423 "prefix", otherwise server cookie updates would not be sent to
3424 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003425
3426 nocache This option is recommended in conjunction with the insert mode
3427 when there is a cache between the client and HAProxy, as it
3428 ensures that a cacheable response will be tagged non-cacheable if
3429 a cookie needs to be inserted. This is important because if all
3430 persistence cookies are added on a cacheable home page for
3431 instance, then all customers will then fetch the page from an
3432 outer cache and will all share the same persistence cookie,
3433 leading to one server receiving much more traffic than others.
3434 See also the "insert" and "postonly" options.
3435
3436 postonly This option ensures that cookie insertion will only be performed
3437 on responses to POST requests. It is an alternative to the
3438 "nocache" option, because POST responses are not cacheable, so
3439 this ensures that the persistence cookie will never get cached.
3440 Since most sites do not need any sort of persistence before the
3441 first POST which generally is a login request, this is a very
3442 efficient method to optimize caching without risking to find a
3443 persistence cookie in the cache.
3444 See also the "insert" and "nocache" options.
3445
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003446 preserve This option may only be used with "insert" and/or "indirect". It
3447 allows the server to emit the persistence cookie itself. In this
3448 case, if a cookie is found in the response, haproxy will leave it
3449 untouched. This is useful in order to end persistence after a
3450 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003451 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003452 the past. By combining this mechanism with the "disable-on-404"
3453 check option, it is possible to perform a completely graceful
3454 shutdown because users will definitely leave the server after
3455 they logout.
3456
Willy Tarreau4992dd22012-05-31 21:02:17 +02003457 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3458 when a cookie is inserted. This attribute is used so that a
3459 user agent doesn't share the cookie with non-HTTP components.
3460 Please check RFC6265 for more information on this attribute.
3461
3462 secure This option tells haproxy to add a "Secure" cookie attribute when
3463 a cookie is inserted. This attribute is used so that a user agent
3464 never emits this cookie over non-secure channels, which means
3465 that a cookie learned with this flag will be presented only over
3466 SSL/TLS connections. Please check RFC6265 for more information on
3467 this attribute.
3468
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003469 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003470 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003471 name. If the domain begins with a dot, the browser is allowed to
3472 use it for any host ending with that name. It is also possible to
3473 specify several domain names by invoking this option multiple
3474 times. Some browsers might have small limits on the number of
3475 domains, so be careful when doing that. For the record, sending
3476 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003477
Willy Tarreau996a92c2010-10-13 19:30:47 +02003478 maxidle This option allows inserted cookies to be ignored after some idle
3479 time. It only works with insert-mode cookies. When a cookie is
3480 sent to the client, the date this cookie was emitted is sent too.
3481 Upon further presentations of this cookie, if the date is older
3482 than the delay indicated by the parameter (in seconds), it will
3483 be ignored. Otherwise, it will be refreshed if needed when the
3484 response is sent to the client. This is particularly useful to
3485 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003486 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003487 this option is set and a cookie has no date, it is always
3488 accepted, but gets refreshed in the response. This maintains the
3489 ability for admins to access their sites. Cookies that have a
3490 date in the future further than 24 hours are ignored. Doing so
3491 lets admins fix timezone issues without risking kicking users off
3492 the site.
3493
3494 maxlife This option allows inserted cookies to be ignored after some life
3495 time, whether they're in use or not. It only works with insert
3496 mode cookies. When a cookie is first sent to the client, the date
3497 this cookie was emitted is sent too. Upon further presentations
3498 of this cookie, if the date is older than the delay indicated by
3499 the parameter (in seconds), it will be ignored. If the cookie in
3500 the request has no date, it is accepted and a date will be set.
3501 Cookies that have a date in the future further than 24 hours are
3502 ignored. Doing so lets admins fix timezone issues without risking
3503 kicking users off the site. Contrary to maxidle, this value is
3504 not refreshed, only the first visit date counts. Both maxidle and
3505 maxlife may be used at the time. This is particularly useful to
3506 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003507 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003508 is stronger than the maxidle method in that it forces a
3509 redispatch after some absolute delay.
3510
Olivier Houchard4e694042017-03-14 20:01:29 +01003511 dynamic Activate dynamic cookies. When used, a session cookie is
3512 dynamically created for each server, based on the IP and port
3513 of the server, and a secret key, specified in the
3514 "dynamic-cookie-key" backend directive.
3515 The cookie will be regenerated each time the IP address change,
3516 and is only generated for IPv4/IPv6.
3517
Christopher Fauletdb2cdbb2020-01-21 11:06:48 +01003518 attr This option tells haproxy to add an extra attribute when a
3519 cookie is inserted. The attribute value can contain any
3520 characters except control ones or ";". This option may be
3521 repeated.
3522
Willy Tarreau0ba27502007-12-24 16:55:16 +01003523 There can be only one persistence cookie per HTTP backend, and it can be
3524 declared in a defaults section. The value of the cookie will be the value
3525 indicated after the "cookie" keyword in a "server" statement. If no cookie
3526 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003527
Willy Tarreau0ba27502007-12-24 16:55:16 +01003528 Examples :
3529 cookie JSESSIONID prefix
3530 cookie SRV insert indirect nocache
3531 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003532 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003533
Willy Tarreau294d0f02015-08-10 19:40:12 +02003534 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003535
Willy Tarreau983e01e2010-01-11 18:42:06 +01003536
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003537declare capture [ request | response ] len <length>
3538 Declares a capture slot.
3539 May be used in sections : defaults | frontend | listen | backend
3540 no | yes | yes | no
3541 Arguments:
3542 <length> is the length allowed for the capture.
3543
3544 This declaration is only available in the frontend or listen section, but the
3545 reserved slot can be used in the backends. The "request" keyword allocates a
3546 capture slot for use in the request, and "response" allocates a capture slot
3547 for use in the response.
3548
3549 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003550 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003551 "http-request capture" and "http-response capture".
3552
3553
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003554default-server [param*]
3555 Change default options for a server in a backend
3556 May be used in sections : defaults | frontend | listen | backend
3557 yes | no | yes | yes
3558 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003559 <param*> is a list of parameters for this server. The "default-server"
3560 keyword accepts an important number of options and has a complete
3561 section dedicated to it. Please refer to section 5 for more
3562 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003563
Willy Tarreau983e01e2010-01-11 18:42:06 +01003564 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003565 default-server inter 1000 weight 13
3566
3567 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003568
Willy Tarreau983e01e2010-01-11 18:42:06 +01003569
Willy Tarreau0ba27502007-12-24 16:55:16 +01003570default_backend <backend>
3571 Specify the backend to use when no "use_backend" rule has been matched.
3572 May be used in sections : defaults | frontend | listen | backend
3573 yes | yes | yes | no
3574 Arguments :
3575 <backend> is the name of the backend to use.
3576
3577 When doing content-switching between frontend and backends using the
3578 "use_backend" keyword, it is often useful to indicate which backend will be
3579 used when no rule has matched. It generally is the dynamic backend which
3580 will catch all undetermined requests.
3581
Willy Tarreau0ba27502007-12-24 16:55:16 +01003582 Example :
3583
3584 use_backend dynamic if url_dyn
3585 use_backend static if url_css url_img extension_img
3586 default_backend dynamic
3587
Willy Tarreau98d04852015-05-26 12:18:29 +02003588 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003589
Willy Tarreau0ba27502007-12-24 16:55:16 +01003590
Baptiste Assmann27f51342013-10-09 06:51:49 +02003591description <string>
3592 Describe a listen, frontend or backend.
3593 May be used in sections : defaults | frontend | listen | backend
3594 no | yes | yes | yes
3595 Arguments : string
3596
3597 Allows to add a sentence to describe the related object in the HAProxy HTML
3598 stats page. The description will be printed on the right of the object name
3599 it describes.
3600 No need to backslash spaces in the <string> arguments.
3601
3602
Willy Tarreau0ba27502007-12-24 16:55:16 +01003603disabled
3604 Disable a proxy, frontend or backend.
3605 May be used in sections : defaults | frontend | listen | backend
3606 yes | yes | yes | yes
3607 Arguments : none
3608
3609 The "disabled" keyword is used to disable an instance, mainly in order to
3610 liberate a listening port or to temporarily disable a service. The instance
3611 will still be created and its configuration will be checked, but it will be
3612 created in the "stopped" state and will appear as such in the statistics. It
3613 will not receive any traffic nor will it send any health-checks or logs. It
3614 is possible to disable many instances at once by adding the "disabled"
3615 keyword in a "defaults" section.
3616
3617 See also : "enabled"
3618
3619
Willy Tarreau5ce94572010-06-07 14:35:41 +02003620dispatch <address>:<port>
3621 Set a default server address
3622 May be used in sections : defaults | frontend | listen | backend
3623 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003624 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003625
3626 <address> is the IPv4 address of the default server. Alternatively, a
3627 resolvable hostname is supported, but this name will be resolved
3628 during start-up.
3629
3630 <ports> is a mandatory port specification. All connections will be sent
3631 to this port, and it is not permitted to use port offsets as is
3632 possible with normal servers.
3633
Willy Tarreau787aed52011-04-15 06:45:37 +02003634 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003635 server can take the connection. In the past it was used to forward non
3636 persistent connections to an auxiliary load balancer. Due to its simple
3637 syntax, it has also been used for simple TCP relays. It is recommended not to
3638 use it for more clarity, and to use the "server" directive instead.
3639
3640 See also : "server"
3641
Olivier Houchard4e694042017-03-14 20:01:29 +01003642
3643dynamic-cookie-key <string>
3644 Set the dynamic cookie secret key for a backend.
3645 May be used in sections : defaults | frontend | listen | backend
3646 yes | no | yes | yes
3647 Arguments : The secret key to be used.
3648
3649 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003650 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003651 specified on the "server" line), using a hash of the IP address of the
3652 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003653 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003654 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003655
Willy Tarreau0ba27502007-12-24 16:55:16 +01003656enabled
3657 Enable a proxy, frontend or backend.
3658 May be used in sections : defaults | frontend | listen | backend
3659 yes | yes | yes | yes
3660 Arguments : none
3661
3662 The "enabled" keyword is used to explicitly enable an instance, when the
3663 defaults has been set to "disabled". This is very rarely used.
3664
3665 See also : "disabled"
3666
3667
3668errorfile <code> <file>
3669 Return a file contents instead of errors generated by HAProxy
3670 May be used in sections : defaults | frontend | listen | backend
3671 yes | yes | yes | yes
3672 Arguments :
3673 <code> is the HTTP status code. Currently, HAProxy is capable of
Anthonin Bonnefoyb1e94072020-06-22 09:17:01 +02003674 generating codes 200, 400, 403, 404, 405, 408, 410, 413, 425, 429,
3675 500, 502, 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003676
3677 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003678 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003679 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003680 error pages, and to use absolute paths, since files are read
3681 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003682
3683 It is important to understand that this keyword is not meant to rewrite
3684 errors returned by the server, but errors detected and returned by HAProxy.
3685 This is why the list of supported errors is limited to a small set.
3686
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003687 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3688
Willy Tarreau0ba27502007-12-24 16:55:16 +01003689 The files are returned verbatim on the TCP socket. This allows any trick such
3690 as redirections to another URL or site, as well as tricks to clean cookies,
3691 force enable or disable caching, etc... The package provides default error
3692 files returning the same contents as default errors.
3693
Willy Tarreau59140a22009-02-22 12:02:30 +01003694 The files should not exceed the configured buffer size (BUFSIZE), which
3695 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003696 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003697 loops between the client and HAProxy when all servers are down, causing an
3698 error to be returned instead of an image. For better HTTP compliance, it is
3699 recommended that all header lines end with CR-LF and not LF alone.
3700
Willy Tarreau0ba27502007-12-24 16:55:16 +01003701 The files are read at the same time as the configuration and kept in memory.
3702 For this reason, the errors continue to be returned even when the process is
3703 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003704 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003705 403 status code and interrogating a blocked URL.
3706
3707 See also : "errorloc", "errorloc302", "errorloc303"
3708
Willy Tarreau59140a22009-02-22 12:02:30 +01003709 Example :
3710 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003711 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003712 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3713 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3714
Willy Tarreau2769aa02007-12-27 18:26:09 +01003715
3716errorloc <code> <url>
3717errorloc302 <code> <url>
3718 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3719 May be used in sections : defaults | frontend | listen | backend
3720 yes | yes | yes | yes
3721 Arguments :
3722 <code> is the HTTP status code. Currently, HAProxy is capable of
Anthonin Bonnefoyb1e94072020-06-22 09:17:01 +02003723 generating codes 200, 400, 403, 404, 405, 408, 410, 413, 425, 429,
3724 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003725
3726 <url> it is the exact contents of the "Location" header. It may contain
3727 either a relative URI to an error page hosted on the same site,
3728 or an absolute URI designating an error page on another site.
3729 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003730 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003731
3732 It is important to understand that this keyword is not meant to rewrite
3733 errors returned by the server, but errors detected and returned by HAProxy.
3734 This is why the list of supported errors is limited to a small set.
3735
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003736 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3737
Willy Tarreau2769aa02007-12-27 18:26:09 +01003738 Note that both keyword return the HTTP 302 status code, which tells the
3739 client to fetch the designated URL using the same HTTP method. This can be
3740 quite problematic in case of non-GET methods such as POST, because the URL
3741 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003742 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003743 status code, indicating to the client that the URL must be fetched with a GET
3744 request.
3745
3746 See also : "errorfile", "errorloc303"
3747
3748
3749errorloc303 <code> <url>
3750 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3751 May be used in sections : defaults | frontend | listen | backend
3752 yes | yes | yes | yes
3753 Arguments :
3754 <code> is the HTTP status code. Currently, HAProxy is capable of
Anthonin Bonnefoyb1e94072020-06-22 09:17:01 +02003755 generating codes 200, 400, 403, 404, 405, 408, 410, 413, 425, 429,
3756 500, 502, 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003757
3758 <url> it is the exact contents of the "Location" header. It may contain
3759 either a relative URI to an error page hosted on the same site,
3760 or an absolute URI designating an error page on another site.
3761 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003762 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003763
3764 It is important to understand that this keyword is not meant to rewrite
3765 errors returned by the server, but errors detected and returned by HAProxy.
3766 This is why the list of supported errors is limited to a small set.
3767
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003768 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3769
Willy Tarreau2769aa02007-12-27 18:26:09 +01003770 Note that both keyword return the HTTP 303 status code, which tells the
3771 client to fetch the designated URL using the same HTTP GET method. This
3772 solves the usual problems associated with "errorloc" and the 302 code. It is
3773 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003774 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003775
3776 See also : "errorfile", "errorloc", "errorloc302"
3777
3778
Simon Horman51a1cf62015-02-03 13:00:44 +09003779email-alert from <emailaddr>
3780 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003781 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003782 May be used in sections: defaults | frontend | listen | backend
3783 yes | yes | yes | yes
3784
3785 Arguments :
3786
3787 <emailaddr> is the from email address to use when sending email alerts
3788
3789 Also requires "email-alert mailers" and "email-alert to" to be set
3790 and if so sending email alerts is enabled for the proxy.
3791
Simon Horman64e34162015-02-06 11:11:57 +09003792 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003793 "email-alert myhostname", "email-alert to", section 3.6 about
3794 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003795
3796
3797email-alert level <level>
3798 Declare the maximum log level of messages for which email alerts will be
3799 sent. This acts as a filter on the sending of email alerts.
3800 May be used in sections: defaults | frontend | listen | backend
3801 yes | yes | yes | yes
3802
3803 Arguments :
3804
3805 <level> One of the 8 syslog levels:
3806 emerg alert crit err warning notice info debug
3807 The above syslog levels are ordered from lowest to highest.
3808
3809 By default level is alert
3810
3811 Also requires "email-alert from", "email-alert mailers" and
3812 "email-alert to" to be set and if so sending email alerts is enabled
3813 for the proxy.
3814
Simon Horman1421e212015-04-30 13:10:35 +09003815 Alerts are sent when :
3816
3817 * An un-paused server is marked as down and <level> is alert or lower
3818 * A paused server is marked as down and <level> is notice or lower
3819 * A server is marked as up or enters the drain state and <level>
3820 is notice or lower
3821 * "option log-health-checks" is enabled, <level> is info or lower,
3822 and a health check status update occurs
3823
Simon Horman64e34162015-02-06 11:11:57 +09003824 See also : "email-alert from", "email-alert mailers",
3825 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003826 section 3.6 about mailers.
3827
3828
3829email-alert mailers <mailersect>
3830 Declare the mailers to be used when sending email alerts
3831 May be used in sections: defaults | frontend | listen | backend
3832 yes | yes | yes | yes
3833
3834 Arguments :
3835
3836 <mailersect> is the name of the mailers section to send email alerts.
3837
3838 Also requires "email-alert from" and "email-alert to" to be set
3839 and if so sending email alerts is enabled for the proxy.
3840
Simon Horman64e34162015-02-06 11:11:57 +09003841 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3842 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003843
3844
3845email-alert myhostname <hostname>
3846 Declare the to hostname address to be used when communicating with
3847 mailers.
3848 May be used in sections: defaults | frontend | listen | backend
3849 yes | yes | yes | yes
3850
3851 Arguments :
3852
Baptiste Assmann738bad92015-12-21 15:27:53 +01003853 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003854
3855 By default the systems hostname is used.
3856
3857 Also requires "email-alert from", "email-alert mailers" and
3858 "email-alert to" to be set and if so sending email alerts is enabled
3859 for the proxy.
3860
Simon Horman64e34162015-02-06 11:11:57 +09003861 See also : "email-alert from", "email-alert level", "email-alert mailers",
3862 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003863
3864
3865email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003866 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003867 header of email alerts. This is the address that email alerts are sent to.
3868 May be used in sections: defaults | frontend | listen | backend
3869 yes | yes | yes | yes
3870
3871 Arguments :
3872
3873 <emailaddr> is the to email address to use when sending email alerts
3874
3875 Also requires "email-alert mailers" and "email-alert to" to be set
3876 and if so sending email alerts is enabled for the proxy.
3877
Simon Horman64e34162015-02-06 11:11:57 +09003878 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003879 "email-alert myhostname", section 3.6 about mailers.
3880
3881
Willy Tarreau4de91492010-01-22 19:10:05 +01003882force-persist { if | unless } <condition>
3883 Declare a condition to force persistence on down servers
3884 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003885 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003886
3887 By default, requests are not dispatched to down servers. It is possible to
3888 force this using "option persist", but it is unconditional and redispatches
3889 to a valid server if "option redispatch" is set. That leaves with very little
3890 possibilities to force some requests to reach a server which is artificially
3891 marked down for maintenance operations.
3892
3893 The "force-persist" statement allows one to declare various ACL-based
3894 conditions which, when met, will cause a request to ignore the down status of
3895 a server and still try to connect to it. That makes it possible to start a
3896 server, still replying an error to the health checks, and run a specially
3897 configured browser to test the service. Among the handy methods, one could
3898 use a specific source IP address, or a specific cookie. The cookie also has
3899 the advantage that it can easily be added/removed on the browser from a test
3900 page. Once the service is validated, it is then possible to open the service
3901 to the world by returning a valid response to health checks.
3902
3903 The forced persistence is enabled when an "if" condition is met, or unless an
3904 "unless" condition is met. The final redispatch is always disabled when this
3905 is used.
3906
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003907 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003908 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003909
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003910
3911filter <name> [param*]
3912 Add the filter <name> in the filter list attached to the proxy.
3913 May be used in sections : defaults | frontend | listen | backend
3914 no | yes | yes | yes
3915 Arguments :
3916 <name> is the name of the filter. Officially supported filters are
3917 referenced in section 9.
3918
Tim Düsterhus4896c442016-11-29 02:15:19 +01003919 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003920 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003921 filter. Please refer to the documentation of the corresponding
3922 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003923
3924 Multiple occurrences of the filter line can be used for the same proxy. The
3925 same filter can be referenced many times if needed.
3926
3927 Example:
3928 listen
3929 bind *:80
3930
3931 filter trace name BEFORE-HTTP-COMP
3932 filter compression
3933 filter trace name AFTER-HTTP-COMP
3934
3935 compression algo gzip
3936 compression offload
3937
3938 server srv1 192.168.0.1:80
3939
3940 See also : section 9.
3941
Willy Tarreau4de91492010-01-22 19:10:05 +01003942
Willy Tarreau2769aa02007-12-27 18:26:09 +01003943fullconn <conns>
3944 Specify at what backend load the servers will reach their maxconn
3945 May be used in sections : defaults | frontend | listen | backend
3946 yes | no | yes | yes
3947 Arguments :
3948 <conns> is the number of connections on the backend which will make the
3949 servers use the maximal number of connections.
3950
Willy Tarreau198a7442008-01-17 12:05:32 +01003951 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003952 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003953 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003954 load. The server will then always accept at least <minconn> connections,
3955 never more than <maxconn>, and the limit will be on the ramp between both
3956 values when the backend has less than <conns> concurrent connections. This
3957 makes it possible to limit the load on the servers during normal loads, but
3958 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003959 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003960
Willy Tarreaufbb78422011-06-05 15:38:35 +02003961 Since it's hard to get this value right, haproxy automatically sets it to
3962 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003963 backend (based on "use_backend" and "default_backend" rules). That way it's
3964 safe to leave it unset. However, "use_backend" involving dynamic names are
3965 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003966
Willy Tarreau2769aa02007-12-27 18:26:09 +01003967 Example :
3968 # The servers will accept between 100 and 1000 concurrent connections each
3969 # and the maximum of 1000 will be reached when the backend reaches 10000
3970 # connections.
3971 backend dynamic
3972 fullconn 10000
3973 server srv1 dyn1:80 minconn 100 maxconn 1000
3974 server srv2 dyn2:80 minconn 100 maxconn 1000
3975
3976 See also : "maxconn", "server"
3977
3978
3979grace <time>
3980 Maintain a proxy operational for some time after a soft stop
3981 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003982 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003983 Arguments :
3984 <time> is the time (by default in milliseconds) for which the instance
3985 will remain operational with the frontend sockets still listening
3986 when a soft-stop is received via the SIGUSR1 signal.
3987
3988 This may be used to ensure that the services disappear in a certain order.
3989 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003990 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003991 needed by the equipment to detect the failure.
3992
3993 Note that currently, there is very little benefit in using this parameter,
3994 and it may in fact complicate the soft-reconfiguration process more than
3995 simplify it.
3996
Willy Tarreau0ba27502007-12-24 16:55:16 +01003997
Andrew Rodland17be45e2016-10-25 17:04:12 -04003998hash-balance-factor <factor>
3999 Specify the balancing factor for bounded-load consistent hashing
4000 May be used in sections : defaults | frontend | listen | backend
4001 yes | no | no | yes
4002 Arguments :
4003 <factor> is the control for the maximum number of concurrent requests to
4004 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01004005 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04004006
4007 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
4008 enables an algorithm that prevents any one server from getting too many
4009 requests at once, even if some hash buckets receive many more requests than
4010 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
4011 <factor> is a percentage greater than 100. For example, if <factor> is 150,
4012 then no server will be allowed to have a load more than 1.5 times the average.
4013 If server weights are used, they will be respected.
4014
4015 If the first-choice server is disqualified, the algorithm will choose another
4016 server based on the request hash, until a server with additional capacity is
4017 found. A higher <factor> allows more imbalance between the servers, while a
4018 lower <factor> means that more servers will be checked on average, affecting
4019 performance. Reasonable values are from 125 to 200.
4020
Willy Tarreau760e81d2018-05-03 07:20:40 +02004021 This setting is also used by "balance random" which internally relies on the
4022 consistent hashing mechanism.
4023
Andrew Rodland17be45e2016-10-25 17:04:12 -04004024 See also : "balance" and "hash-type".
4025
4026
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004027hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004028 Specify a method to use for mapping hashes to servers
4029 May be used in sections : defaults | frontend | listen | backend
4030 yes | no | yes | yes
4031 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04004032 <method> is the method used to select a server from the hash computed by
4033 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004034
Bhaskar98634f02013-10-29 23:30:51 -04004035 map-based the hash table is a static array containing all alive servers.
4036 The hashes will be very smooth, will consider weights, but
4037 will be static in that weight changes while a server is up
4038 will be ignored. This means that there will be no slow start.
4039 Also, since a server is selected by its position in the array,
4040 most mappings are changed when the server count changes. This
4041 means that when a server goes up or down, or when a server is
4042 added to a farm, most connections will be redistributed to
4043 different servers. This can be inconvenient with caches for
4044 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01004045
Bhaskar98634f02013-10-29 23:30:51 -04004046 consistent the hash table is a tree filled with many occurrences of each
4047 server. The hash key is looked up in the tree and the closest
4048 server is chosen. This hash is dynamic, it supports changing
4049 weights while the servers are up, so it is compatible with the
4050 slow start feature. It has the advantage that when a server
4051 goes up or down, only its associations are moved. When a
4052 server is added to the farm, only a few part of the mappings
4053 are redistributed, making it an ideal method for caches.
4054 However, due to its principle, the distribution will never be
4055 very smooth and it may sometimes be necessary to adjust a
4056 server's weight or its ID to get a more balanced distribution.
4057 In order to get the same distribution on multiple load
4058 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004059 same IDs. Note: consistent hash uses sdbm and avalanche if no
4060 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04004061
4062 <function> is the hash function to be used :
4063
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004064 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04004065 reimplementation of ndbm) database library. It was found to do
4066 well in scrambling bits, causing better distribution of the keys
4067 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004068 function with good distribution, unless the total server weight
4069 is a multiple of 64, in which case applying the avalanche
4070 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04004071
4072 djb2 this function was first proposed by Dan Bernstein many years ago
4073 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004074 function provides a better distribution than sdbm. It generally
4075 works well with text-based inputs though it can perform extremely
4076 poorly with numeric-only input or when the total server weight is
4077 a multiple of 33, unless the avalanche modifier is also used.
4078
Willy Tarreaua0f42712013-11-14 14:30:35 +01004079 wt6 this function was designed for haproxy while testing other
4080 functions in the past. It is not as smooth as the other ones, but
4081 is much less sensible to the input data set or to the number of
4082 servers. It can make sense as an alternative to sdbm+avalanche or
4083 djb2+avalanche for consistent hashing or when hashing on numeric
4084 data such as a source IP address or a visitor identifier in a URL
4085 parameter.
4086
Willy Tarreau324f07f2015-01-20 19:44:50 +01004087 crc32 this is the most common CRC32 implementation as used in Ethernet,
4088 gzip, PNG, etc. It is slower than the other ones but may provide
4089 a better distribution or less predictable results especially when
4090 used on strings.
4091
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004092 <modifier> indicates an optional method applied after hashing the key :
4093
4094 avalanche This directive indicates that the result from the hash
4095 function above should not be used in its raw form but that
4096 a 4-byte full avalanche hash must be applied first. The
4097 purpose of this step is to mix the resulting bits from the
4098 previous hash in order to avoid any undesired effect when
4099 the input contains some limited values or when the number of
4100 servers is a multiple of one of the hash's components (64
4101 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
4102 result less predictable, but it's also not as smooth as when
4103 using the original function. Some testing might be needed
4104 with some workloads. This hash is one of the many proposed
4105 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004106
Bhaskar98634f02013-10-29 23:30:51 -04004107 The default hash type is "map-based" and is recommended for most usages. The
4108 default function is "sdbm", the selection of a function should be based on
4109 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004110
Andrew Rodland17be45e2016-10-25 17:04:12 -04004111 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004112
4113
Willy Tarreau0ba27502007-12-24 16:55:16 +01004114http-check disable-on-404
4115 Enable a maintenance mode upon HTTP/404 response to health-checks
4116 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01004117 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01004118 Arguments : none
4119
4120 When this option is set, a server which returns an HTTP code 404 will be
4121 excluded from further load-balancing, but will still receive persistent
4122 connections. This provides a very convenient method for Web administrators
4123 to perform a graceful shutdown of their servers. It is also important to note
4124 that a server which is detected as failed while it was in this mode will not
4125 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
4126 will immediately be reinserted into the farm. The status on the stats page
4127 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01004128 option only works in conjunction with the "httpchk" option. If this option
4129 is used with "http-check expect", then it has precedence over it so that 404
4130 responses will still be considered as soft-stop.
4131
4132 See also : "option httpchk", "http-check expect"
4133
4134
4135http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004136 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01004137 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02004138 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01004139 Arguments :
4140 <match> is a keyword indicating how to look for a specific pattern in the
4141 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004142 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01004143 exclamation mark ("!") to negate the match. Spaces are allowed
4144 between the exclamation mark and the keyword. See below for more
4145 details on the supported keywords.
4146
4147 <pattern> is the pattern to look for. It may be a string or a regular
4148 expression. If the pattern contains spaces, they must be escaped
4149 with the usual backslash ('\').
4150
4151 By default, "option httpchk" considers that response statuses 2xx and 3xx
4152 are valid, and that others are invalid. When "http-check expect" is used,
4153 it defines what is considered valid or invalid. Only one "http-check"
4154 statement is supported in a backend. If a server fails to respond or times
4155 out, the check obviously fails. The available matches are :
4156
4157 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004158 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004159 response's status code is exactly this string. If the
4160 "status" keyword is prefixed with "!", then the response
4161 will be considered invalid if the status code matches.
4162
4163 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004164 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004165 response's status code matches the expression. If the
4166 "rstatus" keyword is prefixed with "!", then the response
4167 will be considered invalid if the status code matches.
4168 This is mostly used to check for multiple codes.
4169
4170 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004171 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004172 response's body contains this exact string. If the
4173 "string" keyword is prefixed with "!", then the response
4174 will be considered invalid if the body contains this
4175 string. This can be used to look for a mandatory word at
4176 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004177 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01004178 trace).
4179
4180 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004181 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004182 response's body matches this expression. If the "rstring"
4183 keyword is prefixed with "!", then the response will be
4184 considered invalid if the body matches the expression.
4185 This can be used to look for a mandatory word at the end
4186 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01004187 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01004188
4189 It is important to note that the responses will be limited to a certain size
4190 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
4191 Thus, too large responses may not contain the mandatory pattern when using
4192 "string" or "rstring". If a large response is absolutely required, it is
4193 possible to change the default max size by setting the global variable.
4194 However, it is worth keeping in mind that parsing very large responses can
4195 waste some CPU cycles, especially when regular expressions are used, and that
4196 it is always better to focus the checks on smaller resources.
4197
Cyril Bonté32602d22015-01-30 00:07:07 +01004198 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
4199 will automatically append a "Connection: close" header, meaning that this
4200 header should not be present in the request provided by "option httpchk".
4201
Willy Tarreaubd741542010-03-16 18:46:54 +01004202 Last, if "http-check expect" is combined with "http-check disable-on-404",
4203 then this last one has precedence when the server responds with 404.
4204
4205 Examples :
4206 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004207 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01004208
4209 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004210 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01004211
4212 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004213 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01004214
4215 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03004216 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01004217
Willy Tarreaubd741542010-03-16 18:46:54 +01004218 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004219
4220
Christopher Fauletf304ad32020-04-09 08:44:06 +02004221http-check send [hdr <name> <value>]* [body <string>]
4222 Add a possible list of headers and/or a body to the request sent during HTTP
4223 health checks.
4224 May be used in sections : defaults | frontend | listen | backend
4225 yes | no | yes | yes
4226 Arguments :
4227 hdr <name> <value> adds the HTTP header field whose name is specified in
4228 <name> and whose value is defined by <value> to the
4229 request sent during HTTP health checks.
4230
4231 body <string> add the body defined by <string> to the request sent
4232 sent during HTTP health checks. If defined, the
4233 "Content-Length" header is thus automatically added
4234 to the request.
4235
4236 In addition to the request line defined by the "option httpchk" directive,
4237 this one is the valid way to add some headers and optionally a body to the
4238 request sent during HTTP health checks. If a body is defined, the associate
4239 "Content-Length" header is automatically added. The old trick consisting to
4240 add headers after the version string on the "option httpchk" line is now
4241 deprecated. Note also the "Connection: close" header is still added if a
4242 "http-check expect" direcive is defined independently of this directive, just
4243 like the state header if the directive "http-check send-state" is defined.
4244
4245 See also : "option httpchk", "http-check send-state" and "http-check expect"
4246
4247
Willy Tarreauef781042010-01-27 11:53:01 +01004248http-check send-state
4249 Enable emission of a state header with HTTP health checks
4250 May be used in sections : defaults | frontend | listen | backend
4251 yes | no | yes | yes
4252 Arguments : none
4253
4254 When this option is set, haproxy will systematically send a special header
4255 "X-Haproxy-Server-State" with a list of parameters indicating to each server
4256 how they are seen by haproxy. This can be used for instance when a server is
4257 manipulated without access to haproxy and the operator needs to know whether
4258 haproxy still sees it up or not, or if the server is the last one in a farm.
4259
4260 The header is composed of fields delimited by semi-colons, the first of which
4261 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
4262 checks on the total number before transition, just as appears in the stats
4263 interface. Next headers are in the form "<variable>=<value>", indicating in
4264 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08004265 - a variable "address", containing the address of the backend server.
4266 This corresponds to the <address> field in the server declaration. For
4267 unix domain sockets, it will read "unix".
4268
4269 - a variable "port", containing the port of the backend server. This
4270 corresponds to the <port> field in the server declaration. For unix
4271 domain sockets, it will read "unix".
4272
Willy Tarreauef781042010-01-27 11:53:01 +01004273 - a variable "name", containing the name of the backend followed by a slash
4274 ("/") then the name of the server. This can be used when a server is
4275 checked in multiple backends.
4276
4277 - a variable "node" containing the name of the haproxy node, as set in the
4278 global "node" variable, otherwise the system's hostname if unspecified.
4279
4280 - a variable "weight" indicating the weight of the server, a slash ("/")
4281 and the total weight of the farm (just counting usable servers). This
4282 helps to know if other servers are available to handle the load when this
4283 one fails.
4284
4285 - a variable "scur" indicating the current number of concurrent connections
4286 on the server, followed by a slash ("/") then the total number of
4287 connections on all servers of the same backend.
4288
4289 - a variable "qcur" indicating the current number of requests in the
4290 server's queue.
4291
4292 Example of a header received by the application server :
4293 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
4294 scur=13/22; qcur=0
4295
4296 See also : "option httpchk", "http-check disable-on-404"
4297
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004298
4299http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004300 Access control for Layer 7 requests
4301
4302 May be used in sections: defaults | frontend | listen | backend
4303 no | yes | yes | yes
4304
Willy Tarreau20b0de52012-12-24 15:45:22 +01004305 The http-request statement defines a set of rules which apply to layer 7
4306 processing. The rules are evaluated in their declaration order when they are
4307 met in a frontend, listen or backend section. Any rule may optionally be
4308 followed by an ACL-based condition, in which case it will only be evaluated
4309 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004310
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004311 The first keyword is the rule's action. The supported actions are described
4312 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004313
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004314 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004315
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004316 It is important to know that http-request rules are processed very early in
4317 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
4318 or "reqadd" rules. That way, headers added by "add-header"/"set-header" are
4319 visible by almost all further ACL rules.
Willy Tarreau53275e82017-11-24 07:52:01 +01004320
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004321 Using "reqadd"/"reqdel"/"reqrep" to manipulate request headers is discouraged
4322 in newer versions (>= 1.5). But if you need to use regular expression to
4323 delete headers, you can still use "reqdel". Also please use
4324 "http-request deny/allow/tarpit" instead of "reqdeny"/"reqpass"/"reqtarpit".
Willy Tarreauccbcc372012-12-27 12:37:57 +01004325
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004326 Example:
4327 acl nagios src 192.168.129.3
4328 acl local_net src 192.168.0.0/16
4329 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004330
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004331 http-request allow if nagios
4332 http-request allow if local_net auth_ok
4333 http-request auth realm Gimme if local_net auth_ok
4334 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004335
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004336 Example:
4337 acl key req.hdr(X-Add-Acl-Key) -m found
4338 acl add path /addacl
4339 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004340
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004341 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004342
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004343 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4344 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004345
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004346 Example:
4347 acl value req.hdr(X-Value) -m found
4348 acl setmap path /setmap
4349 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004350
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004351 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004352
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004353 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4354 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004355
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004356 See also : "stats http-request", section 3.4 about userlists and section 7
4357 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004358
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004359http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004360
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004361 This is used to add a new entry into an ACL. The ACL must be loaded from a
4362 file (even a dummy empty file). The file name of the ACL to be updated is
4363 passed between parentheses. It takes one argument: <key fmt>, which follows
4364 log-format rules, to collect content of the new entry. It performs a lookup
4365 in the ACL before insertion, to avoid duplicated (or more) values. This
4366 lookup is done by a linear search and can be expensive with large lists!
4367 It is the equivalent of the "add acl" command from the stats socket, but can
4368 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004369
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004370http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004371
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004372 This appends an HTTP header field whose name is specified in <name> and
4373 whose value is defined by <fmt> which follows the log-format rules (see
4374 Custom Log Format in section 8.2.4). This is particularly useful to pass
4375 connection-specific information to the server (e.g. the client's SSL
4376 certificate), or to combine several headers into one. This rule is not
4377 final, so it is possible to add other similar rules. Note that header
4378 addition is performed immediately, so one rule might reuse the resulting
4379 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004380
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004381http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004382
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004383 This stops the evaluation of the rules and lets the request pass the check.
4384 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004385
Sasha Pachev218f0642014-06-16 12:05:59 -06004386
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004387http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004388
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004389 This stops the evaluation of the rules and immediately responds with an
4390 HTTP 401 or 407 error code to invite the user to present a valid user name
4391 and password. No further "http-request" rules are evaluated. An optional
4392 "realm" parameter is supported, it sets the authentication realm that is
4393 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004394
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004395 Example:
4396 acl auth_ok http_auth_group(L1) G1
4397 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004398
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004399http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004400
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004401 See section 10.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004402
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004403http-request capture <sample> [ len <length> | id <id> ]
4404 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004405
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004406 This captures sample expression <sample> from the request buffer, and
4407 converts it to a string of at most <len> characters. The resulting string is
4408 stored into the next request "capture" slot, so it will possibly appear next
4409 to some captured HTTP headers. It will then automatically appear in the logs,
4410 and it will be possible to extract it using sample fetch rules to feed it
4411 into headers or anything. The length should be limited given that this size
4412 will be allocated for each capture during the whole session life.
4413 Please check section 7.3 (Fetching samples) and "capture request header" for
4414 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004415
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004416 If the keyword "id" is used instead of "len", the action tries to store the
4417 captured string in a previously declared capture slot. This is useful to run
4418 captures in backends. The slot id can be declared by a previous directive
Baptiste Assmann63b220d2020-01-16 14:34:22 +01004419 "http-request capture" or with the "declare capture" keyword.
4420
4421 When using this action in a backend, double check that the relevant
4422 frontend(s) have the required capture slots otherwise, this rule will be
4423 ignored at run time. This can't be detected at configuration parsing time
4424 due to HAProxy's ability to dynamically resolve backend name at runtime.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004425
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004426http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004427
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004428 This is used to delete an entry from an ACL. The ACL must be loaded from a
4429 file (even a dummy empty file). The file name of the ACL to be updated is
4430 passed between parentheses. It takes one argument: <key fmt>, which follows
4431 log-format rules, to collect content of the entry to delete.
4432 It is the equivalent of the "del acl" command from the stats socket, but can
4433 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004434
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004435http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004436
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004437 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004438
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004439http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004440
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004441 This is used to delete an entry from a MAP. The MAP must be loaded from a
4442 file (even a dummy empty file). The file name of the MAP to be updated is
4443 passed between parentheses. It takes one argument: <key fmt>, which follows
4444 log-format rules, to collect content of the entry to delete.
4445 It takes one argument: "file name" It is the equivalent of the "del map"
4446 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004447
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004448http-request deny [deny_status <status>] [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004449
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004450 This stops the evaluation of the rules and immediately rejects the request
4451 and emits an HTTP 403 error, or optionally the status code specified as an
4452 argument to "deny_status". The list of permitted status codes is limited to
4453 those that can be overridden by the "errorfile" directive.
4454 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004455
Olivier Houchard602bf7d2019-05-10 13:59:15 +02004456http-request disable-l7-retry [ { if | unless } <condition> ]
4457 This disables any attempt to retry the request if it fails for any other
4458 reason than a connection failure. This can be useful for example to make
4459 sure POST requests aren't retried on failure.
4460
Baptiste Assmann333939c2019-01-21 08:34:50 +01004461http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
4462
4463 This action performs a DNS resolution of the output of <expr> and stores
4464 the result in the variable <var>. It uses the DNS resolvers section
4465 pointed by <resolvers>.
4466 It is possible to choose a resolution preference using the optional
4467 arguments 'ipv4' or 'ipv6'.
4468 When performing the DNS resolution, the client side connection is on
4469 pause waiting till the end of the resolution.
4470 If an IP address can be found, it is stored into <var>. If any kind of
4471 error occurs, then <var> is not set.
4472 One can use this action to discover a server IP address at run time and
4473 based on information found in the request (IE a Host header).
4474 If this action is used to find the server's IP address (using the
4475 "set-dst" action), then the server IP address in the backend must be set
4476 to 0.0.0.0.
4477
4478 Example:
4479 resolvers mydns
4480 nameserver local 127.0.0.53:53
4481 nameserver google 8.8.8.8:53
4482 timeout retry 1s
4483 hold valid 10s
4484 hold nx 3s
4485 hold other 3s
4486 hold obsolete 0s
4487 accepted_payload_size 8192
4488
4489 frontend fe
4490 bind 10.42.0.1:80
4491 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
4492 http-request capture var(txn.myip) len 40
4493
4494 # return 503 when the variable is not set,
4495 # which mean DNS resolution error
4496 use_backend b_503 unless { var(txn.myip) -m found }
4497
4498 default_backend be
4499
4500 backend b_503
4501 # dummy backend used to return 503.
4502 # one can use the errorfile directive to send a nice
4503 # 503 error page to end users
4504
4505 backend be
4506 # rule to prevent HAProxy from reconnecting to services
4507 # on the local network (forged DNS name used to scan the network)
4508 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
4509 http-request set-dst var(txn.myip)
4510 server clear 0.0.0.0:0
4511
4512 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
4513 be used to scan the network or worst won't loop over itself...
4514
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004515http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4516
4517 This is used to build an HTTP 103 Early Hints response prior to any other one.
4518 This appends an HTTP header field to this response whose name is specified in
4519 <name> and whose value is defined by <fmt> which follows the log-format rules
4520 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004521 to the client some Link headers to preload resources required to render the
4522 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004523
4524 See RFC 8297 for more information.
4525
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004526http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004527
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004528 This performs an HTTP redirection based on a redirect rule. This is exactly
4529 the same as the "redirect" statement except that it inserts a redirect rule
4530 which can be processed in the middle of other "http-request" rules and that
4531 these rules use the "log-format" strings. See the "redirect" keyword for the
4532 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004533
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004534http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004535
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004536 This stops the evaluation of the rules and immediately closes the connection
4537 without sending any response. It acts similarly to the
4538 "tcp-request content reject" rules. It can be useful to force an immediate
4539 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004540
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004541http-request replace-header <name> <match-regex> <replace-fmt>
4542 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004543
Ilya Shipitsin5c836fd2020-02-29 12:34:59 +05004544 This matches the value of all occurrences of header field <name> against
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01004545 <match-regex>. Matching is performed case-sensitively. Matching values are
4546 completely replaced by <replace-fmt>. Format characters are allowed in
4547 <replace-fmt> and work like <fmt> arguments in "http-request add-header".
4548 Standard back-references using the backslash ('\') followed by a number are
4549 supported.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004550
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01004551 This action acts on whole header lines, regardless of the number of values
4552 they may contain. Thus it is well-suited to process headers naturally
4553 containing commas in their value, such as If-Modified-Since. Headers that
4554 contain a comma-separated list of values, such as Accept, should be processed
4555 using "http-request replace-value".
William Lallemand86d0df02017-11-24 21:36:45 +01004556
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01004557 Example:
4558 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
4559
4560 # applied to:
4561 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4562
4563 # outputs:
4564 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
4565
4566 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004567
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01004568 http-request replace-header User-Agent curl foo
4569
4570 # applied to:
4571 User-Agent: curl/7.47.0
Willy Tarreau09448f72014-06-25 18:12:15 +02004572
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01004573 # outputs:
4574 User-Agent: foo
Willy Tarreau09448f72014-06-25 18:12:15 +02004575
Willy Tarreaudfc85772019-12-17 06:52:51 +01004576http-request replace-path <match-regex> <replace-fmt>
4577 [ { if | unless } <condition> ]
4578
4579 This works like "replace-header" except that it works on the request's path
4580 component instead of a header. The path component starts at the first '/'
4581 after an optional scheme+authority. It does contain the query string if any
4582 is present. The replacement does not modify the scheme nor authority.
4583
4584 It is worth noting that regular expressions may be more expensive to evaluate
4585 than certain ACLs, so rare replacements may benefit from a condition to avoid
4586 performing the evaluation at all if it does not match.
4587
4588 Example:
4589 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
4590 http-request replace-path (.*) /foo\1
4591
4592 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
4593 http-request replace-path ([^?]*)(\?(.*))? \1/foo\2
4594
4595 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
4596 http-request replace-path /foo/(.*) /\1
4597 # or more efficient if only some requests match :
4598 http-request replace-path /foo/(.*) /\1 if { url_beg /foo/ }
4599
Willy Tarreau33810222019-06-12 17:44:02 +02004600http-request replace-uri <match-regex> <replace-fmt>
4601 [ { if | unless } <condition> ]
4602
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01004603 This works like "replace-header" except that it works on the request's URI part
4604 instead of a header. The URI part may contain an optional scheme, authority or
4605 query string. These are considered to be part of the value that is matched
4606 against.
4607
4608 It is worth noting that regular expressions may be more expensive to evaluate
4609 than certain ACLs, so rare replacements may benefit from a condition to avoid
4610 performing the evaluation at all if it does not match.
Willy Tarreau33810222019-06-12 17:44:02 +02004611
Willy Tarreaud41821d2019-12-17 06:51:20 +01004612 IMPORTANT NOTE: historically in HTTP/1.x, the vast majority of requests sent
4613 by browsers use the "origin form", which differs from the "absolute form" in
4614 that they do not contain a scheme nor authority in the URI portion. Mostly
4615 only requests sent to proxies, those forged by hand and some emitted by
4616 certain applications use the absolute form. As such, "replace-uri" usually
4617 works fine most of the time in HTTP/1.x with rules starting with a "/". But
4618 with HTTP/2, clients are encouraged to send absolute URIs only, which look
4619 like the ones HTTP/1 clients use to talk to proxies. Such partial replace-uri
4620 rules may then fail in HTTP/2 when they work in HTTP/1. Either the rules need
Willy Tarreaudfc85772019-12-17 06:52:51 +01004621 to be adapted to optionally match a scheme and authority, or replace-path
4622 should be used.
Willy Tarreau33810222019-06-12 17:44:02 +02004623
Willy Tarreaud41821d2019-12-17 06:51:20 +01004624 Example:
4625 # rewrite all "http" absolute requests to "https":
4626 http-request replace-uri ^http://(.*) https://\1
Willy Tarreau33810222019-06-12 17:44:02 +02004627
Willy Tarreaud41821d2019-12-17 06:51:20 +01004628 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
4629 http-request replace-uri ([^/:]*://[^/]*)?(.*) \1/foo\2
Willy Tarreau33810222019-06-12 17:44:02 +02004630
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004631http-request replace-value <name> <match-regex> <replace-fmt>
4632 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004633
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01004634 This works like "replace-header" except that it matches the regex against
4635 every comma-delimited value of the header field <name> instead of the
4636 entire header. This is suited for all headers which are allowed to carry
4637 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004638
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01004639 Example:
4640 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004641
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01004642 # applied to:
4643 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004644
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01004645 # outputs:
4646 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004647
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004648http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4649http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004650
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004651 This actions increments the GPC0 or GPC1 counter according with the sticky
4652 counter designated by <sc-id>. If an error occurs, this action silently fails
4653 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004654
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004655http-request sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004656
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004657 This action sets the GPT0 tag according to the sticky counter designated by
4658 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4659 occurs, this action silently fails and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004660
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004661http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004662
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004663 This is used to set the destination IP address to the value of specified
4664 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4665 but provides the correct IP in a HTTP header; or you want to mask the IP for
4666 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4667 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004668
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004669 Arguments:
4670 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4671 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004672
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004673 Example:
4674 http-request set-dst hdr(x-dst)
4675 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004676
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004677 When possible, set-dst preserves the original destination port as long as the
4678 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004679
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004680http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004681
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004682 This is used to set the destination port address to the value of specified
4683 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4684 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004685
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004686 Arguments:
4687 <expr> Is a standard HAProxy expression formed by a sample-fetch
4688 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004689
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004690 Example:
4691 http-request set-dst-port hdr(x-port)
4692 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004693
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004694 When possible, set-dst-port preserves the original destination address as
4695 long as the address family supports a port, otherwise it forces the
4696 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004697
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004698http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004699
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004700 This does the same as "http-request add-header" except that the header name
4701 is first removed if it existed. This is useful when passing security
4702 information to the server, where the header must not be manipulated by
4703 external users. Note that the new value is computed before the removal so it
4704 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004705
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004706 Example:
4707 http-request set-header X-Haproxy-Current-Date %T
4708 http-request set-header X-SSL %[ssl_fc]
4709 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4710 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4711 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4712 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4713 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4714 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4715 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02004716
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004717http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004718
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004719 This is used to change the log level of the current request when a certain
4720 condition is met. Valid levels are the 8 syslog levels (see the "log"
4721 keyword) plus the special level "silent" which disables logging for this
4722 request. This rule is not final so the last matching rule wins. This rule
4723 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004724
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004725http-request set-map(<file-name>) <key fmt> <value fmt>
4726 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004727
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004728 This is used to add a new entry into a MAP. The MAP must be loaded from a
4729 file (even a dummy empty file). The file name of the MAP to be updated is
4730 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4731 log-format rules, used to collect MAP key, and <value fmt>, which follows
4732 log-format rules, used to collect content for the new entry.
4733 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4734 more) values. This lookup is done by a linear search and can be expensive
4735 with large lists! It is the equivalent of the "set map" command from the
4736 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004737
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004738http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004739
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004740 This is used to set the Netfilter MARK on all packets sent to the client to
4741 the value passed in <mark> on platforms which support it. This value is an
4742 unsigned 32 bit value which can be matched by netfilter and by the routing
4743 table. It can be expressed both in decimal or hexadecimal format (prefixed by
4744 "0x"). This can be useful to force certain packets to take a different route
4745 (for example a cheaper network path for bulk downloads). This works on Linux
4746 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02004747
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004748http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004749
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004750 This rewrites the request method with the result of the evaluation of format
4751 string <fmt>. There should be very few valid reasons for having to do so as
4752 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004753
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004754http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004755
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004756 This sets the "nice" factor of the current request being processed. It only
4757 has effect against the other requests being processed at the same time.
4758 The default value is 0, unless altered by the "nice" setting on the "bind"
4759 line. The accepted range is -1024..1024. The higher the value, the nicest
4760 the request will be. Lower values will make the request more important than
4761 other ones. This can be useful to improve the speed of some requests, or
4762 lower the priority of non-important requests. Using this setting without
4763 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004764
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004765http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02004766
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004767 This rewrites the request path with the result of the evaluation of format
4768 string <fmt>. The query string, if any, is left intact. If a scheme and
4769 authority is found before the path, they are left intact as well. If the
4770 request doesn't have a path ("*"), this one is replaced with the format.
4771 This can be used to prepend a directory component in front of a path for
4772 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02004773
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004774 Example :
4775 # prepend the host name before the path
4776 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004777
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004778http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004779
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004780 This is used to set the queue priority class of the current request.
4781 The value must be a sample expression which converts to an integer in the
4782 range -2047..2047. Results outside this range will be truncated.
4783 The priority class determines the order in which queued requests are
4784 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004785
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004786http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004787
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004788 This is used to set the queue priority timestamp offset of the current
4789 request. The value must be a sample expression which converts to an integer
4790 in the range -524287..524287. Results outside this range will be truncated.
4791 When a request is queued, it is ordered first by the priority class, then by
4792 the current timestamp adjusted by the given offset in milliseconds. Lower
4793 values have higher priority.
4794 Note that the resulting timestamp is is only tracked with enough precision
4795 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
4796 adjusted timestamp exceeds this value, it will be misidentified as highest
4797 priority. Thus it is important to set "timeout queue" to a value, where when
4798 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004799
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004800http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004801
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004802 This rewrites the request's query string which appears after the first
4803 question mark ("?") with the result of the evaluation of format string <fmt>.
4804 The part prior to the question mark is left intact. If the request doesn't
4805 contain a question mark and the new value is not empty, then one is added at
4806 the end of the URI, followed by the new value. If a question mark was
4807 present, it will never be removed even if the value is empty. This can be
4808 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004809
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004810 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004811
4812 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004813 # replace "%3D" with "=" in the query string
4814 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004815
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004816http-request set-src <expr> [ { if | unless } <condition> ]
4817 This is used to set the source IP address to the value of specified
4818 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
4819 provides the correct IP in a HTTP header; or you want to mask source IP for
Olivier Doucet46c517f2020-04-21 09:32:56 +02004820 privacy. All subsequent calls to "src" fetch will return this value
4821 (see example).
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004822
4823 Arguments :
4824 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4825 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004826
Olivier Doucet46c517f2020-04-21 09:32:56 +02004827 See also "option forwardfor".
4828
Cyril Bonté78caf842010-03-10 22:41:43 +01004829 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004830 http-request set-src hdr(x-forwarded-for)
4831 http-request set-src src,ipmask(24)
4832
Olivier Doucet46c517f2020-04-21 09:32:56 +02004833 # After the masking this will track connections
4834 # based on the IP address with the last byte zeroed out.
4835 http-request track-sc0 src
4836
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004837 When possible, set-src preserves the original source port as long as the
4838 address family allows it, otherwise the source port is set to 0.
4839
4840http-request set-src-port <expr> [ { if | unless } <condition> ]
4841
4842 This is used to set the source port address to the value of specified
4843 expression.
4844
4845 Arguments:
4846 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4847 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004848
Willy Tarreau20b0de52012-12-24 15:45:22 +01004849 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004850 http-request set-src-port hdr(x-port)
4851 http-request set-src-port int(4000)
4852
4853 When possible, set-src-port preserves the original source address as long as
4854 the address family supports a port, otherwise it forces the source address to
4855 IPv4 "0.0.0.0" before rewriting the port.
4856
4857http-request set-tos <tos> [ { if | unless } <condition> ]
4858
4859 This is used to set the TOS or DSCP field value of packets sent to the client
4860 to the value passed in <tos> on platforms which support this. This value
4861 represents the whole 8 bits of the IP TOS field, and can be expressed both in
4862 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
4863 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
4864 be used to adjust some routing behavior on border routers based on some
4865 information from the request.
4866
4867 See RFC 2474, 2597, 3260 and 4594 for more information.
4868
4869http-request set-uri <fmt> [ { if | unless } <condition> ]
4870
4871 This rewrites the request URI with the result of the evaluation of format
4872 string <fmt>. The scheme, authority, path and query string are all replaced
4873 at once. This can be used to rewrite hosts in front of proxies, or to
4874 perform complex modifications to the URI such as moving parts between the
4875 path and the query string.
4876 See also "http-request set-path" and "http-request set-query".
4877
4878http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4879
4880 This is used to set the contents of a variable. The variable is declared
4881 inline.
4882
4883 Arguments:
4884 <var-name> The name of the variable starts with an indication about its
4885 scope. The scopes allowed are:
4886 "proc" : the variable is shared with the whole process
4887 "sess" : the variable is shared with the whole session
4888 "txn" : the variable is shared with the transaction
4889 (request and response)
4890 "req" : the variable is shared only during request
4891 processing
4892 "res" : the variable is shared only during response
4893 processing
4894 This prefix is followed by a name. The separator is a '.'.
4895 The name may only contain characters 'a-z', 'A-Z', '0-9'
4896 and '_'.
4897
4898 <expr> Is a standard HAProxy expression formed by a sample-fetch
4899 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004900
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004901 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004902 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004903
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004904http-request send-spoe-group <engine-name> <group-name>
4905 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004906
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004907 This action is used to trigger sending of a group of SPOE messages. To do so,
4908 the SPOE engine used to send messages must be defined, as well as the SPOE
4909 group to send. Of course, the SPOE engine must refer to an existing SPOE
4910 filter. If not engine name is provided on the SPOE filter line, the SPOE
4911 agent name must be used.
4912
4913 Arguments:
4914 <engine-name> The SPOE engine name.
4915
4916 <group-name> The SPOE group name as specified in the engine
4917 configuration.
4918
4919http-request silent-drop [ { if | unless } <condition> ]
4920
4921 This stops the evaluation of the rules and makes the client-facing connection
4922 suddenly disappear using a system-dependent way that tries to prevent the
4923 client from being notified. The effect it then that the client still sees an
4924 established connection while there's none on HAProxy. The purpose is to
4925 achieve a comparable effect to "tarpit" except that it doesn't use any local
4926 resource at all on the machine running HAProxy. It can resist much higher
4927 loads than "tarpit", and slow down stronger attackers. It is important to
4928 understand the impact of using this mechanism. All stateful equipment placed
4929 between the client and HAProxy (firewalls, proxies, load balancers) will also
4930 keep the established connection for a long time and may suffer from this
4931 action.
4932 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4933 option is used to block the emission of a TCP reset. On other systems, the
4934 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4935 router, though it's still delivered to local networks. Do not use it unless
4936 you fully understand how it works.
4937
4938http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
4939
4940 This stops the evaluation of the rules and immediately blocks the request
4941 without responding for a delay specified by "timeout tarpit" or
4942 "timeout connect" if the former is not set. After that delay, if the client
4943 is still connected, an HTTP error 500 (or optionally the status code
4944 specified as an argument to "deny_status") is returned so that the client
4945 does not suspect it has been tarpitted. Logs will report the flags "PT".
4946 The goal of the tarpit rule is to slow down robots during an attack when
4947 they're limited on the number of concurrent requests. It can be very
4948 efficient against very dumb robots, and will significantly reduce the load
4949 on firewalls compared to a "deny" rule. But when facing "correctly"
4950 developed robots, it can make things worse by forcing haproxy and the front
4951 firewall to support insane number of concurrent connections.
4952 See also the "silent-drop" action.
4953
4954http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4955http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4956http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
4957
4958 This enables tracking of sticky counters from current request. These rules do
4959 not stop evaluation and do not change default action. The number of counters
4960 that may be simultaneously tracked by the same connection is set in
4961 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
4962 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
4963 "track-sc0" rule executed enables tracking of the counters of the specified
4964 table as the first set. The first "track-sc1" rule executed enables tracking
4965 of the counters of the specified table as the second set. The first
4966 "track-sc2" rule executed enables tracking of the counters of the specified
4967 table as the third set. It is a recommended practice to use the first set of
4968 counters for the per-frontend counters and the second set for the per-backend
4969 ones. But this is just a guideline, all may be used everywhere.
4970
4971 Arguments :
4972 <key> is mandatory, and is a sample expression rule as described in
4973 section 7.3. It describes what elements of the incoming request or
4974 connection will be analyzed, extracted, combined, and used to
4975 select which table entry to update the counters.
4976
4977 <table> is an optional table to be used instead of the default one, which
4978 is the stick-table declared in the current proxy. All the counters
4979 for the matches and updates for the key will then be performed in
4980 that table until the session ends.
4981
4982 Once a "track-sc*" rule is executed, the key is looked up in the table and if
4983 it is not found, an entry is allocated for it. Then a pointer to that entry
4984 is kept during all the session's life, and this entry's counters are updated
4985 as often as possible, every time the session's counters are updated, and also
4986 systematically when the session ends. Counters are only updated for events
4987 that happen after the tracking has been started. As an exception, connection
4988 counters and request counters are systematically updated so that they reflect
4989 useful information.
4990
4991 If the entry tracks concurrent connection counters, one connection is counted
4992 for as long as the entry is tracked, and the entry will not expire during
4993 that time. Tracking counters also provides a performance advantage over just
4994 checking the keys, because only one table lookup is performed for all ACL
4995 checks that make use of it.
4996
4997http-request unset-var(<var-name>) [ { if | unless } <condition> ]
4998
4999 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005000
5001 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005002 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005003
Christopher Faulet6bd406e2019-11-22 15:34:17 +01005004http-request use-service <service-name> [ { if | unless } <condition> ]
5005
5006 This directive executes the configured HTTP service to reply to the request
5007 and stops the evaluation of the rules. An HTTP service may choose to reply by
5008 sending any valid HTTP response or it may immediately close the connection
5009 without sending any response. Outside natives services, for instance the
5010 Prometheus exporter, it is possible to write your own services in Lua. No
5011 further "http-request" rules are evaluated.
5012
5013 Arguments :
5014 <service-name> is mandatory. It is the service to call
5015
5016 Example:
5017 http-request use-service prometheus-exporter if { path /metrics }
5018
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005019http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005020
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02005021 This will delay the processing of the request until the SSL handshake
5022 happened. This is mostly useful to delay processing early data until we're
5023 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005024
Willy Tarreauef781042010-01-27 11:53:01 +01005025
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005026http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005027 Access control for Layer 7 responses
5028
5029 May be used in sections: defaults | frontend | listen | backend
5030 no | yes | yes | yes
5031
5032 The http-response statement defines a set of rules which apply to layer 7
5033 processing. The rules are evaluated in their declaration order when they are
5034 met in a frontend, listen or backend section. Any rule may optionally be
5035 followed by an ACL-based condition, in which case it will only be evaluated
5036 if the condition is true. Since these rules apply on responses, the backend
5037 rules are applied first, followed by the frontend's rules.
5038
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005039 The first keyword is the rule's action. The supported actions are described
5040 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005041
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005042 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005043
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005044 It is important to know that http-response rules are processed very early in
5045 the HTTP processing, before "rspdel" or "rsprep" or "rspadd" rules. That way,
5046 headers added by "add-header"/"set-header" are visible by almost all further
5047 ACL rules.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005048
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005049 Using "rspadd"/"rspdel"/"rsprep" to manipulate request headers is discouraged
5050 in newer versions (>= 1.5). But if you need to use regular expression to
5051 delete headers, you can still use "rspdel". Also please use
5052 "http-response deny" instead of "rspdeny".
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005053
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005054 Example:
5055 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02005056
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005057 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06005058
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005059 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
5060 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06005061
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005062 Example:
5063 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06005064
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005065 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06005066
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005067 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
5068 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06005069
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005070 See also : "http-request", section 3.4 about userlists and section 7 about
5071 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06005072
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005073http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005074
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005075 This is used to add a new entry into an ACL. The ACL must be loaded from a
5076 file (even a dummy empty file). The file name of the ACL to be updated is
5077 passed between parentheses. It takes one argument: <key fmt>, which follows
5078 log-format rules, to collect content of the new entry. It performs a lookup
5079 in the ACL before insertion, to avoid duplicated (or more) values.
5080 This lookup is done by a linear search and can be expensive with large lists!
5081 It is the equivalent of the "add acl" command from the stats socket, but can
5082 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06005083
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005084http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005085
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005086 This appends an HTTP header field whose name is specified in <name> and whose
5087 value is defined by <fmt> which follows the log-format rules (see Custom Log
5088 Format in section 8.2.4). This may be used to send a cookie to a client for
5089 example, or to pass some internal information.
5090 This rule is not final, so it is possible to add other similar rules.
5091 Note that header addition is performed immediately, so one rule might reuse
5092 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06005093
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005094http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005095
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005096 This stops the evaluation of the rules and lets the response pass the check.
5097 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06005098
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02005099http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005100
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005101 See section 10.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06005102
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005103http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005104
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005105 This captures sample expression <sample> from the response buffer, and
5106 converts it to a string. The resulting string is stored into the next request
5107 "capture" slot, so it will possibly appear next to some captured HTTP
5108 headers. It will then automatically appear in the logs, and it will be
5109 possible to extract it using sample fetch rules to feed it into headers or
5110 anything. Please check section 7.3 (Fetching samples) and
5111 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005112
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005113 The keyword "id" is the id of the capture slot which is used for storing the
5114 string. The capture slot must be defined in an associated frontend.
5115 This is useful to run captures in backends. The slot id can be declared by a
5116 previous directive "http-response capture" or with the "declare capture"
5117 keyword.
Baptiste Assmann63b220d2020-01-16 14:34:22 +01005118
5119 When using this action in a backend, double check that the relevant
5120 frontend(s) have the required capture slots otherwise, this rule will be
5121 ignored at run time. This can't be detected at configuration parsing time
5122 due to HAProxy's ability to dynamically resolve backend name at runtime.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005123
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005124http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005125
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005126 This is used to delete an entry from an ACL. The ACL must be loaded from a
5127 file (even a dummy empty file). The file name of the ACL to be updated is
5128 passed between parentheses. It takes one argument: <key fmt>, which follows
5129 log-format rules, to collect content of the entry to delete.
5130 It is the equivalent of the "del acl" command from the stats socket, but can
5131 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02005132
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005133http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02005134
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005135 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02005136
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005137http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02005138
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005139 This is used to delete an entry from a MAP. The MAP must be loaded from a
5140 file (even a dummy empty file). The file name of the MAP to be updated is
5141 passed between parentheses. It takes one argument: <key fmt>, which follows
5142 log-format rules, to collect content of the entry to delete.
5143 It takes one argument: "file name" It is the equivalent of the "del map"
5144 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005145
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005146http-response deny [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005147
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005148 This stops the evaluation of the rules and immediately rejects the response
5149 and emits an HTTP 502 error. No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005150
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005151http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005152
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005153 This performs an HTTP redirection based on a redirect rule.
5154 This supports a format string similarly to "http-request redirect" rules,
5155 with the exception that only the "location" type of redirect is possible on
5156 the response. See the "redirect" keyword for the rule's syntax. When a
5157 redirect rule is applied during a response, connections to the server are
5158 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02005159
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005160http-response replace-header <name> <regex-match> <replace-fmt>
5161 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02005162
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01005163 This works like "http-request replace-header" except that it works on the
5164 server's response instead of the client's request.
William Lallemand86d0df02017-11-24 21:36:45 +01005165
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005166 Example:
5167 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02005168
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005169 # applied to:
5170 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005171
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005172 # outputs:
5173 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 +02005174
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005175 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005176
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005177http-response replace-value <name> <regex-match> <replace-fmt>
5178 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005179
Tim Duesterhus2f7045c2019-10-29 00:05:13 +01005180 This works like "http-response replace-value" except that it works on the
5181 server's response instead of the client's request.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005182
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005183 Example:
5184 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01005185
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005186 # applied to:
5187 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01005188
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005189 # outputs:
5190 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01005191
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005192http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
5193http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08005194
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005195 This action increments the GPC0 or GPC1 counter according with the sticky
5196 counter designated by <sc-id>. If an error occurs, this action silently fails
5197 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02005198
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005199http-response sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02005200
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005201 This action sets the GPT0 tag according to the sticky counter designated by
5202 <sc-id> and the value of <int>. The expected result is a boolean. If an error
5203 occurs, this action silently fails and the actions evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01005204
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005205http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02005206
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005207 This action is used to trigger sending of a group of SPOE messages. To do so,
5208 the SPOE engine used to send messages must be defined, as well as the SPOE
5209 group to send. Of course, the SPOE engine must refer to an existing SPOE
5210 filter. If not engine name is provided on the SPOE filter line, the SPOE
5211 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005212
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005213 Arguments:
5214 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005215
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005216 <group-name> The SPOE group name as specified in the engine
5217 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005218
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005219http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005220
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005221 This does the same as "add-header" except that the header name is first
5222 removed if it existed. This is useful when passing security information to
5223 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005224
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005225http-response set-log-level <level> [ { if | unless } <condition> ]
5226
5227 This is used to change the log level of the current request when a certain
5228 condition is met. Valid levels are the 8 syslog levels (see the "log"
5229 keyword) plus the special level "silent" which disables logging for this
5230 request. This rule is not final so the last matching rule wins. This rule can
5231 be useful to disable health checks coming from another equipment.
5232
5233http-response set-map(<file-name>) <key fmt> <value fmt>
5234
5235 This is used to add a new entry into a MAP. The MAP must be loaded from a
5236 file (even a dummy empty file). The file name of the MAP to be updated is
5237 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
5238 log-format rules, used to collect MAP key, and <value fmt>, which follows
5239 log-format rules, used to collect content for the new entry. It performs a
5240 lookup in the MAP before insertion, to avoid duplicated (or more) values.
5241 This lookup is done by a linear search and can be expensive with large lists!
5242 It is the equivalent of the "set map" command from the stats socket, but can
5243 be triggered by an HTTP response.
5244
5245http-response set-mark <mark> [ { if | unless } <condition> ]
5246
5247 This is used to set the Netfilter MARK on all packets sent to the client to
5248 the value passed in <mark> on platforms which support it. This value is an
5249 unsigned 32 bit value which can be matched by netfilter and by the routing
5250 table. It can be expressed both in decimal or hexadecimal format (prefixed
5251 by "0x"). This can be useful to force certain packets to take a different
5252 route (for example a cheaper network path for bulk downloads). This works on
5253 Linux kernels 2.6.32 and above and requires admin privileges.
5254
5255http-response set-nice <nice> [ { if | unless } <condition> ]
5256
5257 This sets the "nice" factor of the current request being processed.
5258 It only has effect against the other requests being processed at the same
5259 time. The default value is 0, unless altered by the "nice" setting on the
5260 "bind" line. The accepted range is -1024..1024. The higher the value, the
5261 nicest the request will be. Lower values will make the request more important
5262 than other ones. This can be useful to improve the speed of some requests, or
5263 lower the priority of non-important requests. Using this setting without
5264 prior experimentation can cause some major slowdown.
5265
5266http-response set-status <status> [reason <str>]
5267 [ { if | unless } <condition> ]
5268
5269 This replaces the response status code with <status> which must be an integer
5270 between 100 and 999. Optionally, a custom reason text can be provided defined
5271 by <str>, or the default reason for the specified code will be used as a
5272 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08005273
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005274 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005275 # return "431 Request Header Fields Too Large"
5276 http-response set-status 431
5277 # return "503 Slow Down", custom reason
5278 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005279
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005280http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005281
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005282 This is used to set the TOS or DSCP field value of packets sent to the client
5283 to the value passed in <tos> on platforms which support this.
5284 This value represents the whole 8 bits of the IP TOS field, and can be
5285 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
5286 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
5287 always 0. This can be used to adjust some routing behavior on border routers
5288 based on some information from the request.
5289
5290 See RFC 2474, 2597, 3260 and 4594 for more information.
5291
5292http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5293
5294 This is used to set the contents of a variable. The variable is declared
5295 inline.
5296
5297 Arguments:
5298 <var-name> The name of the variable starts with an indication about its
5299 scope. The scopes allowed are:
5300 "proc" : the variable is shared with the whole process
5301 "sess" : the variable is shared with the whole session
5302 "txn" : the variable is shared with the transaction
5303 (request and response)
5304 "req" : the variable is shared only during request
5305 processing
5306 "res" : the variable is shared only during response
5307 processing
5308 This prefix is followed by a name. The separator is a '.'.
5309 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5310 and '_'.
5311
5312 <expr> Is a standard HAProxy expression formed by a sample-fetch
5313 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005314
5315 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005316 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005317
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005318http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005319
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005320 This stops the evaluation of the rules and makes the client-facing connection
5321 suddenly disappear using a system-dependent way that tries to prevent the
5322 client from being notified. The effect it then that the client still sees an
5323 established connection while there's none on HAProxy. The purpose is to
5324 achieve a comparable effect to "tarpit" except that it doesn't use any local
5325 resource at all on the machine running HAProxy. It can resist much higher
5326 loads than "tarpit", and slow down stronger attackers. It is important to
5327 understand the impact of using this mechanism. All stateful equipment placed
5328 between the client and HAProxy (firewalls, proxies, load balancers) will also
5329 keep the established connection for a long time and may suffer from this
5330 action.
5331 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
5332 option is used to block the emission of a TCP reset. On other systems, the
5333 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
5334 router, though it's still delivered to local networks. Do not use it unless
5335 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005336
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005337http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
5338http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
5339http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005340
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005341 This enables tracking of sticky counters from current response. Please refer
5342 to "http-request track-sc" for a complete description. The only difference
5343 from "http-request track-sc" is the <key> sample expression can only make use
5344 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
5345 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
5346 supported, haproxy will fail and warn while parsing the config.
5347
5348http-response unset-var(<var-name>) [ { if | unless } <condition> ]
5349
5350 This is used to unset a variable. See "http-response set-var" for details
5351 about <var-name>.
5352
5353 Example:
5354 http-response unset-var(sess.last_redir)
5355
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02005356
Willy Tarreau30631952015-08-06 15:05:24 +02005357http-reuse { never | safe | aggressive | always }
5358 Declare how idle HTTP connections may be shared between requests
5359
5360 May be used in sections: defaults | frontend | listen | backend
5361 yes | no | yes | yes
5362
5363 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01005364 which is considered safe for reuse is moved back to the server's idle
5365 connections pool so that any other request can make use of it. This is the
5366 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02005367
5368 The argument indicates the desired connection reuse strategy :
5369
Olivier Houchard86006a52018-12-14 19:37:49 +01005370 - "never" : idle connections are never shared between sessions. This mode
5371 may be enforced to cancel a different strategy inherited from
5372 a defaults section or for troubleshooting. For example, if an
5373 old bogus application considers that multiple requests over
5374 the same connection come from the same client and it is not
5375 possible to fix the application, it may be desirable to
5376 disable connection sharing in a single backend. An example of
5377 such an application could be an old haproxy using cookie
5378 insertion in tunnel mode and not checking any request past the
5379 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02005380
Olivier Houchard86006a52018-12-14 19:37:49 +01005381 - "safe" : this is the default and the recommended strategy. The first
5382 request of a session is always sent over its own connection,
5383 and only subsequent requests may be dispatched over other
5384 existing connections. This ensures that in case the server
5385 closes the connection when the request is being sent, the
5386 browser can decide to silently retry it. Since it is exactly
5387 equivalent to regular keep-alive, there should be no side
5388 effects.
Willy Tarreau30631952015-08-06 15:05:24 +02005389
5390 - "aggressive" : this mode may be useful in webservices environments where
5391 all servers are not necessarily known and where it would be
5392 appreciable to deliver most first requests over existing
5393 connections. In this case, first requests are only delivered
5394 over existing connections that have been reused at least once,
5395 proving that the server correctly supports connection reuse.
5396 It should only be used when it's sure that the client can
5397 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02005398 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02005399 downsides of rare connection failures.
5400
5401 - "always" : this mode is only recommended when the path to the server is
5402 known for never breaking existing connections quickly after
5403 releasing them. It allows the first request of a session to be
5404 sent to an existing connection. This can provide a significant
5405 performance increase over the "safe" strategy when the backend
5406 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005407 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02005408 sharing. It is recommended that the "http-keep-alive" timeout
5409 remains low in this mode so that no dead connections remain
5410 usable. In most cases, this will lead to the same performance
5411 gains as "aggressive" but with more risks. It should only be
5412 used when it improves the situation over "aggressive".
5413
5414 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005415 connection properties and compatibility. Specifically :
5416 - connections made with "usesrc" followed by a client-dependent value
5417 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005418
5419 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01005420 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005421
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02005422 - connections with certain bogus authentication schemes (relying on the
5423 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005424
Lukas Tribus79a56932019-11-06 11:50:25 +01005425 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02005426
5427 Note: connection reuse improves the accuracy of the "server maxconn" setting,
5428 because almost no new connection will be established while idle connections
5429 remain available. This is particularly true with the "always" strategy.
5430
5431 See also : "option http-keep-alive", "server maxconn"
5432
5433
Mark Lamourinec2247f02012-01-04 13:02:01 -05005434http-send-name-header [<header>]
5435 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05005436 May be used in sections: defaults | frontend | listen | backend
5437 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05005438 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05005439 <header> The header string to use to send the server name
5440
Willy Tarreaue0e32792019-10-07 14:58:02 +02005441 The "http-send-name-header" statement causes the header field named <header>
5442 to be set to the name of the target server at the moment the request is about
5443 to be sent on the wire. Any existing occurrences of this header are removed.
5444 Upon retries and redispatches, the header field is updated to always reflect
5445 the server being attempted to connect to. Given that this header is modified
5446 very late in the connection setup, it may have unexpected effects on already
5447 modified headers. For example using it with transport-level header such as
5448 connection, content-length, transfer-encoding and so on will likely result in
5449 invalid requests being sent to the server. Additionally it has been reported
5450 that this directive is currently being used as a way to overwrite the Host
5451 header field in outgoing requests; while this trick has been known to work
5452 as a side effect of the feature for some time, it is not officially supported
5453 and might possibly not work anymore in a future version depending on the
5454 technical difficulties this feature induces. A long-term solution instead
5455 consists in fixing the application which required this trick so that it binds
5456 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05005457
5458 See also : "server"
5459
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005460id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02005461 Set a persistent ID to a proxy.
5462 May be used in sections : defaults | frontend | listen | backend
5463 no | yes | yes | yes
5464 Arguments : none
5465
5466 Set a persistent ID for the proxy. This ID must be unique and positive.
5467 An unused ID will automatically be assigned if unset. The first assigned
5468 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005469
5470
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005471ignore-persist { if | unless } <condition>
5472 Declare a condition to ignore persistence
5473 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005474 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005475
5476 By default, when cookie persistence is enabled, every requests containing
5477 the cookie are unconditionally persistent (assuming the target server is up
5478 and running).
5479
5480 The "ignore-persist" statement allows one to declare various ACL-based
5481 conditions which, when met, will cause a request to ignore persistence.
5482 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005483 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005484 persistence for a specific User-Agent (for example, some web crawler bots).
5485
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005486 The persistence is ignored when an "if" condition is met, or unless an
5487 "unless" condition is met.
5488
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005489 Example:
5490 acl url_static path_beg /static /images /img /css
5491 acl url_static path_end .gif .png .jpg .css .js
5492 ignore-persist if url_static
5493
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005494 See also : "force-persist", "cookie", and section 7 about ACL usage.
5495
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005496load-server-state-from-file { global | local | none }
5497 Allow seamless reload of HAProxy
5498 May be used in sections: defaults | frontend | listen | backend
5499 yes | no | yes | yes
5500
5501 This directive points HAProxy to a file where server state from previous
5502 running process has been saved. That way, when starting up, before handling
5503 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005504 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005505 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5506 loading state or load states from a file containing all backends and servers.
5507 The state file can be generated by running the command "show servers state"
5508 over the stats socket and redirect output.
5509
Davor Ocelice9ed2812017-12-25 17:49:28 +01005510 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005511 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005512 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005513
5514 Arguments:
5515 global load the content of the file pointed by the global directive
5516 named "server-state-file".
5517
5518 local load the content of the file pointed by the directive
5519 "server-state-file-name" if set. If not set, then the backend
5520 name is used as a file name.
5521
5522 none don't load any stat for this backend
5523
5524 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005525 - server's IP address is preserved across reloads by default, but the
5526 order can be changed thanks to the server's "init-addr" setting. This
5527 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005528 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005529 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005530
5531 - server's weight is applied from previous running process unless it has
5532 has changed between previous and new configuration files.
5533
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005534 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005535
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005536 global
5537 stats socket /tmp/socket
5538 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005539
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005540 defaults
5541 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005542
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005543 backend bk
5544 server s1 127.0.0.1:22 check weight 11
5545 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005546
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005547
5548 Then one can run :
5549
5550 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5551
5552 Content of the file /tmp/server_state would be like this:
5553
5554 1
5555 # <field names skipped for the doc example>
5556 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5557 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5558
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005559 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005560
5561 global
5562 stats socket /tmp/socket
5563 server-state-base /etc/haproxy/states
5564
5565 defaults
5566 load-server-state-from-file local
5567
5568 backend bk
5569 server s1 127.0.0.1:22 check weight 11
5570 server s2 127.0.0.1:22 check weight 12
5571
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005572
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005573 Then one can run :
5574
5575 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5576
5577 Content of the file /etc/haproxy/states/bk would be like this:
5578
5579 1
5580 # <field names skipped for the doc example>
5581 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5582 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5583
5584 See also: "server-state-file", "server-state-file-name", and
5585 "show servers state"
5586
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005587
Willy Tarreau2769aa02007-12-27 18:26:09 +01005588log global
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005589log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
5590 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005591no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005592 Enable per-instance logging of events and traffic.
5593 May be used in sections : defaults | frontend | listen | backend
5594 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005595
5596 Prefix :
5597 no should be used when the logger list must be flushed. For example,
5598 if you don't want to inherit from the default logger list. This
5599 prefix does not allow arguments.
5600
Willy Tarreau2769aa02007-12-27 18:26:09 +01005601 Arguments :
5602 global should be used when the instance's logging parameters are the
5603 same as the global ones. This is the most common usage. "global"
5604 replaces <address>, <facility> and <level> with those of the log
5605 entries found in the "global" section. Only one "log global"
5606 statement may be used per instance, and this form takes no other
5607 parameter.
5608
5609 <address> indicates where to send the logs. It takes the same format as
5610 for the "global" section's logs, and can be one of :
5611
5612 - An IPv4 address optionally followed by a colon (':') and a UDP
5613 port. If no port is specified, 514 is used by default (the
5614 standard syslog port).
5615
David du Colombier24bb5f52011-03-17 10:40:23 +01005616 - An IPv6 address followed by a colon (':') and optionally a UDP
5617 port. If no port is specified, 514 is used by default (the
5618 standard syslog port).
5619
Willy Tarreau2769aa02007-12-27 18:26:09 +01005620 - A filesystem path to a UNIX domain socket, keeping in mind
5621 considerations for chroot (be sure the path is accessible
5622 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005623 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005624
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005625 - A file descriptor number in the form "fd@<number>", which may
5626 point to a pipe, terminal, or socket. In this case unbuffered
5627 logs are used and one writev() call per log is performed. This
5628 is a bit expensive but acceptable for most workloads. Messages
5629 sent this way will not be truncated but may be dropped, in
5630 which case the DroppedLogs counter will be incremented. The
5631 writev() call is atomic even on pipes for messages up to
5632 PIPE_BUF size, which POSIX recommends to be at least 512 and
5633 which is 4096 bytes on most modern operating systems. Any
5634 larger message may be interleaved with messages from other
5635 processes. Exceptionally for debugging purposes the file
5636 descriptor may also be directed to a file, but doing so will
5637 significantly slow haproxy down as non-blocking calls will be
5638 ignored. Also there will be no way to purge nor rotate this
5639 file without restarting the process. Note that the configured
5640 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01005641 with a TCP syslog server. See also the "short" and "raw"
5642 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005643
5644 - "stdout" / "stderr", which are respectively aliases for "fd@1"
5645 and "fd@2", see above.
5646
5647 You may want to reference some environment variables in the
5648 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005649
Willy Tarreau18324f52014-06-27 18:10:07 +02005650 <length> is an optional maximum line length. Log lines larger than this
5651 value will be truncated before being sent. The reason is that
5652 syslog servers act differently on log line length. All servers
5653 support the default value of 1024, but some servers simply drop
5654 larger lines while others do log them. If a server supports long
5655 lines, it may make sense to set this value here in order to avoid
5656 truncating long lines. Similarly, if a server drops long lines,
5657 it is preferable to truncate them before sending them. Accepted
5658 values are 80 to 65535 inclusive. The default value of 1024 is
5659 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005660 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005661
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005662 <ranges> A list of comma-separated ranges to identify the logs to sample.
5663 This is used to balance the load of the logs to send to the log
5664 server. The limits of the ranges cannot be null. They are numbered
5665 from 1. The size or period (in number of logs) of the sample must
5666 be set with <sample_size> parameter.
5667
5668 <sample_size>
5669 The size of the sample in number of logs to consider when balancing
5670 their logging loads. It is used to balance the load of the logs to
5671 send to the syslog server. This size must be greater or equal to the
5672 maximum of the high limits of the ranges.
5673 (see also <ranges> parameter).
5674
Willy Tarreauadb345d2018-11-12 07:56:13 +01005675 <format> is the log format used when generating syslog messages. It may be
5676 one of the following :
5677
5678 rfc3164 The RFC3164 syslog message format. This is the default.
5679 (https://tools.ietf.org/html/rfc3164)
5680
5681 rfc5424 The RFC5424 syslog message format.
5682 (https://tools.ietf.org/html/rfc5424)
5683
Willy Tarreaue8746a02018-11-12 08:45:00 +01005684 short A message containing only a level between angle brackets such as
5685 '<3>', followed by the text. The PID, date, time, process name
5686 and system name are omitted. This is designed to be used with a
5687 local log server. This format is compatible with what the
5688 systemd logger consumes.
5689
Willy Tarreauc1b06452018-11-12 11:57:56 +01005690 raw A message containing only the text. The level, PID, date, time,
5691 process name and system name are omitted. This is designed to
5692 be used in containers or during development, where the severity
5693 only depends on the file descriptor used (stdout/stderr).
5694
Willy Tarreau2769aa02007-12-27 18:26:09 +01005695 <facility> must be one of the 24 standard syslog facilities :
5696
Willy Tarreaue8746a02018-11-12 08:45:00 +01005697 kern user mail daemon auth syslog lpr news
5698 uucp cron auth2 ftp ntp audit alert cron2
5699 local0 local1 local2 local3 local4 local5 local6 local7
5700
Willy Tarreauc1b06452018-11-12 11:57:56 +01005701 Note that the facility is ignored for the "short" and "raw"
5702 formats, but still required as a positional field. It is
5703 recommended to use "daemon" in this case to make it clear that
5704 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005705
5706 <level> is optional and can be specified to filter outgoing messages. By
5707 default, all messages are sent. If a level is specified, only
5708 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005709 will be sent. An optional minimum level can be specified. If it
5710 is set, logs emitted with a more severe level than this one will
5711 be capped to this level. This is used to avoid sending "emerg"
5712 messages on all terminals on some default syslog configurations.
5713 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005714
5715 emerg alert crit err warning notice info debug
5716
William Lallemand0f99e342011-10-12 17:50:54 +02005717 It is important to keep in mind that it is the frontend which decides what to
5718 log from a connection, and that in case of content switching, the log entries
5719 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005720
5721 However, backend log declaration define how and where servers status changes
5722 will be logged. Level "notice" will be used to indicate a server going up,
5723 "warning" will be used for termination signals and definitive service
5724 termination, and "alert" will be used for when a server goes down.
5725
5726 Note : According to RFC3164, messages are truncated to 1024 bytes before
5727 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005728
5729 Example :
5730 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01005731 log stdout format short daemon # send log to systemd
5732 log stdout format raw daemon # send everything to stdout
5733 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005734 log 127.0.0.1:514 local0 notice # only send important events
5735 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005736 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005737
Willy Tarreau2769aa02007-12-27 18:26:09 +01005738
William Lallemand48940402012-01-30 16:47:22 +01005739log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005740 Specifies the log format string to use for traffic logs
5741 May be used in sections: defaults | frontend | listen | backend
5742 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005743
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005744 This directive specifies the log format string that will be used for all logs
5745 resulting from traffic passing through the frontend using this line. If the
5746 directive is used in a defaults section, all subsequent frontends will use
5747 the same log format. Please see section 8.2.4 which covers the log format
5748 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005749
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005750 "log-format" directive overrides previous "option tcplog", "log-format" and
5751 "option httplog" directives.
5752
Dragan Dosen7ad31542015-09-28 17:16:47 +02005753log-format-sd <string>
5754 Specifies the RFC5424 structured-data log format string
5755 May be used in sections: defaults | frontend | listen | backend
5756 yes | yes | yes | no
5757
5758 This directive specifies the RFC5424 structured-data log format string that
5759 will be used for all logs resulting from traffic passing through the frontend
5760 using this line. If the directive is used in a defaults section, all
5761 subsequent frontends will use the same log format. Please see section 8.2.4
5762 which covers the log format string in depth.
5763
5764 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5765 about the RFC5424 structured-data part.
5766
5767 Note : This log format string will be used only for loggers that have set
5768 log format to "rfc5424".
5769
5770 Example :
5771 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5772
5773
Willy Tarreau094af4e2015-01-07 15:03:42 +01005774log-tag <string>
5775 Specifies the log tag to use for all outgoing logs
5776 May be used in sections: defaults | frontend | listen | backend
5777 yes | yes | yes | yes
5778
5779 Sets the tag field in the syslog header to this string. It defaults to the
5780 log-tag set in the global section, otherwise the program name as launched
5781 from the command line, which usually is "haproxy". Sometimes it can be useful
5782 to differentiate between multiple processes running on the same host, or to
5783 differentiate customer instances running in the same process. In the backend,
5784 logs about servers up/down will use this tag. As a hint, it can be convenient
5785 to set a log-tag related to a hosted customer in a defaults section then put
5786 all the frontends and backends for that customer, then start another customer
5787 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005788
Willy Tarreauc35362a2014-04-25 13:58:37 +02005789max-keep-alive-queue <value>
5790 Set the maximum server queue size for maintaining keep-alive connections
5791 May be used in sections: defaults | frontend | listen | backend
5792 yes | no | yes | yes
5793
5794 HTTP keep-alive tries to reuse the same server connection whenever possible,
5795 but sometimes it can be counter-productive, for example if a server has a lot
5796 of connections while other ones are idle. This is especially true for static
5797 servers.
5798
5799 The purpose of this setting is to set a threshold on the number of queued
5800 connections at which haproxy stops trying to reuse the same server and prefers
5801 to find another one. The default value, -1, means there is no limit. A value
5802 of zero means that keep-alive requests will never be queued. For very close
5803 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005804 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005805 use a value of 10 or less). For remote servers suffering from a high latency,
5806 higher values might be needed to cover for the latency and/or the cost of
5807 picking a different server.
5808
5809 Note that this has no impact on responses which are maintained to the same
5810 server consecutively to a 401 response. They will still go to the same server
5811 even if they have to be queued.
5812
5813 See also : "option http-server-close", "option prefer-last-server", server
5814 "maxconn" and cookie persistence.
5815
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01005816max-session-srv-conns <nb>
5817 Set the maximum number of outgoing connections we can keep idling for a given
5818 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
5819 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02005820
Willy Tarreau2769aa02007-12-27 18:26:09 +01005821maxconn <conns>
5822 Fix the maximum number of concurrent connections on a frontend
5823 May be used in sections : defaults | frontend | listen | backend
5824 yes | yes | yes | no
5825 Arguments :
5826 <conns> is the maximum number of concurrent connections the frontend will
5827 accept to serve. Excess connections will be queued by the system
5828 in the socket's listen queue and will be served once a connection
5829 closes.
5830
5831 If the system supports it, it can be useful on big sites to raise this limit
5832 very high so that haproxy manages connection queues, instead of leaving the
5833 clients with unanswered connection attempts. This value should not exceed the
5834 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005835 of tune.bufsize (16kB by default) each, as well as some other data resulting
5836 in about 33 kB of RAM being consumed per established connection. That means
5837 that a medium system equipped with 1GB of RAM can withstand around
5838 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005839
5840 Also, when <conns> is set to large values, it is possible that the servers
5841 are not sized to accept such loads, and for this reason it is generally wise
5842 to assign them some reasonable connection limits.
5843
Willy Tarreauc8d5b952019-02-27 17:25:52 +01005844 When this value is set to zero, which is the default, the global "maxconn"
5845 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02005846
Willy Tarreau2769aa02007-12-27 18:26:09 +01005847 See also : "server", global section's "maxconn", "fullconn"
5848
5849
5850mode { tcp|http|health }
5851 Set the running mode or protocol of the instance
5852 May be used in sections : defaults | frontend | listen | backend
5853 yes | yes | yes | yes
5854 Arguments :
5855 tcp The instance will work in pure TCP mode. A full-duplex connection
5856 will be established between clients and servers, and no layer 7
5857 examination will be performed. This is the default mode. It
5858 should be used for SSL, SSH, SMTP, ...
5859
5860 http The instance will work in HTTP mode. The client request will be
5861 analyzed in depth before connecting to any server. Any request
5862 which is not RFC-compliant will be rejected. Layer 7 filtering,
5863 processing and switching will be possible. This is the mode which
5864 brings HAProxy most of its value.
5865
5866 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005867 to incoming connections and close the connection. Alternatively,
5868 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5869 instead. Nothing will be logged in either case. This mode is used
5870 to reply to external components health checks. This mode is
5871 deprecated and should not be used anymore as it is possible to do
5872 the same and even better by combining TCP or HTTP modes with the
5873 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005874
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005875 When doing content switching, it is mandatory that the frontend and the
5876 backend are in the same mode (generally HTTP), otherwise the configuration
5877 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005878
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005879 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005880 defaults http_instances
5881 mode http
5882
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005883 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005884
Willy Tarreau0ba27502007-12-24 16:55:16 +01005885
Cyril Bontéf0c60612010-02-06 14:44:47 +01005886monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005887 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005888 May be used in sections : defaults | frontend | listen | backend
5889 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005890 Arguments :
5891 if <cond> the monitor request will fail if the condition is satisfied,
5892 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005893 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005894 are met, for instance a low number of servers both in a
5895 backend and its backup.
5896
5897 unless <cond> the monitor request will succeed only if the condition is
5898 satisfied, and will fail otherwise. Such a condition may be
5899 based on a test on the presence of a minimum number of active
5900 servers in a list of backends.
5901
5902 This statement adds a condition which can force the response to a monitor
5903 request to report a failure. By default, when an external component queries
5904 the URI dedicated to monitoring, a 200 response is returned. When one of the
5905 conditions above is met, haproxy will return 503 instead of 200. This is
5906 very useful to report a site failure to an external component which may base
5907 routing advertisements between multiple sites on the availability reported by
5908 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005909 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5910 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005911
5912 Example:
5913 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005914 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005915 acl site_dead nbsrv(dynamic) lt 2
5916 acl site_dead nbsrv(static) lt 2
5917 monitor-uri /site_alive
5918 monitor fail if site_dead
5919
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005920 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005921
5922
5923monitor-net <source>
5924 Declare a source network which is limited to monitor requests
5925 May be used in sections : defaults | frontend | listen | backend
5926 yes | yes | yes | no
5927 Arguments :
5928 <source> is the source IPv4 address or network which will only be able to
5929 get monitor responses to any request. It can be either an IPv4
5930 address, a host name, or an address followed by a slash ('/')
5931 followed by a mask.
5932
5933 In TCP mode, any connection coming from a source matching <source> will cause
5934 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005935 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005936 forwarding the connection to a remote server.
5937
5938 In HTTP mode, a connection coming from a source matching <source> will be
5939 accepted, the following response will be sent without waiting for a request,
5940 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5941 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005942 running without forwarding the request to a backend server. Note that this
5943 response is sent in raw format, without any transformation. This is important
5944 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005945
Willy Tarreau82569f92012-09-27 23:48:56 +02005946 Monitor requests are processed very early, just after tcp-request connection
5947 ACLs which are the only ones able to block them. These connections are short
5948 lived and never wait for any data from the client. They cannot be logged, and
5949 it is the intended purpose. They are only used to report HAProxy's health to
5950 an upper component, nothing more. Please note that "monitor fail" rules do
5951 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005952
Willy Tarreau95cd2832010-03-04 23:36:33 +01005953 Last, please note that only one "monitor-net" statement can be specified in
5954 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005955
Willy Tarreau2769aa02007-12-27 18:26:09 +01005956 Example :
5957 # addresses .252 and .253 are just probing us.
5958 frontend www
5959 monitor-net 192.168.0.252/31
5960
5961 See also : "monitor fail", "monitor-uri"
5962
5963
5964monitor-uri <uri>
5965 Intercept a URI used by external components' monitor requests
5966 May be used in sections : defaults | frontend | listen | backend
5967 yes | yes | yes | no
5968 Arguments :
5969 <uri> is the exact URI which we want to intercept to return HAProxy's
5970 health status instead of forwarding the request.
5971
5972 When an HTTP request referencing <uri> will be received on a frontend,
5973 HAProxy will not forward it nor log it, but instead will return either
5974 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5975 conditions defined with "monitor fail". This is normally enough for any
5976 front-end HTTP probe to detect that the service is UP and running without
5977 forwarding the request to a backend server. Note that the HTTP method, the
5978 version and all headers are ignored, but the request must at least be valid
5979 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5980
Willy Tarreau721d8e02017-12-01 18:25:08 +01005981 Monitor requests are processed very early, just after the request is parsed
5982 and even before any "http-request" or "block" rulesets. The only rulesets
5983 applied before are the tcp-request ones. They cannot be logged either, and it
5984 is the intended purpose. They are only used to report HAProxy's health to an
5985 upper component, nothing more. However, it is possible to add any number of
5986 conditions using "monitor fail" and ACLs so that the result can be adjusted
5987 to whatever check can be imagined (most often the number of available servers
5988 in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005989
5990 Example :
5991 # Use /haproxy_test to report haproxy's status
5992 frontend www
5993 mode http
5994 monitor-uri /haproxy_test
5995
5996 See also : "monitor fail", "monitor-net"
5997
Willy Tarreau0ba27502007-12-24 16:55:16 +01005998
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005999option abortonclose
6000no option abortonclose
6001 Enable or disable early dropping of aborted requests pending in queues.
6002 May be used in sections : defaults | frontend | listen | backend
6003 yes | no | yes | yes
6004 Arguments : none
6005
6006 In presence of very high loads, the servers will take some time to respond.
6007 The per-instance connection queue will inflate, and the response time will
6008 increase respective to the size of the queue times the average per-session
6009 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01006010 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006011 the queue, and slowing down other users, and the servers as well, because the
6012 request will eventually be served, then aborted at the first error
6013 encountered while delivering the response.
6014
6015 As there is no way to distinguish between a full STOP and a simple output
6016 close on the client side, HTTP agents should be conservative and consider
6017 that the client might only have closed its output channel while waiting for
6018 the response. However, this introduces risks of congestion when lots of users
6019 do the same, and is completely useless nowadays because probably no client at
6020 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01006021 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006022 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01006023 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006024 of being the single component to break rare but valid traffic is extremely
6025 low, which adds to the temptation to be able to abort a session early while
6026 still not served and not pollute the servers.
6027
Davor Ocelice9ed2812017-12-25 17:49:28 +01006028 In HAProxy, the user can choose the desired behavior using the option
6029 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006030 compliant and aborted requests will be served. But when the option is
6031 specified, a session with an incoming channel closed will be aborted while
6032 it is still possible, either pending in the queue for a connection slot, or
6033 during the connection establishment if the server has not yet acknowledged
6034 the connection request. This considerably reduces the queue size and the load
6035 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01006036 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006037
6038 If this option has been enabled in a "defaults" section, it can be disabled
6039 in a specific instance by prepending the "no" keyword before it.
6040
6041 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
6042
6043
Willy Tarreau4076a152009-04-02 15:18:36 +02006044option accept-invalid-http-request
6045no option accept-invalid-http-request
6046 Enable or disable relaxing of HTTP request parsing
6047 May be used in sections : defaults | frontend | listen | backend
6048 yes | yes | yes | no
6049 Arguments : none
6050
Willy Tarreau91852eb2015-05-01 13:26:00 +02006051 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02006052 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01006053 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02006054 forbidden characters are essentially used to build attacks exploiting server
6055 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
6056 server will emit invalid header names for whatever reason (configuration,
6057 implementation) and the issue will not be immediately fixed. In such a case,
6058 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01006059 even if that does not make sense, by specifying this option. Similarly, the
6060 list of characters allowed to appear in a URI is well defined by RFC3986, and
6061 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
6062 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01006063 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02006064 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02006065 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
6066 to pass through (no version specified) and multiple digits for both the major
6067 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02006068
6069 This option should never be enabled by default as it hides application bugs
6070 and open security breaches. It should only be deployed after a problem has
6071 been confirmed.
6072
6073 When this option is enabled, erroneous header names will still be accepted in
6074 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01006075 analysis using the "show errors" request on the UNIX stats socket. Similarly,
6076 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02006077 also helps confirming that the issue has been solved.
6078
6079 If this option has been enabled in a "defaults" section, it can be disabled
6080 in a specific instance by prepending the "no" keyword before it.
6081
6082 See also : "option accept-invalid-http-response" and "show errors" on the
6083 stats socket.
6084
6085
6086option accept-invalid-http-response
6087no option accept-invalid-http-response
6088 Enable or disable relaxing of HTTP response parsing
6089 May be used in sections : defaults | frontend | listen | backend
6090 yes | no | yes | yes
6091 Arguments : none
6092
Willy Tarreau91852eb2015-05-01 13:26:00 +02006093 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02006094 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01006095 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02006096 forbidden characters are essentially used to build attacks exploiting server
6097 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
6098 server will emit invalid header names for whatever reason (configuration,
6099 implementation) and the issue will not be immediately fixed. In such a case,
6100 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02006101 even if that does not make sense, by specifying this option. This option also
6102 relaxes the test on the HTTP version format, it allows multiple digits for
6103 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02006104
6105 This option should never be enabled by default as it hides application bugs
6106 and open security breaches. It should only be deployed after a problem has
6107 been confirmed.
6108
6109 When this option is enabled, erroneous header names will still be accepted in
6110 responses, but the complete response will be captured in order to permit
6111 later analysis using the "show errors" request on the UNIX stats socket.
6112 Doing this also helps confirming that the issue has been solved.
6113
6114 If this option has been enabled in a "defaults" section, it can be disabled
6115 in a specific instance by prepending the "no" keyword before it.
6116
6117 See also : "option accept-invalid-http-request" and "show errors" on the
6118 stats socket.
6119
6120
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006121option allbackups
6122no option allbackups
6123 Use either all backup servers at a time or only the first one
6124 May be used in sections : defaults | frontend | listen | backend
6125 yes | no | yes | yes
6126 Arguments : none
6127
6128 By default, the first operational backup server gets all traffic when normal
6129 servers are all down. Sometimes, it may be preferred to use multiple backups
6130 at once, because one will not be enough. When "option allbackups" is enabled,
6131 the load balancing will be performed among all backup servers when all normal
6132 ones are unavailable. The same load balancing algorithm will be used and the
6133 servers' weights will be respected. Thus, there will not be any priority
6134 order between the backup servers anymore.
6135
6136 This option is mostly used with static server farms dedicated to return a
6137 "sorry" page when an application is completely offline.
6138
6139 If this option has been enabled in a "defaults" section, it can be disabled
6140 in a specific instance by prepending the "no" keyword before it.
6141
6142
6143option checkcache
6144no option checkcache
Godbach7056a352013-12-11 20:01:07 +08006145 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006146 May be used in sections : defaults | frontend | listen | backend
6147 yes | no | yes | yes
6148 Arguments : none
6149
6150 Some high-level frameworks set application cookies everywhere and do not
6151 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006152 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006153 high risk of session crossing or stealing between users traversing the same
6154 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02006155 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006156
6157 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006158 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01006159 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006160 response to check if there's a risk of caching a cookie on a client-side
6161 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01006162 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01006163 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01006164 - all those with a return code other than 200, 203, 204, 206, 300, 301,
6165 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01006166 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01006167 - all those that result from a request using a method other than GET, HEAD,
6168 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01006169 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006170 - those with a 'Pragma: no-cache' header
6171 - those with a 'Cache-control: private' header
6172 - those with a 'Cache-control: no-store' header
6173 - those with a 'Cache-control: max-age=0' header
6174 - those with a 'Cache-control: s-maxage=0' header
6175 - those with a 'Cache-control: no-cache' header
6176 - those with a 'Cache-control: no-cache="set-cookie"' header
6177 - those with a 'Cache-control: no-cache="set-cookie,' header
6178 (allowing other fields after set-cookie)
6179
6180 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01006181 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006182 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006183 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006184 that admins are informed that there's something to be fixed.
6185
6186 Due to the high impact on the application, the application should be tested
6187 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006188 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01006189 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006190
6191 If this option has been enabled in a "defaults" section, it can be disabled
6192 in a specific instance by prepending the "no" keyword before it.
6193
6194
6195option clitcpka
6196no option clitcpka
6197 Enable or disable the sending of TCP keepalive packets on the client side
6198 May be used in sections : defaults | frontend | listen | backend
6199 yes | yes | yes | no
6200 Arguments : none
6201
6202 When there is a firewall or any session-aware component between a client and
6203 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006204 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006205 components decides to expire a session which has remained idle for too long.
6206
6207 Enabling socket-level TCP keep-alives makes the system regularly send packets
6208 to the other end of the connection, leaving it active. The delay between
6209 keep-alive probes is controlled by the system only and depends both on the
6210 operating system and its tuning parameters.
6211
6212 It is important to understand that keep-alive packets are neither emitted nor
6213 received at the application level. It is only the network stacks which sees
6214 them. For this reason, even if one side of the proxy already uses keep-alives
6215 to maintain its connection alive, those keep-alive packets will not be
6216 forwarded to the other side of the proxy.
6217
6218 Please note that this has nothing to do with HTTP keep-alive.
6219
6220 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
6221 client side of a connection, which should help when session expirations are
6222 noticed between HAProxy and a client.
6223
6224 If this option has been enabled in a "defaults" section, it can be disabled
6225 in a specific instance by prepending the "no" keyword before it.
6226
6227 See also : "option srvtcpka", "option tcpka"
6228
6229
Willy Tarreau0ba27502007-12-24 16:55:16 +01006230option contstats
6231 Enable continuous traffic statistics updates
6232 May be used in sections : defaults | frontend | listen | backend
6233 yes | yes | yes | no
6234 Arguments : none
6235
6236 By default, counters used for statistics calculation are incremented
6237 only when a session finishes. It works quite well when serving small
6238 objects, but with big ones (for example large images or archives) or
6239 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01006240 a hedgehog. With this option enabled counters get incremented frequently
6241 along the session, typically every 5 seconds, which is often enough to
6242 produce clean graphs. Recounting touches a hotpath directly so it is not
6243 not enabled by default, as it can cause a lot of wakeups for very large
6244 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01006245
6246
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006247option dontlog-normal
6248no option dontlog-normal
6249 Enable or disable logging of normal, successful connections
6250 May be used in sections : defaults | frontend | listen | backend
6251 yes | yes | yes | no
6252 Arguments : none
6253
6254 There are large sites dealing with several thousand connections per second
6255 and for which logging is a major pain. Some of them are even forced to turn
6256 logs off and cannot debug production issues. Setting this option ensures that
6257 normal connections, those which experience no error, no timeout, no retry nor
6258 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
6259 mode, the response status code is checked and return codes 5xx will still be
6260 logged.
6261
6262 It is strongly discouraged to use this option as most of the time, the key to
6263 complex issues is in the normal logs which will not be logged here. If you
6264 need to separate logs, see the "log-separate-errors" option instead.
6265
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006266 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006267 logging.
6268
6269
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006270option dontlognull
6271no option dontlognull
6272 Enable or disable logging of null connections
6273 May be used in sections : defaults | frontend | listen | backend
6274 yes | yes | yes | no
6275 Arguments : none
6276
6277 In certain environments, there are components which will regularly connect to
6278 various systems to ensure that they are still alive. It can be the case from
6279 another load balancer as well as from monitoring systems. By default, even a
6280 simple port probe or scan will produce a log. If those connections pollute
6281 the logs too much, it is possible to enable option "dontlognull" to indicate
6282 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02006283 which typically corresponds to those probes. Note that errors will still be
6284 returned to the client and accounted for in the stats. If this is not what is
6285 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006286
6287 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01006288 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006289 would not be logged.
6290
6291 If this option has been enabled in a "defaults" section, it can be disabled
6292 in a specific instance by prepending the "no" keyword before it.
6293
Willy Tarreau0f228a02015-05-01 15:37:53 +02006294 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
6295 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006296
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006297
Willy Tarreau87cf5142011-08-19 22:57:24 +02006298option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006299 Enable insertion of the X-Forwarded-For header to requests sent to servers
6300 May be used in sections : defaults | frontend | listen | backend
6301 yes | yes | yes | yes
6302 Arguments :
6303 <network> is an optional argument used to disable this option for sources
6304 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02006305 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01006306 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006307
6308 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
6309 their client address. This is sometimes annoying when the client's IP address
6310 is expected in server logs. To solve this problem, the well-known HTTP header
6311 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
6312 This header contains a value representing the client's IP address. Since this
6313 header is always appended at the end of the existing header list, the server
6314 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02006315 the server's manual to find how to enable use of this standard header. Note
6316 that only the last occurrence of the header must be used, since it is really
6317 possible that the client has already brought one.
6318
Willy Tarreaud72758d2010-01-12 10:42:19 +01006319 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02006320 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01006321 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01006322 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006323 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02006324 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01006325
6326 Sometimes, a same HAProxy instance may be shared between a direct client
6327 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6328 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6329 header for a known source address or network by adding the "except" keyword
6330 followed by the network address. In this case, any source IP matching the
6331 network will not cause an addition of this header. Most common uses are with
6332 private networks or 127.0.0.1.
6333
Willy Tarreau87cf5142011-08-19 22:57:24 +02006334 Alternatively, the keyword "if-none" states that the header will only be
6335 added if it is not present. This should only be used in perfectly trusted
6336 environment, as this might cause a security issue if headers reaching haproxy
6337 are under the control of the end-user.
6338
Willy Tarreauc27debf2008-01-06 08:57:02 +01006339 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02006340 least one of them uses it, the header will be added. Note that the backend's
6341 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02006342 both are defined. In the case of the "if-none" argument, if at least one of
6343 the frontend or the backend does not specify it, it wants the addition to be
6344 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006345
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006346 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01006347 # Public HTTP address also used by stunnel on the same machine
6348 frontend www
6349 mode http
6350 option forwardfor except 127.0.0.1 # stunnel already adds the header
6351
Ross Westaf72a1d2008-08-03 10:51:45 +02006352 # Those servers want the IP Address in X-Client
6353 backend www
6354 mode http
6355 option forwardfor header X-Client
6356
Willy Tarreau87cf5142011-08-19 22:57:24 +02006357 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006358 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01006359
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006360
Christopher Fauleta99ff4d2019-07-22 16:18:24 +02006361option h1-case-adjust-bogus-client
6362no option h1-case-adjust-bogus-client
6363 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
6364 May be used in sections : defaults | frontend | listen | backend
6365 yes | yes | yes | no
6366 Arguments : none
6367
6368 There is no standard case for header names because, as stated in RFC7230,
6369 they are case-insensitive. So applications must handle them in a case-
6370 insensitive manner. But some bogus applications violate the standards and
6371 erroneously rely on the cases most commonly used by browsers. This problem
6372 becomes critical with HTTP/2 because all header names must be exchanged in
6373 lower case, and HAProxy follows the same convention. All header names are
6374 sent in lower case to clients and servers, regardless of the HTTP version.
6375
6376 When HAProxy receives an HTTP/1 response, its header names are converted to
6377 lower case and manipulated and sent this way to the clients. If a client is
6378 known to violate the HTTP standards and to fail to process a response coming
6379 from HAProxy, it is possible to transform the lower case header names to a
6380 different format when the response is formatted and sent to the client, by
6381 enabling this option and specifying the list of headers to be reformatted
6382 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6383 must only be a temporary workaround for the time it takes the client to be
6384 fixed, because clients which require such workarounds might be vulnerable to
6385 content smuggling attacks and must absolutely be fixed.
6386
6387 Please note that this option will not affect standards-compliant clients.
6388
6389 If this option has been enabled in a "defaults" section, it can be disabled
6390 in a specific instance by prepending the "no" keyword before it.
6391
6392 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
6393 "h1-case-adjust-file".
6394
6395
6396option h1-case-adjust-bogus-server
6397no option h1-case-adjust-bogus-server
6398 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
6399 May be used in sections : defaults | frontend | listen | backend
6400 yes | no | yes | yes
6401 Arguments : none
6402
6403 There is no standard case for header names because, as stated in RFC7230,
6404 they are case-insensitive. So applications must handle them in a case-
6405 insensitive manner. But some bogus applications violate the standards and
6406 erroneously rely on the cases most commonly used by browsers. This problem
6407 becomes critical with HTTP/2 because all header names must be exchanged in
6408 lower case, and HAProxy follows the same convention. All header names are
6409 sent in lower case to clients and servers, regardless of the HTTP version.
6410
6411 When HAProxy receives an HTTP/1 request, its header names are converted to
6412 lower case and manipulated and sent this way to the servers. If a server is
6413 known to violate the HTTP standards and to fail to process a request coming
6414 from HAProxy, it is possible to transform the lower case header names to a
6415 different format when the request is formatted and sent to the server, by
6416 enabling this option and specifying the list of headers to be reformatted
6417 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6418 must only be a temporary workaround for the time it takes the server to be
6419 fixed, because servers which require such workarounds might be vulnerable to
6420 content smuggling attacks and must absolutely be fixed.
6421
6422 Please note that this option will not affect standards-compliant servers.
6423
6424 If this option has been enabled in a "defaults" section, it can be disabled
6425 in a specific instance by prepending the "no" keyword before it.
6426
6427 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
6428 "h1-case-adjust-file".
6429
6430
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006431option http-buffer-request
6432no option http-buffer-request
6433 Enable or disable waiting for whole HTTP request body before proceeding
6434 May be used in sections : defaults | frontend | listen | backend
6435 yes | yes | yes | yes
6436 Arguments : none
6437
6438 It is sometimes desirable to wait for the body of an HTTP request before
6439 taking a decision. This is what is being done by "balance url_param" for
6440 example. The first use case is to buffer requests from slow clients before
6441 connecting to the server. Another use case consists in taking the routing
6442 decision based on the request body's contents. This option placed in a
6443 frontend or backend forces the HTTP processing to wait until either the whole
6444 body is received, or the request buffer is full, or the first chunk is
6445 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01006446 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006447 the frontend and the backend, so this should definitely not be used by
6448 default.
6449
Baptiste Assmanneccdf432015-10-28 13:49:01 +01006450 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006451
6452
Willy Tarreau0f228a02015-05-01 15:37:53 +02006453option http-ignore-probes
6454no option http-ignore-probes
6455 Enable or disable logging of null connections and request timeouts
6456 May be used in sections : defaults | frontend | listen | backend
6457 yes | yes | yes | no
6458 Arguments : none
6459
6460 Recently some browsers started to implement a "pre-connect" feature
6461 consisting in speculatively connecting to some recently visited web sites
6462 just in case the user would like to visit them. This results in many
6463 connections being established to web sites, which end up in 408 Request
6464 Timeout if the timeout strikes first, or 400 Bad Request when the browser
6465 decides to close them first. These ones pollute the log and feed the error
6466 counters. There was already "option dontlognull" but it's insufficient in
6467 this case. Instead, this option does the following things :
6468 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01006469 was received over a connection before it was closed;
6470 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02006471 - prevent any error counter from being incremented
6472
6473 That way the empty connection is silently ignored. Note that it is better
6474 not to use this unless it is clear that it is needed, because it will hide
6475 real problems. The most common reason for not receiving a request and seeing
6476 a 408 is due to an MTU inconsistency between the client and an intermediary
6477 element such as a VPN, which blocks too large packets. These issues are
6478 generally seen with POST requests as well as GET with large cookies. The logs
6479 are often the only way to detect them.
6480
6481 If this option has been enabled in a "defaults" section, it can be disabled
6482 in a specific instance by prepending the "no" keyword before it.
6483
6484 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
6485
6486
Willy Tarreau16bfb022010-01-16 19:48:41 +01006487option http-keep-alive
6488no option http-keep-alive
6489 Enable or disable HTTP keep-alive from client to server
6490 May be used in sections : defaults | frontend | listen | backend
6491 yes | yes | yes | yes
6492 Arguments : none
6493
Willy Tarreau70dffda2014-01-30 03:07:23 +01006494 By default HAProxy operates in keep-alive mode with regards to persistent
6495 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02006496 leaves the connection idle on both sides between the end of a response and
6497 the start of a new request. This mode may be changed by several options such
6498 as "option http-server-close", "option httpclose" or "option http-tunnel".
6499 This option allows to set back the keep-alive mode, which can be useful when
6500 another mode was used in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01006501
6502 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
6503 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01006504 network) and the fastest session reuse on the server side at the expense
6505 of maintaining idle connections to the servers. In general, it is possible
6506 with this option to achieve approximately twice the request rate that the
6507 "http-server-close" option achieves on small objects. There are mainly two
6508 situations where this option may be useful :
6509
6510 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01006511 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01006512
6513 - when the cost of establishing the connection to the server is significant
6514 compared to the cost of retrieving the associated object from the server.
6515
6516 This last case can happen when the server is a fast static server of cache.
6517 In this case, the server will need to be properly tuned to support high enough
6518 connection counts because connections will last until the client sends another
6519 request.
6520
6521 If the client request has to go to another backend or another server due to
6522 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01006523 immediately be closed and a new one re-opened. Option "prefer-last-server" is
6524 available to try optimize server selection so that if the server currently
6525 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01006526
Willy Tarreau16bfb022010-01-16 19:48:41 +01006527 At the moment, logs will not indicate whether requests came from the same
6528 session or not. The accept date reported in the logs corresponds to the end
6529 of the previous request, and the request time corresponds to the time spent
6530 waiting for a new request. The keep-alive request time is still bound to the
6531 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6532 not set.
6533
Cyril Bonté653dcd62014-02-20 00:13:15 +01006534 This option disables and replaces any previous "option httpclose", "option
Christopher Faulet315b39c2018-09-21 16:26:19 +02006535 http-server-close" or "option http-tunnel". When backend and frontend options
6536 differ, all of these 4 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006537
Christopher Faulet315b39c2018-09-21 16:26:19 +02006538 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01006539 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01006540 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006541
6542
Willy Tarreau96e31212011-05-30 18:10:30 +02006543option http-no-delay
6544no option http-no-delay
6545 Instruct the system to favor low interactive delays over performance in HTTP
6546 May be used in sections : defaults | frontend | listen | backend
6547 yes | yes | yes | yes
6548 Arguments : none
6549
6550 In HTTP, each payload is unidirectional and has no notion of interactivity.
6551 Any agent is expected to queue data somewhat for a reasonably low delay.
6552 There are some very rare server-to-server applications that abuse the HTTP
6553 protocol and expect the payload phase to be highly interactive, with many
6554 interleaved data chunks in both directions within a single request. This is
6555 absolutely not supported by the HTTP specification and will not work across
6556 most proxies or servers. When such applications attempt to do this through
6557 haproxy, it works but they will experience high delays due to the network
6558 optimizations which favor performance by instructing the system to wait for
6559 enough data to be available in order to only send full packets. Typical
6560 delays are around 200 ms per round trip. Note that this only happens with
6561 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6562 affected.
6563
6564 When "option http-no-delay" is present in either the frontend or the backend
6565 used by a connection, all such optimizations will be disabled in order to
6566 make the exchanges as fast as possible. Of course this offers no guarantee on
6567 the functionality, as it may break at any other place. But if it works via
6568 HAProxy, it will work as fast as possible. This option should never be used
6569 by default, and should never be used at all unless such a buggy application
6570 is discovered. The impact of using this option is an increase of bandwidth
6571 usage and CPU usage, which may significantly lower performance in high
6572 latency environments.
6573
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006574 See also : "option http-buffer-request"
6575
Willy Tarreau96e31212011-05-30 18:10:30 +02006576
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006577option http-pretend-keepalive
6578no option http-pretend-keepalive
6579 Define whether haproxy will announce keepalive to the server or not
6580 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006581 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006582 Arguments : none
6583
Christopher Faulet315b39c2018-09-21 16:26:19 +02006584 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006585 adds a "Connection: close" header to the request forwarded to the server.
6586 Unfortunately, when some servers see this header, they automatically refrain
6587 from using the chunked encoding for responses of unknown length, while this
6588 is totally unrelated. The immediate effect is that this prevents haproxy from
6589 maintaining the client connection alive. A second effect is that a client or
6590 a cache could receive an incomplete response without being aware of it, and
6591 consider the response complete.
6592
6593 By setting "option http-pretend-keepalive", haproxy will make the server
6594 believe it will keep the connection alive. The server will then not fall back
6595 to the abnormal undesired above. When haproxy gets the whole response, it
6596 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006597 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006598 connection is correctly closed on the server side.
6599
6600 It is recommended not to enable this option by default, because most servers
6601 will more efficiently close the connection themselves after the last packet,
6602 and release its buffers slightly earlier. Also, the added packet on the
6603 network could slightly reduce the overall peak performance. However it is
6604 worth noting that when this option is enabled, haproxy will have slightly
6605 less work to do. So if haproxy is the bottleneck on the whole architecture,
6606 enabling this option might save a few CPU cycles.
6607
Christopher Faulet98db9762018-09-21 10:25:19 +02006608 This option may be set in backend and listen sections. Using it in a frontend
6609 section will be ignored and a warning will be reported during startup. It is
6610 a backend related option, so there is no real reason to set it on a
6611 frontend. This option may be combined with "option httpclose", which will
6612 cause keepalive to be announced to the server and close to be announced to
6613 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006614
6615 If this option has been enabled in a "defaults" section, it can be disabled
6616 in a specific instance by prepending the "no" keyword before it.
6617
Christopher Faulet315b39c2018-09-21 16:26:19 +02006618 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01006619 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006620
Willy Tarreauc27debf2008-01-06 08:57:02 +01006621
Willy Tarreaub608feb2010-01-02 22:47:18 +01006622option http-server-close
6623no option http-server-close
6624 Enable or disable HTTP connection closing on the server side
6625 May be used in sections : defaults | frontend | listen | backend
6626 yes | yes | yes | yes
6627 Arguments : none
6628
Willy Tarreau70dffda2014-01-30 03:07:23 +01006629 By default HAProxy operates in keep-alive mode with regards to persistent
6630 connections: for each connection it processes each request and response, and
6631 leaves the connection idle on both sides between the end of a response and
6632 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006633 as "option http-server-close", "option httpclose" or "option http-tunnel".
6634 Setting "option http-server-close" enables HTTP connection-close mode on the
6635 server side while keeping the ability to support HTTP keep-alive and
6636 pipelining on the client side. This provides the lowest latency on the client
6637 side (slow network) and the fastest session reuse on the server side to save
6638 server resources, similarly to "option httpclose". It also permits
6639 non-keepalive capable servers to be served in keep-alive mode to the clients
6640 if they conform to the requirements of RFC7230. Please note that some servers
6641 do not always conform to those requirements when they see "Connection: close"
6642 in the request. The effect will be that keep-alive will never be used. A
6643 workaround consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006644
6645 At the moment, logs will not indicate whether requests came from the same
6646 session or not. The accept date reported in the logs corresponds to the end
6647 of the previous request, and the request time corresponds to the time spent
6648 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01006649 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6650 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006651
6652 This option may be set both in a frontend and in a backend. It is enabled if
6653 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet315b39c2018-09-21 16:26:19 +02006654 It disables and replaces any previous "option httpclose", "option http-tunnel"
6655 or "option http-keep-alive". Please check section 4 ("Proxies") to see how
6656 this option combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006657
6658 If this option has been enabled in a "defaults" section, it can be disabled
6659 in a specific instance by prepending the "no" keyword before it.
6660
Christopher Faulet315b39c2018-09-21 16:26:19 +02006661 See also : "option httpclose", "option http-pretend-keepalive",
6662 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006663
6664
Christopher Faulet6c9bbb22019-03-26 21:37:23 +01006665option http-tunnel (deprecated)
6666no option http-tunnel (deprecated)
6667 Disable or enable HTTP connection processing after first transaction.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006668 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet4212a302018-09-21 10:42:19 +02006669 yes | yes | yes | no
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006670 Arguments : none
6671
Christopher Faulet6c9bbb22019-03-26 21:37:23 +01006672 Warning : Because it cannot work in HTTP/2, this option is deprecated and it
6673 is only supported on legacy HTTP frontends. In HTX, it is ignored and a
6674 warning is emitted during HAProxy startup.
6675
Willy Tarreau70dffda2014-01-30 03:07:23 +01006676 By default HAProxy operates in keep-alive mode with regards to persistent
6677 connections: for each connection it processes each request and response, and
6678 leaves the connection idle on both sides between the end of a response and
6679 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006680 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006681
6682 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006683 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01006684 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
6685 is normally not needed anymore unless in very specific cases such as when
6686 using an in-house protocol that looks like HTTP but is not compatible, or
6687 just to log one request per client in order to reduce log size. Note that
6688 everything which works at the HTTP level, including header parsing/addition,
6689 cookie processing or content switching will only work for the first request
6690 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006691
Christopher Faulet4212a302018-09-21 10:42:19 +02006692 This option may be set on frontend and listen sections. Using it on a backend
6693 section will be ignored and a warning will be reported during the startup. It
6694 is a frontend related option, so there is no real reason to set it on a
6695 backend.
6696
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006697 If this option has been enabled in a "defaults" section, it can be disabled
6698 in a specific instance by prepending the "no" keyword before it.
6699
Christopher Faulet315b39c2018-09-21 16:26:19 +02006700 See also : "option httpclose", "option http-server-close",
6701 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006702
6703
Willy Tarreau88d349d2010-01-25 12:15:43 +01006704option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01006705no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01006706 Make use of non-standard Proxy-Connection header instead of Connection
6707 May be used in sections : defaults | frontend | listen | backend
6708 yes | yes | yes | no
6709 Arguments : none
6710
Lukas Tribus23953682017-04-28 13:24:30 +00006711 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006712 Connection header to indicate their wish of persistent or non-persistent
6713 connections, both browsers and proxies ignore this header for proxied
6714 connections and make use of the undocumented, non-standard Proxy-Connection
6715 header instead. The issue begins when trying to put a load balancer between
6716 browsers and such proxies, because there will be a difference between what
6717 haproxy understands and what the client and the proxy agree on.
6718
6719 By setting this option in a frontend, haproxy can automatically switch to use
6720 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006721 defined here as one where the URI begins with neither a '/' nor a '*'. This
6722 is incompatible with the HTTP tunnel mode. Note that this option can only be
6723 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006724
Willy Tarreau844a7e72010-01-31 21:46:18 +01006725 Also, when this option is set, a request which requires authentication will
6726 automatically switch to use proxy authentication headers if it is itself a
6727 proxied request. That makes it possible to check or enforce authentication in
6728 front of an existing proxy.
6729
Willy Tarreau88d349d2010-01-25 12:15:43 +01006730 This option should normally never be used, except in front of a proxy.
6731
Christopher Faulet315b39c2018-09-21 16:26:19 +02006732 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006733
6734
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006735option http-use-htx
6736no option http-use-htx
6737 Switch to the new HTX internal representation for HTTP protocol elements
6738 May be used in sections : defaults | frontend | listen | backend
6739 yes | yes | yes | yes
6740 Arguments : none
6741
Christopher Faulet1d2b5862019-04-12 16:10:51 +02006742 Historically, the HTTP protocol is processed as-is. Inserting, deleting, or
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006743 modifying a header field requires to rewrite the affected part in the buffer
Christopher Faulet1d2b5862019-04-12 16:10:51 +02006744 and to move the buffer's tail accordingly. This mode is known as the legacy
6745 HTTP mode. Since this principle has deep roots in haproxy, the HTTP/2
6746 protocol is converted to HTTP/1.1 before being processed this way. It also
6747 results in the inability to establish HTTP/2 connections to servers because
6748 of the loss of HTTP/2 semantics in the HTTP/1 representation.
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006749
6750 HTX is the name of a totally new native internal representation for the HTTP
6751 protocol, that is agnostic to the version and aims at preserving semantics
6752 all along the chain. It relies on a fast parsing, tokenizing and indexing of
6753 the protocol elements so that no more memory moves are necessary and that
Christopher Faulet1d2b5862019-04-12 16:10:51 +02006754 most elements are directly accessed. It supports using either HTTP/1 or
6755 HTTP/2 on any side regardless of the other side's version. It also supports
6756 upgrades from TCP to HTTP and implicit ones from HTTP/1 to HTTP/2 (matching
6757 the HTTP/2 preface).
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006758
Christopher Faulet1d2b5862019-04-12 16:10:51 +02006759 This option indicates that HTX needs to be used. Since the version 2.0-dev3,
6760 the HTX is the default mode. To switch back on the legacy HTTP mode, the
6761 option must be explicitly disabled using the "no" prefix. For prior versions,
6762 the feature has incomplete functional coverage, so it is not enabled by
6763 default.
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006764
6765 See also : "mode http"
6766
6767
Willy Tarreaud63335a2010-02-26 12:56:52 +01006768option httpchk
6769option httpchk <uri>
6770option httpchk <method> <uri>
6771option httpchk <method> <uri> <version>
6772 Enable HTTP protocol to check on the servers health
6773 May be used in sections : defaults | frontend | listen | backend
6774 yes | no | yes | yes
6775 Arguments :
6776 <method> is the optional HTTP method used with the requests. When not set,
6777 the "OPTIONS" method is used, as it generally requires low server
6778 processing and is easy to filter out from the logs. Any method
6779 may be used, though it is not recommended to invent non-standard
6780 ones.
6781
6782 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6783 which is accessible by default on almost any server, but may be
6784 changed to any other URI. Query strings are permitted.
6785
6786 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6787 but some servers might behave incorrectly in HTTP 1.0, so turning
6788 it to HTTP/1.1 may sometimes help. Note that the Host field is
Christopher Fauletf304ad32020-04-09 08:44:06 +02006789 mandatory in HTTP/1.1, use "http-check send" directive to add it.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006790
6791 By default, server health checks only consist in trying to establish a TCP
6792 connection. When "option httpchk" is specified, a complete HTTP request is
6793 sent once the TCP connection is established, and responses 2xx and 3xx are
6794 considered valid, while all other ones indicate a server failure, including
6795 the lack of any response.
6796
6797 The port and interval are specified in the server configuration.
6798
6799 This option does not necessarily require an HTTP backend, it also works with
6800 plain TCP backends. This is particularly useful to check simple scripts bound
6801 to some dedicated ports using the inetd daemon.
6802
Christopher Fauletf304ad32020-04-09 08:44:06 +02006803 Note : For a while, there was no way to add headers or body in the request
6804 used for HTTP health checks. So a workaround was to hide it at the end
6805 of the version string with a "\r\n" after the version. It is now
6806 deprecated. The directive "http-check send" must be used instead.
6807
Willy Tarreaud63335a2010-02-26 12:56:52 +01006808 Examples :
6809 # Relay HTTPS traffic to Apache instance and check service availability
6810 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6811 backend https_relay
6812 mode tcp
Christopher Fauletf304ad32020-04-09 08:44:06 +02006813 option httpchk OPTIONS * HTTP/1.1
6814 http-check send hdr Host www
Willy Tarreaud63335a2010-02-26 12:56:52 +01006815 server apache1 192.168.1.1:443 check port 80
6816
Simon Hormanafc47ee2013-11-25 10:46:35 +09006817 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6818 "option pgsql-check", "http-check" and the "check", "port" and
6819 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006820
6821
Willy Tarreauc27debf2008-01-06 08:57:02 +01006822option httpclose
6823no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006824 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006825 May be used in sections : defaults | frontend | listen | backend
6826 yes | yes | yes | yes
6827 Arguments : none
6828
Willy Tarreau70dffda2014-01-30 03:07:23 +01006829 By default HAProxy operates in keep-alive mode with regards to persistent
6830 connections: for each connection it processes each request and response, and
6831 leaves the connection idle on both sides between the end of a response and
6832 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006833 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006834
Christopher Faulet315b39c2018-09-21 16:26:19 +02006835 If "option httpclose" is set, HAProxy will close connections with the server
6836 and the client as soon as the request and the response are received. It will
John Roesler27c754c2019-07-10 15:45:51 -05006837 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02006838 and will add one if missing. Any "Connection" header different from "close"
6839 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006840
Christopher Faulet315b39c2018-09-21 16:26:19 +02006841 This option may also be combined with "option http-pretend-keepalive", which
6842 will disable sending of the "Connection: close" header, but will still cause
6843 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006844
6845 This option may be set both in a frontend and in a backend. It is enabled if
6846 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01006847 It disables and replaces any previous "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006848 "option http-keep-alive" or "option http-tunnel". Please check section 4
6849 ("Proxies") to see how this option combines with others when frontend and
6850 backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006851
6852 If this option has been enabled in a "defaults" section, it can be disabled
6853 in a specific instance by prepending the "no" keyword before it.
6854
Christopher Faulet315b39c2018-09-21 16:26:19 +02006855 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006856
6857
Emeric Brun3a058f32009-06-30 18:26:00 +02006858option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006859 Enable logging of HTTP request, session state and timers
6860 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006861 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006862 Arguments :
6863 clf if the "clf" argument is added, then the output format will be
6864 the CLF format instead of HAProxy's default HTTP format. You can
6865 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006866 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006867 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006868
6869 By default, the log output format is very poor, as it only contains the
6870 source and destination addresses, and the instance name. By specifying
6871 "option httplog", each log line turns into a much richer format including,
6872 but not limited to, the HTTP request, the connection timers, the session
6873 status, the connections numbers, the captured headers and cookies, the
6874 frontend, backend and server name, and of course the source address and
6875 ports.
6876
PiBa-NLbd556bf2014-12-11 21:31:54 +01006877 Specifying only "option httplog" will automatically clear the 'clf' mode
6878 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006879
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006880 "option httplog" overrides any previous "log-format" directive.
6881
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006882 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006883
Willy Tarreau55165fe2009-05-10 12:02:55 +02006884
6885option http_proxy
6886no option http_proxy
6887 Enable or disable plain HTTP proxy mode
6888 May be used in sections : defaults | frontend | listen | backend
6889 yes | yes | yes | yes
6890 Arguments : none
6891
6892 It sometimes happens that people need a pure HTTP proxy which understands
6893 basic proxy requests without caching nor any fancy feature. In this case,
6894 it may be worth setting up an HAProxy instance with the "option http_proxy"
6895 set. In this mode, no server is declared, and the connection is forwarded to
6896 the IP address and port found in the URL after the "http://" scheme.
6897
6898 No host address resolution is performed, so this only works when pure IP
6899 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006900 it will probably be used only by experts who know they need exactly it. This
6901 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006902
6903 If this option has been enabled in a "defaults" section, it can be disabled
6904 in a specific instance by prepending the "no" keyword before it.
6905
6906 Example :
6907 # this backend understands HTTP proxy requests and forwards them directly.
6908 backend direct_forward
6909 option httpclose
6910 option http_proxy
6911
6912 See also : "option httpclose"
6913
Willy Tarreau211ad242009-10-03 21:45:07 +02006914
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006915option independent-streams
6916no option independent-streams
6917 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006918 May be used in sections : defaults | frontend | listen | backend
6919 yes | yes | yes | yes
6920 Arguments : none
6921
6922 By default, when data is sent over a socket, both the write timeout and the
6923 read timeout for that socket are refreshed, because we consider that there is
6924 activity on that socket, and we have no other means of guessing if we should
6925 receive data or not.
6926
Davor Ocelice9ed2812017-12-25 17:49:28 +01006927 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006928 exists a situation where it is desirable to disable it, and only refresh the
6929 read timeout if there are incoming data. This happens on sessions with large
6930 timeouts and low amounts of exchanged data such as telnet session. If the
6931 server suddenly disappears, the output data accumulates in the system's
6932 socket buffers, both timeouts are correctly refreshed, and there is no way
6933 to know the server does not receive them, so we don't timeout. However, when
6934 the underlying protocol always echoes sent data, it would be enough by itself
6935 to detect the issue using the read timeout. Note that this problem does not
6936 happen with more verbose protocols because data won't accumulate long in the
6937 socket buffers.
6938
6939 When this option is set on the frontend, it will disable read timeout updates
6940 on data sent to the client. There probably is little use of this case. When
6941 the option is set on the backend, it will disable read timeout updates on
6942 data sent to the server. Doing so will typically break large HTTP posts from
6943 slow lines, so use it with caution.
6944
Lukas Tribus745f15e2018-11-08 12:41:42 +01006945 Note: older versions used to call this setting "option independant-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006946 with a spelling mistake. This spelling is still supported but
6947 deprecated.
6948
Willy Tarreauce887fd2012-05-12 12:50:00 +02006949 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006950
6951
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006952option ldap-check
6953 Use LDAPv3 health checks for server testing
6954 May be used in sections : defaults | frontend | listen | backend
6955 yes | no | yes | yes
6956 Arguments : none
6957
6958 It is possible to test that the server correctly talks LDAPv3 instead of just
6959 testing that it accepts the TCP connection. When this option is set, an
6960 LDAPv3 anonymous simple bind message is sent to the server, and the response
6961 is analyzed to find an LDAPv3 bind response message.
6962
6963 The server is considered valid only when the LDAP response contains success
6964 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6965
6966 Logging of bind requests is server dependent see your documentation how to
6967 configure it.
6968
6969 Example :
6970 option ldap-check
6971
6972 See also : "option httpchk"
6973
6974
Simon Horman98637e52014-06-20 12:30:16 +09006975option external-check
6976 Use external processes for server health checks
6977 May be used in sections : defaults | frontend | listen | backend
6978 yes | no | yes | yes
6979
6980 It is possible to test the health of a server using an external command.
6981 This is achieved by running the executable set using "external-check
6982 command".
6983
6984 Requires the "external-check" global to be set.
6985
6986 See also : "external-check", "external-check command", "external-check path"
6987
6988
Willy Tarreau211ad242009-10-03 21:45:07 +02006989option log-health-checks
6990no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006991 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006992 May be used in sections : defaults | frontend | listen | backend
6993 yes | no | yes | yes
6994 Arguments : none
6995
Willy Tarreaubef1b322014-05-13 21:01:39 +02006996 By default, failed health check are logged if server is UP and successful
6997 health checks are logged if server is DOWN, so the amount of additional
6998 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006999
Willy Tarreaubef1b322014-05-13 21:01:39 +02007000 When this option is enabled, any change of the health check status or to
7001 the server's health will be logged, so that it becomes possible to know
7002 that a server was failing occasional checks before crashing, or exactly when
7003 it failed to respond a valid HTTP status, then when the port started to
7004 reject connections, then when the server stopped responding at all.
7005
Davor Ocelice9ed2812017-12-25 17:49:28 +01007006 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02007007 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02007008
Willy Tarreaubef1b322014-05-13 21:01:39 +02007009 See also: "option httpchk", "option ldap-check", "option mysql-check",
7010 "option pgsql-check", "option redis-check", "option smtpchk",
7011 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02007012
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007013
7014option log-separate-errors
7015no option log-separate-errors
7016 Change log level for non-completely successful connections
7017 May be used in sections : defaults | frontend | listen | backend
7018 yes | yes | yes | no
7019 Arguments : none
7020
7021 Sometimes looking for errors in logs is not easy. This option makes haproxy
7022 raise the level of logs containing potentially interesting information such
7023 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
7024 level changes from "info" to "err". This makes it possible to log them
7025 separately to a different file with most syslog daemons. Be careful not to
7026 remove them from the original file, otherwise you would lose ordering which
7027 provides very important information.
7028
7029 Using this option, large sites dealing with several thousand connections per
7030 second may log normal traffic to a rotating buffer and only archive smaller
7031 error logs.
7032
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007033 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02007034 logging.
7035
Willy Tarreauc27debf2008-01-06 08:57:02 +01007036
7037option logasap
7038no option logasap
Jerome Magnina1d4a732020-04-23 19:01:17 +02007039 Enable or disable early logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007040 May be used in sections : defaults | frontend | listen | backend
7041 yes | yes | yes | no
7042 Arguments : none
7043
Jerome Magnina1d4a732020-04-23 19:01:17 +02007044 By default, logs are emitted when all the log format variables and sample
7045 fetches used in the definition of the log-format string return a value, or
7046 when the session is terminated. This allows the built in log-format strings
7047 to account for the transfer time, or the number of bytes in log messages.
7048
7049 When handling long lived connections such as large file transfers or RDP,
7050 it may take a while for the request or connection to appear in the logs.
7051 Using "option logasap", the log message is created as soon as the server
7052 connection is established in mode tcp, or as soon as the server sends the
7053 complete headers in mode http. Missing information in the logs will be the
7054 total number of bytes which will only indicate the amount of data transfered
7055 before the message was created and the total time which will not take the
7056 remainder of the connection life or transfer time into account. For the case
7057 of HTTP, it is good practice to capture the Content-Length response header
7058 so that the logs at least indicate how many bytes are expected to be
7059 transfered.
Willy Tarreauc27debf2008-01-06 08:57:02 +01007060
Willy Tarreaucc6c8912009-02-22 10:53:55 +01007061 Examples :
7062 listen http_proxy 0.0.0.0:80
7063 mode http
7064 option httplog
7065 option logasap
7066 log 192.168.2.200 local3
7067
7068 >>> Feb 6 12:14:14 localhost \
7069 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
7070 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
7071 "GET /image.iso HTTP/1.0"
7072
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007073 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01007074 logging.
7075
7076
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02007077option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007078 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007079 May be used in sections : defaults | frontend | listen | backend
7080 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007081 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02007082 <username> This is the username which will be used when connecting to MySQL
7083 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02007084 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007085
7086 If you specify a username, the check consists of sending two MySQL packet,
7087 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01007088 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007089 Error packet. It is a basic but useful test which does not produce error nor
7090 aborted connect on the server. However, it requires adding an authorization
7091 in the MySQL table, like this :
7092
7093 USE mysql;
7094 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
7095 FLUSH PRIVILEGES;
7096
7097 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007098 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007099 Error packet, we don't send anything in this mode. It was reported that it
7100 can generate lockout if check is too frequent and/or if there is not enough
7101 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
7102 value as if a connection is established successfully within fewer than MySQL
7103 "max_connect_errors" attempts after a previous connection was interrupted,
7104 the error count for the host is cleared to zero. If HAProxy's server get
7105 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
7106
7107 Remember that this does not check database presence nor database consistency.
7108 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007109
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02007110 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007111
7112 Most often, an incoming MySQL server needs to see the client's IP address for
7113 various purposes, including IP privilege matching and connection logging.
7114 When possible, it is often wise to masquerade the client's IP address when
7115 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007116 which requires the transparent proxy feature to be compiled in, and the MySQL
7117 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007118
7119 See also: "option httpchk"
7120
7121
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007122option nolinger
7123no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007124 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007125 May be used in sections: defaults | frontend | listen | backend
7126 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007127 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007128
Davor Ocelice9ed2812017-12-25 17:49:28 +01007129 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007130 physically disconnected), the session timeouts triggers and the session is
7131 closed. But it will remain in FIN_WAIT1 state for some time in the system,
7132 using some resources and possibly limiting the ability to establish newer
7133 connections.
7134
7135 When this happens, it is possible to activate "option nolinger" which forces
7136 the system to immediately remove any socket's pending data on close. Thus,
7137 the session is instantly purged from the system's tables. This usually has
7138 side effects such as increased number of TCP resets due to old retransmits
7139 getting immediately rejected. Some firewalls may sometimes complain about
7140 this too.
7141
7142 For this reason, it is not recommended to use this option when not absolutely
7143 needed. You know that you need it when you have thousands of FIN_WAIT1
7144 sessions on your system (TIME_WAIT ones do not count).
7145
7146 This option may be used both on frontends and backends, depending on the side
7147 where it is required. Use it on the frontend for clients, and on the backend
7148 for servers.
7149
7150 If this option has been enabled in a "defaults" section, it can be disabled
7151 in a specific instance by prepending the "no" keyword before it.
7152
7153
Willy Tarreau55165fe2009-05-10 12:02:55 +02007154option originalto [ except <network> ] [ header <name> ]
7155 Enable insertion of the X-Original-To header to requests sent to servers
7156 May be used in sections : defaults | frontend | listen | backend
7157 yes | yes | yes | yes
7158 Arguments :
7159 <network> is an optional argument used to disable this option for sources
7160 matching <network>
7161 <name> an optional argument to specify a different "X-Original-To"
7162 header name.
7163
7164 Since HAProxy can work in transparent mode, every request from a client can
7165 be redirected to the proxy and HAProxy itself can proxy every request to a
7166 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
7167 be lost. This is annoying when you want access rules based on destination ip
7168 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
7169 added by HAProxy to all requests sent to the server. This header contains a
7170 value representing the original destination IP address. Since this must be
7171 configured to always use the last occurrence of this header only. Note that
7172 only the last occurrence of the header must be used, since it is really
7173 possible that the client has already brought one.
7174
7175 The keyword "header" may be used to supply a different header name to replace
7176 the default "X-Original-To". This can be useful where you might already
7177 have a "X-Original-To" header from a different application, and you need
7178 preserve it. Also if your backend server doesn't use the "X-Original-To"
7179 header and requires different one.
7180
7181 Sometimes, a same HAProxy instance may be shared between a direct client
7182 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
7183 used to decrypt HTTPS traffic). It is possible to disable the addition of the
7184 header for a known source address or network by adding the "except" keyword
7185 followed by the network address. In this case, any source IP matching the
7186 network will not cause an addition of this header. Most common uses are with
7187 private networks or 127.0.0.1.
7188
7189 This option may be specified either in the frontend or in the backend. If at
7190 least one of them uses it, the header will be added. Note that the backend's
7191 setting of the header subargument takes precedence over the frontend's if
7192 both are defined.
7193
Willy Tarreau55165fe2009-05-10 12:02:55 +02007194 Examples :
7195 # Original Destination address
7196 frontend www
7197 mode http
7198 option originalto except 127.0.0.1
7199
7200 # Those servers want the IP Address in X-Client-Dst
7201 backend www
7202 mode http
7203 option originalto header X-Client-Dst
7204
Christopher Faulet315b39c2018-09-21 16:26:19 +02007205 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02007206
7207
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007208option persist
7209no option persist
7210 Enable or disable forced persistence on down servers
7211 May be used in sections: defaults | frontend | listen | backend
7212 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007213 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007214
7215 When an HTTP request reaches a backend with a cookie which references a dead
7216 server, by default it is redispatched to another server. It is possible to
7217 force the request to be sent to the dead server first using "option persist"
7218 if absolutely needed. A common use case is when servers are under extreme
7219 load and spend their time flapping. In this case, the users would still be
7220 directed to the server they opened the session on, in the hope they would be
7221 correctly served. It is recommended to use "option redispatch" in conjunction
7222 with this option so that in the event it would not be possible to connect to
7223 the server at all (server definitely dead), the client would finally be
7224 redirected to another valid server.
7225
7226 If this option has been enabled in a "defaults" section, it can be disabled
7227 in a specific instance by prepending the "no" keyword before it.
7228
Willy Tarreau4de91492010-01-22 19:10:05 +01007229 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007230
7231
Willy Tarreau0c122822013-12-15 18:49:01 +01007232option pgsql-check [ user <username> ]
7233 Use PostgreSQL health checks for server testing
7234 May be used in sections : defaults | frontend | listen | backend
7235 yes | no | yes | yes
7236 Arguments :
7237 <username> This is the username which will be used when connecting to
7238 PostgreSQL server.
7239
7240 The check sends a PostgreSQL StartupMessage and waits for either
7241 Authentication request or ErrorResponse message. It is a basic but useful
7242 test which does not produce error nor aborted connect on the server.
7243 This check is identical with the "mysql-check".
7244
7245 See also: "option httpchk"
7246
7247
Willy Tarreau9420b122013-12-15 18:58:25 +01007248option prefer-last-server
7249no option prefer-last-server
7250 Allow multiple load balanced requests to remain on the same server
7251 May be used in sections: defaults | frontend | listen | backend
7252 yes | no | yes | yes
7253 Arguments : none
7254
7255 When the load balancing algorithm in use is not deterministic, and a previous
7256 request was sent to a server to which haproxy still holds a connection, it is
7257 sometimes desirable that subsequent requests on a same session go to the same
7258 server as much as possible. Note that this is different from persistence, as
7259 we only indicate a preference which haproxy tries to apply without any form
7260 of warranty. The real use is for keep-alive connections sent to servers. When
7261 this option is used, haproxy will try to reuse the same connection that is
7262 attached to the server instead of rebalancing to another server, causing a
7263 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01007264 not make much sense to use this in combination with hashing algorithms. Note,
7265 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02007266 to a proxy which sends a 407 (authentication required), when the load
7267 balancing algorithm is not deterministic. This is mandatory for use with the
7268 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01007269 troubleshooting some faulty applications. Option prefer-last-server might be
7270 desirable in these environments as well, to avoid redistributing the traffic
7271 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01007272
7273 If this option has been enabled in a "defaults" section, it can be disabled
7274 in a specific instance by prepending the "no" keyword before it.
7275
7276 See also: "option http-keep-alive"
7277
7278
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007279option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07007280option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007281no option redispatch
7282 Enable or disable session redistribution in case of connection failure
7283 May be used in sections: defaults | frontend | listen | backend
7284 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07007285 Arguments :
7286 <interval> The optional integer value that controls how often redispatches
7287 occur when retrying connections. Positive value P indicates a
7288 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01007289 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07007290 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007291 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07007292 positive value of 1 would indicate a redispatch on every retry,
7293 and a positive value of 3 would indicate a redispatch on every
7294 third retry. You can disable redispatches with a value of 0.
7295
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007296
7297 In HTTP mode, if a server designated by a cookie is down, clients may
7298 definitely stick to it because they cannot flush the cookie, so they will not
7299 be able to access the service anymore.
7300
Willy Tarreau59884a62019-01-02 14:48:31 +01007301 Specifying "option redispatch" will allow the proxy to break cookie or
7302 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007303
Joseph Lynch726ab712015-05-11 23:25:34 -07007304 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007305 connection failures. Of course, it requires having "retries" set to a nonzero
7306 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007307
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007308 This form is the preferred form, which replaces both the "redispatch" and
7309 "redisp" keywords.
7310
7311 If this option has been enabled in a "defaults" section, it can be disabled
7312 in a specific instance by prepending the "no" keyword before it.
7313
Willy Tarreau4de91492010-01-22 19:10:05 +01007314 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007315
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007316
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007317option redis-check
7318 Use redis health checks for server testing
7319 May be used in sections : defaults | frontend | listen | backend
7320 yes | no | yes | yes
7321 Arguments : none
7322
7323 It is possible to test that the server correctly talks REDIS protocol instead
7324 of just testing that it accepts the TCP connection. When this option is set,
7325 a PING redis command is sent to the server, and the response is analyzed to
7326 find the "+PONG" response message.
7327
7328 Example :
7329 option redis-check
7330
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03007331 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007332
7333
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007334option smtpchk
7335option smtpchk <hello> <domain>
7336 Use SMTP health checks for server testing
7337 May be used in sections : defaults | frontend | listen | backend
7338 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01007339 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007340 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02007341 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007342 values will be turned into the default command ("HELO").
7343
7344 <domain> is the domain name to present to the server. It may only be
7345 specified (and is mandatory) if the hello command has been
7346 specified. By default, "localhost" is used.
7347
7348 When "option smtpchk" is set, the health checks will consist in TCP
7349 connections followed by an SMTP command. By default, this command is
7350 "HELO localhost". The server's return code is analyzed and only return codes
7351 starting with a "2" will be considered as valid. All other responses,
7352 including a lack of response will constitute an error and will indicate a
7353 dead server.
7354
7355 This test is meant to be used with SMTP servers or relays. Depending on the
7356 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01007357 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007358 25 is often easier than adjusting the configuration.
7359
7360 Most often, an incoming SMTP server needs to see the client's IP address for
7361 various purposes, including spam filtering, anti-spoofing and logging. When
7362 possible, it is often wise to masquerade the client's IP address when
7363 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007364 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007365
7366 Example :
7367 option smtpchk HELO mydomain.org
7368
7369 See also : "option httpchk", "source"
7370
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007371
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02007372option socket-stats
7373no option socket-stats
7374
7375 Enable or disable collecting & providing separate statistics for each socket.
7376 May be used in sections : defaults | frontend | listen | backend
7377 yes | yes | yes | no
7378
7379 Arguments : none
7380
7381
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007382option splice-auto
7383no option splice-auto
7384 Enable or disable automatic kernel acceleration on sockets in both directions
7385 May be used in sections : defaults | frontend | listen | backend
7386 yes | yes | yes | yes
7387 Arguments : none
7388
7389 When this option is enabled either on a frontend or on a backend, haproxy
7390 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007391 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007392 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007393 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007394 are not much aggressive in order to limit excessive use of splicing. This
7395 option requires splicing to be enabled at compile time, and may be globally
7396 disabled with the global option "nosplice". Since splice uses pipes, using it
7397 requires that there are enough spare pipes.
7398
7399 Important note: kernel-based TCP splicing is a Linux-specific feature which
7400 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
7401 transfer data between sockets without copying these data to user-space, thus
7402 providing noticeable performance gains and CPU cycles savings. Since many
7403 early implementations are buggy, corrupt data and/or are inefficient, this
7404 feature is not enabled by default, and it should be used with extreme care.
7405 While it is not possible to detect the correctness of an implementation,
7406 2.6.29 is the first version offering a properly working implementation. In
7407 case of doubt, splicing may be globally disabled using the global "nosplice"
7408 keyword.
7409
7410 Example :
7411 option splice-auto
7412
7413 If this option has been enabled in a "defaults" section, it can be disabled
7414 in a specific instance by prepending the "no" keyword before it.
7415
7416 See also : "option splice-request", "option splice-response", and global
7417 options "nosplice" and "maxpipes"
7418
7419
7420option splice-request
7421no option splice-request
7422 Enable or disable automatic kernel acceleration on sockets for requests
7423 May be used in sections : defaults | frontend | listen | backend
7424 yes | yes | yes | yes
7425 Arguments : none
7426
7427 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007428 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007429 the client to the server. It might still use the recv/send scheme if there
7430 are no spare pipes left. This option requires splicing to be enabled at
7431 compile time, and may be globally disabled with the global option "nosplice".
7432 Since splice uses pipes, using it requires that there are enough spare pipes.
7433
7434 Important note: see "option splice-auto" for usage limitations.
7435
7436 Example :
7437 option splice-request
7438
7439 If this option has been enabled in a "defaults" section, it can be disabled
7440 in a specific instance by prepending the "no" keyword before it.
7441
7442 See also : "option splice-auto", "option splice-response", and global options
7443 "nosplice" and "maxpipes"
7444
7445
7446option splice-response
7447no option splice-response
7448 Enable or disable automatic kernel acceleration on sockets for responses
7449 May be used in sections : defaults | frontend | listen | backend
7450 yes | yes | yes | yes
7451 Arguments : none
7452
7453 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007454 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007455 the server to the client. It might still use the recv/send scheme if there
7456 are no spare pipes left. This option requires splicing to be enabled at
7457 compile time, and may be globally disabled with the global option "nosplice".
7458 Since splice uses pipes, using it requires that there are enough spare pipes.
7459
7460 Important note: see "option splice-auto" for usage limitations.
7461
7462 Example :
7463 option splice-response
7464
7465 If this option has been enabled in a "defaults" section, it can be disabled
7466 in a specific instance by prepending the "no" keyword before it.
7467
7468 See also : "option splice-auto", "option splice-request", and global options
7469 "nosplice" and "maxpipes"
7470
7471
Christopher Fauletba7bc162016-11-07 21:07:38 +01007472option spop-check
7473 Use SPOP health checks for server testing
7474 May be used in sections : defaults | frontend | listen | backend
7475 no | no | no | yes
7476 Arguments : none
7477
7478 It is possible to test that the server correctly talks SPOP protocol instead
7479 of just testing that it accepts the TCP connection. When this option is set,
7480 a HELLO handshake is performed between HAProxy and the server, and the
7481 response is analyzed to check no error is reported.
7482
7483 Example :
7484 option spop-check
7485
7486 See also : "option httpchk"
7487
7488
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007489option srvtcpka
7490no option srvtcpka
7491 Enable or disable the sending of TCP keepalive packets on the server side
7492 May be used in sections : defaults | frontend | listen | backend
7493 yes | no | yes | yes
7494 Arguments : none
7495
7496 When there is a firewall or any session-aware component between a client and
7497 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007498 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007499 components decides to expire a session which has remained idle for too long.
7500
7501 Enabling socket-level TCP keep-alives makes the system regularly send packets
7502 to the other end of the connection, leaving it active. The delay between
7503 keep-alive probes is controlled by the system only and depends both on the
7504 operating system and its tuning parameters.
7505
7506 It is important to understand that keep-alive packets are neither emitted nor
7507 received at the application level. It is only the network stacks which sees
7508 them. For this reason, even if one side of the proxy already uses keep-alives
7509 to maintain its connection alive, those keep-alive packets will not be
7510 forwarded to the other side of the proxy.
7511
7512 Please note that this has nothing to do with HTTP keep-alive.
7513
7514 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
7515 server side of a connection, which should help when session expirations are
7516 noticed between HAProxy and a server.
7517
7518 If this option has been enabled in a "defaults" section, it can be disabled
7519 in a specific instance by prepending the "no" keyword before it.
7520
7521 See also : "option clitcpka", "option tcpka"
7522
7523
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007524option ssl-hello-chk
7525 Use SSLv3 client hello health checks for server testing
7526 May be used in sections : defaults | frontend | listen | backend
7527 yes | no | yes | yes
7528 Arguments : none
7529
7530 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
7531 possible to test that the server correctly talks SSL instead of just testing
7532 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
7533 SSLv3 client hello messages are sent once the connection is established to
7534 the server, and the response is analyzed to find an SSL server hello message.
7535 The server is considered valid only when the response contains this server
7536 hello message.
7537
7538 All servers tested till there correctly reply to SSLv3 client hello messages,
7539 and most servers tested do not even log the requests containing only hello
7540 messages, which is appreciable.
7541
Willy Tarreau763a95b2012-10-04 23:15:39 +02007542 Note that this check works even when SSL support was not built into haproxy
7543 because it forges the SSL message. When SSL support is available, it is best
7544 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007545
Willy Tarreau763a95b2012-10-04 23:15:39 +02007546 See also: "option httpchk", "check-ssl"
7547
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007548
Willy Tarreaued179852013-12-16 01:07:00 +01007549option tcp-check
7550 Perform health checks using tcp-check send/expect sequences
7551 May be used in sections: defaults | frontend | listen | backend
7552 yes | no | yes | yes
7553
7554 This health check method is intended to be combined with "tcp-check" command
7555 lists in order to support send/expect types of health check sequences.
7556
7557 TCP checks currently support 4 modes of operations :
7558 - no "tcp-check" directive : the health check only consists in a connection
7559 attempt, which remains the default mode.
7560
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007561 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01007562 used to send a string along with a connection opening. With some
7563 protocols, it helps sending a "QUIT" message for example that prevents
7564 the server from logging a connection error for each health check. The
7565 check result will still be based on the ability to open the connection
7566 only.
7567
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007568 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01007569 The connection is opened and haproxy waits for the server to present some
7570 contents which must validate some rules. The check result will be based
7571 on the matching between the contents and the rules. This is suited for
7572 POP, IMAP, SMTP, FTP, SSH, TELNET.
7573
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007574 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007575 used to test a hello-type protocol. HAProxy sends a message, the server
7576 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007577 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007578 suited for protocols which require a binding or a request/response model.
7579 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7580 already all have their dedicated checks with a deeper understanding of
7581 the respective protocols.
7582 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007583 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007584
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007585 A fifth mode can be used to insert comments in different steps of the
7586 script.
7587
7588 For each tcp-check rule you create, you can add a "comment" directive,
7589 followed by a string. This string will be reported in the log and stderr
7590 in debug mode. It is useful to make user-friendly error reporting.
7591 The "comment" is of course optional.
7592
7593
Willy Tarreaued179852013-12-16 01:07:00 +01007594 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007595 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007596 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007597 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007598
Davor Ocelice9ed2812017-12-25 17:49:28 +01007599 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007600 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007601 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007602
7603 # look for the redis master server after ensuring it speaks well
7604 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007605 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007606 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007607 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007608 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007609 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007610 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007611 tcp-check send info\ replication\r\n
7612 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007613 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007614 tcp-check send QUIT\r\n
7615 tcp-check expect string +OK
7616
Davor Ocelice9ed2812017-12-25 17:49:28 +01007617 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007618 (send many headers before analyzing)
7619 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007620 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007621 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7622 tcp-check send Host:\ www.mydomain.com\r\n
7623 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7624 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007625 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007626
7627
7628 See also : "tcp-check expect", "tcp-check send"
7629
7630
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007631option tcp-smart-accept
7632no option tcp-smart-accept
7633 Enable or disable the saving of one ACK packet during the accept sequence
7634 May be used in sections : defaults | frontend | listen | backend
7635 yes | yes | yes | no
7636 Arguments : none
7637
7638 When an HTTP connection request comes in, the system acknowledges it on
7639 behalf of HAProxy, then the client immediately sends its request, and the
7640 system acknowledges it too while it is notifying HAProxy about the new
7641 connection. HAProxy then reads the request and responds. This means that we
7642 have one TCP ACK sent by the system for nothing, because the request could
7643 very well be acknowledged by HAProxy when it sends its response.
7644
7645 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7646 sending this useless ACK on platforms which support it (currently at least
7647 Linux). It must not cause any problem, because the system will send it anyway
7648 after 40 ms if the response takes more time than expected to come.
7649
7650 During complex network debugging sessions, it may be desirable to disable
7651 this optimization because delayed ACKs can make troubleshooting more complex
7652 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007653 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007654
7655 It is also possible to force it for non-HTTP proxies by simply specifying
7656 "option tcp-smart-accept". For instance, it can make sense with some services
7657 such as SMTP where the server speaks first.
7658
7659 It is recommended to avoid forcing this option in a defaults section. In case
7660 of doubt, consider setting it back to automatic values by prepending the
7661 "default" keyword before it, or disabling it using the "no" keyword.
7662
Willy Tarreaud88edf22009-06-14 15:48:17 +02007663 See also : "option tcp-smart-connect"
7664
7665
7666option tcp-smart-connect
7667no option tcp-smart-connect
7668 Enable or disable the saving of one ACK packet during the connect sequence
7669 May be used in sections : defaults | frontend | listen | backend
7670 yes | no | yes | yes
7671 Arguments : none
7672
7673 On certain systems (at least Linux), HAProxy can ask the kernel not to
7674 immediately send an empty ACK upon a connection request, but to directly
7675 send the buffer request instead. This saves one packet on the network and
7676 thus boosts performance. It can also be useful for some servers, because they
7677 immediately get the request along with the incoming connection.
7678
7679 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7680 It is not enabled by default because it makes network troubleshooting more
7681 complex.
7682
7683 It only makes sense to enable it with protocols where the client speaks first
7684 such as HTTP. In other situations, if there is no data to send in place of
7685 the ACK, a normal ACK is sent.
7686
7687 If this option has been enabled in a "defaults" section, it can be disabled
7688 in a specific instance by prepending the "no" keyword before it.
7689
7690 See also : "option tcp-smart-accept"
7691
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007692
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007693option tcpka
7694 Enable or disable the sending of TCP keepalive packets on both sides
7695 May be used in sections : defaults | frontend | listen | backend
7696 yes | yes | yes | yes
7697 Arguments : none
7698
7699 When there is a firewall or any session-aware component between a client and
7700 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007701 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007702 components decides to expire a session which has remained idle for too long.
7703
7704 Enabling socket-level TCP keep-alives makes the system regularly send packets
7705 to the other end of the connection, leaving it active. The delay between
7706 keep-alive probes is controlled by the system only and depends both on the
7707 operating system and its tuning parameters.
7708
7709 It is important to understand that keep-alive packets are neither emitted nor
7710 received at the application level. It is only the network stacks which sees
7711 them. For this reason, even if one side of the proxy already uses keep-alives
7712 to maintain its connection alive, those keep-alive packets will not be
7713 forwarded to the other side of the proxy.
7714
7715 Please note that this has nothing to do with HTTP keep-alive.
7716
7717 Using option "tcpka" enables the emission of TCP keep-alive probes on both
7718 the client and server sides of a connection. Note that this is meaningful
7719 only in "defaults" or "listen" sections. If this option is used in a
7720 frontend, only the client side will get keep-alives, and if this option is
7721 used in a backend, only the server side will get keep-alives. For this
7722 reason, it is strongly recommended to explicitly use "option clitcpka" and
7723 "option srvtcpka" when the configuration is split between frontends and
7724 backends.
7725
7726 See also : "option clitcpka", "option srvtcpka"
7727
Willy Tarreau844e3c52008-01-11 16:28:18 +01007728
7729option tcplog
7730 Enable advanced logging of TCP connections with session state and timers
7731 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007732 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007733 Arguments : none
7734
7735 By default, the log output format is very poor, as it only contains the
7736 source and destination addresses, and the instance name. By specifying
7737 "option tcplog", each log line turns into a much richer format including, but
7738 not limited to, the connection timers, the session status, the connections
7739 numbers, the frontend, backend and server name, and of course the source
7740 address and ports. This option is useful for pure TCP proxies in order to
7741 find which of the client or server disconnects or times out. For normal HTTP
7742 proxies, it's better to use "option httplog" which is even more complete.
7743
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007744 "option tcplog" overrides any previous "log-format" directive.
7745
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007746 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007747
7748
Willy Tarreau844e3c52008-01-11 16:28:18 +01007749option transparent
7750no option transparent
7751 Enable client-side transparent proxying
7752 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007753 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007754 Arguments : none
7755
7756 This option was introduced in order to provide layer 7 persistence to layer 3
7757 load balancers. The idea is to use the OS's ability to redirect an incoming
7758 connection for a remote address to a local process (here HAProxy), and let
7759 this process know what address was initially requested. When this option is
7760 used, sessions without cookies will be forwarded to the original destination
7761 IP address of the incoming request (which should match that of another
7762 equipment), while requests with cookies will still be forwarded to the
7763 appropriate server.
7764
7765 Note that contrary to a common belief, this option does NOT make HAProxy
7766 present the client's IP to the server when establishing the connection.
7767
Willy Tarreaua1146052011-03-01 09:51:54 +01007768 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007769 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007770
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007771
Simon Horman98637e52014-06-20 12:30:16 +09007772external-check command <command>
7773 Executable to run when performing an external-check
7774 May be used in sections : defaults | frontend | listen | backend
7775 yes | no | yes | yes
7776
7777 Arguments :
7778 <command> is the external command to run
7779
Simon Horman98637e52014-06-20 12:30:16 +09007780 The arguments passed to the to the command are:
7781
Cyril Bonté777be862014-12-02 21:21:35 +01007782 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007783
Cyril Bonté777be862014-12-02 21:21:35 +01007784 The <proxy_address> and <proxy_port> are derived from the first listener
7785 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7786 listener the proxy_address will be the path of the socket and the
7787 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7788 possible to determine a listener, and both <proxy_address> and <proxy_port>
7789 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007790
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007791 Some values are also provided through environment variables.
7792
7793 Environment variables :
7794 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7795 applicable, for example in a "backend" section).
7796
7797 HAPROXY_PROXY_ID The backend id.
7798
7799 HAPROXY_PROXY_NAME The backend name.
7800
7801 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7802 applicable, for example in a "backend" section or
7803 for a UNIX socket).
7804
7805 HAPROXY_SERVER_ADDR The server address.
7806
7807 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7808
7809 HAPROXY_SERVER_ID The server id.
7810
7811 HAPROXY_SERVER_MAXCONN The server max connections.
7812
7813 HAPROXY_SERVER_NAME The server name.
7814
7815 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7816 socket).
7817
7818 PATH The PATH environment variable used when executing
7819 the command may be set using "external-check path".
7820
William Lallemand4d03e432019-06-14 15:35:37 +02007821 See also "2.3. Environment variables" for other variables.
7822
Simon Horman98637e52014-06-20 12:30:16 +09007823 If the command executed and exits with a zero status then the check is
7824 considered to have passed, otherwise the check is considered to have
7825 failed.
7826
7827 Example :
7828 external-check command /bin/true
7829
7830 See also : "external-check", "option external-check", "external-check path"
7831
7832
7833external-check path <path>
7834 The value of the PATH environment variable used when running an external-check
7835 May be used in sections : defaults | frontend | listen | backend
7836 yes | no | yes | yes
7837
7838 Arguments :
7839 <path> is the path used when executing external command to run
7840
7841 The default path is "".
7842
7843 Example :
7844 external-check path "/usr/bin:/bin"
7845
7846 See also : "external-check", "option external-check",
7847 "external-check command"
7848
7849
Emeric Brun647caf12009-06-30 17:57:00 +02007850persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007851persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007852 Enable RDP cookie-based persistence
7853 May be used in sections : defaults | frontend | listen | backend
7854 yes | no | yes | yes
7855 Arguments :
7856 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007857 default cookie name "msts" will be used. There currently is no
7858 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007859
7860 This statement enables persistence based on an RDP cookie. The RDP cookie
7861 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007862 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007863 and if an RDP cookie is found, it is decoded. If it matches a known server
7864 which is still UP (or if "option persist" is set), then the connection is
7865 forwarded to this server.
7866
7867 Note that this only makes sense in a TCP backend, but for this to work, the
7868 frontend must have waited long enough to ensure that an RDP cookie is present
7869 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007870 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007871 a single "listen" section.
7872
Willy Tarreau61e28f22010-05-16 22:31:05 +02007873 Also, it is important to understand that the terminal server will emit this
7874 RDP cookie only if it is configured for "token redirection mode", which means
7875 that the "IP address redirection" option is disabled.
7876
Emeric Brun647caf12009-06-30 17:57:00 +02007877 Example :
7878 listen tse-farm
7879 bind :3389
7880 # wait up to 5s for an RDP cookie in the request
7881 tcp-request inspect-delay 5s
7882 tcp-request content accept if RDP_COOKIE
7883 # apply RDP cookie persistence
7884 persist rdp-cookie
7885 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007886 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007887 balance rdp-cookie
7888 server srv1 1.1.1.1:3389
7889 server srv2 1.1.1.2:3389
7890
Simon Hormanab814e02011-06-24 14:50:20 +09007891 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7892 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007893
7894
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007895rate-limit sessions <rate>
7896 Set a limit on the number of new sessions accepted per second on a frontend
7897 May be used in sections : defaults | frontend | listen | backend
7898 yes | yes | yes | no
7899 Arguments :
7900 <rate> The <rate> parameter is an integer designating the maximum number
7901 of new sessions per second to accept on the frontend.
7902
7903 When the frontend reaches the specified number of new sessions per second, it
7904 stops accepting new connections until the rate drops below the limit again.
7905 During this time, the pending sessions will be kept in the socket's backlog
7906 (in system buffers) and haproxy will not even be aware that sessions are
7907 pending. When applying very low limit on a highly loaded service, it may make
7908 sense to increase the socket's backlog using the "backlog" keyword.
7909
7910 This feature is particularly efficient at blocking connection-based attacks
7911 or service abuse on fragile servers. Since the session rate is measured every
7912 millisecond, it is extremely accurate. Also, the limit applies immediately,
7913 no delay is needed at all to detect the threshold.
7914
7915 Example : limit the connection rate on SMTP to 10 per second max
7916 listen smtp
7917 mode tcp
7918 bind :25
7919 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007920 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007921
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007922 Note : when the maximum rate is reached, the frontend's status is not changed
7923 but its sockets appear as "WAITING" in the statistics if the
7924 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007925
7926 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7927
7928
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007929redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7930redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7931redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007932 Return an HTTP redirection if/unless a condition is matched
7933 May be used in sections : defaults | frontend | listen | backend
7934 no | yes | yes | yes
7935
7936 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007937 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007938
Willy Tarreau0140f252008-11-19 21:07:09 +01007939 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007940 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007941 the HTTP "Location" header. When used in an "http-request" rule,
7942 <loc> value follows the log-format rules and can include some
7943 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007944
7945 <pfx> With "redirect prefix", the "Location" header is built from the
7946 concatenation of <pfx> and the complete URI path, including the
7947 query string, unless the "drop-query" option is specified (see
7948 below). As a special case, if <pfx> equals exactly "/", then
7949 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007950 redirect to the same URL (for instance, to insert a cookie). When
7951 used in an "http-request" rule, <pfx> value follows the log-format
7952 rules and can include some dynamic values (see Custom Log Format
7953 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007954
7955 <sch> With "redirect scheme", then the "Location" header is built by
7956 concatenating <sch> with "://" then the first occurrence of the
7957 "Host" header, and then the URI path, including the query string
7958 unless the "drop-query" option is specified (see below). If no
7959 path is found or if the path is "*", then "/" is used instead. If
7960 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007961 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007962 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007963 HTTPS. When used in an "http-request" rule, <sch> value follows
7964 the log-format rules and can include some dynamic values (see
7965 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007966
7967 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007968 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7969 with 302 used by default if no code is specified. 301 means
7970 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007971 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007972 cache the redirection. 303 is equivalent to 302 except that the
7973 browser will fetch the location with a GET method. 307 is just
7974 like 302 but makes it clear that the same method must be reused.
7975 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007976
7977 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007978 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007979
7980 - "drop-query"
7981 When this keyword is used in a prefix-based redirection, then the
7982 location will be set without any possible query-string, which is useful
7983 for directing users to a non-secure page for instance. It has no effect
7984 with a location-type redirect.
7985
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007986 - "append-slash"
7987 This keyword may be used in conjunction with "drop-query" to redirect
7988 users who use a URL not ending with a '/' to the same one with the '/'.
7989 It can be useful to ensure that search engines will only see one URL.
7990 For this, a return code 301 is preferred.
7991
Willy Tarreau0140f252008-11-19 21:07:09 +01007992 - "set-cookie NAME[=value]"
7993 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7994 to the response. This is sometimes used to indicate that a user has
7995 been seen, for instance to protect against some types of DoS. No other
7996 cookie option is added, so the cookie will be a session cookie. Note
7997 that for a browser, a sole cookie name without an equal sign is
7998 different from a cookie with an equal sign.
7999
8000 - "clear-cookie NAME[=]"
8001 A "Set-Cookie" header will be added with NAME (and optionally "="), but
8002 with the "Max-Age" attribute set to zero. This will tell the browser to
8003 delete this cookie. It is useful for instance on logout pages. It is
8004 important to note that clearing the cookie "NAME" will not remove a
8005 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
8006 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02008007
8008 Example: move the login URL only to HTTPS.
8009 acl clear dst_port 80
8010 acl secure dst_port 8080
8011 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01008012 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01008013 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01008014 acl cookie_set hdr_sub(cookie) SEEN=1
8015
8016 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01008017 redirect prefix https://mysite.com if login_page !secure
8018 redirect prefix http://mysite.com drop-query if login_page !uid_given
8019 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01008020 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02008021
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01008022 Example: send redirects for request for articles without a '/'.
8023 acl missing_slash path_reg ^/article/[^/]*$
8024 redirect code 301 prefix / drop-query append-slash if missing_slash
8025
Willy Tarreau2e1dca82012-09-12 08:43:15 +02008026 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01008027 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02008028
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01008029 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02008030 http-request redirect code 301 location \
8031 http://www.%[hdr(host)]%[capture.req.uri] \
8032 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01008033
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008034 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02008035
8036
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008037redisp (deprecated)
8038redispatch (deprecated)
8039 Enable or disable session redistribution in case of connection failure
8040 May be used in sections: defaults | frontend | listen | backend
8041 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008042 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008043
8044 In HTTP mode, if a server designated by a cookie is down, clients may
8045 definitely stick to it because they cannot flush the cookie, so they will not
8046 be able to access the service anymore.
8047
8048 Specifying "redispatch" will allow the proxy to break their persistence and
8049 redistribute them to a working server.
8050
8051 It also allows to retry last connection to another server in case of multiple
8052 connection failures. Of course, it requires having "retries" set to a nonzero
8053 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01008054
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008055 This form is deprecated, do not use it in any new configuration, use the new
8056 "option redispatch" instead.
8057
8058 See also : "option redispatch"
8059
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008060
Willy Tarreau96d51952019-05-22 20:34:35 +02008061reqadd <string> [{if | unless} <cond>] (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008062 Add a header at the end of the HTTP request
8063 May be used in sections : defaults | frontend | listen | backend
8064 no | yes | yes | yes
8065 Arguments :
8066 <string> is the complete line to be added. Any space or known delimiter
8067 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008068 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01008069
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01008070 <cond> is an optional matching condition built from ACLs. It makes it
8071 possible to ignore this rule when other conditions are not met.
8072
Willy Tarreau303c0352008-01-17 19:01:39 +01008073 A new line consisting in <string> followed by a line feed will be added after
8074 the last header of an HTTP request.
8075
8076 Header transformations only apply to traffic which passes through HAProxy,
8077 and not to traffic generated by HAProxy, such as health-checks or error
8078 responses.
8079
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01008080 Example : add "X-Proto: SSL" to requests coming via port 81
8081 acl is-ssl dst_port 81
8082 reqadd X-Proto:\ SSL if is-ssl
8083
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008084 See also: "rspadd", "http-request", section 6 about HTTP header manipulation,
8085 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008086
8087
Willy Tarreau96d51952019-05-22 20:34:35 +02008088reqallow <search> [{if | unless} <cond>] (deprecated)
8089reqiallow <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008090 Definitely allow an HTTP request if a line matches a regular expression
8091 May be used in sections : defaults | frontend | listen | backend
8092 no | yes | yes | yes
8093 Arguments :
8094 <search> is the regular expression applied to HTTP headers and to the
8095 request line. This is an extended regular expression. Parenthesis
8096 grouping is supported and no preliminary backslash is required.
8097 Any space or known delimiter must be escaped using a backslash
8098 ('\'). The pattern applies to a full line at a time. The
8099 "reqallow" keyword strictly matches case while "reqiallow"
8100 ignores case.
8101
Willy Tarreau5321c422010-01-28 20:35:13 +01008102 <cond> is an optional matching condition built from ACLs. It makes it
8103 possible to ignore this rule when other conditions are not met.
8104
Willy Tarreau303c0352008-01-17 19:01:39 +01008105 A request containing any line which matches extended regular expression
8106 <search> will mark the request as allowed, even if any later test would
8107 result in a deny. The test applies both to the request line and to request
8108 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01008109 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01008110
8111 It is easier, faster and more powerful to use ACLs to write access policies.
8112 Reqdeny, reqallow and reqpass should be avoided in new designs.
8113
8114 Example :
8115 # allow www.* but refuse *.local
8116 reqiallow ^Host:\ www\.
8117 reqideny ^Host:\ .*\.local
8118
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008119 See also: "reqdeny", "block", "http-request", section 6 about HTTP header
8120 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008121
8122
Willy Tarreau96d51952019-05-22 20:34:35 +02008123reqdel <search> [{if | unless} <cond>] (deprecated)
8124reqidel <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008125 Delete all headers matching a regular expression in an HTTP request
8126 May be used in sections : defaults | frontend | listen | backend
8127 no | yes | yes | yes
8128 Arguments :
8129 <search> is the regular expression applied to HTTP headers and to the
8130 request line. This is an extended regular expression. Parenthesis
8131 grouping is supported and no preliminary backslash is required.
8132 Any space or known delimiter must be escaped using a backslash
8133 ('\'). The pattern applies to a full line at a time. The "reqdel"
8134 keyword strictly matches case while "reqidel" ignores case.
8135
Willy Tarreau5321c422010-01-28 20:35:13 +01008136 <cond> is an optional matching condition built from ACLs. It makes it
8137 possible to ignore this rule when other conditions are not met.
8138
Willy Tarreau303c0352008-01-17 19:01:39 +01008139 Any header line matching extended regular expression <search> in the request
8140 will be completely deleted. Most common use of this is to remove unwanted
8141 and/or dangerous headers or cookies from a request before passing it to the
8142 next servers.
8143
8144 Header transformations only apply to traffic which passes through HAProxy,
8145 and not to traffic generated by HAProxy, such as health-checks or error
8146 responses. Keep in mind that header names are not case-sensitive.
8147
8148 Example :
8149 # remove X-Forwarded-For header and SERVER cookie
8150 reqidel ^X-Forwarded-For:.*
8151 reqidel ^Cookie:.*SERVER=
8152
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008153 See also: "reqadd", "reqrep", "rspdel", "http-request", section 6 about
8154 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008155
8156
Willy Tarreau96d51952019-05-22 20:34:35 +02008157reqdeny <search> [{if | unless} <cond>] (deprecated)
8158reqideny <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008159 Deny an HTTP request if a line matches a regular expression
8160 May be used in sections : defaults | frontend | listen | backend
8161 no | yes | yes | yes
8162 Arguments :
8163 <search> is the regular expression applied to HTTP headers and to the
8164 request line. This is an extended regular expression. Parenthesis
8165 grouping is supported and no preliminary backslash is required.
8166 Any space or known delimiter must be escaped using a backslash
8167 ('\'). The pattern applies to a full line at a time. The
8168 "reqdeny" keyword strictly matches case while "reqideny" ignores
8169 case.
8170
Willy Tarreau5321c422010-01-28 20:35:13 +01008171 <cond> is an optional matching condition built from ACLs. It makes it
8172 possible to ignore this rule when other conditions are not met.
8173
Willy Tarreau303c0352008-01-17 19:01:39 +01008174 A request containing any line which matches extended regular expression
8175 <search> will mark the request as denied, even if any later test would
8176 result in an allow. The test applies both to the request line and to request
8177 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01008178 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01008179
Willy Tarreauced27012008-01-17 20:35:34 +01008180 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008181 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01008182 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01008183
Willy Tarreau303c0352008-01-17 19:01:39 +01008184 It is easier, faster and more powerful to use ACLs to write access policies.
8185 Reqdeny, reqallow and reqpass should be avoided in new designs.
8186
8187 Example :
8188 # refuse *.local, then allow www.*
8189 reqideny ^Host:\ .*\.local
8190 reqiallow ^Host:\ www\.
8191
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008192 See also: "reqallow", "rspdeny", "block", "http-request", section 6 about
8193 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008194
8195
Willy Tarreau96d51952019-05-22 20:34:35 +02008196reqpass <search> [{if | unless} <cond>] (deprecated)
8197reqipass <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008198 Ignore any HTTP request line matching a regular expression in next rules
8199 May be used in sections : defaults | frontend | listen | backend
8200 no | yes | yes | yes
8201 Arguments :
8202 <search> is the regular expression applied to HTTP headers and to the
8203 request line. This is an extended regular expression. Parenthesis
8204 grouping is supported and no preliminary backslash is required.
8205 Any space or known delimiter must be escaped using a backslash
8206 ('\'). The pattern applies to a full line at a time. The
8207 "reqpass" keyword strictly matches case while "reqipass" ignores
8208 case.
8209
Willy Tarreau5321c422010-01-28 20:35:13 +01008210 <cond> is an optional matching condition built from ACLs. It makes it
8211 possible to ignore this rule when other conditions are not met.
8212
Willy Tarreau303c0352008-01-17 19:01:39 +01008213 A request containing any line which matches extended regular expression
8214 <search> will skip next rules, without assigning any deny or allow verdict.
8215 The test applies both to the request line and to request headers. Keep in
8216 mind that URLs in request line are case-sensitive while header names are not.
8217
8218 It is easier, faster and more powerful to use ACLs to write access policies.
8219 Reqdeny, reqallow and reqpass should be avoided in new designs.
8220
8221 Example :
8222 # refuse *.local, then allow www.*, but ignore "www.private.local"
8223 reqipass ^Host:\ www.private\.local
8224 reqideny ^Host:\ .*\.local
8225 reqiallow ^Host:\ www\.
8226
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008227 See also: "reqallow", "reqdeny", "block", "http-request", section 6 about
8228 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008229
8230
Willy Tarreau96d51952019-05-22 20:34:35 +02008231reqrep <search> <string> [{if | unless} <cond>] (deprecated)
8232reqirep <search> <string> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008233 Replace a regular expression with a string in an HTTP request line
8234 May be used in sections : defaults | frontend | listen | backend
8235 no | yes | yes | yes
8236 Arguments :
8237 <search> is the regular expression applied to HTTP headers and to the
8238 request line. This is an extended regular expression. Parenthesis
8239 grouping is supported and no preliminary backslash is required.
8240 Any space or known delimiter must be escaped using a backslash
8241 ('\'). The pattern applies to a full line at a time. The "reqrep"
8242 keyword strictly matches case while "reqirep" ignores case.
8243
8244 <string> is the complete line to be added. Any space or known delimiter
8245 must be escaped using a backslash ('\'). References to matched
8246 pattern groups are possible using the common \N form, with N
8247 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008248 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01008249
Willy Tarreau5321c422010-01-28 20:35:13 +01008250 <cond> is an optional matching condition built from ACLs. It makes it
8251 possible to ignore this rule when other conditions are not met.
8252
Willy Tarreau303c0352008-01-17 19:01:39 +01008253 Any line matching extended regular expression <search> in the request (both
8254 the request line and header lines) will be completely replaced with <string>.
8255 Most common use of this is to rewrite URLs or domain names in "Host" headers.
8256
8257 Header transformations only apply to traffic which passes through HAProxy,
8258 and not to traffic generated by HAProxy, such as health-checks or error
8259 responses. Note that for increased readability, it is suggested to add enough
8260 spaces between the request and the response. Keep in mind that URLs in
8261 request line are case-sensitive while header names are not.
8262
8263 Example :
8264 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04008265 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01008266 # replace "www.mydomain.com" with "www" in the host name.
8267 reqirep ^Host:\ www.mydomain.com Host:\ www
8268
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008269 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", "http-request",
8270 section 6 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008271
8272
Willy Tarreau96d51952019-05-22 20:34:35 +02008273reqtarpit <search> [{if | unless} <cond>] (deprecated)
8274reqitarpit <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008275 Tarpit an HTTP request containing a line matching a regular expression
8276 May be used in sections : defaults | frontend | listen | backend
8277 no | yes | yes | yes
8278 Arguments :
8279 <search> is the regular expression applied to HTTP headers and to the
8280 request line. This is an extended regular expression. Parenthesis
8281 grouping is supported and no preliminary backslash is required.
8282 Any space or known delimiter must be escaped using a backslash
8283 ('\'). The pattern applies to a full line at a time. The
8284 "reqtarpit" keyword strictly matches case while "reqitarpit"
8285 ignores case.
8286
Willy Tarreau5321c422010-01-28 20:35:13 +01008287 <cond> is an optional matching condition built from ACLs. It makes it
8288 possible to ignore this rule when other conditions are not met.
8289
Willy Tarreau303c0352008-01-17 19:01:39 +01008290 A request containing any line which matches extended regular expression
8291 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01008292 be kept open for a pre-defined time, then will return an HTTP error 500 so
8293 that the attacker does not suspect it has been tarpitted. The status 500 will
8294 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01008295 delay is defined by "timeout tarpit", or "timeout connect" if the former is
8296 not set.
8297
8298 The goal of the tarpit is to slow down robots attacking servers with
8299 identifiable requests. Many robots limit their outgoing number of connections
8300 and stay connected waiting for a reply which can take several minutes to
8301 come. Depending on the environment and attack, it may be particularly
8302 efficient at reducing the load on the network and firewalls.
8303
Willy Tarreau5321c422010-01-28 20:35:13 +01008304 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01008305 # ignore user-agents reporting any flavor of "Mozilla" or "MSIE", but
Willy Tarreau303c0352008-01-17 19:01:39 +01008306 # block all others.
8307 reqipass ^User-Agent:\.*(Mozilla|MSIE)
8308 reqitarpit ^User-Agent:
8309
Willy Tarreau5321c422010-01-28 20:35:13 +01008310 # block bad guys
8311 acl badguys src 10.1.0.3 172.16.13.20/28
8312 reqitarpit . if badguys
8313
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008314 See also: "reqallow", "reqdeny", "reqpass", "http-request", section 6
8315 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008316
8317
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02008318retries <value>
8319 Set the number of retries to perform on a server after a connection failure
8320 May be used in sections: defaults | frontend | listen | backend
8321 yes | no | yes | yes
8322 Arguments :
8323 <value> is the number of times a connection attempt should be retried on
8324 a server when a connection either is refused or times out. The
8325 default value is 3.
8326
8327 It is important to understand that this value applies to the number of
8328 connection attempts, not full requests. When a connection has effectively
8329 been established to a server, there will be no more retry.
8330
8331 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07008332 a turn-around timer of min("timeout connect", one second) is applied before
8333 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02008334
8335 When "option redispatch" is set, the last retry may be performed on another
8336 server even if a cookie references a different server.
8337
8338 See also : "option redispatch"
8339
8340
Olivier Houcharda254a372019-04-05 15:30:12 +02008341retry-on [list of keywords]
8342 Specify when to attempt to automatically retry a failed request
8343 May be used in sections: defaults | frontend | listen | backend
8344 yes | no | yes | yes
8345 Arguments :
8346 <keywords> is a list of keywords or HTTP status codes, each representing a
8347 type of failure event on which an attempt to retry the request
8348 is desired. Please read the notes at the bottom before changing
8349 this setting. The following keywords are supported :
8350
8351 none never retry
8352
8353 conn-failure retry when the connection or the SSL handshake failed
8354 and the request could not be sent. This is the default.
8355
8356 empty-response retry when the server connection was closed after part
8357 of the request was sent, and nothing was received from
8358 the server. This type of failure may be caused by the
8359 request timeout on the server side, poor network
8360 condition, or a server crash or restart while
8361 processing the request.
8362
Olivier Houcharde3249a92019-05-03 23:01:47 +02008363 junk-response retry when the server returned something not looking
8364 like a complete HTTP response. This includes partial
8365 responses headers as well as non-HTTP contents. It
8366 usually is a bad idea to retry on such events, which
8367 may be caused a configuration issue (wrong server port)
8368 or by the request being harmful to the server (buffer
8369 overflow attack for example).
8370
Olivier Houcharda254a372019-04-05 15:30:12 +02008371 response-timeout the server timeout stroke while waiting for the server
8372 to respond to the request. This may be caused by poor
8373 network condition, the reuse of an idle connection
8374 which has expired on the path, or by the request being
8375 extremely expensive to process. It generally is a bad
8376 idea to retry on such events on servers dealing with
8377 heavy database processing (full scans, etc) as it may
8378 amplify denial of service attacks.
8379
Olivier Houchard865d8392019-05-03 22:46:27 +02008380 0rtt-rejected retry requests which were sent over early data and were
8381 rejected by the server. These requests are generally
8382 considered to be safe to retry.
8383
Olivier Houcharda254a372019-04-05 15:30:12 +02008384 <status> any HTTP status code among "404" (Not Found), "408"
8385 (Request Timeout), "425" (Too Early), "500" (Server
8386 Error), "501" (Not Implemented), "502" (Bad Gateway),
8387 "503" (Service Unavailable), "504" (Gateway Timeout).
8388
Olivier Houchardddf0e032019-05-10 18:05:40 +02008389 all-retryable-errors
8390 retry request for any error that are considered
8391 retryable. This currently activates "conn-failure",
8392 "empty-response", "junk-response", "response-timeout",
8393 "0rtt-rejected", "500", "502", "503", and "504".
8394
Olivier Houcharda254a372019-04-05 15:30:12 +02008395 Using this directive replaces any previous settings with the new ones; it is
8396 not cumulative.
8397
8398 Please note that using anything other than "none" and "conn-failure" requires
8399 to allocate a buffer and copy the whole request into it, so it has memory and
8400 performance impacts. Requests not fitting in a single buffer will never be
8401 retried (see the global tune.bufsize setting).
8402
8403 You have to make sure the application has a replay protection mechanism built
8404 in such as a unique transaction IDs passed in requests, or that replaying the
8405 same request has no consequence, or it is very dangerous to use any retry-on
8406 value beside "conn-failure" and "none". Static file servers and caches are
8407 generally considered safe against any type of retry. Using a status code can
8408 be useful to quickly leave a server showing an abnormal behavior (out of
8409 memory, file system issues, etc), but in this case it may be a good idea to
8410 immediately redispatch the connection to another server (please see "option
8411 redispatch" for this). Last, it is important to understand that most causes
8412 of failures are the requests themselves and that retrying a request causing a
8413 server to misbehave will often make the situation even worse for this server,
8414 or for the whole service in case of redispatch.
8415
8416 Unless you know exactly how the application deals with replayed requests, you
8417 should not use this directive.
8418
8419 The default is "conn-failure".
8420
8421 See also: "retries", "option redispatch", "tune.bufsize"
8422
Willy Tarreau96d51952019-05-22 20:34:35 +02008423rspadd <string> [{if | unless} <cond>] (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008424 Add a header at the end of the HTTP response
8425 May be used in sections : defaults | frontend | listen | backend
8426 no | yes | yes | yes
8427 Arguments :
8428 <string> is the complete line to be added. Any space or known delimiter
8429 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008430 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01008431
Willy Tarreaufdb563c2010-01-31 15:43:27 +01008432 <cond> is an optional matching condition built from ACLs. It makes it
8433 possible to ignore this rule when other conditions are not met.
8434
Willy Tarreau303c0352008-01-17 19:01:39 +01008435 A new line consisting in <string> followed by a line feed will be added after
8436 the last header of an HTTP response.
8437
8438 Header transformations only apply to traffic which passes through HAProxy,
8439 and not to traffic generated by HAProxy, such as health-checks or error
8440 responses.
8441
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008442 See also: "rspdel" "reqadd", "http-response", section 6 about HTTP header
8443 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008444
8445
Willy Tarreau96d51952019-05-22 20:34:35 +02008446rspdel <search> [{if | unless} <cond>] (deprecated)
8447rspidel <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008448 Delete all headers matching a regular expression in an HTTP response
8449 May be used in sections : defaults | frontend | listen | backend
8450 no | yes | yes | yes
8451 Arguments :
8452 <search> is the regular expression applied to HTTP headers and to the
8453 response line. This is an extended regular expression, so
8454 parenthesis grouping is supported and no preliminary backslash
8455 is required. Any space or known delimiter must be escaped using
8456 a backslash ('\'). The pattern applies to a full line at a time.
8457 The "rspdel" keyword strictly matches case while "rspidel"
8458 ignores case.
8459
Willy Tarreaufdb563c2010-01-31 15:43:27 +01008460 <cond> is an optional matching condition built from ACLs. It makes it
8461 possible to ignore this rule when other conditions are not met.
8462
Willy Tarreau303c0352008-01-17 19:01:39 +01008463 Any header line matching extended regular expression <search> in the response
8464 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008465 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01008466 client.
8467
8468 Header transformations only apply to traffic which passes through HAProxy,
8469 and not to traffic generated by HAProxy, such as health-checks or error
8470 responses. Keep in mind that header names are not case-sensitive.
8471
8472 Example :
8473 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02008474 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01008475
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008476 See also: "rspadd", "rsprep", "reqdel", "http-response", section 6 about
8477 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008478
8479
Willy Tarreau96d51952019-05-22 20:34:35 +02008480rspdeny <search> [{if | unless} <cond>] (deprecated)
8481rspideny <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008482 Block an HTTP response if a line matches a regular expression
8483 May be used in sections : defaults | frontend | listen | backend
8484 no | yes | yes | yes
8485 Arguments :
8486 <search> is the regular expression applied to HTTP headers and to the
8487 response line. This is an extended regular expression, so
8488 parenthesis grouping is supported and no preliminary backslash
8489 is required. Any space or known delimiter must be escaped using
8490 a backslash ('\'). The pattern applies to a full line at a time.
8491 The "rspdeny" keyword strictly matches case while "rspideny"
8492 ignores case.
8493
Willy Tarreaufdb563c2010-01-31 15:43:27 +01008494 <cond> is an optional matching condition built from ACLs. It makes it
8495 possible to ignore this rule when other conditions are not met.
8496
Willy Tarreau303c0352008-01-17 19:01:39 +01008497 A response containing any line which matches extended regular expression
8498 <search> will mark the request as denied. The test applies both to the
8499 response line and to response headers. Keep in mind that header names are not
8500 case-sensitive.
8501
8502 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01008503 block the response before it reaches the client. If a response is denied, it
8504 will be replaced with an HTTP 502 error so that the client never retrieves
8505 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01008506
8507 It is easier, faster and more powerful to use ACLs to write access policies.
8508 Rspdeny should be avoided in new designs.
8509
8510 Example :
8511 # Ensure that no content type matching ms-word will leak
8512 rspideny ^Content-type:\.*/ms-word
8513
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008514 See also: "reqdeny", "acl", "block", "http-response", section 6 about
8515 HTTP header manipulation and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008516
8517
Willy Tarreau96d51952019-05-22 20:34:35 +02008518rsprep <search> <string> [{if | unless} <cond>] (deprecated)
8519rspirep <search> <string> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008520 Replace a regular expression with a string in an HTTP response line
8521 May be used in sections : defaults | frontend | listen | backend
8522 no | yes | yes | yes
8523 Arguments :
8524 <search> is the regular expression applied to HTTP headers and to the
8525 response line. This is an extended regular expression, so
8526 parenthesis grouping is supported and no preliminary backslash
8527 is required. Any space or known delimiter must be escaped using
8528 a backslash ('\'). The pattern applies to a full line at a time.
8529 The "rsprep" keyword strictly matches case while "rspirep"
8530 ignores case.
8531
8532 <string> is the complete line to be added. Any space or known delimiter
8533 must be escaped using a backslash ('\'). References to matched
8534 pattern groups are possible using the common \N form, with N
8535 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008536 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01008537
Willy Tarreaufdb563c2010-01-31 15:43:27 +01008538 <cond> is an optional matching condition built from ACLs. It makes it
8539 possible to ignore this rule when other conditions are not met.
8540
Willy Tarreau303c0352008-01-17 19:01:39 +01008541 Any line matching extended regular expression <search> in the response (both
8542 the response line and header lines) will be completely replaced with
8543 <string>. Most common use of this is to rewrite Location headers.
8544
8545 Header transformations only apply to traffic which passes through HAProxy,
8546 and not to traffic generated by HAProxy, such as health-checks or error
8547 responses. Note that for increased readability, it is suggested to add enough
8548 spaces between the request and the response. Keep in mind that header names
8549 are not case-sensitive.
8550
8551 Example :
8552 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
8553 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
8554
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008555 See also: "rspadd", "rspdel", "reqrep", "http-response", section 6 about
8556 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008557
8558
David du Colombier486df472011-03-17 10:40:26 +01008559server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008560 Declare a server in a backend
8561 May be used in sections : defaults | frontend | listen | backend
8562 no | no | yes | yes
8563 Arguments :
8564 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01008565 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05008566 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008567
David du Colombier486df472011-03-17 10:40:26 +01008568 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
8569 resolvable hostname is supported, but this name will be resolved
8570 during start-up. Address "0.0.0.0" or "*" has a special meaning.
8571 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02008572 address as the one from the client connection. This is useful in
8573 transparent proxy architectures where the client's connection is
8574 intercepted and haproxy must forward to the original destination
8575 address. This is more or less what the "transparent" keyword does
8576 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01008577 to report statistics. Optionally, an address family prefix may be
8578 used before the address to force the family regardless of the
8579 address format, which can be useful to specify a path to a unix
8580 socket with no slash ('/'). Currently supported prefixes are :
8581 - 'ipv4@' -> address is always IPv4
8582 - 'ipv6@' -> address is always IPv6
8583 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008584 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02008585 - 'sockpair@' -> address is the FD of a connected unix
8586 socket or of a socketpair. During a connection, the
8587 backend creates a pair of connected sockets, and passes
8588 one of them over the FD. The bind part will use the
8589 received socket as the client FD. Should be used
8590 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02008591 You may want to reference some environment variables in the
8592 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01008593 variables. The "init-addr" setting can be used to modify the way
8594 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008595
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008596 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008597 be sent to this port. If unset, the same port the client
8598 connected to will be used. The port may also be prefixed by a "+"
8599 or a "-". In this case, the server's port will be determined by
8600 adding this value to the client's port.
8601
8602 <param*> is a list of parameters for this server. The "server" keywords
8603 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008604 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008605
8606 Examples :
8607 server first 10.1.1.1:1080 cookie first check inter 1000
8608 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01008609 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02008610 server backup "${SRV_BACKUP}:1080" backup
8611 server www1_dc1 "${LAN_DC1}.101:80"
8612 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008613
Willy Tarreau55dcaf62015-09-27 15:03:15 +02008614 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
8615 sun_path length is used for the address length. Some other programs
8616 such as socat use the string length only by default. Pass the option
8617 ",unix-tightsocklen=0" to any abstract socket definition in socat to
8618 make it compatible with HAProxy's.
8619
Mark Lamourinec2247f02012-01-04 13:02:01 -05008620 See also: "default-server", "http-send-name-header" and section 5 about
8621 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008622
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02008623server-state-file-name [<file>]
8624 Set the server state file to read, load and apply to servers available in
8625 this backend. It only applies when the directive "load-server-state-from-file"
8626 is set to "local". When <file> is not provided or if this directive is not
8627 set, then backend name is used. If <file> starts with a slash '/', then it is
8628 considered as an absolute path. Otherwise, <file> is concatenated to the
8629 global directive "server-state-file-base".
8630
8631 Example: the minimal configuration below would make HAProxy look for the
8632 state server file '/etc/haproxy/states/bk':
8633
8634 global
8635 server-state-file-base /etc/haproxy/states
8636
Willy Tarreau750bb0c2020-03-05 16:03:58 +01008637 backend bk
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02008638 load-server-state-from-file
8639
8640 See also: "server-state-file-base", "load-server-state-from-file", and
8641 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008642
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02008643server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
8644 Set a template to initialize servers with shared parameters.
8645 The names of these servers are built from <prefix> and <num | range> parameters.
8646 May be used in sections : defaults | frontend | listen | backend
8647 no | no | yes | yes
8648
8649 Arguments:
8650 <prefix> A prefix for the server names to be built.
8651
8652 <num | range>
8653 If <num> is provided, this template initializes <num> servers
8654 with 1 up to <num> as server name suffixes. A range of numbers
8655 <num_low>-<num_high> may also be used to use <num_low> up to
8656 <num_high> as server name suffixes.
8657
8658 <fqdn> A FQDN for all the servers this template initializes.
8659
8660 <port> Same meaning as "server" <port> argument (see "server" keyword).
8661
8662 <params*>
8663 Remaining server parameters among all those supported by "server"
8664 keyword.
8665
8666 Examples:
8667 # Initializes 3 servers with srv1, srv2 and srv3 as names,
8668 # google.com as FQDN, and health-check enabled.
8669 server-template srv 1-3 google.com:80 check
8670
8671 # or
8672 server-template srv 3 google.com:80 check
8673
8674 # would be equivalent to:
8675 server srv1 google.com:80 check
8676 server srv2 google.com:80 check
8677 server srv3 google.com:80 check
8678
8679
8680
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008681source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02008682source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008683source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008684 Set the source address for outgoing connections
8685 May be used in sections : defaults | frontend | listen | backend
8686 yes | no | yes | yes
8687 Arguments :
8688 <addr> is the IPv4 address HAProxy will bind to before connecting to a
8689 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01008690
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008691 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01008692 the most appropriate address to reach its destination. Optionally
8693 an address family prefix may be used before the address to force
8694 the family regardless of the address format, which can be useful
8695 to specify a path to a unix socket with no slash ('/'). Currently
8696 supported prefixes are :
8697 - 'ipv4@' -> address is always IPv4
8698 - 'ipv6@' -> address is always IPv6
8699 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008700 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02008701 You may want to reference some environment variables in the
8702 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008703
8704 <port> is an optional port. It is normally not needed but may be useful
8705 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02008706 the system will select a free port. Note that port ranges are not
8707 supported in the backend. If you want to force port ranges, you
8708 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008709
8710 <addr2> is the IP address to present to the server when connections are
8711 forwarded in full transparent proxy mode. This is currently only
8712 supported on some patched Linux kernels. When this address is
8713 specified, clients connecting to the server will be presented
8714 with this address, while health checks will still use the address
8715 <addr>.
8716
8717 <port2> is the optional port to present to the server when connections
8718 are forwarded in full transparent proxy mode (see <addr2> above).
8719 The default value of zero means the system will select a free
8720 port.
8721
Willy Tarreaubce70882009-09-07 11:51:47 +02008722 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
8723 This is the name of a comma-separated header list which can
8724 contain multiple IP addresses. By default, the last occurrence is
8725 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01008726 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02008727 by previous proxy, typically Stunnel. In order to use another
8728 occurrence from the last one, please see the <occ> parameter
8729 below. When the header (or occurrence) is not found, no binding
8730 is performed so that the proxy's default IP address is used. Also
8731 keep in mind that the header name is case insensitive, as for any
8732 HTTP header.
8733
8734 <occ> is the occurrence number of a value to be used in a multi-value
8735 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008736 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02008737 address. Positive values indicate a position from the first
8738 occurrence, 1 being the first one. Negative values indicate
8739 positions relative to the last one, -1 being the last one. This
8740 is helpful for situations where an X-Forwarded-For header is set
8741 at the entry point of an infrastructure and must be used several
8742 proxy layers away. When this value is not specified, -1 is
8743 assumed. Passing a zero here disables the feature.
8744
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008745 <name> is an optional interface name to which to bind to for outgoing
8746 traffic. On systems supporting this features (currently, only
8747 Linux), this allows one to bind all traffic to the server to
8748 this interface even if it is not the one the system would select
8749 based on routing tables. This should be used with extreme care.
8750 Note that using this option requires root privileges.
8751
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008752 The "source" keyword is useful in complex environments where a specific
8753 address only is allowed to connect to the servers. It may be needed when a
8754 private address must be used through a public gateway for instance, and it is
8755 known that the system cannot determine the adequate source address by itself.
8756
8757 An extension which is available on certain patched Linux kernels may be used
8758 through the "usesrc" optional keyword. It makes it possible to connect to the
8759 servers with an IP address which does not belong to the system itself. This
8760 is called "full transparent proxy mode". For this to work, the destination
8761 servers have to route their traffic back to this address through the machine
8762 running HAProxy, and IP forwarding must generally be enabled on this machine.
8763
8764 In this "full transparent proxy" mode, it is possible to force a specific IP
8765 address to be presented to the servers. This is not much used in fact. A more
8766 common use is to tell HAProxy to present the client's IP address. For this,
8767 there are two methods :
8768
8769 - present the client's IP and port addresses. This is the most transparent
8770 mode, but it can cause problems when IP connection tracking is enabled on
8771 the machine, because a same connection may be seen twice with different
8772 states. However, this solution presents the huge advantage of not
8773 limiting the system to the 64k outgoing address+port couples, because all
8774 of the client ranges may be used.
8775
8776 - present only the client's IP address and select a spare port. This
8777 solution is still quite elegant but slightly less transparent (downstream
8778 firewalls logs will not match upstream's). It also presents the downside
8779 of limiting the number of concurrent connections to the usual 64k ports.
8780 However, since the upstream and downstream ports are different, local IP
8781 connection tracking on the machine will not be upset by the reuse of the
8782 same session.
8783
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008784 This option sets the default source for all servers in the backend. It may
8785 also be specified in a "defaults" section. Finer source address specification
8786 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008787 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008788
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008789 In order to work, "usesrc" requires root privileges.
8790
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008791 Examples :
8792 backend private
8793 # Connect to the servers using our 192.168.1.200 source address
8794 source 192.168.1.200
8795
8796 backend transparent_ssl1
8797 # Connect to the SSL farm from the client's source address
8798 source 192.168.1.200 usesrc clientip
8799
8800 backend transparent_ssl2
8801 # Connect to the SSL farm from the client's source address and port
8802 # not recommended if IP conntrack is present on the local machine.
8803 source 192.168.1.200 usesrc client
8804
8805 backend transparent_ssl3
8806 # Connect to the SSL farm from the client's source address. It
8807 # is more conntrack-friendly.
8808 source 192.168.1.200 usesrc clientip
8809
8810 backend transparent_smtp
8811 # Connect to the SMTP farm from the client's source address/port
8812 # with Tproxy version 4.
8813 source 0.0.0.0 usesrc clientip
8814
Willy Tarreaubce70882009-09-07 11:51:47 +02008815 backend transparent_http
8816 # Connect to the servers using the client's IP as seen by previous
8817 # proxy.
8818 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8819
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008820 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008821 the Linux kernel on www.balabit.com, the "bind" keyword.
8822
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008823
Willy Tarreau844e3c52008-01-11 16:28:18 +01008824srvtimeout <timeout> (deprecated)
8825 Set the maximum inactivity time on the server side.
8826 May be used in sections : defaults | frontend | listen | backend
8827 yes | no | yes | yes
8828 Arguments :
8829 <timeout> is the timeout value specified in milliseconds by default, but
8830 can be in any other unit if the number is suffixed by the unit,
8831 as explained at the top of this document.
8832
8833 The inactivity timeout applies when the server is expected to acknowledge or
8834 send data. In HTTP mode, this timeout is particularly important to consider
8835 during the first phase of the server's response, when it has to send the
8836 headers, as it directly represents the server's processing time for the
8837 request. To find out what value to put there, it's often good to start with
8838 what would be considered as unacceptable response times, then check the logs
8839 to observe the response time distribution, and adjust the value accordingly.
8840
8841 The value is specified in milliseconds by default, but can be in any other
8842 unit if the number is suffixed by the unit, as specified at the top of this
8843 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
8844 recommended that the client timeout remains equal to the server timeout in
8845 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008846 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01008847 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +01008848 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01008849
8850 This parameter is specific to backends, but can be specified once for all in
8851 "defaults" sections. This is in fact one of the easiest solutions not to
8852 forget about it. An unspecified timeout results in an infinite timeout, which
8853 is not recommended. Such a usage is accepted and works but reports a warning
8854 during startup because it may results in accumulation of expired sessions in
8855 the system if the system's timeouts are not configured either.
8856
8857 This parameter is provided for compatibility but is currently deprecated.
8858 Please use "timeout server" instead.
8859
Willy Tarreauce887fd2012-05-12 12:50:00 +02008860 See also : "timeout server", "timeout tunnel", "timeout client" and
8861 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01008862
8863
Cyril Bonté66c327d2010-10-12 00:14:37 +02008864stats admin { if | unless } <cond>
8865 Enable statistics admin level if/unless a condition is matched
8866 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008867 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008868
8869 This statement enables the statistics admin level if/unless a condition is
8870 matched.
8871
8872 The admin level allows to enable/disable servers from the web interface. By
8873 default, statistics page is read-only for security reasons.
8874
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008875 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8876 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008877 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008878
Cyril Bonté23b39d92011-02-10 22:54:44 +01008879 Currently, the POST request is limited to the buffer size minus the reserved
8880 buffer space, which means that if the list of servers is too long, the
8881 request won't be processed. It is recommended to alter few servers at a
8882 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008883
8884 Example :
8885 # statistics admin level only for localhost
8886 backend stats_localhost
8887 stats enable
8888 stats admin if LOCALHOST
8889
8890 Example :
8891 # statistics admin level always enabled because of the authentication
8892 backend stats_auth
8893 stats enable
8894 stats auth admin:AdMiN123
8895 stats admin if TRUE
8896
8897 Example :
8898 # statistics admin level depends on the authenticated user
8899 userlist stats-auth
8900 group admin users admin
8901 user admin insecure-password AdMiN123
8902 group readonly users haproxy
8903 user haproxy insecure-password haproxy
8904
8905 backend stats_auth
8906 stats enable
8907 acl AUTH http_auth(stats-auth)
8908 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8909 stats http-request auth unless AUTH
8910 stats admin if AUTH_ADMIN
8911
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008912 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8913 "bind-process", section 3.4 about userlists and section 7 about
8914 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008915
8916
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008917stats auth <user>:<passwd>
8918 Enable statistics with authentication and grant access to an account
8919 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008920 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008921 Arguments :
8922 <user> is a user name to grant access to
8923
8924 <passwd> is the cleartext password associated to this user
8925
8926 This statement enables statistics with default settings, and restricts access
8927 to declared users only. It may be repeated as many times as necessary to
8928 allow as many users as desired. When a user tries to access the statistics
8929 without a valid account, a "401 Forbidden" response will be returned so that
8930 the browser asks the user to provide a valid user and password. The real
8931 which will be returned to the browser is configurable using "stats realm".
8932
8933 Since the authentication method is HTTP Basic Authentication, the passwords
8934 circulate in cleartext on the network. Thus, it was decided that the
8935 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008936 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008937
8938 It is also possible to reduce the scope of the proxies which appear in the
8939 report using "stats scope".
8940
8941 Though this statement alone is enough to enable statistics reporting, it is
8942 recommended to set all other settings in order to avoid relying on default
8943 unobvious parameters.
8944
8945 Example :
8946 # public access (limited to this backend only)
8947 backend public_www
8948 server srv1 192.168.0.1:80
8949 stats enable
8950 stats hide-version
8951 stats scope .
8952 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008953 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008954 stats auth admin1:AdMiN123
8955 stats auth admin2:AdMiN321
8956
8957 # internal monitoring access (unlimited)
8958 backend private_monitoring
8959 stats enable
8960 stats uri /admin?stats
8961 stats refresh 5s
8962
8963 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8964
8965
8966stats enable
8967 Enable statistics reporting with default settings
8968 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008969 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008970 Arguments : none
8971
8972 This statement enables statistics reporting with default settings defined
8973 at build time. Unless stated otherwise, these settings are used :
8974 - stats uri : /haproxy?stats
8975 - stats realm : "HAProxy Statistics"
8976 - stats auth : no authentication
8977 - stats scope : no restriction
8978
8979 Though this statement alone is enough to enable statistics reporting, it is
8980 recommended to set all other settings in order to avoid relying on default
8981 unobvious parameters.
8982
8983 Example :
8984 # public access (limited to this backend only)
8985 backend public_www
8986 server srv1 192.168.0.1:80
8987 stats enable
8988 stats hide-version
8989 stats scope .
8990 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008991 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008992 stats auth admin1:AdMiN123
8993 stats auth admin2:AdMiN321
8994
8995 # internal monitoring access (unlimited)
8996 backend private_monitoring
8997 stats enable
8998 stats uri /admin?stats
8999 stats refresh 5s
9000
9001 See also : "stats auth", "stats realm", "stats uri"
9002
9003
Willy Tarreaud63335a2010-02-26 12:56:52 +01009004stats hide-version
9005 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009006 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009007 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01009008 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009009
Willy Tarreaud63335a2010-02-26 12:56:52 +01009010 By default, the stats page reports some useful status information along with
9011 the statistics. Among them is HAProxy's version. However, it is generally
9012 considered dangerous to report precise version to anyone, as it can help them
9013 target known weaknesses with specific attacks. The "stats hide-version"
9014 statement removes the version from the statistics report. This is recommended
9015 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009016
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02009017 Though this statement alone is enough to enable statistics reporting, it is
9018 recommended to set all other settings in order to avoid relying on default
9019 unobvious parameters.
9020
Willy Tarreaud63335a2010-02-26 12:56:52 +01009021 Example :
9022 # public access (limited to this backend only)
9023 backend public_www
9024 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02009025 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01009026 stats hide-version
9027 stats scope .
9028 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009029 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01009030 stats auth admin1:AdMiN123
9031 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009032
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009033 # internal monitoring access (unlimited)
9034 backend private_monitoring
9035 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01009036 stats uri /admin?stats
9037 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01009038
Willy Tarreaud63335a2010-02-26 12:56:52 +01009039 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02009040
Willy Tarreau983e01e2010-01-11 18:42:06 +01009041
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02009042stats http-request { allow | deny | auth [realm <realm>] }
9043 [ { if | unless } <condition> ]
9044 Access control for statistics
9045
9046 May be used in sections: defaults | frontend | listen | backend
9047 no | no | yes | yes
9048
9049 As "http-request", these set of options allow to fine control access to
9050 statistics. Each option may be followed by if/unless and acl.
9051 First option with matched condition (or option without condition) is final.
9052 For "deny" a 403 error will be returned, for "allow" normal processing is
9053 performed, for "auth" a 401/407 error code is returned so the client
9054 should be asked to enter a username and password.
9055
9056 There is no fixed limit to the number of http-request statements per
9057 instance.
9058
9059 See also : "http-request", section 3.4 about userlists and section 7
9060 about ACL usage.
9061
9062
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009063stats realm <realm>
9064 Enable statistics and set authentication realm
9065 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009066 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009067 Arguments :
9068 <realm> is the name of the HTTP Basic Authentication realm reported to
9069 the browser. The browser uses it to display it in the pop-up
9070 inviting the user to enter a valid username and password.
9071
9072 The realm is read as a single word, so any spaces in it should be escaped
9073 using a backslash ('\').
9074
9075 This statement is useful only in conjunction with "stats auth" since it is
9076 only related to authentication.
9077
9078 Though this statement alone is enough to enable statistics reporting, it is
9079 recommended to set all other settings in order to avoid relying on default
9080 unobvious parameters.
9081
9082 Example :
9083 # public access (limited to this backend only)
9084 backend public_www
9085 server srv1 192.168.0.1:80
9086 stats enable
9087 stats hide-version
9088 stats scope .
9089 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009090 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009091 stats auth admin1:AdMiN123
9092 stats auth admin2:AdMiN321
9093
9094 # internal monitoring access (unlimited)
9095 backend private_monitoring
9096 stats enable
9097 stats uri /admin?stats
9098 stats refresh 5s
9099
9100 See also : "stats auth", "stats enable", "stats uri"
9101
9102
9103stats refresh <delay>
9104 Enable statistics with automatic refresh
9105 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009106 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009107 Arguments :
9108 <delay> is the suggested refresh delay, specified in seconds, which will
9109 be returned to the browser consulting the report page. While the
9110 browser is free to apply any delay, it will generally respect it
9111 and refresh the page this every seconds. The refresh interval may
9112 be specified in any other non-default time unit, by suffixing the
9113 unit after the value, as explained at the top of this document.
9114
9115 This statement is useful on monitoring displays with a permanent page
9116 reporting the load balancer's activity. When set, the HTML report page will
9117 include a link "refresh"/"stop refresh" so that the user can select whether
9118 he wants automatic refresh of the page or not.
9119
9120 Though this statement alone is enough to enable statistics reporting, it is
9121 recommended to set all other settings in order to avoid relying on default
9122 unobvious parameters.
9123
9124 Example :
9125 # public access (limited to this backend only)
9126 backend public_www
9127 server srv1 192.168.0.1:80
9128 stats enable
9129 stats hide-version
9130 stats scope .
9131 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009132 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009133 stats auth admin1:AdMiN123
9134 stats auth admin2:AdMiN321
9135
9136 # internal monitoring access (unlimited)
9137 backend private_monitoring
9138 stats enable
9139 stats uri /admin?stats
9140 stats refresh 5s
9141
9142 See also : "stats auth", "stats enable", "stats realm", "stats uri"
9143
9144
9145stats scope { <name> | "." }
9146 Enable statistics and limit access scope
9147 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009148 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009149 Arguments :
9150 <name> is the name of a listen, frontend or backend section to be
9151 reported. The special name "." (a single dot) designates the
9152 section in which the statement appears.
9153
9154 When this statement is specified, only the sections enumerated with this
9155 statement will appear in the report. All other ones will be hidden. This
9156 statement may appear as many times as needed if multiple sections need to be
9157 reported. Please note that the name checking is performed as simple string
9158 comparisons, and that it is never checked that a give section name really
9159 exists.
9160
9161 Though this statement alone is enough to enable statistics reporting, it is
9162 recommended to set all other settings in order to avoid relying on default
9163 unobvious parameters.
9164
9165 Example :
9166 # public access (limited to this backend only)
9167 backend public_www
9168 server srv1 192.168.0.1:80
9169 stats enable
9170 stats hide-version
9171 stats scope .
9172 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009173 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009174 stats auth admin1:AdMiN123
9175 stats auth admin2:AdMiN321
9176
9177 # internal monitoring access (unlimited)
9178 backend private_monitoring
9179 stats enable
9180 stats uri /admin?stats
9181 stats refresh 5s
9182
9183 See also : "stats auth", "stats enable", "stats realm", "stats uri"
9184
Willy Tarreaud63335a2010-02-26 12:56:52 +01009185
Willy Tarreauc9705a12010-07-27 20:05:50 +02009186stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01009187 Enable reporting of a description on the statistics page.
9188 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009189 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01009190
Willy Tarreauc9705a12010-07-27 20:05:50 +02009191 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01009192 description from global section is automatically used instead.
9193
9194 This statement is useful for users that offer shared services to their
9195 customers, where node or description should be different for each customer.
9196
9197 Though this statement alone is enough to enable statistics reporting, it is
9198 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01009199 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01009200
9201 Example :
9202 # internal monitoring access (unlimited)
9203 backend private_monitoring
9204 stats enable
9205 stats show-desc Master node for Europe, Asia, Africa
9206 stats uri /admin?stats
9207 stats refresh 5s
9208
9209 See also: "show-node", "stats enable", "stats uri" and "description" in
9210 global section.
9211
9212
9213stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02009214 Enable reporting additional information on the statistics page
9215 May be used in sections : defaults | frontend | listen | backend
9216 yes | yes | yes | yes
9217 Arguments : none
9218
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009219 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01009220 - cap: capabilities (proxy)
9221 - mode: one of tcp, http or health (proxy)
9222 - id: SNMP ID (proxy, socket, server)
9223 - IP (socket, server)
9224 - cookie (backend, server)
9225
9226 Though this statement alone is enough to enable statistics reporting, it is
9227 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01009228 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01009229
9230 See also: "stats enable", "stats uri".
9231
9232
9233stats show-node [ <name> ]
9234 Enable reporting of a host name on the statistics page.
9235 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009236 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01009237 Arguments:
9238 <name> is an optional name to be reported. If unspecified, the
9239 node name from global section is automatically used instead.
9240
9241 This statement is useful for users that offer shared services to their
9242 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01009243 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01009244
9245 Though this statement alone is enough to enable statistics reporting, it is
9246 recommended to set all other settings in order to avoid relying on default
9247 unobvious parameters.
9248
9249 Example:
9250 # internal monitoring access (unlimited)
9251 backend private_monitoring
9252 stats enable
9253 stats show-node Europe-1
9254 stats uri /admin?stats
9255 stats refresh 5s
9256
9257 See also: "show-desc", "stats enable", "stats uri", and "node" in global
9258 section.
9259
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009260
9261stats uri <prefix>
9262 Enable statistics and define the URI prefix to access them
9263 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009264 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009265 Arguments :
9266 <prefix> is the prefix of any URI which will be redirected to stats. This
9267 prefix may contain a question mark ('?') to indicate part of a
9268 query string.
9269
9270 The statistics URI is intercepted on the relayed traffic, so it appears as a
9271 page within the normal application. It is strongly advised to ensure that the
9272 selected URI will never appear in the application, otherwise it will never be
9273 possible to reach it in the application.
9274
9275 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009276 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009277 It is generally a good idea to include a question mark in the URI so that
9278 intermediate proxies refrain from caching the results. Also, since any string
9279 beginning with the prefix will be accepted as a stats request, the question
9280 mark helps ensuring that no valid URI will begin with the same words.
9281
9282 It is sometimes very convenient to use "/" as the URI prefix, and put that
9283 statement in a "listen" instance of its own. That makes it easy to dedicate
9284 an address or a port to statistics only.
9285
9286 Though this statement alone is enough to enable statistics reporting, it is
9287 recommended to set all other settings in order to avoid relying on default
9288 unobvious parameters.
9289
9290 Example :
9291 # public access (limited to this backend only)
9292 backend public_www
9293 server srv1 192.168.0.1:80
9294 stats enable
9295 stats hide-version
9296 stats scope .
9297 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009298 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009299 stats auth admin1:AdMiN123
9300 stats auth admin2:AdMiN321
9301
9302 # internal monitoring access (unlimited)
9303 backend private_monitoring
9304 stats enable
9305 stats uri /admin?stats
9306 stats refresh 5s
9307
9308 See also : "stats auth", "stats enable", "stats realm"
9309
9310
Willy Tarreaud63335a2010-02-26 12:56:52 +01009311stick match <pattern> [table <table>] [{if | unless} <cond>]
9312 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009313 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01009314 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009315
9316 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009317 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009318 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01009319 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009320 stickiness table. This rule is mandatory.
9321
9322 <table> is an optional stickiness table name. If unspecified, the same
9323 backend's table is used. A stickiness table is declared using
9324 the "stick-table" statement.
9325
9326 <cond> is an optional matching condition. It makes it possible to match
9327 on a certain criterion only when other conditions are met (or
9328 not met). For instance, it could be used to match on a source IP
9329 address except when a request passes through a known proxy, in
9330 which case we'd match on a header containing that IP address.
9331
9332 Some protocols or applications require complex stickiness rules and cannot
9333 always simply rely on cookies nor hashing. The "stick match" statement
9334 describes a rule to extract the stickiness criterion from an incoming request
9335 or connection. See section 7 for a complete list of possible patterns and
9336 transformation rules.
9337
9338 The table has to be declared using the "stick-table" statement. It must be of
9339 a type compatible with the pattern. By default it is the one which is present
9340 in the same backend. It is possible to share a table with other backends by
9341 referencing it using the "table" keyword. If another table is referenced,
9342 the server's ID inside the backends are used. By default, all server IDs
9343 start at 1 in each backend, so the server ordering is enough. But in case of
9344 doubt, it is highly recommended to force server IDs using their "id" setting.
9345
9346 It is possible to restrict the conditions where a "stick match" statement
9347 will apply, using "if" or "unless" followed by a condition. See section 7 for
9348 ACL based conditions.
9349
9350 There is no limit on the number of "stick match" statements. The first that
9351 applies and matches will cause the request to be directed to the same server
9352 as was used for the request which created the entry. That way, multiple
9353 matches can be used as fallbacks.
9354
9355 The stick rules are checked after the persistence cookies, so they will not
9356 affect stickiness if a cookie has already been used to select a server. That
9357 way, it becomes very easy to insert cookies and match on IP addresses in
9358 order to maintain stickiness between HTTP and HTTPS.
9359
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009360 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9361 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009362 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009363
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009364 Example :
9365 # forward SMTP users to the same server they just used for POP in the
9366 # last 30 minutes
9367 backend pop
9368 mode tcp
9369 balance roundrobin
9370 stick store-request src
9371 stick-table type ip size 200k expire 30m
9372 server s1 192.168.1.1:110
9373 server s2 192.168.1.1:110
9374
9375 backend smtp
9376 mode tcp
9377 balance roundrobin
9378 stick match src table pop
9379 server s1 192.168.1.1:25
9380 server s2 192.168.1.1:25
9381
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009382 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02009383 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009384
9385
9386stick on <pattern> [table <table>] [{if | unless} <condition>]
9387 Define a request pattern to associate a user to a server
9388 May be used in sections : defaults | frontend | listen | backend
9389 no | no | yes | yes
9390
9391 Note : This form is exactly equivalent to "stick match" followed by
9392 "stick store-request", all with the same arguments. Please refer
9393 to both keywords for details. It is only provided as a convenience
9394 for writing more maintainable configurations.
9395
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009396 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9397 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009398 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009399
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009400 Examples :
9401 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01009402 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009403
9404 # ...is strictly equivalent to this one :
9405 stick match src table pop if !localhost
9406 stick store-request src table pop if !localhost
9407
9408
9409 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
9410 # well as HTTP without cookie. Share the same table between both accesses.
9411 backend http
9412 mode http
9413 balance roundrobin
9414 stick on src table https
9415 cookie SRV insert indirect nocache
9416 server s1 192.168.1.1:80 cookie s1
9417 server s2 192.168.1.1:80 cookie s2
9418
9419 backend https
9420 mode tcp
9421 balance roundrobin
9422 stick-table type ip size 200k expire 30m
9423 stick on src
9424 server s1 192.168.1.1:443
9425 server s2 192.168.1.1:443
9426
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009427 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009428
9429
9430stick store-request <pattern> [table <table>] [{if | unless} <condition>]
9431 Define a request pattern used to create an entry in a stickiness table
9432 May be used in sections : defaults | frontend | listen | backend
9433 no | no | yes | yes
9434
9435 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009436 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009437 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01009438 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009439 server is selected.
9440
9441 <table> is an optional stickiness table name. If unspecified, the same
9442 backend's table is used. A stickiness table is declared using
9443 the "stick-table" statement.
9444
9445 <cond> is an optional storage condition. It makes it possible to store
9446 certain criteria only when some conditions are met (or not met).
9447 For instance, it could be used to store the source IP address
9448 except when the request passes through a known proxy, in which
9449 case we'd store a converted form of a header containing that IP
9450 address.
9451
9452 Some protocols or applications require complex stickiness rules and cannot
9453 always simply rely on cookies nor hashing. The "stick store-request" statement
9454 describes a rule to decide what to extract from the request and when to do
9455 it, in order to store it into a stickiness table for further requests to
9456 match it using the "stick match" statement. Obviously the extracted part must
9457 make sense and have a chance to be matched in a further request. Storing a
9458 client's IP address for instance often makes sense. Storing an ID found in a
9459 URL parameter also makes sense. Storing a source port will almost never make
9460 any sense because it will be randomly matched. See section 7 for a complete
9461 list of possible patterns and transformation rules.
9462
9463 The table has to be declared using the "stick-table" statement. It must be of
9464 a type compatible with the pattern. By default it is the one which is present
9465 in the same backend. It is possible to share a table with other backends by
9466 referencing it using the "table" keyword. If another table is referenced,
9467 the server's ID inside the backends are used. By default, all server IDs
9468 start at 1 in each backend, so the server ordering is enough. But in case of
9469 doubt, it is highly recommended to force server IDs using their "id" setting.
9470
9471 It is possible to restrict the conditions where a "stick store-request"
9472 statement will apply, using "if" or "unless" followed by a condition. This
9473 condition will be evaluated while parsing the request, so any criteria can be
9474 used. See section 7 for ACL based conditions.
9475
9476 There is no limit on the number of "stick store-request" statements, but
9477 there is a limit of 8 simultaneous stores per request or response. This
9478 makes it possible to store up to 8 criteria, all extracted from either the
9479 request or the response, regardless of the number of rules. Only the 8 first
9480 ones which match will be kept. Using this, it is possible to feed multiple
9481 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009482 another protocol or access method. Using multiple store-request rules with
9483 the same table is possible and may be used to find the best criterion to rely
9484 on, by arranging the rules by decreasing preference order. Only the first
9485 extracted criterion for a given table will be stored. All subsequent store-
9486 request rules referencing the same table will be skipped and their ACLs will
9487 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009488
9489 The "store-request" rules are evaluated once the server connection has been
9490 established, so that the table will contain the real server that processed
9491 the request.
9492
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009493 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9494 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009495 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009496
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009497 Example :
9498 # forward SMTP users to the same server they just used for POP in the
9499 # last 30 minutes
9500 backend pop
9501 mode tcp
9502 balance roundrobin
9503 stick store-request src
9504 stick-table type ip size 200k expire 30m
9505 server s1 192.168.1.1:110
9506 server s2 192.168.1.1:110
9507
9508 backend smtp
9509 mode tcp
9510 balance roundrobin
9511 stick match src table pop
9512 server s1 192.168.1.1:25
9513 server s2 192.168.1.1:25
9514
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009515 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02009516 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009517
9518
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009519stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02009520 size <size> [expire <expire>] [nopurge] [peers <peersect>]
9521 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08009522 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009523 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009524 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009525
9526 Arguments :
9527 ip a table declared with "type ip" will only store IPv4 addresses.
9528 This form is very compact (about 50 bytes per entry) and allows
9529 very fast entry lookup and stores with almost no overhead. This
9530 is mainly used to store client source IP addresses.
9531
David du Colombier9a6d3c92011-03-17 10:40:24 +01009532 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
9533 This form is very compact (about 60 bytes per entry) and allows
9534 very fast entry lookup and stores with almost no overhead. This
9535 is mainly used to store client source IP addresses.
9536
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009537 integer a table declared with "type integer" will store 32bit integers
9538 which can represent a client identifier found in a request for
9539 instance.
9540
9541 string a table declared with "type string" will store substrings of up
9542 to <len> characters. If the string provided by the pattern
9543 extractor is larger than <len>, it will be truncated before
9544 being stored. During matching, at most <len> characters will be
9545 compared between the string in the table and the extracted
9546 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009547 to 32 characters.
9548
9549 binary a table declared with "type binary" will store binary blocks
9550 of <len> bytes. If the block provided by the pattern
9551 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02009552 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009553 is shorter than <len>, it will be padded by 0. When not
9554 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009555
9556 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009557 "string" type table (See type "string" above). Or the number
9558 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009559 changing this parameter as memory usage will proportionally
9560 increase.
9561
9562 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01009563 value directly impacts memory usage. Count approximately
9564 50 bytes per entry, plus the size of a string if any. The size
9565 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009566
9567 [nopurge] indicates that we refuse to purge older entries when the table
9568 is full. When not specified and the table is full when haproxy
9569 wants to store an entry in it, it will flush a few of the oldest
9570 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009571 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009572 be desirable to refuse new entries instead of purging the older
9573 ones. That may be the case when the amount of data to store is
9574 far above the hardware limits and we prefer not to offer access
9575 to new clients than to reject the ones already connected. When
9576 using this parameter, be sure to properly set the "expire"
9577 parameter (see below).
9578
Emeric Brunf099e792010-09-27 12:05:28 +02009579 <peersect> is the name of the peers section to use for replication. Entries
9580 which associate keys to server IDs are kept synchronized with
9581 the remote peers declared in this section. All entries are also
9582 automatically learned from the local peer (old process) during a
9583 soft restart.
9584
Willy Tarreau1abc6732015-05-01 19:21:02 +02009585 NOTE : each peers section may be referenced only by tables
9586 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009587
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009588 <expire> defines the maximum duration of an entry in the table since it
9589 was last created, refreshed or matched. The expiration delay is
9590 defined using the standard time format, similarly as the various
9591 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009592 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02009593 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009594 be removed once full. Be sure not to use the "nopurge" parameter
9595 if not expiration delay is specified.
9596
Willy Tarreau08d5f982010-06-06 13:34:54 +02009597 <data_type> is used to store additional information in the stick-table. This
9598 may be used by ACLs in order to control various criteria related
9599 to the activity of the client matching the stick-table. For each
9600 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02009601 that the additional data can fit. Several data types may be
9602 stored with an entry. Multiple data types may be specified after
9603 the "store" keyword, as a comma-separated list. Alternatively,
9604 it is possible to repeat the "store" keyword followed by one or
9605 several data types. Except for the "server_id" type which is
9606 automatically detected and enabled, all data types must be
9607 explicitly declared to be stored. If an ACL references a data
9608 type which is not stored, the ACL will simply not match. Some
9609 data types require an argument which must be passed just after
9610 the type between parenthesis. See below for the supported data
9611 types and their arguments.
9612
9613 The data types that can be stored with an entry are the following :
9614 - server_id : this is an integer which holds the numeric ID of the server a
9615 request was assigned to. It is used by the "stick match", "stick store",
9616 and "stick on" rules. It is automatically enabled when referenced.
9617
9618 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
9619 integer which may be used for anything. Most of the time it will be used
9620 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01009621 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02009622
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009623 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
9624 over a period. It is a positive 32-bit integer integer which may be used
9625 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01009626 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009627 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01009628 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009629
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009630 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
9631 integer which may be used for anything. Most of the time it will be used
9632 to put a special tag on some entries, for instance to note that a
9633 specific behavior was detected and must be known for future matches.
9634
9635 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
9636 over a period. It is a positive 32-bit integer integer which may be used
9637 for anything. Just like <gpc1>, it counts events, but instead of keeping
9638 a cumulative number, it maintains the rate at which the counter is
9639 incremented. Most of the time it will be used to measure the frequency of
9640 occurrence of certain events (e.g. requests to a specific URL).
9641
Willy Tarreauc9705a12010-07-27 20:05:50 +02009642 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
9643 the absolute number of connections received from clients which matched
9644 this entry. It does not mean the connections were accepted, just that
9645 they were received.
9646
9647 - conn_cur : Current Connections. It is a positive 32-bit integer which
9648 stores the concurrent connection counts for the entry. It is incremented
9649 once an incoming connection matches the entry, and decremented once the
9650 connection leaves. That way it is possible to know at any time the exact
9651 number of concurrent connections for an entry.
9652
9653 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9654 integer parameter <period> which indicates in milliseconds the length
9655 of the period over which the average is measured. It reports the average
9656 incoming connection rate over that period, in connections per period. The
9657 result is an integer which can be matched using ACLs.
9658
9659 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
9660 the absolute number of sessions received from clients which matched this
9661 entry. A session is a connection that was accepted by the layer 4 rules.
9662
9663 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9664 integer parameter <period> which indicates in milliseconds the length
9665 of the period over which the average is measured. It reports the average
9666 incoming session rate over that period, in sessions per period. The
9667 result is an integer which can be matched using ACLs.
9668
9669 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
9670 counts the absolute number of HTTP requests received from clients which
9671 matched this entry. It does not matter whether they are valid requests or
9672 not. Note that this is different from sessions when keep-alive is used on
9673 the client side.
9674
9675 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9676 integer parameter <period> which indicates in milliseconds the length
9677 of the period over which the average is measured. It reports the average
9678 HTTP request rate over that period, in requests per period. The result is
9679 an integer which can be matched using ACLs. It does not matter whether
9680 they are valid requests or not. Note that this is different from sessions
9681 when keep-alive is used on the client side.
9682
9683 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
9684 counts the absolute number of HTTP requests errors induced by clients
9685 which matched this entry. Errors are counted on invalid and truncated
9686 requests, as well as on denied or tarpitted requests, and on failed
9687 authentications. If the server responds with 4xx, then the request is
9688 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01009689 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02009690
9691 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9692 integer parameter <period> which indicates in milliseconds the length
9693 of the period over which the average is measured. It reports the average
9694 HTTP request error rate over that period, in requests per period (see
9695 http_err_cnt above for what is accounted as an error). The result is an
9696 integer which can be matched using ACLs.
9697
9698 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009699 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02009700 which matched this entry. Headers are included in the count. This may be
9701 used to limit abuse of upload features on photo or video servers.
9702
9703 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9704 integer parameter <period> which indicates in milliseconds the length
9705 of the period over which the average is measured. It reports the average
9706 incoming bytes rate over that period, in bytes per period. It may be used
9707 to detect users which upload too much and too fast. Warning: with large
9708 uploads, it is possible that the amount of uploaded data will be counted
9709 once upon termination, thus causing spikes in the average transfer speed
9710 instead of having a smooth one. This may partially be smoothed with
9711 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
9712 recommended for better fairness.
9713
9714 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009715 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02009716 matched this entry. Headers are included in the count. This may be used
9717 to limit abuse of bots sucking the whole site.
9718
9719 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
9720 an integer parameter <period> which indicates in milliseconds the length
9721 of the period over which the average is measured. It reports the average
9722 outgoing bytes rate over that period, in bytes per period. It may be used
9723 to detect users which download too much and too fast. Warning: with large
9724 transfers, it is possible that the amount of transferred data will be
9725 counted once upon termination, thus causing spikes in the average
9726 transfer speed instead of having a smooth one. This may partially be
9727 smoothed with "option contstats" though this is not perfect yet. Use of
9728 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02009729
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009730 There is only one stick-table per proxy. At the moment of writing this doc,
9731 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009732 to be required, simply create a dummy backend with a stick-table in it and
9733 reference it.
9734
9735 It is important to understand that stickiness based on learning information
9736 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01009737 lost upon restart unless peers are properly configured to transfer such
9738 information upon restart (recommended). In general it can be good as a
9739 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009740
Willy Tarreauc9705a12010-07-27 20:05:50 +02009741 Last, memory requirements may be important when storing many data types.
9742 Indeed, storing all indicators above at once in each entry requires 116 bytes
9743 per entry, or 116 MB for a 1-million entries table. This is definitely not
9744 something that can be ignored.
9745
9746 Example:
9747 # Keep track of counters of up to 1 million IP addresses over 5 minutes
9748 # and store a general purpose counter and the average connection rate
9749 # computed over a sliding window of 30 seconds.
9750 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
9751
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009752 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01009753 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009754
9755
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009756stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01009757 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009758 May be used in sections : defaults | frontend | listen | backend
9759 no | no | yes | yes
9760
9761 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009762 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009763 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009764 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009765 server is selected.
9766
9767 <table> is an optional stickiness table name. If unspecified, the same
9768 backend's table is used. A stickiness table is declared using
9769 the "stick-table" statement.
9770
9771 <cond> is an optional storage condition. It makes it possible to store
9772 certain criteria only when some conditions are met (or not met).
9773 For instance, it could be used to store the SSL session ID only
9774 when the response is a SSL server hello.
9775
9776 Some protocols or applications require complex stickiness rules and cannot
9777 always simply rely on cookies nor hashing. The "stick store-response"
9778 statement describes a rule to decide what to extract from the response and
9779 when to do it, in order to store it into a stickiness table for further
9780 requests to match it using the "stick match" statement. Obviously the
9781 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009782 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009783 See section 7 for a complete list of possible patterns and transformation
9784 rules.
9785
9786 The table has to be declared using the "stick-table" statement. It must be of
9787 a type compatible with the pattern. By default it is the one which is present
9788 in the same backend. It is possible to share a table with other backends by
9789 referencing it using the "table" keyword. If another table is referenced,
9790 the server's ID inside the backends are used. By default, all server IDs
9791 start at 1 in each backend, so the server ordering is enough. But in case of
9792 doubt, it is highly recommended to force server IDs using their "id" setting.
9793
9794 It is possible to restrict the conditions where a "stick store-response"
9795 statement will apply, using "if" or "unless" followed by a condition. This
9796 condition will be evaluated while parsing the response, so any criteria can
9797 be used. See section 7 for ACL based conditions.
9798
9799 There is no limit on the number of "stick store-response" statements, but
9800 there is a limit of 8 simultaneous stores per request or response. This
9801 makes it possible to store up to 8 criteria, all extracted from either the
9802 request or the response, regardless of the number of rules. Only the 8 first
9803 ones which match will be kept. Using this, it is possible to feed multiple
9804 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009805 another protocol or access method. Using multiple store-response rules with
9806 the same table is possible and may be used to find the best criterion to rely
9807 on, by arranging the rules by decreasing preference order. Only the first
9808 extracted criterion for a given table will be stored. All subsequent store-
9809 response rules referencing the same table will be skipped and their ACLs will
9810 not be evaluated. However, even if a store-request rule references a table, a
9811 store-response rule may also use the same table. This means that each table
9812 may learn exactly one element from the request and one element from the
9813 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009814
9815 The table will contain the real server that processed the request.
9816
9817 Example :
9818 # Learn SSL session ID from both request and response and create affinity.
9819 backend https
9820 mode tcp
9821 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009822 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009823 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009824
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009825 acl clienthello req_ssl_hello_type 1
9826 acl serverhello rep_ssl_hello_type 2
9827
9828 # use tcp content accepts to detects ssl client and server hello.
9829 tcp-request inspect-delay 5s
9830 tcp-request content accept if clienthello
9831
9832 # no timeout on response inspect delay by default.
9833 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009834
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009835 # SSL session ID (SSLID) may be present on a client or server hello.
9836 # Its length is coded on 1 byte at offset 43 and its value starts
9837 # at offset 44.
9838
9839 # Match and learn on request if client hello.
9840 stick on payload_lv(43,1) if clienthello
9841
9842 # Learn on response if server hello.
9843 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009844
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009845 server s1 192.168.1.1:443
9846 server s2 192.168.1.1:443
9847
9848 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9849 extraction.
9850
9851
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009852tcp-check connect [params*]
9853 Opens a new connection
9854 May be used in sections: defaults | frontend | listen | backend
9855 no | no | yes | yes
9856
9857 When an application lies on more than a single TCP port or when HAProxy
9858 load-balance many services in a single backend, it makes sense to probe all
9859 the services individually before considering a server as operational.
9860
9861 When there are no TCP port configured on the server line neither server port
9862 directive, then the 'tcp-check connect port <port>' must be the first step
9863 of the sequence.
9864
9865 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9866 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9867 do.
9868
9869 Parameters :
9870 They are optional and can be used to describe how HAProxy should open and
9871 use the TCP connection.
9872
9873 port if not set, check port or server port is used.
9874 It tells HAProxy where to open the connection to.
9875 <port> must be a valid TCP port source integer, from 1 to 65535.
9876
9877 send-proxy send a PROXY protocol string
9878
9879 ssl opens a ciphered connection
9880
9881 Examples:
9882 # check HTTP and HTTPs services on a server.
9883 # first open port 80 thanks to server line port directive, then
9884 # tcp-check opens port 443, ciphered and run a request on it:
9885 option tcp-check
9886 tcp-check connect
9887 tcp-check send GET\ /\ HTTP/1.0\r\n
9888 tcp-check send Host:\ haproxy.1wt.eu\r\n
9889 tcp-check send \r\n
9890 tcp-check expect rstring (2..|3..)
9891 tcp-check connect port 443 ssl
9892 tcp-check send GET\ /\ HTTP/1.0\r\n
9893 tcp-check send Host:\ haproxy.1wt.eu\r\n
9894 tcp-check send \r\n
9895 tcp-check expect rstring (2..|3..)
9896 server www 10.0.0.1 check port 80
9897
9898 # check both POP and IMAP from a single server:
9899 option tcp-check
9900 tcp-check connect port 110
9901 tcp-check expect string +OK\ POP3\ ready
9902 tcp-check connect port 143
9903 tcp-check expect string *\ OK\ IMAP4\ ready
9904 server mail 10.0.0.1 check
9905
9906 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9907
9908
9909tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009910 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009911 May be used in sections: defaults | frontend | listen | backend
9912 no | no | yes | yes
9913
9914 Arguments :
9915 <match> is a keyword indicating how to look for a specific pattern in the
9916 response. The keyword may be one of "string", "rstring" or
9917 binary.
9918 The keyword may be preceded by an exclamation mark ("!") to negate
9919 the match. Spaces are allowed between the exclamation mark and the
9920 keyword. See below for more details on the supported keywords.
9921
9922 <pattern> is the pattern to look for. It may be a string or a regular
9923 expression. If the pattern contains spaces, they must be escaped
9924 with the usual backslash ('\').
9925 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009926 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009927 two digits will represent a byte. The hexadecimal digits may be
9928 used upper or lower case.
9929
9930
9931 The available matches are intentionally similar to their http-check cousins :
9932
9933 string <string> : test the exact string matches in the response buffer.
9934 A health check response will be considered valid if the
9935 response's buffer contains this exact string. If the
9936 "string" keyword is prefixed with "!", then the response
9937 will be considered invalid if the body contains this
9938 string. This can be used to look for a mandatory pattern
9939 in a protocol response, or to detect a failure when a
9940 specific error appears in a protocol banner.
9941
9942 rstring <regex> : test a regular expression on the response buffer.
9943 A health check response will be considered valid if the
9944 response's buffer matches this expression. If the
9945 "rstring" keyword is prefixed with "!", then the response
9946 will be considered invalid if the body matches the
9947 expression.
9948
9949 binary <hexstring> : test the exact string in its hexadecimal form matches
9950 in the response buffer. A health check response will
9951 be considered valid if the response's buffer contains
9952 this exact hexadecimal string.
9953 Purpose is to match data on binary protocols.
9954
9955 It is important to note that the responses will be limited to a certain size
9956 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9957 Thus, too large responses may not contain the mandatory pattern when using
9958 "string", "rstring" or binary. If a large response is absolutely required, it
9959 is possible to change the default max size by setting the global variable.
9960 However, it is worth keeping in mind that parsing very large responses can
9961 waste some CPU cycles, especially when regular expressions are used, and that
9962 it is always better to focus the checks on smaller resources. Also, in its
9963 current state, the check will not find any string nor regex past a null
9964 character in the response. Similarly it is not possible to request matching
9965 the null character.
9966
9967 Examples :
9968 # perform a POP check
9969 option tcp-check
9970 tcp-check expect string +OK\ POP3\ ready
9971
9972 # perform an IMAP check
9973 option tcp-check
9974 tcp-check expect string *\ OK\ IMAP4\ ready
9975
9976 # look for the redis master server
9977 option tcp-check
9978 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009979 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009980 tcp-check send info\ replication\r\n
9981 tcp-check expect string role:master
9982 tcp-check send QUIT\r\n
9983 tcp-check expect string +OK
9984
9985
9986 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9987 "tcp-check send-binary", "http-check expect", tune.chksize
9988
9989
9990tcp-check send <data>
9991 Specify a string to be sent as a question during a generic health check
9992 May be used in sections: defaults | frontend | listen | backend
9993 no | no | yes | yes
9994
9995 <data> : the data to be sent as a question during a generic health check
9996 session. For now, <data> must be a string.
9997
9998 Examples :
9999 # look for the redis master server
10000 option tcp-check
10001 tcp-check send info\ replication\r\n
10002 tcp-check expect string role:master
10003
10004 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
10005 "tcp-check send-binary", tune.chksize
10006
10007
Davor Ocelice9ed2812017-12-25 17:49:28 +010010008tcp-check send-binary <hexstring>
10009 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010010 tcp health check
10011 May be used in sections: defaults | frontend | listen | backend
10012 no | no | yes | yes
10013
10014 <data> : the data to be sent as a question during a generic health check
10015 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +010010016 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +020010017 response buffer. A health check response will be considered
10018 valid if the response's buffer contains this exact
10019 hexadecimal string.
10020 Purpose is to send binary data to ask on binary protocols.
10021
10022 Examples :
10023 # redis check in binary
10024 option tcp-check
10025 tcp-check send-binary 50494e470d0a # PING\r\n
10026 tcp-check expect binary 2b504F4e47 # +PONG
10027
10028
10029 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
10030 "tcp-check send", tune.chksize
10031
10032
Willy Tarreaue9656522010-08-17 15:40:09 +020010033tcp-request connection <action> [{if | unless} <condition>]
10034 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +020010035 May be used in sections : defaults | frontend | listen | backend
10036 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +020010037 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020010038 <action> defines the action to perform if the condition applies. See
10039 below.
Willy Tarreau1a687942010-05-23 22:40:30 +020010040
Willy Tarreaue9656522010-08-17 15:40:09 +020010041 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010042
10043 Immediately after acceptance of a new incoming connection, it is possible to
10044 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +020010045 or dropped or have its counters tracked. Those conditions cannot make use of
10046 any data contents because the connection has not been read from yet, and the
10047 buffers are not yet allocated. This is used to selectively and very quickly
10048 accept or drop connections from various sources with a very low overhead. If
10049 some contents need to be inspected in order to take the decision, the
10050 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010051
Willy Tarreaue9656522010-08-17 15:40:09 +020010052 The "tcp-request connection" rules are evaluated in their exact declaration
10053 order. If no rule matches or if there is no rule, the default action is to
10054 accept the incoming connection. There is no specific limit to the number of
10055 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010056
Willy Tarreaua9083d02015-05-08 15:27:59 +020010057 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +020010058 - accept :
10059 accepts the connection if the condition is true (when used with "if")
10060 or false (when used with "unless"). The first such rule executed ends
10061 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010062
Willy Tarreaue9656522010-08-17 15:40:09 +020010063 - reject :
10064 rejects the connection if the condition is true (when used with "if")
10065 or false (when used with "unless"). The first such rule executed ends
10066 the rules evaluation. Rejected connections do not even become a
10067 session, which is why they are accounted separately for in the stats,
10068 as "denied connections". They are not considered for the session
10069 rate-limit and are not logged either. The reason is that these rules
10070 should only be used to filter extremely high connection rates such as
10071 the ones encountered during a massive DDoS attack. Under these extreme
10072 conditions, the simple action of logging each event would make the
10073 system collapse and would considerably lower the filtering capacity. If
10074 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +020010075 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010076
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010077 - expect-proxy layer4 :
10078 configures the client-facing connection to receive a PROXY protocol
10079 header before any byte is read from the socket. This is equivalent to
10080 having the "accept-proxy" keyword on the "bind" line, except that using
10081 the TCP rule allows the PROXY protocol to be accepted only for certain
10082 IP address ranges using an ACL. This is convenient when multiple layers
10083 of load balancers are passed through by traffic coming from public
10084 hosts.
10085
Bertrand Jacquin90759682016-06-06 15:35:39 +010010086 - expect-netscaler-cip layer4 :
10087 configures the client-facing connection to receive a NetScaler Client
10088 IP insertion protocol header before any byte is read from the socket.
10089 This is equivalent to having the "accept-netscaler-cip" keyword on the
10090 "bind" line, except that using the TCP rule allows the PROXY protocol
10091 to be accepted only for certain IP address ranges using an ACL. This
10092 is convenient when multiple layers of load balancers are passed
10093 through by traffic coming from public hosts.
10094
Willy Tarreau18bf01e2014-06-13 16:18:52 +020010095 - capture <sample> len <length> :
10096 This only applies to "tcp-request content" rules. It captures sample
10097 expression <sample> from the request buffer, and converts it to a
10098 string of at most <len> characters. The resulting string is stored into
10099 the next request "capture" slot, so it will possibly appear next to
10100 some captured HTTP headers. It will then automatically appear in the
10101 logs, and it will be possible to extract it using sample fetch rules to
10102 feed it into headers or anything. The length should be limited given
10103 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +020010104 session life. Please check section 7.3 (Fetching samples) and "capture
10105 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +020010106
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010107 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +020010108 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020010109 rules do not stop evaluation and do not change default action. The
10110 number of counters that may be simultaneously tracked by the same
10111 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roesler27c754c2019-07-10 15:45:51 -050010112 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020010113 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
10114 tracking of the counters of the specified table as the first set. The
10115 first "track-sc1" rule executed enables tracking of the counters of the
10116 specified table as the second set. The first "track-sc2" rule executed
10117 enables tracking of the counters of the specified table as the third
10118 set. It is a recommended practice to use the first set of counters for
10119 the per-frontend counters and the second set for the per-backend ones.
10120 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010121
Willy Tarreaue9656522010-08-17 15:40:09 +020010122 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010123 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +020010124 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +010010125 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010126 and used to select which table entry to update the counters.
10127 Note that "tcp-request connection" cannot use content-based
10128 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010129
Willy Tarreaue9656522010-08-17 15:40:09 +020010130 <table> is an optional table to be used instead of the default one,
10131 which is the stick-table declared in the current proxy. All
10132 the counters for the matches and updates for the key will
10133 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010134
Willy Tarreaue9656522010-08-17 15:40:09 +020010135 Once a "track-sc*" rule is executed, the key is looked up in the table
10136 and if it is not found, an entry is allocated for it. Then a pointer to
10137 that entry is kept during all the session's life, and this entry's
10138 counters are updated as often as possible, every time the session's
10139 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010140 Counters are only updated for events that happen after the tracking has
10141 been started. For example, connection counters will not be updated when
10142 tracking layer 7 information, since the connection event happens before
10143 layer7 information is extracted.
10144
Willy Tarreaue9656522010-08-17 15:40:09 +020010145 If the entry tracks concurrent connection counters, one connection is
10146 counted for as long as the entry is tracked, and the entry will not
10147 expire during that time. Tracking counters also provides a performance
10148 advantage over just checking the keys, because only one table lookup is
10149 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010150
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020010151 - sc-inc-gpc0(<sc-id>):
10152 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
10153 counter designated by <sc-id>. If an error occurs, this action silently
10154 fails and the actions evaluation continues.
10155
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010156 - sc-inc-gpc1(<sc-id>):
10157 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
10158 counter designated by <sc-id>. If an error occurs, this action silently
10159 fails and the actions evaluation continues.
10160
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010161 - sc-set-gpt0(<sc-id>) <int>:
10162 This action sets the GPT0 tag according to the sticky counter designated
10163 by <sc-id> and the value of <int>. The expected result is a boolean. If
10164 an error occurs, this action silently fails and the actions evaluation
10165 continues.
10166
William Lallemand2e785f22016-05-25 01:48:42 +020010167 - set-src <expr> :
10168 Is used to set the source IP address to the value of specified
10169 expression. Useful if you want to mask source IP for privacy.
10170 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020010171 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +020010172
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020010173 Arguments:
10174 <expr> Is a standard HAProxy expression formed by a sample-fetch
10175 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +020010176
10177 Example:
William Lallemand2e785f22016-05-25 01:48:42 +020010178 tcp-request connection set-src src,ipmask(24)
10179
Willy Tarreau0c630532016-10-21 17:52:58 +020010180 When possible, set-src preserves the original source port as long as the
10181 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +020010182
William Lallemand44be6402016-05-25 01:51:35 +020010183 - set-src-port <expr> :
10184 Is used to set the source port address to the value of specified
10185 expression.
10186
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020010187 Arguments:
10188 <expr> Is a standard HAProxy expression formed by a sample-fetch
10189 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +020010190
10191 Example:
William Lallemand44be6402016-05-25 01:51:35 +020010192 tcp-request connection set-src-port int(4000)
10193
Willy Tarreau0c630532016-10-21 17:52:58 +020010194 When possible, set-src-port preserves the original source address as long
10195 as the address family supports a port, otherwise it forces the source
10196 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +020010197
William Lallemand13e9b0c2016-05-25 02:34:07 +020010198 - set-dst <expr> :
10199 Is used to set the destination IP address to the value of specified
10200 expression. Useful if you want to mask IP for privacy in log.
10201 If you want to provide an IP from a HTTP header use "http-request
10202 set-dst". If you want to connect to the new address/port, use
10203 '0.0.0.0:0' as a server address in the backend.
10204
10205 <expr> Is a standard HAProxy expression formed by a sample-fetch
10206 followed by some converters.
10207
10208 Example:
10209
10210 tcp-request connection set-dst dst,ipmask(24)
10211 tcp-request connection set-dst ipv4(10.0.0.1)
10212
Willy Tarreau0c630532016-10-21 17:52:58 +020010213 When possible, set-dst preserves the original destination port as long as
10214 the address family allows it, otherwise the destination port is set to 0.
10215
William Lallemand13e9b0c2016-05-25 02:34:07 +020010216 - set-dst-port <expr> :
10217 Is used to set the destination port address to the value of specified
10218 expression. If you want to connect to the new address/port, use
10219 '0.0.0.0:0' as a server address in the backend.
10220
10221
10222 <expr> Is a standard HAProxy expression formed by a sample-fetch
10223 followed by some converters.
10224
10225 Example:
10226
10227 tcp-request connection set-dst-port int(4000)
10228
Willy Tarreau0c630532016-10-21 17:52:58 +020010229 When possible, set-dst-port preserves the original destination address as
10230 long as the address family supports a port, otherwise it forces the
10231 destination address to IPv4 "0.0.0.0" before rewriting the port.
10232
Willy Tarreau2d392c22015-08-24 01:43:45 +020010233 - "silent-drop" :
10234 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010010235 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020010236 to prevent the client from being notified. The effect it then that the
10237 client still sees an established connection while there's none on
10238 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
10239 except that it doesn't use any local resource at all on the machine
10240 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010010241 slow down stronger attackers. It is important to understand the impact
10242 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010243 client and HAProxy (firewalls, proxies, load balancers) will also keep
10244 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010245 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010246 TCP_REPAIR socket option is used to block the emission of a TCP
10247 reset. On other systems, the socket's TTL is reduced to 1 so that the
10248 TCP reset doesn't pass the first router, though it's still delivered to
10249 local networks. Do not use it unless you fully understand how it works.
10250
Willy Tarreaue9656522010-08-17 15:40:09 +020010251 Note that the "if/unless" condition is optional. If no condition is set on
10252 the action, it is simply performed unconditionally. That can be useful for
10253 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010254
Willy Tarreaue9656522010-08-17 15:40:09 +020010255 Example: accept all connections from white-listed hosts, reject too fast
10256 connection without counting them, and track accepted connections.
10257 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010258
Willy Tarreaue9656522010-08-17 15:40:09 +020010259 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010260 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010261 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010262
Willy Tarreaue9656522010-08-17 15:40:09 +020010263 Example: accept all connections from white-listed hosts, count all other
10264 connections and reject too fast ones. This results in abusive ones
10265 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010266
Willy Tarreaue9656522010-08-17 15:40:09 +020010267 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010268 tcp-request connection track-sc0 src
10269 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010270
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010271 Example: enable the PROXY protocol for traffic coming from all known proxies.
10272
10273 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10274
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010275 See section 7 about ACL usage.
10276
Willy Tarreau4f614292016-10-21 17:49:36 +020010277 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010278
10279
Willy Tarreaue9656522010-08-17 15:40:09 +020010280tcp-request content <action> [{if | unless} <condition>]
10281 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +020010282 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020010283 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +020010284 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020010285 <action> defines the action to perform if the condition applies. See
10286 below.
Willy Tarreau62644772008-07-16 18:36:06 +020010287
Willy Tarreaue9656522010-08-17 15:40:09 +020010288 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +020010289
Davor Ocelice9ed2812017-12-25 17:49:28 +010010290 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +020010291 called "TCP content inspection". During this stage, ACL-based rules are
10292 evaluated every time the request contents are updated, until either an
10293 "accept" or a "reject" rule matches, or the TCP request inspection delay
10294 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +020010295
Willy Tarreaue9656522010-08-17 15:40:09 +020010296 The first difference between these rules and "tcp-request connection" rules
10297 is that "tcp-request content" rules can make use of contents to take a
10298 decision. Most often, these decisions will consider a protocol recognition or
10299 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +010010300 both frontends and backends. In case of HTTP keep-alive with the client, all
10301 tcp-request content rules are evaluated again, so haproxy keeps a record of
10302 what sticky counters were assigned by a "tcp-request connection" versus a
10303 "tcp-request content" rule, and flushes all the content-related ones after
10304 processing an HTTP request, so that they may be evaluated again by the rules
10305 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010306 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +010010307 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +020010308
Willy Tarreaue9656522010-08-17 15:40:09 +020010309 Content-based rules are evaluated in their exact declaration order. If no
10310 rule matches or if there is no rule, the default action is to accept the
10311 contents. There is no specific limit to the number of rules which may be
10312 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +020010313
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010314 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +020010315 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +010010316 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +020010317 - reject : the request is rejected and the connection is closed
10318 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -040010319 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010320 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020010321 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010322 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +020010323 - sc-set-gpt0(<sc-id>) <int>
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020010324 - set-dst <expr>
10325 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010326 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010327 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +020010328 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +010010329 - send-spoe-group <engine-name> <group-name>
Christopher Faulet6bd406e2019-11-22 15:34:17 +010010330 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +020010331
Willy Tarreaue9656522010-08-17 15:40:09 +020010332 They have the same meaning as their counter-parts in "tcp-request connection"
10333 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +010010334 For "do-resolve" action, please check the "http-request do-resolve"
10335 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +020010336
Willy Tarreauf3338342014-01-28 21:40:28 +010010337 While there is nothing mandatory about it, it is recommended to use the
10338 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
10339 content" rules in the frontend, and track-sc2 for "tcp-request content"
10340 rules in the backend, because that makes the configuration more readable
10341 and easier to troubleshoot, but this is just a guideline and all counters
10342 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +020010343
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010344 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +020010345 the action, it is simply performed unconditionally. That can be useful for
10346 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +020010347
Willy Tarreaue9656522010-08-17 15:40:09 +020010348 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +020010349 rules, since HTTP-specific ACL matches are able to preliminarily parse the
10350 contents of a buffer before extracting the required data. If the buffered
10351 contents do not parse as a valid HTTP message, then the ACL does not match.
10352 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +010010353 processing, so there is no risk of parsing something differently. In an HTTP
10354 backend connected to from an HTTP frontend, it is guaranteed that HTTP
10355 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +020010356
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010357 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020010358 are present when the rule is processed. The rule processing engine is able to
10359 wait until the inspect delay expires when the data to be tracked is not yet
10360 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010361
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020010362 The "set-dst" and "set-dst-port" are used to set respectively the destination
10363 IP and port. More information on how to use it at "http-request set-dst".
10364
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010365 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +020010366 declared inline. For "tcp-request session" rules, only session-level
10367 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010368
Daniel Schneller0b547052016-03-21 20:46:57 +010010369 <var-name> The name of the variable starts with an indication about
10370 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010010371 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010010372 "sess" : the variable is shared with the whole session
10373 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010374 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010010375 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010376 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010010377 "res" : the variable is shared only during response
10378 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010379 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010010380 The name may only contain characters 'a-z', 'A-Z', '0-9',
10381 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010382
10383 <expr> Is a standard HAProxy expression formed by a sample-fetch
10384 followed by some converters.
10385
Christopher Faulet85d79c92016-11-09 16:54:56 +010010386 The "unset-var" is used to unset a variable. See above for details about
10387 <var-name>.
10388
Patrick Hemmer268a7072018-05-11 12:52:31 -040010389 The "set-priority-class" is used to set the queue priority class of the
10390 current request. The value must be a sample expression which converts to an
10391 integer in the range -2047..2047. Results outside this range will be
10392 truncated. The priority class determines the order in which queued requests
10393 are processed. Lower values have higher priority.
10394
10395 The "set-priority-offset" is used to set the queue priority timestamp offset
10396 of the current request. The value must be a sample expression which converts
10397 to an integer in the range -524287..524287. Results outside this range will be
10398 truncated. When a request is queued, it is ordered first by the priority
10399 class, then by the current timestamp adjusted by the given offset in
10400 milliseconds. Lower values have higher priority.
10401 Note that the resulting timestamp is is only tracked with enough precision for
10402 524,287ms (8m44s287ms). If the request is queued long enough to where the
10403 adjusted timestamp exceeds this value, it will be misidentified as highest
10404 priority. Thus it is important to set "timeout queue" to a value, where when
10405 combined with the offset, does not exceed this limit.
10406
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010407 The "send-spoe-group" is used to trigger sending of a group of SPOE
10408 messages. To do so, the SPOE engine used to send messages must be defined, as
10409 well as the SPOE group to send. Of course, the SPOE engine must refer to an
10410 existing SPOE filter. If not engine name is provided on the SPOE filter line,
10411 the SPOE agent name must be used.
10412
10413 <engine-name> The SPOE engine name.
10414
10415 <group-name> The SPOE group name as specified in the engine configuration.
10416
Christopher Faulet6bd406e2019-11-22 15:34:17 +010010417 The "use-service" is used to executes a TCP service which will reply to the
10418 request and stop the evaluation of the rules. This service may choose to
10419 reply by sending any valid response or it may immediately close the
10420 connection without sending anything. Outside natives services, it is possible
10421 to write your own services in Lua. No further "tcp-request" rules are
10422 evaluated.
10423
10424 Example:
10425 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
10426
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010427 Example:
10428
10429 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +010010430 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010431
Willy Tarreau62644772008-07-16 18:36:06 +020010432 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +020010433 # Accept HTTP requests containing a Host header saying "example.com"
10434 # and reject everything else.
10435 acl is_host_com hdr(Host) -i example.com
10436 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +020010437 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +020010438 tcp-request content reject
10439
10440 Example:
Willy Tarreau62644772008-07-16 18:36:06 +020010441 # reject SMTP connection if client speaks first
10442 tcp-request inspect-delay 30s
10443 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010444 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +020010445
10446 # Forward HTTPS connection only if client speaks
10447 tcp-request inspect-delay 30s
10448 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010449 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +020010450 tcp-request content reject
10451
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010452 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010453 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010454 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020010455 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010456 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
10457 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010458
10459 Example:
10460 # track request counts per "base" (concatenation of Host+URL)
10461 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020010462 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010463
Willy Tarreaue9656522010-08-17 15:40:09 +020010464 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010465 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +020010466
10467 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +010010468 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +020010469 # protecting all our sites
10470 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010471 tcp-request connection track-sc0 src
10472 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +020010473 ...
10474 use_backend http_dynamic if { path_end .php }
10475
10476 backend http_dynamic
10477 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010478 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +020010479 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010480 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010481 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010482 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +020010483 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +020010484
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010485 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +020010486
Jarno Huuskonen95b012b2017-04-06 13:59:14 +030010487 See also : "tcp-request connection", "tcp-request session",
10488 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +020010489
10490
10491tcp-request inspect-delay <timeout>
10492 Set the maximum allowed time to wait for data during content inspection
10493 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020010494 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +020010495 Arguments :
10496 <timeout> is the timeout value specified in milliseconds by default, but
10497 can be in any other unit if the number is suffixed by the unit,
10498 as explained at the top of this document.
10499
10500 People using haproxy primarily as a TCP relay are often worried about the
10501 risk of passing any type of protocol to a server without any analysis. In
10502 order to be able to analyze the request contents, we must first withhold
10503 the data then analyze them. This statement simply enables withholding of
10504 data for at most the specified amount of time.
10505
Willy Tarreaufb356202010-08-03 14:02:05 +020010506 TCP content inspection applies very early when a connection reaches a
10507 frontend, then very early when the connection is forwarded to a backend. This
10508 means that a connection may experience a first delay in the frontend and a
10509 second delay in the backend if both have tcp-request rules.
10510
Willy Tarreau62644772008-07-16 18:36:06 +020010511 Note that when performing content inspection, haproxy will evaluate the whole
10512 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010513 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +020010514 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +010010515 contents are definitive. If no delay is set, haproxy will not wait at all
10516 and will immediately apply a verdict based on the available information.
10517 Obviously this is unlikely to be very useful and might even be racy, so such
10518 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020010519
10520 As soon as a rule matches, the request is released and continues as usual. If
10521 the timeout is reached and no rule matches, the default policy will be to let
10522 it pass through unaffected.
10523
10524 For most protocols, it is enough to set it to a few seconds, as most clients
10525 send the full request immediately upon connection. Add 3 or more seconds to
10526 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +010010527 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +010010528 before the server (e.g. SMTP), or to wait for a client to talk before passing
10529 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +020010530 least the inspection delay, otherwise it will expire first. If the client
10531 closes the connection or if the buffer is full, the delay immediately expires
10532 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +020010533
Willy Tarreau55165fe2009-05-10 12:02:55 +020010534 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020010535 "timeout client".
10536
10537
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010538tcp-response content <action> [{if | unless} <condition>]
10539 Perform an action on a session response depending on a layer 4-7 condition
10540 May be used in sections : defaults | frontend | listen | backend
10541 no | no | yes | yes
10542 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020010543 <action> defines the action to perform if the condition applies. See
10544 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010545
10546 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
10547
Davor Ocelice9ed2812017-12-25 17:49:28 +010010548 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010549 called "TCP content inspection". During this stage, ACL-based rules are
10550 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010551 "accept", "close" or a "reject" rule matches, or a TCP response inspection
10552 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010553
10554 Most often, these decisions will consider a protocol recognition or validity.
10555
10556 Content-based rules are evaluated in their exact declaration order. If no
10557 rule matches or if there is no rule, the default action is to accept the
10558 contents. There is no specific limit to the number of rules which may be
10559 inserted.
10560
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010561 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010562 - accept :
10563 accepts the response if the condition is true (when used with "if")
10564 or false (when used with "unless"). The first such rule executed ends
10565 the rules evaluation.
10566
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010567 - close :
10568 immediately closes the connection with the server if the condition is
10569 true (when used with "if"), or false (when used with "unless"). The
10570 first such rule executed ends the rules evaluation. The main purpose of
10571 this action is to force a connection to be finished between a client
10572 and a server after an exchange when the application protocol expects
10573 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010574 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010575 protocols.
10576
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010577 - reject :
10578 rejects the response if the condition is true (when used with "if")
10579 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010580 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010581
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010582 - set-var(<var-name>) <expr>
10583 Sets a variable.
10584
Christopher Faulet85d79c92016-11-09 16:54:56 +010010585 - unset-var(<var-name>)
10586 Unsets a variable.
10587
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020010588 - sc-inc-gpc0(<sc-id>):
10589 This action increments the GPC0 counter according to the sticky
10590 counter designated by <sc-id>. If an error occurs, this action fails
10591 silently and the actions evaluation continues.
10592
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010593 - sc-inc-gpc1(<sc-id>):
10594 This action increments the GPC1 counter according to the sticky
10595 counter designated by <sc-id>. If an error occurs, this action fails
10596 silently and the actions evaluation continues.
10597
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010598 - sc-set-gpt0(<sc-id>) <int> :
10599 This action sets the GPT0 tag according to the sticky counter designated
10600 by <sc-id> and the value of <int>. The expected result is a boolean. If
10601 an error occurs, this action silently fails and the actions evaluation
10602 continues.
10603
Willy Tarreau2d392c22015-08-24 01:43:45 +020010604 - "silent-drop" :
10605 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010010606 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020010607 to prevent the client from being notified. The effect it then that the
10608 client still sees an established connection while there's none on
10609 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
10610 except that it doesn't use any local resource at all on the machine
10611 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010010612 slow down stronger attackers. It is important to understand the impact
10613 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010614 client and HAProxy (firewalls, proxies, load balancers) will also keep
10615 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010616 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010617 TCP_REPAIR socket option is used to block the emission of a TCP
10618 reset. On other systems, the socket's TTL is reduced to 1 so that the
10619 TCP reset doesn't pass the first router, though it's still delivered to
10620 local networks. Do not use it unless you fully understand how it works.
10621
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010622 - send-spoe-group <engine-name> <group-name>
10623 Send a group of SPOE messages.
10624
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010625 Note that the "if/unless" condition is optional. If no condition is set on
10626 the action, it is simply performed unconditionally. That can be useful for
10627 for changing the default action to a reject.
10628
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010629 It is perfectly possible to match layer 7 contents with "tcp-response
10630 content" rules, but then it is important to ensure that a full response has
10631 been buffered, otherwise no contents will match. In order to achieve this,
10632 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010633 period.
10634
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010635 The "set-var" is used to set the content of a variable. The variable is
10636 declared inline.
10637
Daniel Schneller0b547052016-03-21 20:46:57 +010010638 <var-name> The name of the variable starts with an indication about
10639 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010010640 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010010641 "sess" : the variable is shared with the whole session
10642 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010643 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010010644 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010645 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010010646 "res" : the variable is shared only during response
10647 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010648 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010010649 The name may only contain characters 'a-z', 'A-Z', '0-9',
10650 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010651
10652 <expr> Is a standard HAProxy expression formed by a sample-fetch
10653 followed by some converters.
10654
10655 Example:
10656
10657 tcp-request content set-var(sess.my_var) src
10658
Christopher Faulet85d79c92016-11-09 16:54:56 +010010659 The "unset-var" is used to unset a variable. See above for details about
10660 <var-name>.
10661
10662 Example:
10663
10664 tcp-request content unset-var(sess.my_var)
10665
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010666 The "send-spoe-group" is used to trigger sending of a group of SPOE
10667 messages. To do so, the SPOE engine used to send messages must be defined, as
10668 well as the SPOE group to send. Of course, the SPOE engine must refer to an
10669 existing SPOE filter. If not engine name is provided on the SPOE filter line,
10670 the SPOE agent name must be used.
10671
10672 <engine-name> The SPOE engine name.
10673
10674 <group-name> The SPOE group name as specified in the engine configuration.
10675
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010676 See section 7 about ACL usage.
10677
10678 See also : "tcp-request content", "tcp-response inspect-delay"
10679
10680
Willy Tarreau4f614292016-10-21 17:49:36 +020010681tcp-request session <action> [{if | unless} <condition>]
10682 Perform an action on a validated session depending on a layer 5 condition
10683 May be used in sections : defaults | frontend | listen | backend
10684 no | yes | yes | no
10685 Arguments :
10686 <action> defines the action to perform if the condition applies. See
10687 below.
10688
10689 <condition> is a standard layer5-only ACL-based condition (see section 7).
10690
Davor Ocelice9ed2812017-12-25 17:49:28 +010010691 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020010692 it is possible to evaluate some conditions to decide whether this session
10693 must be accepted or dropped or have its counters tracked. Those conditions
10694 cannot make use of any data contents because no buffers are allocated yet and
10695 the processing cannot wait at this stage. The main use case it to copy some
10696 early information into variables (since variables are accessible in the
10697 session), or to keep track of some information collected after the handshake,
10698 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010010699 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020010700 extracted information can thus be copied to a variable or tracked using
10701 "track-sc" rules. Of course it is also possible to decide to accept/reject as
10702 with other rulesets. Most operations performed here could also be performed
10703 in "tcp-request content" rules, except that in HTTP these rules are evaluated
10704 for each new request, and that might not always be acceptable. For example a
10705 rule might increment a counter on each evaluation. It would also be possible
10706 that a country is resolved by geolocation from the source IP address,
10707 assigned to a session-wide variable, then the source address rewritten from
10708 an HTTP header for all requests. If some contents need to be inspected in
10709 order to take the decision, the "tcp-request content" statements must be used
10710 instead.
10711
10712 The "tcp-request session" rules are evaluated in their exact declaration
10713 order. If no rule matches or if there is no rule, the default action is to
10714 accept the incoming session. There is no specific limit to the number of
10715 rules which may be inserted.
10716
10717 Several types of actions are supported :
10718 - accept : the request is accepted
10719 - reject : the request is rejected and the connection is closed
10720 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
10721 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010722 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010723 - sc-set-gpt0(<sc-id>) <int>
10724 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010725 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010726 - silent-drop
10727
10728 These actions have the same meaning as their respective counter-parts in
10729 "tcp-request connection" and "tcp-request content", so please refer to these
10730 sections for a complete description.
10731
10732 Note that the "if/unless" condition is optional. If no condition is set on
10733 the action, it is simply performed unconditionally. That can be useful for
10734 "track-sc*" actions as well as for changing the default action to a reject.
10735
10736 Example: track the original source address by default, or the one advertised
10737 in the PROXY protocol header for connection coming from the local
10738 proxies. The first connection-level rule enables receipt of the
10739 PROXY protocol for these ones, the second rule tracks whatever
10740 address we decide to keep after optional decoding.
10741
10742 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10743 tcp-request session track-sc0 src
10744
10745 Example: accept all sessions from white-listed hosts, reject too fast
10746 sessions without counting them, and track accepted sessions.
10747 This results in session rate being capped from abusive sources.
10748
10749 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10750 tcp-request session reject if { src_sess_rate gt 10 }
10751 tcp-request session track-sc0 src
10752
10753 Example: accept all sessions from white-listed hosts, count all other
10754 sessions and reject too fast ones. This results in abusive ones
10755 being blocked as long as they don't slow down.
10756
10757 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10758 tcp-request session track-sc0 src
10759 tcp-request session reject if { sc0_sess_rate gt 10 }
10760
10761 See section 7 about ACL usage.
10762
10763 See also : "tcp-request connection", "tcp-request content", "stick-table"
10764
10765
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010766tcp-response inspect-delay <timeout>
10767 Set the maximum allowed time to wait for a response during content inspection
10768 May be used in sections : defaults | frontend | listen | backend
10769 no | no | yes | yes
10770 Arguments :
10771 <timeout> is the timeout value specified in milliseconds by default, but
10772 can be in any other unit if the number is suffixed by the unit,
10773 as explained at the top of this document.
10774
10775 See also : "tcp-response content", "tcp-request inspect-delay".
10776
10777
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010778timeout check <timeout>
10779 Set additional check timeout, but only after a connection has been already
10780 established.
10781
10782 May be used in sections: defaults | frontend | listen | backend
10783 yes | no | yes | yes
10784 Arguments:
10785 <timeout> is the timeout value specified in milliseconds by default, but
10786 can be in any other unit if the number is suffixed by the unit,
10787 as explained at the top of this document.
10788
10789 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
10790 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010791 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010792 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010010793 (Please also note that there is no valid reason to have such long connect
10794 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
10795 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010796
10797 If "timeout check" is not set haproxy uses "inter" for complete check
10798 timeout (connect + read) exactly like all <1.3.15 version.
10799
10800 In most cases check request is much simpler and faster to handle than normal
10801 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010010802 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010803
10804 This parameter is specific to backends, but can be specified once for all in
10805 "defaults" sections. This is in fact one of the easiest solutions not to
10806 forget about it.
10807
Willy Tarreau41a340d2008-01-22 12:25:31 +010010808 See also: "timeout connect", "timeout queue", "timeout server",
10809 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010810
10811
Willy Tarreau0ba27502007-12-24 16:55:16 +010010812timeout client <timeout>
10813timeout clitimeout <timeout> (deprecated)
10814 Set the maximum inactivity time on the client side.
10815 May be used in sections : defaults | frontend | listen | backend
10816 yes | yes | yes | no
10817 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010818 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010819 can be in any other unit if the number is suffixed by the unit,
10820 as explained at the top of this document.
10821
10822 The inactivity timeout applies when the client is expected to acknowledge or
10823 send data. In HTTP mode, this timeout is particularly important to consider
10824 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010825 response while it is reading data sent by the server. That said, for the
10826 first phase, it is preferable to set the "timeout http-request" to better
10827 protect HAProxy from Slowloris like attacks. The value is specified in
10828 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010829 suffixed by the unit, as specified at the top of this document. In TCP mode
10830 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10831 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010832 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010833 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010834 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10835 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010836 which overrides "timeout client" and "timeout server" for tunnels, as well as
10837 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010838
10839 This parameter is specific to frontends, but can be specified once for all in
10840 "defaults" sections. This is in fact one of the easiest solutions not to
10841 forget about it. An unspecified timeout results in an infinite timeout, which
10842 is not recommended. Such a usage is accepted and works but reports a warning
John Roesler27c754c2019-07-10 15:45:51 -050010843 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010844 the system if the system's timeouts are not configured either.
10845
Willy Tarreau95c4e142017-11-26 12:18:55 +010010846 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010847
Willy Tarreau0ba27502007-12-24 16:55:16 +010010848 This parameter replaces the old, deprecated "clitimeout". It is recommended
10849 to use it to write new configurations. The form "timeout clitimeout" is
10850 provided only by backwards compatibility but its use is strongly discouraged.
10851
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010852 See also : "clitimeout", "timeout server", "timeout tunnel",
10853 "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010854
10855
Willy Tarreau05cdd962014-05-10 14:30:07 +020010856timeout client-fin <timeout>
10857 Set the inactivity timeout on the client side for half-closed connections.
10858 May be used in sections : defaults | frontend | listen | backend
10859 yes | yes | yes | no
10860 Arguments :
10861 <timeout> is the timeout value specified in milliseconds by default, but
10862 can be in any other unit if the number is suffixed by the unit,
10863 as explained at the top of this document.
10864
10865 The inactivity timeout applies when the client is expected to acknowledge or
10866 send data while one direction is already shut down. This timeout is different
10867 from "timeout client" in that it only applies to connections which are closed
10868 in one direction. This is particularly useful to avoid keeping connections in
10869 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10870 problem is particularly common long connections such as RDP or WebSocket.
10871 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010872 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10873 frame was sent, often indicating an expectation that the connection quickly
10874 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010875
10876 This parameter is specific to frontends, but can be specified once for all in
10877 "defaults" sections. By default it is not set, so half-closed connections
10878 will use the other timeouts (timeout.client or timeout.tunnel).
10879
10880 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10881
10882
Willy Tarreau0ba27502007-12-24 16:55:16 +010010883timeout connect <timeout>
10884timeout contimeout <timeout> (deprecated)
10885 Set the maximum time to wait for a connection attempt to a server to succeed.
10886 May be used in sections : defaults | frontend | listen | backend
10887 yes | no | yes | yes
10888 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010889 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010890 can be in any other unit if the number is suffixed by the unit,
10891 as explained at the top of this document.
10892
10893 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010894 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010895 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010896 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010897 connect timeout also presets both queue and tarpit timeouts to the same value
10898 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010899
10900 This parameter is specific to backends, but can be specified once for all in
10901 "defaults" sections. This is in fact one of the easiest solutions not to
10902 forget about it. An unspecified timeout results in an infinite timeout, which
10903 is not recommended. Such a usage is accepted and works but reports a warning
John Roesler27c754c2019-07-10 15:45:51 -050010904 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010905 the system if the system's timeouts are not configured either.
10906
10907 This parameter replaces the old, deprecated "contimeout". It is recommended
10908 to use it to write new configurations. The form "timeout contimeout" is
10909 provided only by backwards compatibility but its use is strongly discouraged.
10910
Willy Tarreau41a340d2008-01-22 12:25:31 +010010911 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
10912 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010913
10914
Willy Tarreaub16a5742010-01-10 14:46:16 +010010915timeout http-keep-alive <timeout>
10916 Set the maximum allowed time to wait for a new HTTP request to appear
10917 May be used in sections : defaults | frontend | listen | backend
10918 yes | yes | yes | yes
10919 Arguments :
10920 <timeout> is the timeout value specified in milliseconds by default, but
10921 can be in any other unit if the number is suffixed by the unit,
10922 as explained at the top of this document.
10923
10924 By default, the time to wait for a new request in case of keep-alive is set
10925 by "timeout http-request". However this is not always convenient because some
10926 people want very short keep-alive timeouts in order to release connections
10927 faster, and others prefer to have larger ones but still have short timeouts
10928 once the request has started to present itself.
10929
10930 The "http-keep-alive" timeout covers these needs. It will define how long to
10931 wait for a new HTTP request to start coming after a response was sent. Once
10932 the first byte of request has been seen, the "http-request" timeout is used
10933 to wait for the complete request to come. Note that empty lines prior to a
10934 new request do not refresh the timeout and are not counted as a new request.
10935
10936 There is also another difference between the two timeouts : when a connection
10937 expires during timeout http-keep-alive, no error is returned, the connection
10938 just closes. If the connection expires in "http-request" while waiting for a
10939 connection to complete, a HTTP 408 error is returned.
10940
10941 In general it is optimal to set this value to a few tens to hundreds of
10942 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010943 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010944 1 millisecond) it will probably only accept pipelined requests but not the
10945 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010946 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010947
10948 If this parameter is not set, the "http-request" timeout applies, and if both
10949 are not set, "timeout client" still applies at the lower level. It should be
10950 set in the frontend to take effect, unless the frontend is in TCP mode, in
10951 which case the HTTP backend's timeout will be used.
10952
Willy Tarreau95c4e142017-11-26 12:18:55 +010010953 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10954 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010955 (where we only have one connection per client and a connection setup).
10956
Willy Tarreaub16a5742010-01-10 14:46:16 +010010957 See also : "timeout http-request", "timeout client".
10958
10959
Willy Tarreau036fae02008-01-06 13:24:40 +010010960timeout http-request <timeout>
10961 Set the maximum allowed time to wait for a complete HTTP request
10962 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010963 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010964 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010965 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010966 can be in any other unit if the number is suffixed by the unit,
10967 as explained at the top of this document.
10968
10969 In order to offer DoS protection, it may be required to lower the maximum
10970 accepted time to receive a complete HTTP request without affecting the client
10971 timeout. This helps protecting against established connections on which
10972 nothing is sent. The client timeout cannot offer a good protection against
10973 this abuse because it is an inactivity timeout, which means that if the
10974 attacker sends one character every now and then, the timeout will not
10975 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010976 types, the request will be aborted if it does not complete in time. When the
10977 timeout expires, an HTTP 408 response is sent to the client to inform it
10978 about the problem, and the connection is closed. The logs will report
10979 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010980 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010981 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10982 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010983
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010984 By default, this timeout only applies to the header part of the request,
10985 and not to any data. As soon as the empty line is received, this timeout is
10986 not used anymore. When combined with "option http-buffer-request", this
10987 timeout also applies to the body of the request..
10988 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010989 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010990
10991 Generally it is enough to set it to a few seconds, as most clients send the
10992 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010993 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010994 generally work on local networks as long as there are no packet losses. This
10995 will prevent people from sending bare HTTP requests using telnet.
10996
10997 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010998 chunk of the incoming request. It should be set in the frontend to take
10999 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
11000 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010011001
Willy Tarreau0f228a02015-05-01 15:37:53 +020011002 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010011003 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010011004
Willy Tarreau844e3c52008-01-11 16:28:18 +010011005
11006timeout queue <timeout>
11007 Set the maximum time to wait in the queue for a connection slot to be free
11008 May be used in sections : defaults | frontend | listen | backend
11009 yes | no | yes | yes
11010 Arguments :
11011 <timeout> is the timeout value specified in milliseconds by default, but
11012 can be in any other unit if the number is suffixed by the unit,
11013 as explained at the top of this document.
11014
11015 When a server's maxconn is reached, connections are left pending in a queue
11016 which may be server-specific or global to the backend. In order not to wait
11017 indefinitely, a timeout is applied to requests pending in the queue. If the
11018 timeout is reached, it is considered that the request will almost never be
11019 served, so it is dropped and a 503 error is returned to the client.
11020
11021 The "timeout queue" statement allows to fix the maximum time for a request to
11022 be left pending in a queue. If unspecified, the same value as the backend's
11023 connection timeout ("timeout connect") is used, for backwards compatibility
11024 with older versions with no "timeout queue" parameter.
11025
11026 See also : "timeout connect", "contimeout".
11027
11028
11029timeout server <timeout>
11030timeout srvtimeout <timeout> (deprecated)
11031 Set the maximum inactivity time on the server side.
11032 May be used in sections : defaults | frontend | listen | backend
11033 yes | no | yes | yes
11034 Arguments :
11035 <timeout> is the timeout value specified in milliseconds by default, but
11036 can be in any other unit if the number is suffixed by the unit,
11037 as explained at the top of this document.
11038
11039 The inactivity timeout applies when the server is expected to acknowledge or
11040 send data. In HTTP mode, this timeout is particularly important to consider
11041 during the first phase of the server's response, when it has to send the
11042 headers, as it directly represents the server's processing time for the
11043 request. To find out what value to put there, it's often good to start with
11044 what would be considered as unacceptable response times, then check the logs
11045 to observe the response time distribution, and adjust the value accordingly.
11046
11047 The value is specified in milliseconds by default, but can be in any other
11048 unit if the number is suffixed by the unit, as specified at the top of this
11049 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
11050 recommended that the client timeout remains equal to the server timeout in
11051 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010011052 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010011053 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010011054 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
11055 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020011056 "timeout tunnel", which overrides "timeout client" and "timeout server" for
11057 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011058
11059 This parameter is specific to backends, but can be specified once for all in
11060 "defaults" sections. This is in fact one of the easiest solutions not to
11061 forget about it. An unspecified timeout results in an infinite timeout, which
11062 is not recommended. Such a usage is accepted and works but reports a warning
John Roesler27c754c2019-07-10 15:45:51 -050011063 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010011064 the system if the system's timeouts are not configured either.
11065
11066 This parameter replaces the old, deprecated "srvtimeout". It is recommended
11067 to use it to write new configurations. The form "timeout srvtimeout" is
11068 provided only by backwards compatibility but its use is strongly discouraged.
11069
Willy Tarreauce887fd2012-05-12 12:50:00 +020011070 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010011071
Willy Tarreau05cdd962014-05-10 14:30:07 +020011072
11073timeout server-fin <timeout>
11074 Set the inactivity timeout on the server side for half-closed connections.
11075 May be used in sections : defaults | frontend | listen | backend
11076 yes | no | yes | yes
11077 Arguments :
11078 <timeout> is the timeout value specified in milliseconds by default, but
11079 can be in any other unit if the number is suffixed by the unit,
11080 as explained at the top of this document.
11081
11082 The inactivity timeout applies when the server is expected to acknowledge or
11083 send data while one direction is already shut down. This timeout is different
11084 from "timeout server" in that it only applies to connections which are closed
11085 in one direction. This is particularly useful to avoid keeping connections in
11086 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
11087 This problem is particularly common long connections such as RDP or WebSocket.
11088 Note that this timeout can override "timeout tunnel" when a connection shuts
11089 down in one direction. This setting was provided for completeness, but in most
11090 situations, it should not be needed.
11091
11092 This parameter is specific to backends, but can be specified once for all in
11093 "defaults" sections. By default it is not set, so half-closed connections
11094 will use the other timeouts (timeout.server or timeout.tunnel).
11095
11096 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
11097
Willy Tarreau844e3c52008-01-11 16:28:18 +010011098
11099timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010011100 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010011101 May be used in sections : defaults | frontend | listen | backend
11102 yes | yes | yes | yes
11103 Arguments :
11104 <timeout> is the tarpit duration specified in milliseconds by default, but
11105 can be in any other unit if the number is suffixed by the unit,
11106 as explained at the top of this document.
11107
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011108 When a connection is tarpitted using "http-request tarpit" or
11109 "reqtarpit", it is maintained open with no activity for a certain
11110 amount of time, then closed. "timeout tarpit" defines how long it will
11111 be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011112
11113 The value is specified in milliseconds by default, but can be in any other
11114 unit if the number is suffixed by the unit, as specified at the top of this
11115 document. If unspecified, the same value as the backend's connection timeout
11116 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010011117 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011118
11119 See also : "timeout connect", "contimeout".
11120
11121
Willy Tarreauce887fd2012-05-12 12:50:00 +020011122timeout tunnel <timeout>
11123 Set the maximum inactivity time on the client and server side for tunnels.
11124 May be used in sections : defaults | frontend | listen | backend
11125 yes | no | yes | yes
11126 Arguments :
11127 <timeout> is the timeout value specified in milliseconds by default, but
11128 can be in any other unit if the number is suffixed by the unit,
11129 as explained at the top of this document.
11130
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011131 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020011132 between a client and a server, and the connection remains inactive in both
11133 directions. This timeout supersedes both the client and server timeouts once
11134 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010011135 analyzer remains attached to either connection (e.g. tcp content rules are
11136 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020011137 when switching to the WebSocket protocol, or forwarding a CONNECT request
11138 to a proxy), or after the first response when no keepalive/close option is
11139 specified.
11140
Willy Tarreau05cdd962014-05-10 14:30:07 +020011141 Since this timeout is usually used in conjunction with long-lived connections,
11142 it usually is a good idea to also set "timeout client-fin" to handle the
11143 situation where a client suddenly disappears from the net and does not
11144 acknowledge a close, or sends a shutdown and does not acknowledge pending
11145 data anymore. This can happen in lossy networks where firewalls are present,
11146 and is detected by the presence of large amounts of sessions in a FIN_WAIT
11147 state.
11148
Willy Tarreauce887fd2012-05-12 12:50:00 +020011149 The value is specified in milliseconds by default, but can be in any other
11150 unit if the number is suffixed by the unit, as specified at the top of this
11151 document. Whatever the expected normal idle time, it is a good practice to
11152 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010011153 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020011154
11155 This parameter is specific to backends, but can be specified once for all in
11156 "defaults" sections. This is in fact one of the easiest solutions not to
11157 forget about it.
11158
11159 Example :
11160 defaults http
11161 option http-server-close
11162 timeout connect 5s
11163 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020011164 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020011165 timeout server 30s
11166 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
11167
Willy Tarreau05cdd962014-05-10 14:30:07 +020011168 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020011169
11170
Willy Tarreau844e3c52008-01-11 16:28:18 +010011171transparent (deprecated)
11172 Enable client-side transparent proxying
11173 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010011174 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010011175 Arguments : none
11176
11177 This keyword was introduced in order to provide layer 7 persistence to layer
11178 3 load balancers. The idea is to use the OS's ability to redirect an incoming
11179 connection for a remote address to a local process (here HAProxy), and let
11180 this process know what address was initially requested. When this option is
11181 used, sessions without cookies will be forwarded to the original destination
11182 IP address of the incoming request (which should match that of another
11183 equipment), while requests with cookies will still be forwarded to the
11184 appropriate server.
11185
11186 The "transparent" keyword is deprecated, use "option transparent" instead.
11187
11188 Note that contrary to a common belief, this option does NOT make HAProxy
11189 present the client's IP to the server when establishing the connection.
11190
Willy Tarreau844e3c52008-01-11 16:28:18 +010011191 See also: "option transparent"
11192
William Lallemanda73203e2012-03-12 12:48:57 +010011193unique-id-format <string>
11194 Generate a unique ID for each request.
11195 May be used in sections : defaults | frontend | listen | backend
11196 yes | yes | yes | no
11197 Arguments :
11198 <string> is a log-format string.
11199
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011200 This keyword creates a ID for each request using the custom log format. A
11201 unique ID is useful to trace a request passing through many components of
11202 a complex infrastructure. The newly created ID may also be logged using the
11203 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010011204
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011205 The format should be composed from elements that are guaranteed to be
11206 unique when combined together. For instance, if multiple haproxy instances
11207 are involved, it might be important to include the node name. It is often
11208 needed to log the incoming connection's source and destination addresses
11209 and ports. Note that since multiple requests may be performed over the same
11210 connection, including a request counter may help differentiate them.
11211 Similarly, a timestamp may protect against a rollover of the counter.
11212 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010011213
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011214 It is recommended to use hexadecimal notation for many fields since it
11215 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010011216
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011217 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010011218
Julien Vehentf21be322014-03-07 08:27:34 -050011219 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010011220
11221 will generate:
11222
11223 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
11224
11225 See also: "unique-id-header"
11226
11227unique-id-header <name>
11228 Add a unique ID header in the HTTP request.
11229 May be used in sections : defaults | frontend | listen | backend
11230 yes | yes | yes | no
11231 Arguments :
11232 <name> is the name of the header.
11233
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011234 Add a unique-id header in the HTTP request sent to the server, using the
11235 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010011236
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011237 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010011238
Julien Vehentf21be322014-03-07 08:27:34 -050011239 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010011240 unique-id-header X-Unique-ID
11241
11242 will generate:
11243
11244 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
11245
11246 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010011247
Willy Tarreauf51658d2014-04-23 01:21:56 +020011248use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020011249 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011250 May be used in sections : defaults | frontend | listen | backend
11251 no | yes | yes | no
11252 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010011253 <backend> is the name of a valid backend or "listen" section, or a
11254 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011255
Willy Tarreauf51658d2014-04-23 01:21:56 +020011256 <condition> is a condition composed of ACLs, as described in section 7. If
11257 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011258
11259 When doing content-switching, connections arrive on a frontend and are then
11260 dispatched to various backends depending on a number of conditions. The
11261 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020011262 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011263 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020011264 source address validation) or combined with a "tcp-request" rule to wait for
11265 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011266
11267 There may be as many "use_backend" rules as desired. All of these rules are
11268 evaluated in their declaration order, and the first one which matches will
11269 assign the backend.
11270
11271 In the first form, the backend will be used if the condition is met. In the
11272 second form, the backend will be used if the condition is not met. If no
11273 condition is valid, the backend defined with "default_backend" will be used.
11274 If no default backend is defined, either the servers in the same section are
11275 used (in case of a "listen" section) or, in case of a frontend, no server is
11276 used and a 503 service unavailable response is returned.
11277
Willy Tarreau51aecc72009-07-12 09:47:04 +020011278 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011279 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020011280 and backend processing will immediately follow, or the backend will wait for
11281 a complete HTTP request to get in. This feature is useful when a frontend
11282 must decode several protocols on a unique port, one of them being HTTP.
11283
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010011284 When <backend> is a simple name, it is resolved at configuration time, and an
11285 error is reported if the specified backend does not exist. If <backend> is
11286 a log-format string instead, no check may be done at configuration time, so
11287 the backend name is resolved dynamically at run time. If the resulting
11288 backend name does not correspond to any valid backend, no other rule is
11289 evaluated, and the default_backend directive is applied instead. Note that
11290 when using dynamic backend names, it is highly recommended to use a prefix
11291 that no other backend uses in order to ensure that an unauthorized backend
11292 cannot be forced from the request.
11293
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011294 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010011295 used to detect the association between frontends and backends to compute the
11296 backend's "fullconn" setting. This cannot be done for dynamic names.
11297
11298 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
11299 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010011300
Willy Tarreau036fae02008-01-06 13:24:40 +010011301
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011302use-server <server> if <condition>
11303use-server <server> unless <condition>
11304 Only use a specific server if/unless an ACL-based condition is matched.
11305 May be used in sections : defaults | frontend | listen | backend
11306 no | no | yes | yes
11307 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011308 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011309
11310 <condition> is a condition composed of ACLs, as described in section 7.
11311
11312 By default, connections which arrive to a backend are load-balanced across
11313 the available servers according to the configured algorithm, unless a
11314 persistence mechanism such as a cookie is used and found in the request.
11315
11316 Sometimes it is desirable to forward a particular request to a specific
11317 server without having to declare a dedicated backend for this server. This
11318 can be achieved using the "use-server" rules. These rules are evaluated after
11319 the "redirect" rules and before evaluating cookies, and they have precedence
11320 on them. There may be as many "use-server" rules as desired. All of these
11321 rules are evaluated in their declaration order, and the first one which
11322 matches will assign the server.
11323
11324 If a rule designates a server which is down, and "option persist" is not used
11325 and no force-persist rule was validated, it is ignored and evaluation goes on
11326 with the next rules until one matches.
11327
11328 In the first form, the server will be used if the condition is met. In the
11329 second form, the server will be used if the condition is not met. If no
11330 condition is valid, the processing continues and the server will be assigned
11331 according to other persistence mechanisms.
11332
11333 Note that even if a rule is matched, cookie processing is still performed but
11334 does not assign the server. This allows prefixed cookies to have their prefix
11335 stripped.
11336
11337 The "use-server" statement works both in HTTP and TCP mode. This makes it
11338 suitable for use with content-based inspection. For instance, a server could
11339 be selected in a farm according to the TLS SNI field. And if these servers
11340 have their weight set to zero, they will not be used for other traffic.
11341
11342 Example :
11343 # intercept incoming TLS requests based on the SNI field
11344 use-server www if { req_ssl_sni -i www.example.com }
11345 server www 192.168.0.1:443 weight 0
11346 use-server mail if { req_ssl_sni -i mail.example.com }
11347 server mail 192.168.0.1:587 weight 0
11348 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000011349 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011350 # all the rest is forwarded to this server
11351 server default 192.168.0.2:443 check
11352
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011353 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011354
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011355
Davor Ocelice9ed2812017-12-25 17:49:28 +0100113565. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011357--------------------------
11358
11359The "bind", "server" and "default-server" keywords support a number of settings
11360depending on some build options and on the system HAProxy was built on. These
11361settings generally each consist in one word sometimes followed by a value,
11362written on the same line as the "bind" or "server" line. All these options are
11363described in this section.
11364
11365
113665.1. Bind options
11367-----------------
11368
11369The "bind" keyword supports a certain number of settings which are all passed
11370as arguments on the same line. The order in which those arguments appear makes
11371no importance, provided that they appear after the bind address. All of these
11372parameters are optional. Some of them consist in a single words (booleans),
11373while other ones expect a value after them. In this case, the value must be
11374provided immediately after the setting name.
11375
11376The currently supported settings are the following ones.
11377
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011378accept-netscaler-cip <magic number>
11379 Enforces the use of the NetScaler Client IP insertion protocol over any
11380 connection accepted by any of the TCP sockets declared on the same line. The
11381 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
11382 the incoming connection to be used everywhere an address is used, with the
11383 only exception of "tcp-request connection" rules which will only see the
11384 real connection address. Logs will reflect the addresses indicated in the
11385 protocol, unless it is violated, in which case the real address will still
11386 be used. This keyword combined with support from external components can be
11387 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010011388 mechanism which is not always reliable and not even always usable. See also
11389 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
11390 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011391
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011392accept-proxy
11393 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020011394 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
11395 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011396 3/4 addresses of the incoming connection to be used everywhere an address is
11397 used, with the only exception of "tcp-request connection" rules which will
11398 only see the real connection address. Logs will reflect the addresses
11399 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010011400 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011401 components can be used as an efficient and reliable alternative to the
11402 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011403 usable. See also "tcp-request connection expect-proxy" for a finer-grained
11404 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011405
Olivier Houchardc2aae742017-09-22 18:26:28 +020011406allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010011407 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010011408 due to security considerations. Because it is vulnerable to replay attacks,
John Roesler27c754c2019-07-10 15:45:51 -050011409 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010011410 that are idempotent. You can use the "wait-for-handshake" action for any
11411 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020011412
Willy Tarreauab861d32013-04-02 02:30:41 +020011413alpn <protocols>
11414 This enables the TLS ALPN extension and advertises the specified protocol
11415 list as supported on top of ALPN. The protocol list consists in a comma-
11416 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roesler27c754c2019-07-10 15:45:51 -050011417 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020011418 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010011419 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
11420 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
11421 now obsolete NPN extension. At the time of writing this, most browsers still
11422 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
11423 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
11424 are expected to be supported, both versions can be advertised, in order of
11425 preference, like below :
11426
11427 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020011428
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011429backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010011430 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011431 backlog is used instead, which generally defaults to the maxconn value.
11432
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010011433curves <curves>
11434 This setting is only available when support for OpenSSL was built in. It sets
11435 the string describing the list of elliptic curves algorithms ("curve suite")
11436 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
11437 string is a colon-delimited list of curve name.
11438 Example: "X25519:P-256" (without quote)
11439 When "curves" is set, "ecdhe" parameter is ignored.
11440
Emeric Brun7fb34422012-09-28 15:26:15 +020011441ecdhe <named curve>
11442 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010011443 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
11444 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020011445
Emeric Brunfd33a262012-10-11 16:28:27 +020011446ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020011447 This setting is only available when support for OpenSSL was built in. It
11448 designates a PEM file from which to load CA certificates used to verify
11449 client's certificate.
11450
Emeric Brunb6dc9342012-09-28 17:55:37 +020011451ca-ignore-err [all|<errorID>,...]
11452 This setting is only available when support for OpenSSL was built in.
11453 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
11454 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
11455 error is ignored.
11456
Christopher Faulet31af49d2015-06-09 17:29:50 +020011457ca-sign-file <cafile>
11458 This setting is only available when support for OpenSSL was built in. It
11459 designates a PEM file containing both the CA certificate and the CA private
11460 key used to create and sign server's certificates. This is a mandatory
11461 setting when the dynamic generation of certificates is enabled. See
11462 'generate-certificates' for details.
11463
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000011464ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020011465 This setting is only available when support for OpenSSL was built in. It is
11466 the CA private key passphrase. This setting is optional and used only when
11467 the dynamic generation of certificates is enabled. See
11468 'generate-certificates' for details.
11469
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011470ciphers <ciphers>
11471 This setting is only available when support for OpenSSL was built in. It sets
11472 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000011473 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011474 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011475 information and recommendations see e.g.
11476 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
11477 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
11478 cipher configuration, please check the "ciphersuites" keyword.
11479
11480ciphersuites <ciphersuites>
11481 This setting is only available when support for OpenSSL was built in and
11482 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
11483 the list of cipher algorithms ("cipher suite") that are negotiated during the
11484 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011485 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
11486 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011487
Emeric Brunfd33a262012-10-11 16:28:27 +020011488crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020011489 This setting is only available when support for OpenSSL was built in. It
11490 designates a PEM file from which to load certificate revocation list used
11491 to verify client's certificate.
11492
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011493crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000011494 This setting is only available when support for OpenSSL was built in. It
11495 designates a PEM file containing both the required certificates and any
11496 associated private keys. This file can be built by concatenating multiple
11497 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
11498 requires an intermediate certificate, this can also be concatenated into this
11499 file.
11500
11501 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
11502 are loaded.
11503
11504 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010011505 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010011506 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
11507 specified multiple times in order to load certificates from multiple files or
11508 directories. The certificates will be presented to clients who provide a
11509 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010011510 subjects. Wildcards are supported, where a wildcard character '*' is used
11511 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010011512 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000011513
11514 If no SNI is provided by the client or if the SSL library does not support
11515 TLS extensions, or if the client provides an SNI hostname which does not
11516 match any certificate, then the first loaded certificate will be presented.
11517 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010011518 recommended to load the default one first as a file or to ensure that it will
11519 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000011520
Emeric Brune032bfa2012-09-28 13:01:45 +020011521 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011522
Davor Ocelice9ed2812017-12-25 17:49:28 +010011523 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000011524 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010011525 choose a web server that the CA believes requires an intermediate CA (for
11526 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000011527 others, e.g. nginx, result in a wrong bundle that will not work for some
11528 clients).
11529
Emeric Brun4147b2e2014-06-16 18:36:30 +020011530 For each PEM file, haproxy checks for the presence of file at the same path
11531 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
11532 Status Request extension (also known as "OCSP stapling") is automatically
11533 enabled. The content of this file is optional. If not empty, it must contain
11534 a valid OCSP Response in DER format. In order to be valid an OCSP Response
11535 must comply with the following rules: it has to indicate a good status,
11536 it has to be a single response for the certificate of the PEM file, and it
11537 has to be valid at the moment of addition. If these rules are not respected
11538 the OCSP Response is ignored and a warning is emitted. In order to identify
11539 which certificate an OCSP Response applies to, the issuer's certificate is
11540 necessary. If the issuer's certificate is not found in the PEM file, it will
11541 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
11542 if it exists otherwise it will fail with an error.
11543
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010011544 For each PEM file, haproxy also checks for the presence of file at the same
11545 path suffixed by ".sctl". If such file is found, support for Certificate
11546 Transparency (RFC6962) TLS extension is enabled. The file must contain a
11547 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
11548 to check basic syntax, but no signatures are verified.
11549
yanbzhu6c25e9e2016-01-05 12:52:02 -050011550 There are cases where it is desirable to support multiple key types, e.g. RSA
11551 and ECDSA in the cipher suites offered to the clients. This allows clients
11552 that support EC certificates to be able to use EC ciphers, while
11553 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050011554
11555 In order to provide this functionality, multiple PEM files, each with a
11556 different key type, are required. To associate these PEM files into a
11557 "cert bundle" that is recognized by haproxy, they must be named in the
11558 following way: All PEM files that are to be bundled must have the same base
11559 name, with a suffix indicating the key type. Currently, three suffixes are
11560 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
11561 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
11562 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
11563 suffix matters. To load this bundle into haproxy, specify the base name only:
11564
11565 Example : bind :8443 ssl crt example.pem
11566
yanbzhu6c25e9e2016-01-05 12:52:02 -050011567 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050011568 a cert bundle.
11569
Davor Ocelice9ed2812017-12-25 17:49:28 +010011570 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050011571 support multiple key types. PEM files are combined based on Common Name
11572 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
11573 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
11574 entries in the certificates in that bundle, haproxy will not be able to
11575 provide multi-cert support.
11576
11577 Assuming bundle in the example above contained the following:
11578
11579 Filename | CN | SAN
11580 -------------------+-----------------+-------------------
11581 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050011582 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050011583 example.pem.ecdsa | www.example.com | ecdsa.example.com
11584 -------------------+-----------------+-------------------
11585
11586 Users connecting with an SNI of "www.example.com" will be able
11587 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
11588 "rsa.example.com" will only be able to use RSA cipher suites, and users
11589 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020011590 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
11591 no need to bundle certificates. ECDSA certificate will be preferred if client
11592 support it.
yanbzhud19630c2015-12-14 15:10:25 -050011593
11594 If a directory name is given as the <cert> argument, haproxy will
11595 automatically search and load bundled files in that directory.
11596
11597 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
11598 bundling. Each certificate can have its own .ocsp and .issuer file. At this
11599 time, sctl is not supported in multi-certificate bundling.
11600
Emeric Brunb6dc9342012-09-28 17:55:37 +020011601crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000011602 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010011603 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011604 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000011605 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020011606
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011607crt-list <file>
11608 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011609 designates a list of PEM file with an optional ssl configuration and a SNI
11610 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011611
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011612 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
11613
William Lallemand0b77c182020-06-30 16:11:36 +020011614 sslbindconf supports "allow-0rtt", "alpn", "ca-file", "ciphers",
11615 "ciphersuites", "crl-file", "curves", "ecdhe", "no-ca-names", "npn",
11616 "verify" configuration. With BoringSSL and Openssl >= 1.1.1
11617 "ssl-min-ver" and "ssl-max-ver" are also supported. It overrides the
11618 configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011619
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020011620 Wildcards are supported in the SNI filter. Negative filter are also supported,
11621 only useful in combination with a wildcard filter to exclude a particular SNI.
11622 The certificates will be presented to clients who provide a valid TLS Server
11623 Name Indication field matching one of the SNI filters. If no SNI filter is
11624 specified, the CN and alt subjects are used. This directive may be specified
11625 multiple times. See the "crt" option for more information. The default
11626 certificate is still needed to meet OpenSSL expectations. If it is not used,
11627 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011628
yanbzhu6c25e9e2016-01-05 12:52:02 -050011629 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020011630 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020011631 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
11632 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
11633 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050011634
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011635 crt-list file example:
11636 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011637 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011638 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011639 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011640
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011641defer-accept
11642 Is an optional keyword which is supported only on certain Linux kernels. It
11643 states that a connection will only be accepted once some data arrive on it,
11644 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010011645 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011646 performance by ensuring that most of the request is already available when
11647 the connection is accepted. On the other hand, it will not be able to detect
11648 connections which don't talk. It is important to note that this option is
11649 broken in all kernels up to 2.6.31, as the connection is never accepted until
11650 the client talks. This can cause issues with front firewalls which would see
11651 an established connection while the proxy will only see it in SYN_RECV. This
11652 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
11653
William Lallemandf6975e92017-05-26 17:42:10 +020011654expose-fd listeners
11655 This option is only usable with the stats socket. It gives your stats socket
11656 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020011657 During a reload with the master-worker mode, the process is automatically
11658 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011659 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020011660
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011661force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011662 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011663 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011664 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011665 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011666
11667force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011668 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011669 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011670 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011671
11672force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011673 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011674 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011675 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011676
11677force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011678 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011679 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011680 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011681
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011682force-tlsv13
11683 This option enforces use of TLSv1.3 only on SSL connections instantiated from
11684 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011685 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011686
Christopher Faulet31af49d2015-06-09 17:29:50 +020011687generate-certificates
11688 This setting is only available when support for OpenSSL was built in. It
11689 enables the dynamic SSL certificates generation. A CA certificate and its
11690 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
11691 a transparent forward proxy, SSL requests generate errors because of a common
11692 name mismatch on the certificate presented to the client. With this option
11693 enabled, HAProxy will try to forge a certificate using the SNI hostname
11694 indicated by the client. This is done only if no certificate matches the SNI
11695 hostname (see 'crt-list'). If an error occurs, the default certificate is
11696 used, else the 'strict-sni' option is set.
11697 It can also be used when HAProxy is configured as a reverse proxy to ease the
11698 deployment of an architecture with many backends.
11699
11700 Creating a SSL certificate is an expensive operation, so a LRU cache is used
11701 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010011702 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020011703 certificate is used many times.
11704
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011705gid <gid>
11706 Sets the group of the UNIX sockets to the designated system gid. It can also
11707 be set by default in the global section's "unix-bind" statement. Note that
11708 some platforms simply ignore this. This setting is equivalent to the "group"
11709 setting except that the group ID is used instead of its name. This setting is
11710 ignored by non UNIX sockets.
11711
11712group <group>
11713 Sets the group of the UNIX sockets to the designated system group. It can
11714 also be set by default in the global section's "unix-bind" statement. Note
11715 that some platforms simply ignore this. This setting is equivalent to the
11716 "gid" setting except that the group name is used instead of its gid. This
11717 setting is ignored by non UNIX sockets.
11718
11719id <id>
11720 Fixes the socket ID. By default, socket IDs are automatically assigned, but
11721 sometimes it is more convenient to fix them to ease monitoring. This value
11722 must be strictly positive and unique within the listener/frontend. This
11723 option can only be used when defining only a single socket.
11724
11725interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010011726 Restricts the socket to a specific interface. When specified, only packets
11727 received from that particular interface are processed by the socket. This is
11728 currently only supported on Linux. The interface must be a primary system
11729 interface, not an aliased interface. It is also possible to bind multiple
11730 frontends to the same address if they are bound to different interfaces. Note
11731 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010011732 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
11733 uses the same interface as inbound traffic, and its associated routing table,
11734 even if there are explicit routes through different interfaces configured.
11735 This can prove useful to address asymmetric routing issues when the same
11736 client IP addresses need to be able to reach frontends hosted on different
11737 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011738
Willy Tarreauabb175f2012-09-24 12:43:26 +020011739level <level>
11740 This setting is used with the stats sockets only to restrict the nature of
11741 the commands that can be issued on the socket. It is ignored by other
11742 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010011743 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020011744 read, and no change is allowed. It would make sense on systems where it
11745 is not easy to restrict access to the socket.
11746 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011747 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020011748 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010011749 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020011750 all counters).
11751
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020011752severity-output <format>
11753 This setting is used with the stats sockets only to configure severity
11754 level output prepended to informational feedback messages. Severity
11755 level of messages can range between 0 and 7, conforming to syslog
11756 rfc5424. Valid and successful socket commands requesting data
11757 (i.e. "show map", "get acl foo" etc.) will never have a severity level
11758 prepended. It is ignored by other sockets. <format> can be one of :
11759 - "none" (default) no severity level is prepended to feedback messages.
11760 - "number" severity level is prepended as a number.
11761 - "string" severity level is prepended as a string following the
11762 rfc5424 convention.
11763
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011764maxconn <maxconn>
11765 Limits the sockets to this number of concurrent connections. Extraneous
11766 connections will remain in the system's backlog until a connection is
11767 released. If unspecified, the limit will be the same as the frontend's
11768 maxconn. Note that in case of port ranges or multiple addresses, the same
11769 value will be applied to each socket. This setting enables different
11770 limitations on expensive sockets, for instance SSL entries which may easily
11771 eat all memory.
11772
11773mode <mode>
11774 Sets the octal mode used to define access permissions on the UNIX socket. It
11775 can also be set by default in the global section's "unix-bind" statement.
11776 Note that some platforms simply ignore this. This setting is ignored by non
11777 UNIX sockets.
11778
11779mss <maxseg>
11780 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
11781 connections. This can be used to force a lower MSS for certain specific
11782 ports, for instance for connections passing through a VPN. Note that this
11783 relies on a kernel feature which is theoretically supported under Linux but
11784 was buggy in all versions prior to 2.6.28. It may or may not work on other
11785 operating systems. It may also not change the advertised value but change the
11786 effective size of outgoing segments. The commonly advertised value for TCPv4
11787 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
11788 positive, it will be used as the advertised MSS. If it is negative, it will
11789 indicate by how much to reduce the incoming connection's advertised MSS for
11790 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
11791
11792name <name>
11793 Sets an optional name for these sockets, which will be reported on the stats
11794 page.
11795
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011796namespace <name>
11797 On Linux, it is possible to specify which network namespace a socket will
11798 belong to. This directive makes it possible to explicitly bind a listener to
11799 a namespace different from the default one. Please refer to your operating
11800 system's documentation to find more details about network namespaces.
11801
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011802nice <nice>
11803 Sets the 'niceness' of connections initiated from the socket. Value must be
11804 in the range -1024..1024 inclusive, and defaults to zero. Positive values
11805 means that such connections are more friendly to others and easily offer
11806 their place in the scheduler. On the opposite, negative values mean that
11807 connections want to run with a higher priority than others. The difference
11808 only happens under high loads when the system is close to saturation.
11809 Negative values are appropriate for low-latency or administration services,
11810 and high values are generally recommended for CPU intensive tasks such as SSL
11811 processing or bulk transfers which are less sensible to latency. For example,
11812 it may make sense to use a positive value for an SMTP socket and a negative
11813 one for an RDP socket.
11814
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011815no-ca-names
11816 This setting is only available when support for OpenSSL was built in. It
11817 prevents from send CA names in server hello message when ca-file is used.
11818
Emeric Brun9b3009b2012-10-05 11:55:06 +020011819no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011820 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011821 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011822 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011823 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011824 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11825 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011826
Emeric Brun90ad8722012-10-02 14:00:59 +020011827no-tls-tickets
11828 This setting is only available when support for OpenSSL was built in. It
11829 disables the stateless session resumption (RFC 5077 TLS Ticket
11830 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011831 session resumption is more expensive in CPU usage. This option is also
11832 available on global statement "ssl-default-bind-options".
Lukas Tribusd8fd6362020-03-10 00:56:09 +010011833 The TLS ticket mechanism is only used up to TLS 1.2.
11834 Forward Secrecy is compromised with TLS tickets, unless ticket keys
11835 are periodically rotated (via reload or by using "tls-ticket-keys").
Emeric Brun90ad8722012-10-02 14:00:59 +020011836
Emeric Brun9b3009b2012-10-05 11:55:06 +020011837no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011838 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011839 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011840 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011841 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011842 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11843 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011844
Emeric Brun9b3009b2012-10-05 11:55:06 +020011845no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011846 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011847 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011848 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011849 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011850 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11851 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011852
Emeric Brun9b3009b2012-10-05 11:55:06 +020011853no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011854 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011855 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011856 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011857 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011858 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11859 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011860
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011861no-tlsv13
11862 This setting is only available when support for OpenSSL was built in. It
11863 disables support for TLSv1.3 on any sockets instantiated from the listener
11864 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11865 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011866 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11867 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011868
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011869npn <protocols>
11870 This enables the NPN TLS extension and advertises the specified protocol list
11871 as supported on top of NPN. The protocol list consists in a comma-delimited
11872 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roesler27c754c2019-07-10 15:45:51 -050011873 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011874 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011875 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11876 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11877 version of OpenSSL, NPN might still be used as most clients still support it
11878 at the time of writing this. It is possible to enable both NPN and ALPN
11879 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011880
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011881prefer-client-ciphers
11882 Use the client's preference when selecting the cipher suite, by default
11883 the server's preference is enforced. This option is also available on
11884 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011885 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11886 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11887 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011888
Christopher Fauletc644fa92017-11-23 22:44:11 +010011889process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010011890 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010011891 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011892 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011893 between the two is applied. If in the end the listener is not allowed to run
11894 on any remaining process, a warning is emitted, and the listener will either
11895 run on the first process of the listener if a single process was specified,
11896 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011897 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010011898 connections for this listener, for the the process set. If multiple processes
11899 and threads are configured, a warning is emitted, as it either results from a
11900 configuration error or a misunderstanding of these models. For the unlikely
11901 case where several ranges are needed, this directive may be repeated.
11902 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010011903
11904 all | odd | even | number[-[number]]
11905
11906 Ranges can be partially defined. The higher bound can be omitted. In such
11907 case, it is replaced by the corresponding maximum value. The main purpose of
11908 this directive is to be used with the stats sockets and have one different
11909 socket per process. The second purpose is to have multiple bind lines sharing
11910 the same IP:port but not the same process in a listener, so that the system
11911 can distribute the incoming connections into multiple queues and allow a
11912 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11913 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011914
Christopher Fauleta717b992018-04-10 14:43:00 +020011915proto <name>
11916 Forces the multiplexer's protocol to use for the incoming connections. It
11917 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11918 be usable on the frontend side. The list of available protocols is reported
11919 in haproxy -vv.
11920 Idea behind this optipon is to bypass the selection of the best multiplexer's
11921 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011922 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011923 h2" on the bind line.
11924
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011925ssl
11926 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011927 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011928 certificate is necessary (see "crt" above). All contents in the buffers will
11929 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011930 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11931 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011932
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011933ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11934 This option enforces use of <version> or lower on SSL connections instantiated
11935 from this listener. This option is also available on global statement
11936 "ssl-default-bind-options". See also "ssl-min-ver".
11937
11938ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11939 This option enforces use of <version> or upper on SSL connections instantiated
11940 from this listener. This option is also available on global statement
11941 "ssl-default-bind-options". See also "ssl-max-ver".
11942
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011943strict-sni
11944 This setting is only available when support for OpenSSL was built in. The
11945 SSL/TLS negotiation is allow only if the client provided an SNI which match
11946 a certificate. The default certificate is not used.
11947 See the "crt" option for more information.
11948
Willy Tarreau2af207a2015-02-04 00:45:58 +010011949tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011950 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011951 listening socket. This option is available on Linux since version 2.6.37. It
11952 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011953 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011954 useful on long-lived connections experiencing long idle periods such as
11955 remote terminals or database connection pools, where the client and server
11956 timeouts must remain high to allow a long period of idle, but where it is
11957 important to detect that the client has disappeared in order to release all
11958 resources associated with its connection (and the server's session). The
11959 argument is a delay expressed in milliseconds by default. This only works
11960 for regular TCP connections, and is ignored for other protocols.
11961
Willy Tarreau1c862c52012-10-05 16:21:00 +020011962tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011963 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011964 enables TCP Fast Open on the listening socket, which means that clients which
11965 support this feature will be able to send a request and receive a response
11966 during the 3-way handshake starting from second connection, thus saving one
11967 round-trip after the first connection. This only makes sense with protocols
11968 that use high connection rates and where each round trip matters. This can
11969 possibly cause issues with many firewalls which do not accept data on SYN
11970 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011971 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11972 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11973 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011974
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011975tls-ticket-keys <keyfile>
11976 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010011977 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
11978 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
11979 The first key determines the key length used for next keys: you can't mix
11980 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
11981 build option (default 3) and at least as many keys need to be present in
11982 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
11983 penultimate one for encryption. This enables easy key rotation by just
11984 appending new key to the file and reloading the process. Keys must be
11985 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
11986 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011987 storage such as hard drives (hint: use tmpfs and don't swap those files).
11988 Lifetime hint can be changed using tune.ssl.timeout.
11989
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011990transparent
11991 Is an optional keyword which is supported only on certain Linux kernels. It
11992 indicates that the addresses will be bound even if they do not belong to the
11993 local machine, and that packets targeting any of these addresses will be
11994 intercepted just as if the addresses were locally configured. This normally
11995 requires that IP forwarding is enabled. Caution! do not use this with the
11996 default address '*', as it would redirect any traffic for the specified port.
11997 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11998 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11999 kernel version. Some distribution kernels include backports of the feature,
12000 so check for support with your vendor.
12001
Willy Tarreau77e3af92012-11-24 15:07:23 +010012002v4v6
12003 Is an optional keyword which is supported only on most recent systems
12004 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
12005 and IPv6 when it uses the default address. Doing so is sometimes necessary
12006 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012007 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010012008
Willy Tarreau9b6700f2012-11-24 11:55:28 +010012009v6only
12010 Is an optional keyword which is supported only on most recent systems
12011 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
12012 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010012013 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
12014 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010012015
Willy Tarreaub6205fd2012-09-24 12:27:33 +020012016uid <uid>
12017 Sets the owner of the UNIX sockets to the designated system uid. It can also
12018 be set by default in the global section's "unix-bind" statement. Note that
12019 some platforms simply ignore this. This setting is equivalent to the "user"
12020 setting except that the user numeric ID is used instead of its name. This
12021 setting is ignored by non UNIX sockets.
12022
12023user <user>
12024 Sets the owner of the UNIX sockets to the designated system user. It can also
12025 be set by default in the global section's "unix-bind" statement. Note that
12026 some platforms simply ignore this. This setting is equivalent to the "uid"
12027 setting except that the user name is used instead of its uid. This setting is
12028 ignored by non UNIX sockets.
12029
Emeric Brun1a073b42012-09-28 17:07:34 +020012030verify [none|optional|required]
12031 This setting is only available when support for OpenSSL was built in. If set
12032 to 'none', client certificate is not requested. This is the default. In other
12033 cases, a client certificate is requested. If the client does not provide a
12034 certificate after the request and if 'verify' is set to 'required', then the
12035 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020012036 certificate provided by the client is always verified using CAs from
12037 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
12038 is aborted, regardless of the 'verify' option, unless the error code exactly
12039 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020012040
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200120415.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010012042------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012043
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010012044The "server" and "default-server" keywords support a certain number of settings
12045which are all passed as arguments on the server line. The order in which those
12046arguments appear does not count, and they are all optional. Some of those
12047settings are single words (booleans) while others expect one or several values
12048after them. In this case, the values must immediately follow the setting name.
12049Except default-server, all those settings must be specified after the server's
12050address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020012051
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012052 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010012053 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020012054
Frédéric Lécailled2376272017-03-21 18:52:12 +010012055Note that all these settings are supported both by "server" and "default-server"
12056keywords, except "id" which is only supported by "server".
12057
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012058The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010012059
Willy Tarreauceb4ac92012-04-28 00:41:46 +020012060addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012061 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010012062 to send health-checks or to probe the agent-check. On some servers, it may be
12063 desirable to dedicate an IP address to specific component able to perform
12064 complex tests which are more suitable to health-checks than the application.
12065 This parameter is ignored if the "check" parameter is not set. See also the
12066 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012067
Simon Hormand60d6912013-11-25 10:46:36 +090012068agent-check
12069 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010012070 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010012071 to the port set by the "agent-port" parameter and reading an ASCII string
12072 terminated by the first '\r' or '\n' met. The string is made of a series of
12073 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090012074
Willy Tarreau81f5d942013-12-09 20:51:51 +010012075 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090012076 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020012077 weight of a server as configured when haproxy starts. Note that a zero
12078 weight is reported on the stats page as "DRAIN" since it has the same
12079 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090012080
Davor Ocelice9ed2812017-12-25 17:49:28 +010012081 - The string "maxconn:" followed by an integer (no space between). Values
12082 in this format will set the maxconn of a server. The maximum number of
12083 connections advertised needs to be multiplied by the number of load
12084 balancers and different backends that use this health check to get the
12085 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020012086
Willy Tarreau81f5d942013-12-09 20:51:51 +010012087 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012088 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090012089
Willy Tarreau81f5d942013-12-09 20:51:51 +010012090 - The word "drain". This will turn the server's administrative state to the
12091 DRAIN mode, thus it will not accept any new connections other than those
12092 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090012093
Willy Tarreau81f5d942013-12-09 20:51:51 +010012094 - The word "maint". This will turn the server's administrative state to the
12095 MAINT mode, thus it will not accept any new connections at all, and health
12096 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090012097
Willy Tarreau81f5d942013-12-09 20:51:51 +010012098 - The words "down", "failed", or "stopped", optionally followed by a
12099 description string after a sharp ('#'). All of these mark the server's
12100 operating state as DOWN, but since the word itself is reported on the stats
12101 page, the difference allows an administrator to know if the situation was
12102 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010012103 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010012104 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090012105
Willy Tarreau81f5d942013-12-09 20:51:51 +010012106 - The word "up" sets back the server's operating state as UP if health checks
12107 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090012108
Willy Tarreau81f5d942013-12-09 20:51:51 +010012109 Parameters which are not advertised by the agent are not changed. For
12110 example, an agent might be designed to monitor CPU usage and only report a
12111 relative weight and never interact with the operating status. Similarly, an
12112 agent could be designed as an end-user interface with 3 radio buttons
12113 allowing an administrator to change only the administrative state. However,
12114 it is important to consider that only the agent may revert its own actions,
12115 so if a server is set to DRAIN mode or to DOWN state using the agent, the
12116 agent must implement the other equivalent actions to bring the service into
12117 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090012118
Simon Horman2f1f9552013-11-25 10:46:37 +090012119 Failure to connect to the agent is not considered an error as connectivity
12120 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010012121 parameter. Warning though, it is not a good idea to stop an agent after it
12122 reports "down", since only an agent reporting "up" will be able to turn the
12123 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010012124 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090012125
Willy Tarreau81f5d942013-12-09 20:51:51 +010012126 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010012127 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090012128
James Brown55f9ff12015-10-21 18:19:05 -070012129agent-send <string>
12130 If this option is specified, haproxy will send the given string (verbatim)
12131 to the agent server upon connection. You could, for example, encode
12132 the backend name into this string, which would enable your agent to send
12133 different responses based on the backend. Make sure to include a '\n' if
12134 you want to terminate your request with a newline.
12135
Simon Hormand60d6912013-11-25 10:46:36 +090012136agent-inter <delay>
12137 The "agent-inter" parameter sets the interval between two agent checks
12138 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
12139
12140 Just as with every other time-based parameter, it may be entered in any
12141 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
12142 parameter also serves as a timeout for agent checks "timeout check" is
12143 not set. In order to reduce "resonance" effects when multiple servers are
12144 hosted on the same hardware, the agent and health checks of all servers
12145 are started with a small time offset between them. It is also possible to
12146 add some random noise in the agent and health checks interval using the
12147 global "spread-checks" keyword. This makes sense for instance when a lot
12148 of backends use the same servers.
12149
12150 See also the "agent-check" and "agent-port" parameters.
12151
Misiek768d8602017-01-09 09:52:43 +010012152agent-addr <addr>
12153 The "agent-addr" parameter sets address for agent check.
12154
12155 You can offload agent-check to another target, so you can make single place
12156 managing status and weights of servers defined in haproxy in case you can't
12157 make self-aware and self-managing services. You can specify both IP or
12158 hostname, it will be resolved.
12159
Simon Hormand60d6912013-11-25 10:46:36 +090012160agent-port <port>
12161 The "agent-port" parameter sets the TCP port used for agent checks.
12162
12163 See also the "agent-check" and "agent-inter" parameters.
12164
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020012165allow-0rtt
12166 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020012167 Note that early data will be sent only if the client used early data, or
12168 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020012169
Olivier Houchardc7566002018-11-20 23:33:50 +010012170alpn <protocols>
12171 This enables the TLS ALPN extension and advertises the specified protocol
12172 list as supported on top of ALPN. The protocol list consists in a comma-
12173 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roesler27c754c2019-07-10 15:45:51 -050012174 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010012175 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
12176 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
12177 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
12178 now obsolete NPN extension.
12179 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
12180 be advertised, in order of preference, like below :
12181
12182 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
12183
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012184backup
12185 When "backup" is present on a server line, the server is only used in load
12186 balancing when all other non-backup servers are unavailable. Requests coming
12187 with a persistence cookie referencing the server will always be served
12188 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010012189 the "allbackups" option is set in the backend. See also the "no-backup" and
12190 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012191
Emeric Brunef42d922012-10-11 16:11:36 +020012192ca-file <cafile>
12193 This setting is only available when support for OpenSSL was built in. It
12194 designates a PEM file from which to load CA certificates used to verify
12195 server's certificate.
12196
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012197check
12198 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010012199 always considered available. If "check" is set, the server is available when
12200 accepting periodic TCP connections, to ensure that it is really able to serve
12201 requests. The default address and port to send the tests to are those of the
12202 server, and the default source is the same as the one defined in the
12203 backend. It is possible to change the address using the "addr" parameter, the
12204 port using the "port" parameter, the source address using the "source"
12205 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090012206 parameters. The request method is define in the backend using the "httpchk",
12207 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010012208 refer to those options and parameters for more information. See also
12209 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012210
Willy Tarreau6c16adc2012-10-05 00:04:16 +020012211check-send-proxy
12212 This option forces emission of a PROXY protocol line with outgoing health
12213 checks, regardless of whether the server uses send-proxy or not for the
12214 normal traffic. By default, the PROXY protocol is enabled for health checks
12215 if it is already enabled for normal traffic and if no "port" nor "addr"
12216 directive is present. However, if such a directive is present, the
12217 "check-send-proxy" option needs to be used to force the use of the
12218 protocol. See also the "send-proxy" option for more information.
12219
Olivier Houchard92150142018-12-21 19:47:01 +010012220check-alpn <protocols>
12221 Defines which protocols to advertise with ALPN. The protocol list consists in
12222 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
12223 (without quotes). If it is not set, the server ALPN is used.
12224
Jérôme Magninae9bb762018-12-09 16:08:26 +010012225check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020012226 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010012227 over SSL. It is only possible to use a string to set <sni>. If you want to
12228 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020012229
Willy Tarreau763a95b2012-10-04 23:15:39 +020012230check-ssl
12231 This option forces encryption of all health checks over SSL, regardless of
12232 whether the server uses SSL or not for the normal traffic. This is generally
12233 used when an explicit "port" or "addr" directive is specified and SSL health
12234 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012235 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020012236 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
12237 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012238 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010012239 See the "ssl" option for more information and "no-check-ssl" to disable
12240 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012241
Alexander Liu2a54bb72019-05-22 19:44:48 +080012242check-via-socks4
John Roesler27c754c2019-07-10 15:45:51 -050012243 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080012244 default, the health checks won't go through socks tunnel even it was enabled
12245 for normal traffic.
12246
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012247ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012248 This setting is only available when support for OpenSSL was built in. This
12249 option sets the string describing the list of cipher algorithms that is
12250 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012251 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
12252 information and recommendations see e.g.
12253 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
12254 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
12255 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012256
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012257ciphersuites <ciphersuites>
12258 This setting is only available when support for OpenSSL was built in and
12259 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
12260 describing the list of cipher algorithms that is negotiated during the TLS
12261 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012262 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
12263 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
12264 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012265
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012266cookie <value>
12267 The "cookie" parameter sets the cookie value assigned to the server to
12268 <value>. This value will be checked in incoming requests, and the first
12269 operational server possessing the same value will be selected. In return, in
12270 cookie insertion or rewrite modes, this value will be assigned to the cookie
12271 sent to the client. There is nothing wrong in having several servers sharing
12272 the same cookie value, and it is in fact somewhat common between normal and
12273 backup servers. See also the "cookie" keyword in backend section.
12274
Emeric Brunef42d922012-10-11 16:11:36 +020012275crl-file <crlfile>
12276 This setting is only available when support for OpenSSL was built in. It
12277 designates a PEM file from which to load certificate revocation list used
12278 to verify server's certificate.
12279
Emeric Bruna7aa3092012-10-26 12:58:00 +020012280crt <cert>
12281 This setting is only available when support for OpenSSL was built in.
12282 It designates a PEM file from which to load both a certificate and the
12283 associated private key. This file can be built by concatenating both PEM
12284 files into one. This certificate will be sent if the server send a client
12285 certificate request.
12286
Willy Tarreau96839092010-03-29 10:02:24 +020012287disabled
12288 The "disabled" keyword starts the server in the "disabled" state. That means
12289 that it is marked down in maintenance mode, and no connection other than the
12290 ones allowed by persist mode will reach it. It is very well suited to setup
12291 new servers, because normal traffic will never reach them, while it is still
12292 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012293 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020012294
Frédéric Lécailled2376272017-03-21 18:52:12 +010012295enabled
12296 This option may be used as 'server' setting to reset any 'disabled'
12297 setting which would have been inherited from 'default-server' directive as
12298 default value.
12299 It may also be used as 'default-server' setting to reset any previous
12300 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020012301
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012302error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010012303 If health observing is enabled, the "error-limit" parameter specifies the
12304 number of consecutive errors that triggers event selected by the "on-error"
12305 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012306
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012307 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012308
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012309fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012310 The "fall" parameter states that a server will be considered as dead after
12311 <count> consecutive unsuccessful health checks. This value defaults to 3 if
12312 unspecified. See also the "check", "inter" and "rise" parameters.
12313
Emeric Brun8694b9a2012-10-05 14:39:07 +020012314force-sslv3
12315 This option enforces use of SSLv3 only when SSL is used to communicate with
12316 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012317 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012318 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012319
12320force-tlsv10
12321 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012322 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012323 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012324
12325force-tlsv11
12326 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012327 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012328 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012329
12330force-tlsv12
12331 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012332 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012333 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012334
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012335force-tlsv13
12336 This option enforces use of TLSv1.3 only when SSL is used to communicate with
12337 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012338 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012339
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012340id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020012341 Set a persistent ID for the server. This ID must be positive and unique for
12342 the proxy. An unused ID will automatically be assigned if unset. The first
12343 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012344
Willy Tarreau6a031d12016-11-07 19:42:35 +010012345init-addr {last | libc | none | <ip>},[...]*
12346 Indicate in what order the server's address should be resolved upon startup
12347 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012348 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010012349 method which succeeds is used. If the end of the list is reached without
12350 finding a working method, an error is thrown. Method "last" suggests to pick
12351 the address which appears in the state file (see "server-state-file"). Method
12352 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
12353 depending on the operating system and build options). Method "none"
12354 specifically indicates that the server should start without any valid IP
12355 address in a down state. It can be useful to ignore some DNS issues upon
12356 startup, waiting for the situation to get fixed later. Finally, an IP address
12357 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012358 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010012359 server used to catch old sessions and present them with a decent error
12360 message for example. When the "first" load balancing algorithm is used, this
12361 IP address could point to a fake server used to trigger the creation of new
12362 instances on the fly. This option defaults to "last,libc" indicating that the
12363 previous address found in the state file (if any) is used first, otherwise
12364 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012365 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010012366
12367 Example:
12368 defaults
12369 # never fail on address resolution
12370 default-server init-addr last,libc,none
12371
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012372inter <delay>
12373fastinter <delay>
12374downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012375 The "inter" parameter sets the interval between two consecutive health checks
12376 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
12377 It is also possible to use "fastinter" and "downinter" to optimize delays
12378 between checks depending on the server state :
12379
Pieter Baauw44fc9df2015-09-17 21:30:46 +020012380 Server state | Interval used
12381 ----------------------------------------+----------------------------------
12382 UP 100% (non-transitional) | "inter"
12383 ----------------------------------------+----------------------------------
12384 Transitionally UP (going down "fall"), | "fastinter" if set,
12385 Transitionally DOWN (going up "rise"), | "inter" otherwise.
12386 or yet unchecked. |
12387 ----------------------------------------+----------------------------------
12388 DOWN 100% (non-transitional) | "downinter" if set,
12389 | "inter" otherwise.
12390 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010012391
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012392 Just as with every other time-based parameter, they can be entered in any
12393 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
12394 serves as a timeout for health checks sent to servers if "timeout check" is
12395 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090012396 hosted on the same hardware, the agent and health checks of all servers
12397 are started with a small time offset between them. It is also possible to
12398 add some random noise in the agent and health checks interval using the
12399 global "spread-checks" keyword. This makes sense for instance when a lot
12400 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012401
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012402maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012403 The "maxconn" parameter specifies the maximal number of concurrent
12404 connections that will be sent to this server. If the number of incoming
Tim Duesterhus50cfb312019-11-27 22:35:27 +010012405 concurrent connections goes higher than this value, they will be queued,
12406 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012407 save fragile servers from going down under extreme loads. If a "minconn"
12408 parameter is specified, the limit becomes dynamic. The default value is "0"
12409 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
12410 the backend's "fullconn" keyword.
12411
Tim Duesterhus50cfb312019-11-27 22:35:27 +010012412 In HTTP mode this parameter limits the number of concurrent requests instead
12413 of the number of connections. Multiple requests might be multiplexed over a
12414 single TCP connection to the server. As an example if you specify a maxconn
12415 of 50 you might see between 1 and 50 actual server connections, but no more
12416 than 50 concurrent requests.
12417
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012418maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012419 The "maxqueue" parameter specifies the maximal number of connections which
12420 will wait in the queue for this server. If this limit is reached, next
12421 requests will be redispatched to other servers instead of indefinitely
12422 waiting to be served. This will break persistence but may allow people to
12423 quickly re-log in when the server they try to connect to is dying. The
12424 default value is "0" which means the queue is unlimited. See also the
12425 "maxconn" and "minconn" parameters.
12426
Willy Tarreau9c538e02019-01-23 10:21:49 +010012427max-reuse <count>
12428 The "max-reuse" argument indicates the HTTP connection processors that they
12429 should not reuse a server connection more than this number of times to send
12430 new requests. Permitted values are -1 (the default), which disables this
12431 limit, or any positive value. Value zero will effectively disable keep-alive.
12432 This is only used to work around certain server bugs which cause them to leak
12433 resources over time. The argument is not necessarily respected by the lower
12434 layers as there might be technical limitations making it impossible to
12435 enforce. At least HTTP/2 connections to servers will respect it.
12436
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012437minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012438 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
12439 limit following the backend's load. The server will always accept at least
12440 <minconn> connections, never more than <maxconn>, and the limit will be on
12441 the ramp between both values when the backend has less than <fullconn>
12442 concurrent connections. This makes it possible to limit the load on the
12443 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012444 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012445 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012446
Willy Tarreaud72f0f32015-10-13 14:50:22 +020012447namespace <name>
12448 On Linux, it is possible to specify which network namespace a socket will
12449 belong to. This directive makes it possible to explicitly bind a server to
12450 a namespace different from the default one. Please refer to your operating
12451 system's documentation to find more details about network namespaces.
12452
Frédéric Lécailled2376272017-03-21 18:52:12 +010012453no-agent-check
12454 This option may be used as "server" setting to reset any "agent-check"
12455 setting which would have been inherited from "default-server" directive as
12456 default value.
12457 It may also be used as "default-server" setting to reset any previous
12458 "default-server" "agent-check" setting.
12459
12460no-backup
12461 This option may be used as "server" setting to reset any "backup"
12462 setting which would have been inherited from "default-server" directive as
12463 default value.
12464 It may also be used as "default-server" setting to reset any previous
12465 "default-server" "backup" setting.
12466
12467no-check
12468 This option may be used as "server" setting to reset any "check"
12469 setting which would have been inherited from "default-server" directive as
12470 default value.
12471 It may also be used as "default-server" setting to reset any previous
12472 "default-server" "check" setting.
12473
12474no-check-ssl
12475 This option may be used as "server" setting to reset any "check-ssl"
12476 setting which would have been inherited from "default-server" directive as
12477 default value.
12478 It may also be used as "default-server" setting to reset any previous
12479 "default-server" "check-ssl" setting.
12480
Frédéric Lécailled2376272017-03-21 18:52:12 +010012481no-send-proxy
12482 This option may be used as "server" setting to reset any "send-proxy"
12483 setting which would have been inherited from "default-server" directive as
12484 default value.
12485 It may also be used as "default-server" setting to reset any previous
12486 "default-server" "send-proxy" setting.
12487
12488no-send-proxy-v2
12489 This option may be used as "server" setting to reset any "send-proxy-v2"
12490 setting which would have been inherited from "default-server" directive as
12491 default value.
12492 It may also be used as "default-server" setting to reset any previous
12493 "default-server" "send-proxy-v2" setting.
12494
12495no-send-proxy-v2-ssl
12496 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
12497 setting which would have been inherited from "default-server" directive as
12498 default value.
12499 It may also be used as "default-server" setting to reset any previous
12500 "default-server" "send-proxy-v2-ssl" setting.
12501
12502no-send-proxy-v2-ssl-cn
12503 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
12504 setting which would have been inherited from "default-server" directive as
12505 default value.
12506 It may also be used as "default-server" setting to reset any previous
12507 "default-server" "send-proxy-v2-ssl-cn" setting.
12508
12509no-ssl
12510 This option may be used as "server" setting to reset any "ssl"
12511 setting which would have been inherited from "default-server" directive as
12512 default value.
12513 It may also be used as "default-server" setting to reset any previous
12514 "default-server" "ssl" setting.
12515
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010012516no-ssl-reuse
12517 This option disables SSL session reuse when SSL is used to communicate with
12518 the server. It will force the server to perform a full handshake for every
12519 new connection. It's probably only useful for benchmarking, troubleshooting,
12520 and for paranoid users.
12521
Emeric Brun9b3009b2012-10-05 11:55:06 +020012522no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012523 This option disables support for SSLv3 when SSL is used to communicate with
12524 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012525 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012526
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012527 Supported in default-server: No
12528
Emeric Brunf9c5c472012-10-11 15:28:34 +020012529no-tls-tickets
12530 This setting is only available when support for OpenSSL was built in. It
12531 disables the stateless session resumption (RFC 5077 TLS Ticket
12532 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012533 session resumption is more expensive in CPU usage for servers. This option
12534 is also available on global statement "ssl-default-server-options".
Lukas Tribusd8fd6362020-03-10 00:56:09 +010012535 The TLS ticket mechanism is only used up to TLS 1.2.
12536 Forward Secrecy is compromised with TLS tickets, unless ticket keys
12537 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010012538 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020012539
Emeric Brun9b3009b2012-10-05 11:55:06 +020012540no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020012541 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020012542 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12543 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012544 often makes sense to disable it when communicating with local servers. This
12545 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012546 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012547
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012548 Supported in default-server: No
12549
Emeric Brun9b3009b2012-10-05 11:55:06 +020012550no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020012551 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020012552 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12553 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012554 often makes sense to disable it when communicating with local servers. This
12555 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012556 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012557
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012558 Supported in default-server: No
12559
Emeric Brun9b3009b2012-10-05 11:55:06 +020012560no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020012561 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012562 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12563 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012564 often makes sense to disable it when communicating with local servers. This
12565 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012566 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012567
12568 Supported in default-server: No
12569
12570no-tlsv13
12571 This option disables support for TLSv1.3 when SSL is used to communicate with
12572 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12573 using any configuration option. TLSv1 is more expensive than SSLv3 so it
12574 often makes sense to disable it when communicating with local servers. This
12575 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012576 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012577
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012578 Supported in default-server: No
12579
Frédéric Lécailled2376272017-03-21 18:52:12 +010012580no-verifyhost
12581 This option may be used as "server" setting to reset any "verifyhost"
12582 setting which would have been inherited from "default-server" directive as
12583 default value.
12584 It may also be used as "default-server" setting to reset any previous
12585 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012586
Frédéric Lécailleaeeb1c92019-07-04 14:19:06 +020012587no-tfo
12588 This option may be used as "server" setting to reset any "tfo"
12589 setting which would have been inherited from "default-server" directive as
12590 default value.
12591 It may also be used as "default-server" setting to reset any previous
12592 "default-server" "tfo" setting.
12593
Simon Hormanfa461682011-06-25 09:39:49 +090012594non-stick
12595 Never add connections allocated to this sever to a stick-table.
12596 This may be used in conjunction with backup to ensure that
12597 stick-table persistence is disabled for backup servers.
12598
Olivier Houchardc7566002018-11-20 23:33:50 +010012599npn <protocols>
12600 This enables the NPN TLS extension and advertises the specified protocol list
12601 as supported on top of NPN. The protocol list consists in a comma-delimited
12602 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roesler27c754c2019-07-10 15:45:51 -050012603 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010012604 enabled (check with haproxy -vv). Note that the NPN extension has been
12605 replaced with the ALPN extension (see the "alpn" keyword), though this one is
12606 only available starting with OpenSSL 1.0.2.
12607
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012608observe <mode>
12609 This option enables health adjusting based on observing communication with
12610 the server. By default this functionality is disabled and enabling it also
12611 requires to enable health checks. There are two supported modes: "layer4" and
12612 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
12613 significant. In layer7, which is only allowed for http proxies, responses
12614 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010012615 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012616
12617 See also the "check", "on-error" and "error-limit".
12618
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012619on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012620 Select what should happen when enough consecutive errors are detected.
12621 Currently, four modes are available:
12622 - fastinter: force fastinter
12623 - fail-check: simulate a failed check, also forces fastinter (default)
12624 - sudden-death: simulate a pre-fatal failed health check, one more failed
12625 check will mark a server down, forces fastinter
12626 - mark-down: mark the server immediately down and force fastinter
12627
12628 See also the "check", "observe" and "error-limit".
12629
Simon Hormane0d1bfb2011-06-21 14:34:58 +090012630on-marked-down <action>
12631 Modify what occurs when a server is marked down.
12632 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012633 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
12634 all connections to the server are immediately terminated when the server
12635 goes down. It might be used if the health check detects more complex cases
12636 than a simple connection status, and long timeouts would cause the service
12637 to remain unresponsive for too long a time. For instance, a health check
12638 might detect that a database is stuck and that there's no chance to reuse
12639 existing connections anymore. Connections killed this way are logged with
12640 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090012641
12642 Actions are disabled by default
12643
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012644on-marked-up <action>
12645 Modify what occurs when a server is marked up.
12646 Currently one action is available:
12647 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
12648 done only if the server is not in backup state and if it is not disabled
12649 (it must have an effective weight > 0). This can be used sometimes to force
12650 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012651 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
12652 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012653 with extreme care. Sessions killed because a server comes up are logged
12654 with an 'U' termination code (for "Up").
12655
12656 Actions are disabled by default
12657
Olivier Houchard006e3102018-12-10 18:30:32 +010012658pool-max-conn <max>
12659 Set the maximum number of idling connections for a server. -1 means unlimited
12660 connections, 0 means no idle connections. The default is -1. When idle
12661 connections are enabled, orphaned idle connections which do not belong to any
12662 client session anymore are moved to a dedicated pool so that they remain
12663 usable by future clients. This only applies to connections that can be shared
12664 according to the same principles as those applying to "http-reuse".
12665
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010012666pool-purge-delay <delay>
12667 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010012668 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020012669 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010012670
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012671port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012672 Using the "port" parameter, it becomes possible to use a different port to
12673 send health-checks. On some servers, it may be desirable to dedicate a port
12674 to a specific component able to perform complex tests which are more suitable
12675 to health-checks than the application. It is common to run a simple script in
12676 inetd for instance. This parameter is ignored if the "check" parameter is not
12677 set. See also the "addr" parameter.
12678
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020012679proto <name>
12680
12681 Forces the multiplexer's protocol to use for the outgoing connections to this
12682 server. It must be compatible with the mode of the backend (TCP or HTTP). It
12683 must also be usable on the backend side. The list of available protocols is
12684 reported in haproxy -vv.
12685 Idea behind this optipon is to bypass the selection of the best multiplexer's
12686 protocol for all connections established to this server.
12687
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012688redir <prefix>
12689 The "redir" parameter enables the redirection mode for all GET and HEAD
12690 requests addressing this server. This means that instead of having HAProxy
12691 forward the request to the server, it will send an "HTTP 302" response with
12692 the "Location" header composed of this prefix immediately followed by the
12693 requested URI beginning at the leading '/' of the path component. That means
12694 that no trailing slash should be used after <prefix>. All invalid requests
12695 will be rejected, and all non-GET or HEAD requests will be normally served by
12696 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012697 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012698 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012699 users to a remote location in case of local disaster. Main use consists in
12700 increasing bandwidth for static servers by having the clients directly
12701 connect to them. Note: never use a relative location here, it would cause a
12702 loop between the client and HAProxy!
12703
12704 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
12705
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012706rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012707 The "rise" parameter states that a server will be considered as operational
12708 after <count> consecutive successful health checks. This value defaults to 2
12709 if unspecified. See also the "check", "inter" and "fall" parameters.
12710
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020012711resolve-opts <option>,<option>,...
12712 Comma separated list of options to apply to DNS resolution linked to this
12713 server.
12714
12715 Available options:
12716
12717 * allow-dup-ip
12718 By default, HAProxy prevents IP address duplication in a backend when DNS
12719 resolution at runtime is in operation.
12720 That said, for some cases, it makes sense that two servers (in the same
12721 backend, being resolved by the same FQDN) have the same IP address.
12722 For such case, simply enable this option.
12723 This is the opposite of prevent-dup-ip.
12724
12725 * prevent-dup-ip
12726 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
12727 an IP address already set to a server in the same backend and sharing the
12728 same fqdn.
12729 This is the opposite of allow-dup-ip.
12730
12731 Example:
12732 backend b_myapp
12733 default-server init-addr none resolvers dns
12734 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
12735 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
12736
12737 With the option allow-dup-ip set:
12738 * if the nameserver returns a single IP address, then both servers will use
12739 it
12740 * If the nameserver returns 2 IP addresses, then each server will pick up a
12741 different address
12742
12743 Default value: not set
12744
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012745resolve-prefer <family>
12746 When DNS resolution is enabled for a server and multiple IP addresses from
12747 different families are returned, HAProxy will prefer using an IP address
12748 from the family mentioned in the "resolve-prefer" parameter.
12749 Available families: "ipv4" and "ipv6"
12750
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020012751 Default value: ipv6
12752
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012753 Example:
12754
12755 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012756
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012757resolve-net <network>[,<network[,...]]
John Roesler27c754c2019-07-10 15:45:51 -050012758 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012759 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010012760 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010012761 different datacenters. The latency between datacenter is not negligible, so
12762 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012763 configured network, another address is selected.
12764
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012765 Example:
12766
12767 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012768
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012769resolvers <id>
12770 Points to an existing "resolvers" section to resolve current server's
12771 hostname.
12772
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012773 Example:
12774
12775 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012776
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012777 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012778
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012779send-proxy
12780 The "send-proxy" parameter enforces use of the PROXY protocol over any
12781 connection established to this server. The PROXY protocol informs the other
12782 end about the layer 3/4 addresses of the incoming connection, so that it can
12783 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012784 upper layer protocol. For connections accepted by an "accept-proxy" or
12785 "accept-netscaler-cip" listener, the advertised address will be used. Only
12786 TCPv4 and TCPv6 address families are supported. Other families such as
12787 Unix sockets, will report an UNKNOWN family. Servers using this option can
12788 fully be chained to another instance of haproxy listening with an
12789 "accept-proxy" setting. This setting must not be used if the server isn't
12790 aware of the protocol. When health checks are sent to the server, the PROXY
12791 protocol is automatically used when this option is set, unless there is an
12792 explicit "port" or "addr" directive, in which case an explicit
12793 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012794 See also the "no-send-proxy" option of this section and "accept-proxy" and
12795 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012796
David Safb76832014-05-08 23:42:08 -040012797send-proxy-v2
12798 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
12799 over any connection established to this server. The PROXY protocol informs
12800 the other end about the layer 3/4 addresses of the incoming connection, so
12801 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020012802 whatever the upper layer protocol. It also send ALPN information if an alpn
12803 have been negotiated. This setting must not be used if the server isn't aware
12804 of this version of the protocol. See also the "no-send-proxy-v2" option of
12805 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012806
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012807proxy-v2-options <option>[,<option>]*
12808 The "proxy-v2-options" parameter add option to send in PROXY protocol version
12809 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010012810 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
12811 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010012812 certificate, "cert-key": key algorithm of the used certificate), "authority":
12813 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010012814 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012815
David Safb76832014-05-08 23:42:08 -040012816send-proxy-v2-ssl
12817 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12818 2 over any connection established to this server. The PROXY protocol informs
12819 the other end about the layer 3/4 addresses of the incoming connection, so
12820 that it can know the client's address or the public address it accessed to,
12821 whatever the upper layer protocol. In addition, the SSL information extension
12822 of the PROXY protocol is added to the PROXY protocol header. This setting
12823 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012824 See also the "no-send-proxy-v2-ssl" option of this section and the
12825 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012826
12827send-proxy-v2-ssl-cn
12828 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12829 2 over any connection established to this server. The PROXY protocol informs
12830 the other end about the layer 3/4 addresses of the incoming connection, so
12831 that it can know the client's address or the public address it accessed to,
12832 whatever the upper layer protocol. In addition, the SSL information extension
12833 of the PROXY protocol, along along with the Common Name from the subject of
12834 the client certificate (if any), is added to the PROXY protocol header. This
12835 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012836 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
12837 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012838
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012839slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012840 The "slowstart" parameter for a server accepts a value in milliseconds which
12841 indicates after how long a server which has just come back up will run at
12842 full speed. Just as with every other time-based parameter, it can be entered
12843 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
12844 linearly from 0 to 100% during this time. The limitation applies to two
12845 parameters :
12846
12847 - maxconn: the number of connections accepted by the server will grow from 1
12848 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
12849
12850 - weight: when the backend uses a dynamic weighted algorithm, the weight
12851 grows linearly from 1 to 100%. In this case, the weight is updated at every
12852 health-check. For this reason, it is important that the "inter" parameter
12853 is smaller than the "slowstart", in order to maximize the number of steps.
12854
12855 The slowstart never applies when haproxy starts, otherwise it would cause
12856 trouble to running servers. It only applies when a server has been previously
12857 seen as failed.
12858
Willy Tarreau732eac42015-07-09 11:40:25 +020012859sni <expression>
12860 The "sni" parameter evaluates the sample fetch expression, converts it to a
12861 string and uses the result as the host name sent in the SNI TLS extension to
12862 the server. A typical use case is to send the SNI received from the client in
12863 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012864 expression, though alternatives such as req.hdr(host) can also make sense. If
12865 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012866 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010012867 "verify" directive for more details. If you want to set a SNI for health
12868 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012869
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012870source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012871source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012872source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012873 The "source" parameter sets the source address which will be used when
12874 connecting to the server. It follows the exact same parameters and principle
12875 as the backend "source" keyword, except that it only applies to the server
12876 referencing it. Please consult the "source" keyword for details.
12877
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012878 Additionally, the "source" statement on a server line allows one to specify a
12879 source port range by indicating the lower and higher bounds delimited by a
12880 dash ('-'). Some operating systems might require a valid IP address when a
12881 source port range is specified. It is permitted to have the same IP/range for
12882 several servers. Doing so makes it possible to bypass the maximum of 64k
12883 total concurrent connections. The limit will then reach 64k connections per
12884 server.
12885
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012886 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12887 specifying the source address without port(s).
12888
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012889ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012890 This option enables SSL ciphering on outgoing connections to the server. It
12891 is critical to verify server certificates using "verify" when using SSL to
12892 connect to servers, otherwise the communication is prone to trivial man in
12893 the-middle attacks rendering SSL useless. When this option is used, health
12894 checks are automatically sent in SSL too unless there is a "port" or an
12895 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012896 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12897 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012898
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012899ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12900 This option enforces use of <version> or lower when SSL is used to communicate
12901 with the server. This option is also available on global statement
12902 "ssl-default-server-options". See also "ssl-min-ver".
12903
12904ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12905 This option enforces use of <version> or upper when SSL is used to communicate
12906 with the server. This option is also available on global statement
12907 "ssl-default-server-options". See also "ssl-max-ver".
12908
Frédéric Lécailled2376272017-03-21 18:52:12 +010012909ssl-reuse
12910 This option may be used as "server" setting to reset any "no-ssl-reuse"
12911 setting which would have been inherited from "default-server" directive as
12912 default value.
12913 It may also be used as "default-server" setting to reset any previous
12914 "default-server" "no-ssl-reuse" setting.
12915
12916stick
12917 This option may be used as "server" setting to reset any "non-stick"
12918 setting which would have been inherited from "default-server" directive as
12919 default value.
12920 It may also be used as "default-server" setting to reset any previous
12921 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012922
Alexander Liu2a54bb72019-05-22 19:44:48 +080012923socks4 <addr>:<port>
John Roesler27c754c2019-07-10 15:45:51 -050012924 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080012925 server. Using this option won't force the health check to go via socks4 by
12926 default. You will have to use the keyword "check-via-socks4" to enable it.
12927
Willy Tarreau163d4622015-10-13 16:16:41 +020012928tcp-ut <delay>
12929 Sets the TCP User Timeout for all outgoing connections to this server. This
12930 option is available on Linux since version 2.6.37. It allows haproxy to
12931 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012932 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012933 long-lived connections experiencing long idle periods such as remote
12934 terminals or database connection pools, where the client and server timeouts
12935 must remain high to allow a long period of idle, but where it is important to
12936 detect that the server has disappeared in order to release all resources
12937 associated with its connection (and the client's session). One typical use
12938 case is also to force dead server connections to die when health checks are
12939 too slow or during a soft reload since health checks are then disabled. The
12940 argument is a delay expressed in milliseconds by default. This only works for
12941 regular TCP connections, and is ignored for other protocols.
12942
Willy Tarreau034c88c2017-01-23 23:36:45 +010012943tfo
12944 This option enables using TCP fast open when connecting to servers, on
12945 systems that support it (currently only the Linux kernel >= 4.11).
12946 See the "tfo" bind option for more information about TCP fast open.
12947 Please note that when using tfo, you should also use the "conn-failure",
12948 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécailleaeeb1c92019-07-04 14:19:06 +020012949 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010012950
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012951track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012952 This option enables ability to set the current state of the server by tracking
12953 another one. It is possible to track a server which itself tracks another
12954 server, provided that at the end of the chain, a server has health checks
12955 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012956 used, it has to be enabled on both proxies.
12957
Frédéric Lécailled2376272017-03-21 18:52:12 +010012958tls-tickets
12959 This option may be used as "server" setting to reset any "no-tls-tickets"
12960 setting which would have been inherited from "default-server" directive as
12961 default value.
Lukas Tribusd8fd6362020-03-10 00:56:09 +010012962 The TLS ticket mechanism is only used up to TLS 1.2.
12963 Forward Secrecy is compromised with TLS tickets, unless ticket keys
12964 are periodically rotated (via reload or by using "tls-ticket-keys").
Frédéric Lécailled2376272017-03-21 18:52:12 +010012965 It may also be used as "default-server" setting to reset any previous
Bjoern Jacke04037d32020-02-13 14:16:16 +010012966 "default-server" "no-tls-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012967
Emeric Brunef42d922012-10-11 16:11:36 +020012968verify [none|required]
12969 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012970 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012971 certificate provided by the server is verified using CAs from 'ca-file' and
12972 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012973 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012974 the name passed using the "sni" directive, or if not provided, the static
12975 host name passed using the "verifyhost" directive. When no name is found, the
12976 certificate's names are ignored. For this reason, without SNI it's important
12977 to use "verifyhost". On verification failure the handshake is aborted. It is
12978 critically important to verify server certificates when using SSL to connect
12979 to servers, otherwise the communication is prone to trivial man-in-the-middle
12980 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12981 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012982
Evan Broderbe554312013-06-27 00:05:25 -070012983verifyhost <hostname>
12984 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012985 only takes effect if 'verify required' is also specified. This directive sets
12986 a default static hostname to check the server's certificate against when no
12987 SNI was used to connect to the server. If SNI is not used, this is the only
12988 way to enable hostname verification. This static hostname, when set, will
12989 also be used for health checks (which cannot provide an SNI value). If none
12990 of the hostnames in the certificate match the specified hostname, the
12991 handshake is aborted. The hostnames in the server-provided certificate may
12992 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012993
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012994weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012995 The "weight" parameter is used to adjust the server's weight relative to
12996 other servers. All servers will receive a load proportional to their weight
12997 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012998 load. The default weight is 1, and the maximal value is 256. A value of 0
12999 means the server will not participate in load-balancing but will still accept
13000 persistent connections. If this parameter is used to distribute the load
13001 according to server's capacity, it is recommended to start with values which
13002 can both grow and shrink, for instance between 10 and 100 to leave enough
13003 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013004
13005
Cyril Bonté46175dd2015-07-02 22:45:32 +0200130065.3. Server IP address resolution using DNS
13007-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013008
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013009HAProxy allows using a host name on the server line to retrieve its IP address
13010using name servers. By default, HAProxy resolves the name when parsing the
13011configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013012This is not sufficient in some cases, such as in Amazon where a server's IP
13013can change after a reboot or an ELB Virtual IP can change based on current
13014workload.
13015This chapter describes how HAProxy can be configured to process server's name
13016resolution at run time.
13017Whether run time server name resolution has been enable or not, HAProxy will
13018carry on doing the first resolution when parsing the configuration.
13019
13020
Cyril Bonté46175dd2015-07-02 22:45:32 +0200130215.3.1. Global overview
13022----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013023
13024As we've seen in introduction, name resolution in HAProxy occurs at two
13025different steps of the process life:
13026
13027 1. when starting up, HAProxy parses the server line definition and matches a
13028 host name. It uses libc functions to get the host name resolved. This
13029 resolution relies on /etc/resolv.conf file.
13030
Christopher Faulet67957bd2017-09-27 11:00:59 +020013031 2. at run time, HAProxy performs periodically name resolutions for servers
13032 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013033
13034A few other events can trigger a name resolution at run time:
13035 - when a server's health check ends up in a connection timeout: this may be
13036 because the server has a new IP address. So we need to trigger a name
13037 resolution to know this new IP.
13038
Christopher Faulet67957bd2017-09-27 11:00:59 +020013039When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010013040HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020013041SRV label is specified, then the corresponding SRV records will be retrieved
13042from the DNS server, and the provided hostnames will be used. The SRV label
13043will be checked periodically, and if any server are added or removed, haproxy
13044will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020013045
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013046A few things important to notice:
John Roesler27c754c2019-07-10 15:45:51 -050013047 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013048 first valid response.
13049
13050 - a resolution is considered as invalid (NX, timeout, refused), when all the
13051 servers return an error.
13052
13053
Cyril Bonté46175dd2015-07-02 22:45:32 +0200130545.3.2. The resolvers section
13055----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013056
13057This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020013058HAProxy. There can be as many as resolvers section as needed. Each section can
13059contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013060
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013061When multiple name servers are configured in a resolvers section, then HAProxy
13062uses the first valid response. In case of invalid responses, only the last one
13063is treated. Purpose is to give the chance to a slow server to deliver a valid
13064answer after a fast faulty or outdated server.
13065
13066When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020013067used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013068
Christopher Faulet67957bd2017-09-27 11:00:59 +020013069 1. HAProxy retries the same DNS query with a new query type. The A queries are
13070 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
13071 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013072
Christopher Faulet67957bd2017-09-27 11:00:59 +020013073 2. When the fallback on the query type was done (or not applicable), HAProxy
13074 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013075
Christopher Faulet67957bd2017-09-27 11:00:59 +020013076 3. HAProxy retries previous steps <resolve_retires> times. If no valid
13077 response is received after that, it stops the DNS resolution and reports
13078 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013079
Christopher Faulet67957bd2017-09-27 11:00:59 +020013080For example, with 2 name servers configured in a resolvers section, the
13081following scenarios are possible:
13082
13083 - First response is valid and is applied directly, second response is
13084 ignored
13085
13086 - First response is invalid and second one is valid, then second response is
13087 applied
13088
13089 - First response is a NX domain and second one a truncated response, then
13090 HAProxy retries the query with a new type
13091
13092 - First response is a NX domain and second one is a timeout, then HAProxy
13093 retries the query with a new type
13094
13095 - Query timed out for both name servers, then HAProxy retries it with the
13096 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013097
Olivier Houcharda8c6db82017-07-06 18:46:47 +020013098As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
13099a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020013100<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020013101
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013102
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013103resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010013104 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013105
13106A resolvers section accept the following parameters:
13107
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020013108accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010013109 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020013110 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020013111 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
13112 by RFC 6891)
13113
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020013114 Note: the maximum allowed value is 8192.
13115
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013116nameserver <id> <ip>:<port>
13117 DNS server description:
13118 <id> : label of the server, should be unique
13119 <ip> : IP address of the server
13120 <port> : port where the DNS service actually runs
13121
Ben Draut44e609b2018-05-29 15:40:08 -060013122parse-resolv-conf
13123 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
13124 list. Ordered as if each nameserver in /etc/resolv.conf was individually
13125 placed in the resolvers section in place of this directive.
13126
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013127hold <status> <period>
13128 Defines <period> during which the last name resolution should be kept based
13129 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010013130 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020013131 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013132 <period> : interval between two successive name resolution when the last
13133 answer was in <status>. It follows the HAProxy time format.
13134 <period> is in milliseconds by default.
13135
Baptiste Assmann686408b2017-08-18 10:15:42 +020013136 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013137
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013138resolve_retries <nb>
13139 Defines the number <nb> of queries to send to resolve a server name before
13140 giving up.
13141 Default value: 3
13142
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013143 A retry occurs on name server timeout or when the full sequence of DNS query
13144 type failover is over and we need to start up from the default ANY query
13145 type.
13146
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013147timeout <event> <time>
13148 Defines timeouts related to name resolution
13149 <event> : the event on which the <time> timeout period applies to.
13150 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013151 - resolve : default time to trigger name resolutions when no
13152 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020013153 Default value: 1s
13154 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013155 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020013156 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013157 <time> : time related to the event. It follows the HAProxy time format.
13158 <time> is expressed in milliseconds.
13159
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013160 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013161
13162 resolvers mydns
13163 nameserver dns1 10.0.0.1:53
13164 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060013165 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013166 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020013167 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013168 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010013169 hold other 30s
13170 hold refused 30s
13171 hold nx 30s
13172 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013173 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020013174 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013175
13176
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200131776. HTTP header manipulation
13178---------------------------
13179
13180In HTTP mode, it is possible to rewrite, add or delete some of the request and
13181response headers based on regular expressions. It is also possible to block a
13182request or a response if a particular header matches a regular expression,
13183which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010013184against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013185
Willy Tarreau70dffda2014-01-30 03:07:23 +010013186If HAProxy encounters an "Informational Response" (status code 1xx), it is able
13187to process all rsp* rules which can allow, deny, rewrite or delete a header,
13188but it will refuse to add a header to any such messages as this is not
13189HTTP-compliant. The reason for still processing headers in such responses is to
13190stop and/or fix any possible information leak which may happen, for instance
13191because another downstream equipment would unconditionally add a header, or if
13192a server name appears there. When such messages are seen, normal processing
13193still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020013194
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013195This section covers common usage of the following keywords, described in detail
13196in section 4.2 :
13197
13198 - reqadd <string>
13199 - reqallow <search>
13200 - reqiallow <search>
13201 - reqdel <search>
13202 - reqidel <search>
13203 - reqdeny <search>
13204 - reqideny <search>
13205 - reqpass <search>
13206 - reqipass <search>
13207 - reqrep <search> <replace>
13208 - reqirep <search> <replace>
13209 - reqtarpit <search>
13210 - reqitarpit <search>
13211 - rspadd <string>
13212 - rspdel <search>
13213 - rspidel <search>
13214 - rspdeny <search>
13215 - rspideny <search>
13216 - rsprep <search> <replace>
13217 - rspirep <search> <replace>
13218
13219With all these keywords, the same conventions are used. The <search> parameter
13220is a POSIX extended regular expression (regex) which supports grouping through
13221parenthesis (without the backslash). Spaces and other delimiters must be
13222prefixed with a backslash ('\') to avoid confusion with a field delimiter.
13223Other characters may be prefixed with a backslash to change their meaning :
13224
13225 \t for a tab
13226 \r for a carriage return (CR)
13227 \n for a new line (LF)
13228 \ to mark a space and differentiate it from a delimiter
13229 \# to mark a sharp and differentiate it from a comment
13230 \\ to use a backslash in a regex
13231 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
13232 \xXX to write the ASCII hex code XX as in the C language
13233
13234The <replace> parameter contains the string to be used to replace the largest
13235portion of text matching the regex. It can make use of the special characters
13236above, and can reference a substring which is delimited by parenthesis in the
13237regex, by writing a backslash ('\') immediately followed by one digit from 0 to
132389 indicating the group position (0 designating the entire line). This practice
13239is very common to users of the "sed" program.
13240
13241The <string> parameter represents the string which will systematically be added
13242after the last header line. It can also use special character sequences above.
13243
13244Notes related to these keywords :
13245---------------------------------
13246 - these keywords are not always convenient to allow/deny based on header
13247 contents. It is strongly recommended to use ACLs with the "block" keyword
13248 instead, resulting in far more flexible and manageable rules.
13249
13250 - lines are always considered as a whole. It is not possible to reference
13251 a header name only or a value only. This is important because of the way
13252 headers are written (notably the number of spaces after the colon).
13253
13254 - the first line is always considered as a header, which makes it possible to
13255 rewrite or filter HTTP requests URIs or response codes, but in turn makes
13256 it harder to distinguish between headers and request line. The regex prefix
13257 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
13258 ^[^ \t:]*: matches any header name followed by a colon.
13259
13260 - for performances reasons, the number of characters added to a request or to
13261 a response is limited at build time to values between 1 and 4 kB. This
13262 should normally be far more than enough for most usages. If it is too short
13263 on occasional usages, it is possible to gain some space by removing some
13264 useless headers before adding new ones.
13265
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013266 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013267 without the 'i' letter except that they ignore case when matching patterns.
13268
13269 - when a request passes through a frontend then a backend, all req* rules
13270 from the frontend will be evaluated, then all req* rules from the backend
13271 will be evaluated. The reverse path is applied to responses.
13272
13273 - req* statements are applied after "block" statements, so that "block" is
13274 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010013275 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013276
13277
Willy Tarreau74ca5042013-06-11 23:12:07 +0200132787. Using ACLs and fetching samples
13279----------------------------------
13280
Davor Ocelice9ed2812017-12-25 17:49:28 +010013281HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020013282client or server information, from tables, environmental information etc...
13283The action of extracting such data is called fetching a sample. Once retrieved,
13284these samples may be used for various purposes such as a key to a stick-table,
13285but most common usages consist in matching them against predefined constant
13286data called patterns.
13287
13288
132897.1. ACL basics
13290---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013291
13292The use of Access Control Lists (ACL) provides a flexible solution to perform
13293content switching and generally to take decisions based on content extracted
13294from the request, the response or any environmental status. The principle is
13295simple :
13296
Willy Tarreau74ca5042013-06-11 23:12:07 +020013297 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013298 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020013299 - apply one or multiple pattern matching methods on this sample
13300 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013301
Willy Tarreau74ca5042013-06-11 23:12:07 +020013302The actions generally consist in blocking a request, selecting a backend, or
13303adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013304
13305In order to define a test, the "acl" keyword is used. The syntax is :
13306
Willy Tarreau74ca5042013-06-11 23:12:07 +020013307 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013308
13309This creates a new ACL <aclname> or completes an existing one with new tests.
13310Those tests apply to the portion of request/response specified in <criterion>
13311and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013312an operator which may be specified before the set of values. Optionally some
13313conversion operators may be applied to the sample, and they will be specified
13314as a comma-delimited list of keywords just after the first keyword. The values
13315are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013316
13317ACL names must be formed from upper and lower case letters, digits, '-' (dash),
13318'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
13319which means that "my_acl" and "My_Acl" are two different ACLs.
13320
13321There is no enforced limit to the number of ACLs. The unused ones do not affect
13322performance, they just consume a small amount of memory.
13323
Willy Tarreau74ca5042013-06-11 23:12:07 +020013324The criterion generally is the name of a sample fetch method, or one of its ACL
13325specific declinations. The default test method is implied by the output type of
13326this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013327methods of a same sample fetch method. The sample fetch methods are the only
13328ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013329
13330Sample fetch methods return data which can be of the following types :
13331 - boolean
13332 - integer (signed or unsigned)
13333 - IPv4 or IPv6 address
13334 - string
13335 - data block
13336
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013337Converters transform any of these data into any of these. For example, some
13338converters might convert a string to a lower-case string while other ones
13339would turn a string to an IPv4 address, or apply a netmask to an IP address.
13340The resulting sample is of the type of the last converter applied to the list,
13341which defaults to the type of the sample fetch method.
13342
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013343Each sample or converter returns data of a specific type, specified with its
13344keyword in this documentation. When an ACL is declared using a standard sample
13345fetch method, certain types automatically involved a default matching method
13346which are summarized in the table below :
13347
13348 +---------------------+-----------------+
13349 | Sample or converter | Default |
13350 | output type | matching method |
13351 +---------------------+-----------------+
13352 | boolean | bool |
13353 +---------------------+-----------------+
13354 | integer | int |
13355 +---------------------+-----------------+
13356 | ip | ip |
13357 +---------------------+-----------------+
13358 | string | str |
13359 +---------------------+-----------------+
13360 | binary | none, use "-m" |
13361 +---------------------+-----------------+
13362
13363Note that in order to match a binary samples, it is mandatory to specify a
13364matching method, see below.
13365
Willy Tarreau74ca5042013-06-11 23:12:07 +020013366The ACL engine can match these types against patterns of the following types :
13367 - boolean
13368 - integer or integer range
13369 - IP address / network
13370 - string (exact, substring, suffix, prefix, subdir, domain)
13371 - regular expression
13372 - hex block
13373
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013374The following ACL flags are currently supported :
13375
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013376 -i : ignore case during matching of all subsequent patterns.
13377 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013378 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010013379 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010013380 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010013381 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020013382 -- : force end of flags. Useful when a string looks like one of the flags.
13383
Willy Tarreau74ca5042013-06-11 23:12:07 +020013384The "-f" flag is followed by the name of a file from which all lines will be
13385read as individual values. It is even possible to pass multiple "-f" arguments
13386if the patterns are to be loaded from multiple files. Empty lines as well as
13387lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
13388will be stripped. If it is absolutely necessary to insert a valid pattern
13389beginning with a sharp, just prefix it with a space so that it is not taken for
13390a comment. Depending on the data type and match method, haproxy may load the
13391lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
13392exact string matching. In this case, duplicates will automatically be removed.
13393
Thierry FOURNIER9860c412014-01-29 14:23:29 +010013394The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
13395parsed as two column file. The first column contains the patterns used by the
13396ACL, and the second column contain the samples. The sample can be used later by
13397a map. This can be useful in some rare cases where an ACL would just be used to
13398check for the existence of a pattern in a map before a mapping is applied.
13399
Thierry FOURNIER3534d882014-01-20 17:01:44 +010013400The "-u" flag forces the unique id of the ACL. This unique id is used with the
13401socket interface to identify ACL and dynamically change its values. Note that a
13402file is always identified by its name even if an id is set.
13403
Willy Tarreau74ca5042013-06-11 23:12:07 +020013404Also, note that the "-i" flag applies to subsequent entries and not to entries
13405loaded from files preceding it. For instance :
13406
13407 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
13408
13409In this example, each line of "exact-ua.lst" will be exactly matched against
13410the "user-agent" header of the request. Then each line of "generic-ua" will be
13411case-insensitively matched. Then the word "test" will be insensitively matched
13412as well.
13413
13414The "-m" flag is used to select a specific pattern matching method on the input
13415sample. All ACL-specific criteria imply a pattern matching method and generally
13416do not need this flag. However, this flag is useful with generic sample fetch
13417methods to describe how they're going to be matched against the patterns. This
13418is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010013419obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020013420followed by a pattern matching method name, this method is used instead of the
13421default one for the criterion. This makes it possible to match contents in ways
13422that were not initially planned, or with sample fetch methods which return a
13423string. The matching method also affects the way the patterns are parsed.
13424
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010013425The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
13426By default, if the parser cannot parse ip address it considers that the parsed
13427string is maybe a domain name and try dns resolution. The flag "-n" disable this
13428resolution. It is useful for detecting malformed ip lists. Note that if the DNS
13429server is not reachable, the haproxy configuration parsing may last many minutes
John Roesler27c754c2019-07-10 15:45:51 -050013430waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010013431flag "-n" disable this behavior. Note also that during the runtime, this
13432function is disabled for the dynamic acl modifications.
13433
Willy Tarreau74ca5042013-06-11 23:12:07 +020013434There are some restrictions however. Not all methods can be used with all
13435sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
13436be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020013437
13438 - "found" : only check if the requested sample could be found in the stream,
13439 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020013440 to pass any pattern to avoid confusion. This matching method is
13441 particularly useful to detect presence of certain contents such
13442 as headers, cookies, etc... even if they are empty and without
13443 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013444
13445 - "bool" : check the value as a boolean. It can only be applied to fetches
13446 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013447 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013448
13449 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020013450 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013451
13452 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020013453 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013454
Davor Ocelice9ed2812017-12-25 17:49:28 +010013455 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020013456 binary sequence. This may be used with binary or string samples.
13457
13458 - "len" : match the sample's length as an integer. This may be used with
13459 binary or string samples.
13460
Willy Tarreau74ca5042013-06-11 23:12:07 +020013461 - "str" : exact match : match the contents against a string. This may be
13462 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013463
Willy Tarreau74ca5042013-06-11 23:12:07 +020013464 - "sub" : substring match : check that the contents contain at least one of
13465 the provided string patterns. This may be used with binary or
13466 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013467
Willy Tarreau74ca5042013-06-11 23:12:07 +020013468 - "reg" : regex match : match the contents against a list of regular
13469 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013470
Willy Tarreau74ca5042013-06-11 23:12:07 +020013471 - "beg" : prefix match : check that the contents begin like the provided
13472 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013473
Willy Tarreau74ca5042013-06-11 23:12:07 +020013474 - "end" : suffix match : check that the contents end like the provided
13475 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013476
Willy Tarreau74ca5042013-06-11 23:12:07 +020013477 - "dir" : subdir match : check that a slash-delimited portion of the
13478 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013479 This may be used with binary or string samples.
13480
Willy Tarreau74ca5042013-06-11 23:12:07 +020013481 - "dom" : domain match : check that a dot-delimited portion of the contents
13482 exactly match one of the provided string patterns. This may be
13483 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013484
13485For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
13486request, it is possible to do :
13487
13488 acl jsess_present cook(JSESSIONID) -m found
13489
13490In order to apply a regular expression on the 500 first bytes of data in the
13491buffer, one would use the following acl :
13492
13493 acl script_tag payload(0,500) -m reg -i <script>
13494
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013495On systems where the regex library is much slower when using "-i", it is
13496possible to convert the sample to lowercase before matching, like this :
13497
13498 acl script_tag payload(0,500),lower -m reg <script>
13499
Willy Tarreau74ca5042013-06-11 23:12:07 +020013500All ACL-specific criteria imply a default matching method. Most often, these
13501criteria are composed by concatenating the name of the original sample fetch
13502method and the matching method. For example, "hdr_beg" applies the "beg" match
13503to samples retrieved using the "hdr" fetch method. Since all ACL-specific
13504criteria rely on a sample fetch method, it is always possible instead to use
13505the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013506
Willy Tarreau74ca5042013-06-11 23:12:07 +020013507If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013508the matching method is simply applied to the underlying sample fetch method.
13509For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013510
Willy Tarreau74ca5042013-06-11 23:12:07 +020013511 acl short_form hdr_beg(host) www.
13512 acl alternate1 hdr_beg(host) -m beg www.
13513 acl alternate2 hdr_dom(host) -m beg www.
13514 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013515
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013516
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013517The table below summarizes the compatibility matrix between sample or converter
13518types and the pattern types to fetch against. It indicates for each compatible
13519combination the name of the matching method to be used, surrounded with angle
13520brackets ">" and "<" when the method is the default one and will work by
13521default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010013522
Willy Tarreau74ca5042013-06-11 23:12:07 +020013523 +-------------------------------------------------+
13524 | Input sample type |
13525 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013526 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013527 +----------------------+---------+---------+---------+---------+---------+
13528 | none (presence only) | found | found | found | found | found |
13529 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013530 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013531 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013532 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013533 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013534 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013535 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013536 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013537 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013538 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013539 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013540 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013541 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013542 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013543 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013544 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013545 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013546 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013547 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013548 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013549 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013550 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013551 +----------------------+---------+---------+---------+---------+---------+
13552 | hex block | | | | bin | bin |
13553 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020013554
13555
Willy Tarreau74ca5042013-06-11 23:12:07 +0200135567.1.1. Matching booleans
13557------------------------
13558
13559In order to match a boolean, no value is needed and all values are ignored.
13560Boolean matching is used by default for all fetch methods of type "boolean".
13561When boolean matching is used, the fetched value is returned as-is, which means
13562that a boolean "true" will always match and a boolean "false" will never match.
13563
13564Boolean matching may also be enforced using "-m bool" on fetch methods which
13565return an integer value. Then, integer value 0 is converted to the boolean
13566"false" and all other values are converted to "true".
13567
Willy Tarreau6a06a402007-07-15 20:15:28 +020013568
Willy Tarreau74ca5042013-06-11 23:12:07 +0200135697.1.2. Matching integers
13570------------------------
13571
13572Integer matching applies by default to integer fetch methods. It can also be
13573enforced on boolean fetches using "-m int". In this case, "false" is converted
13574to the integer 0, and "true" is converted to the integer 1.
13575
13576Integer matching also supports integer ranges and operators. Note that integer
13577matching only applies to positive values. A range is a value expressed with a
13578lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013579
13580For instance, "1024:65535" is a valid range to represent a range of
13581unprivileged ports, and "1024:" would also work. "0:1023" is a valid
13582representation of privileged ports, and ":1023" would also work.
13583
Willy Tarreau62644772008-07-16 18:36:06 +020013584As a special case, some ACL functions support decimal numbers which are in fact
13585two integers separated by a dot. This is used with some version checks for
13586instance. All integer properties apply to those decimal numbers, including
13587ranges and operators.
13588
Willy Tarreau6a06a402007-07-15 20:15:28 +020013589For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010013590operators with ranges does not make much sense and is strongly discouraged.
13591Similarly, it does not make much sense to perform order comparisons with a set
13592of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013593
Willy Tarreau0ba27502007-12-24 16:55:16 +010013594Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020013595
13596 eq : true if the tested value equals at least one value
13597 ge : true if the tested value is greater than or equal to at least one value
13598 gt : true if the tested value is greater than at least one value
13599 le : true if the tested value is less than or equal to at least one value
13600 lt : true if the tested value is less than at least one value
13601
Willy Tarreau0ba27502007-12-24 16:55:16 +010013602For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020013603
13604 acl negative-length hdr_val(content-length) lt 0
13605
Willy Tarreau62644772008-07-16 18:36:06 +020013606This one matches SSL versions between 3.0 and 3.1 (inclusive) :
13607
13608 acl sslv3 req_ssl_ver 3:3.1
13609
Willy Tarreau6a06a402007-07-15 20:15:28 +020013610
Willy Tarreau74ca5042013-06-11 23:12:07 +0200136117.1.3. Matching strings
13612-----------------------
13613
13614String matching applies to string or binary fetch methods, and exists in 6
13615different forms :
13616
13617 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013618 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020013619
13620 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013621 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020013622
13623 - prefix match (-m beg) : the patterns are compared with the beginning of
13624 the extracted string, and the ACL matches if any of them matches.
13625
13626 - suffix match (-m end) : the patterns are compared with the end of the
13627 extracted string, and the ACL matches if any of them matches.
13628
Baptiste Assmann33db6002016-03-06 23:32:10 +010013629 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020013630 string, delimited with slashes ("/"), and the ACL matches if any of them
13631 matches.
13632
13633 - domain match (-m dom) : the patterns are looked up inside the extracted
13634 string, delimited with dots ("."), and the ACL matches if any of them
13635 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013636
13637String matching applies to verbatim strings as they are passed, with the
13638exception of the backslash ("\") which makes it possible to escape some
13639characters such as the space. If the "-i" flag is passed before the first
13640string, then the matching will be performed ignoring the case. In order
13641to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010013642before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020013643
Mathias Weiersmuellerb2fe2232019-12-02 09:43:40 +010013644Do not use string matches for binary fetches which might contain null bytes
13645(0x00), as the comparison stops at the occurrence of the first null byte.
13646Instead, convert the binary fetch to a hex string with the hex converter first.
13647
13648Example:
13649 # matches if the string <tag> is present in the binary sample
13650 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
13651
Willy Tarreau6a06a402007-07-15 20:15:28 +020013652
Willy Tarreau74ca5042013-06-11 23:12:07 +0200136537.1.4. Matching regular expressions (regexes)
13654---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013655
13656Just like with string matching, regex matching applies to verbatim strings as
13657they are passed, with the exception of the backslash ("\") which makes it
13658possible to escape some characters such as the space. If the "-i" flag is
13659passed before the first regex, then the matching will be performed ignoring
13660the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010013661the "--" flag before the first string. Same principle applies of course to
13662match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020013663
13664
Willy Tarreau74ca5042013-06-11 23:12:07 +0200136657.1.5. Matching arbitrary data blocks
13666-------------------------------------
13667
13668It is possible to match some extracted samples against a binary block which may
13669not safely be represented as a string. For this, the patterns must be passed as
13670a series of hexadecimal digits in an even number, when the match method is set
13671to binary. Each sequence of two digits will represent a byte. The hexadecimal
13672digits may be used upper or lower case.
13673
13674Example :
13675 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
13676 acl hello payload(0,6) -m bin 48656c6c6f0a
13677
13678
136797.1.6. Matching IPv4 and IPv6 addresses
13680---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013681
13682IPv4 addresses values can be specified either as plain addresses or with a
13683netmask appended, in which case the IPv4 address matches whenever it is
13684within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010013685host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010013686difficult to read and debug configurations. If hostnames are used, you should
13687at least ensure that they are present in /etc/hosts so that the configuration
13688does not depend on any random DNS match at the moment the configuration is
13689parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013690
Daniel Schnellereba56342016-04-13 00:26:52 +020013691The dotted IPv4 address notation is supported in both regular as well as the
13692abbreviated form with all-0-octets omitted:
13693
13694 +------------------+------------------+------------------+
13695 | Example 1 | Example 2 | Example 3 |
13696 +------------------+------------------+------------------+
13697 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
13698 | 192.168.1 | 10.12 | 127.1 |
13699 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
13700 | 192.168.1/22 | 10.12/8 | 127.1/8 |
13701 +------------------+------------------+------------------+
13702
13703Notice that this is different from RFC 4632 CIDR address notation in which
13704192.168.42/24 would be equivalent to 192.168.42.0/24.
13705
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013706IPv6 may be entered in their usual form, with or without a netmask appended.
13707Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
13708trouble with randomly resolved IP addresses, host names are never allowed in
13709IPv6 patterns.
13710
13711HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
13712following situations :
13713 - tested address is IPv4, pattern address is IPv4, the match applies
13714 in IPv4 using the supplied mask if any.
13715 - tested address is IPv6, pattern address is IPv6, the match applies
13716 in IPv6 using the supplied mask if any.
13717 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
13718 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
13719 ::IPV4 or ::ffff:IPV4, otherwise it fails.
13720 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
13721 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
13722 applied in IPv6 using the supplied IPv6 mask.
13723
Willy Tarreau74ca5042013-06-11 23:12:07 +020013724
137257.2. Using ACLs to form conditions
13726----------------------------------
13727
13728Some actions are only performed upon a valid condition. A condition is a
13729combination of ACLs with operators. 3 operators are supported :
13730
13731 - AND (implicit)
13732 - OR (explicit with the "or" keyword or the "||" operator)
13733 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020013734
Willy Tarreau74ca5042013-06-11 23:12:07 +020013735A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013736
Willy Tarreau74ca5042013-06-11 23:12:07 +020013737 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020013738
Willy Tarreau74ca5042013-06-11 23:12:07 +020013739Such conditions are generally used after an "if" or "unless" statement,
13740indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020013741
Willy Tarreau74ca5042013-06-11 23:12:07 +020013742For instance, to block HTTP requests to the "*" URL with methods other than
13743"OPTIONS", as well as POST requests without content-length, and GET or HEAD
13744requests with a content-length greater than 0, and finally every request which
13745is not either GET/HEAD/POST/OPTIONS !
13746
13747 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013748 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
13749 http-request deny if METH_GET HTTP_CONTENT
13750 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020013751
13752To select a different backend for requests to static contents on the "www" site
13753and to every request on the "img", "video", "download" and "ftp" hosts :
13754
13755 acl url_static path_beg /static /images /img /css
13756 acl url_static path_end .gif .png .jpg .css .js
13757 acl host_www hdr_beg(host) -i www
13758 acl host_static hdr_beg(host) -i img. video. download. ftp.
13759
Davor Ocelice9ed2812017-12-25 17:49:28 +010013760 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020013761 # of host "www". Use backend "www" for the rest.
13762 use_backend static if host_static or host_www url_static
13763 use_backend www if host_www
13764
13765It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
13766expressions that are built on the fly without needing to be declared. They must
13767be enclosed between braces, with a space before and after each brace (because
13768the braces must be seen as independent words). Example :
13769
13770 The following rule :
13771
13772 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013773 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020013774
13775 Can also be written that way :
13776
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013777 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020013778
13779It is generally not recommended to use this construct because it's a lot easier
13780to leave errors in the configuration when written that way. However, for very
13781simple rules matching only one source IP address for instance, it can make more
13782sense to use them than to declare ACLs with random names. Another example of
13783good use is the following :
13784
13785 With named ACLs :
13786
13787 acl site_dead nbsrv(dynamic) lt 2
13788 acl site_dead nbsrv(static) lt 2
13789 monitor fail if site_dead
13790
13791 With anonymous ACLs :
13792
13793 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
13794
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013795See section 4.2 for detailed help on the "http-request deny" and "use_backend"
13796keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013797
13798
137997.3. Fetching samples
13800---------------------
13801
13802Historically, sample fetch methods were only used to retrieve data to match
13803against patterns using ACLs. With the arrival of stick-tables, a new class of
13804sample fetch methods was created, most often sharing the same syntax as their
13805ACL counterpart. These sample fetch methods are also known as "fetches". As
13806of now, ACLs and fetches have converged. All ACL fetch methods have been made
13807available as fetch methods, and ACLs may use any sample fetch method as well.
13808
13809This section details all available sample fetch methods and their output type.
13810Some sample fetch methods have deprecated aliases that are used to maintain
13811compatibility with existing configurations. They are then explicitly marked as
13812deprecated and should not be used in new setups.
13813
13814The ACL derivatives are also indicated when available, with their respective
13815matching methods. These ones all have a well defined default pattern matching
13816method, so it is never necessary (though allowed) to pass the "-m" option to
13817indicate how the sample will be matched using ACLs.
13818
13819As indicated in the sample type versus matching compatibility matrix above,
13820when using a generic sample fetch method in an ACL, the "-m" option is
13821mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13822the same keyword exists as an ACL keyword and as a standard fetch method, the
13823ACL engine will automatically pick the ACL-only one by default.
13824
13825Some of these keywords support one or multiple mandatory arguments, and one or
13826multiple optional arguments. These arguments are strongly typed and are checked
13827when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013828incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13829are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013830is optional, it will be indicated below between square brackets ('[ ]'). When
13831all arguments are optional, the parenthesis may be omitted.
13832
13833Thus, the syntax of a standard sample fetch method is one of the following :
13834 - name
13835 - name(arg1)
13836 - name(arg1,arg2)
13837
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013838
138397.3.1. Converters
13840-----------------
13841
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013842Sample fetch methods may be combined with transformations to be applied on top
13843of the fetched sample (also called "converters"). These combinations form what
13844is called "sample expressions" and the result is a "sample". Initially this
13845was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013846has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013847unique-id-format, add-header, ...).
13848
13849These transformations are enumerated as a series of specific keywords after the
13850sample fetch method. These keywords may equally be appended immediately after
13851the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013852support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013853
Willy Tarreau97707872015-01-27 15:12:13 +010013854A certain category of converters are bitwise and arithmetic operators which
13855support performing basic operations on integers. Some bitwise operations are
13856supported (and, or, xor, cpl) and some arithmetic operations are supported
13857(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13858bool) which make it possible to report a match without having to write an ACL.
13859
Willy Tarreau74ca5042013-06-11 23:12:07 +020013860The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013861
Ben Shillitof25e8e52016-12-02 14:25:37 +00001386251d.single(<prop>[,<prop>*])
13863 Returns values for the properties requested as a string, where values are
13864 separated by the delimiter specified with "51degrees-property-separator".
13865 The device is identified using the User-Agent header passed to the
13866 converter. The function can be passed up to five property names, and if a
13867 property name can't be found, the value "NoData" is returned.
13868
13869 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013870 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13871 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013872 # User-Agent passed to the converter.
13873 frontend http-in
13874 bind *:8081
13875 default_backend servers
13876 http-request set-header X-51D-DeviceTypeMobileTablet \
13877 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13878
Willy Tarreau97707872015-01-27 15:12:13 +010013879add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013880 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013881 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013882 name. The name of the variable starts with an indication about its scope. The
13883 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013884 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013885 "sess" : the variable is shared with the whole session
13886 "txn" : the variable is shared with the transaction (request and response)
13887 "req" : the variable is shared only during request processing
13888 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013889 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013890 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013891
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010013892aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
13893 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
13894 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
13895 need to be base64 encoded and the returned result is in raw byte format.
13896 If the <aead_tag> validation fails, the converter doesn't return any data.
13897 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
13898 converter requires at least OpenSSL 1.0.1.
13899
13900 Example:
13901 http-response set-header X-Decrypted-Text %[var(txn.enc),\
13902 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
13903
Willy Tarreau97707872015-01-27 15:12:13 +010013904and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013905 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013906 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013907 numeric value or a variable name. The name of the variable starts with an
13908 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013909 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013910 "sess" : the variable is shared with the whole session
13911 "txn" : the variable is shared with the transaction (request and response)
13912 "req" : the variable is shared only during request processing
13913 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013914 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013915 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013916
Holger Just1bfc24b2017-05-06 00:56:53 +020013917b64dec
13918 Converts (decodes) a base64 encoded input string to its binary
13919 representation. It performs the inverse operation of base64().
13920
Emeric Brun53d1a982014-04-30 18:21:37 +020013921base64
13922 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013923 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013924 an SSL ID can be copied in a header).
13925
Willy Tarreau97707872015-01-27 15:12:13 +010013926bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013927 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013928 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013929 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013930 presence of a flag).
13931
Emeric Brun54c4ac82014-11-03 15:32:43 +010013932bytes(<offset>[,<length>])
13933 Extracts some bytes from an input binary sample. The result is a binary
13934 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013935 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013936
Willy Tarreau280f42b2018-02-19 15:34:12 +010013937concat([<start>],[<var>],[<end>])
13938 Concatenates up to 3 fields after the current sample which is then turned to
13939 a string. The first one, <start>, is a constant string, that will be appended
13940 immediately after the existing sample. It may be omitted if not used. The
13941 second one, <var>, is a variable name. The variable will be looked up, its
13942 contents converted to a string, and it will be appended immediately after the
13943 <first> part. If the variable is not found, nothing is appended. It may be
13944 omitted as well. The third field, <end> is a constant string that will be
13945 appended after the variable. It may also be omitted. Together, these elements
13946 allow to concatenate variables with delimiters to an existing set of
13947 variables. This can be used to build new variables made of a succession of
John Roesler27c754c2019-07-10 15:45:51 -050013948 other variables, such as colon-delimited values. Note that due to the config
Willy Tarreau280f42b2018-02-19 15:34:12 +010013949 parser, it is not possible to use a comma nor a closing parenthesis as
John Roesler27c754c2019-07-10 15:45:51 -050013950 delimiters.
Willy Tarreau280f42b2018-02-19 15:34:12 +010013951
13952 Example:
13953 tcp-request session set-var(sess.src) src
13954 tcp-request session set-var(sess.dn) ssl_c_s_dn
13955 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13956 http-request set-header x-hap-sig %[var(txn.sig)]
13957
Willy Tarreau97707872015-01-27 15:12:13 +010013958cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013959 Takes the input value of type signed integer, applies a ones-complement
13960 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013961
Willy Tarreau80599772015-01-20 19:35:24 +010013962crc32([<avalanche>])
13963 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13964 hash function. Optionally, it is possible to apply a full avalanche hash
13965 function to the output if the optional <avalanche> argument equals 1. This
13966 converter uses the same functions as used by the various hash-based load
13967 balancing algorithms, so it will provide exactly the same results. It is
13968 provided for compatibility with other software which want a CRC32 to be
13969 computed on some input keys, so it follows the most common implementation as
13970 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13971 but may provide a better or at least less predictable distribution. It must
13972 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013973 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13974
13975crc32c([<avalanche>])
13976 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13977 hash function. Optionally, it is possible to apply a full avalanche hash
13978 function to the output if the optional <avalanche> argument equals 1. This
13979 converter uses the same functions as described in RFC4960, Appendix B [8].
13980 It is provided for compatibility with other software which want a CRC32C to be
13981 computed on some input keys. It is slower than the other algorithms and it must
13982 not be used for security purposes as a 32-bit hash is trivial to break. See
13983 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013984
David Carlier29b3ca32015-09-25 14:09:21 +010013985da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013986 Asks the DeviceAtlas converter to identify the User Agent string passed on
13987 input, and to emit a string made of the concatenation of the properties
13988 enumerated in argument, delimited by the separator defined by the global
13989 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013990 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013991 configuration language.
13992
13993 Example:
13994 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013995 bind *:8881
13996 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013997 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 +020013998
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013999debug
14000 This converter is used as debug tool. It dumps on screen the content and the
14001 type of the input sample. The sample is returned as is on its output. This
14002 converter only exists when haproxy was built with debugging enabled.
14003
Willy Tarreau97707872015-01-27 15:12:13 +010014004div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014005 Divides the input value of type signed integer by <value>, and returns the
14006 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014007 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010014008 variable name. The name of the variable starts with an indication about its
14009 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014010 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014011 "sess" : the variable is shared with the whole session
14012 "txn" : the variable is shared with the transaction (request and response)
14013 "req" : the variable is shared only during request processing
14014 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014015 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014016 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014017
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014018djb2([<avalanche>])
14019 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
14020 hash function. Optionally, it is possible to apply a full avalanche hash
14021 function to the output if the optional <avalanche> argument equals 1. This
14022 converter uses the same functions as used by the various hash-based load
14023 balancing algorithms, so it will provide exactly the same results. It is
14024 mostly intended for debugging, but can be used as a stick-table entry to
14025 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010014026 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
14027 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014028
Willy Tarreau97707872015-01-27 15:12:13 +010014029even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014030 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010014031 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
14032
Marcin Deranek9631a282018-04-16 14:30:46 +020014033field(<index>,<delimiters>[,<count>])
14034 Extracts the substring at the given index counting from the beginning
14035 (positive index) or from the end (negative index) considering given delimiters
14036 from an input string. Indexes start at 1 or -1 and delimiters are a string
14037 formatted list of chars. Optionally you can specify <count> of fields to
14038 extract (default: 1). Value of 0 indicates extraction of all remaining
14039 fields.
14040
14041 Example :
14042 str(f1_f2_f3__f5),field(5,_) # f5
14043 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
14044 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
14045 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
14046 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010014047
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014048hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010014049 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014050 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010014051 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 +020014052 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010014053
Dragan Dosen3f957b22017-10-24 09:27:34 +020014054hex2i
14055 Converts a hex string containing two hex digits per input byte to an
John Roesler27c754c2019-07-10 15:45:51 -050014056 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020014057
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014058http_date([<offset>])
14059 Converts an integer supposed to contain a date since epoch to a string
14060 representing this date in a format suitable for use in HTTP header fields. If
14061 an offset value is specified, then it is a number of seconds that is added to
14062 the date before the conversion is operated. This is particularly useful to
14063 emit Date header fields, Expires values in responses when combined with a
14064 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014065
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014066in_table(<table>)
14067 Uses the string representation of the input sample to perform a look up in
14068 the specified table. If the key is not found in the table, a boolean false
14069 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010014070 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014071 or not a source IP address or an Authorization header was already seen).
14072
Tim Duesterhus1478aa72018-01-25 16:24:51 +010014073ipmask(<mask4>, [<mask6>])
14074 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020014075 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010014076 table entries and as such use the same server. The mask4 can be passed in
14077 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
14078 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
14079 If no mask6 is given IPv6 addresses will fail to convert for backwards
14080 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020014081
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014082json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010014083 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014084 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020014085 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014086 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
14087 of errors:
14088 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
14089 bytes, ...)
14090 - invalid range (the decoded value is within a UTF-8 prohibited range),
14091 - code overlong (the value is encoded with more bytes than necessary).
14092
14093 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
14094 character is greater than 0xffff because the JSON string escape specification
14095 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
14096 in 4 variants designated by a combination of two suffix letters : "p" for
14097 "permissive" and "s" for "silently ignore". The behaviors of the decoders
14098 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010014099 - "ascii" : never fails;
14100 - "utf8" : fails on any detected errors;
14101 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014102 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010014103 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014104 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
14105 characters corresponding to the other errors.
14106
14107 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010014108 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014109
14110 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014111 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020014112 capture request header user-agent len 150
14113 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014114
14115 Input request from client 127.0.0.1:
14116 GET / HTTP/1.0
14117 User-Agent: Very "Ugly" UA 1/2
14118
14119 Output log:
14120 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
14121
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014122language(<value>[,<default>])
14123 Returns the value with the highest q-factor from a list as extracted from the
14124 "accept-language" header using "req.fhdr". Values with no q-factor have a
14125 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
14126 belong to the list of semi-colon delimited <values> will be considered. The
14127 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
14128 given list and a default value is provided, it is returned. Note that language
14129 names may have a variant after a dash ('-'). If this variant is present in the
14130 list, it will be matched, but if it is not, only the base language is checked.
14131 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010014132 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014133 ordering of the values in the request counts, as the first value among
14134 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020014135
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014136 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020014137
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014138 # this configuration switches to the backend matching a
14139 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020014140
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014141 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
14142 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
14143 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
14144 use_backend spanish if es
14145 use_backend french if fr
14146 use_backend english if en
14147 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020014148
Willy Tarreau60a2ee72017-12-15 07:13:48 +010014149length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010014150 Get the length of the string. This can only be placed after a string
14151 sample fetch function or after a transformation keyword returning a string
14152 type. The result is of type integer.
14153
Willy Tarreauffcb2e42014-07-10 16:29:08 +020014154lower
14155 Convert a string sample to lower case. This can only be placed after a string
14156 sample fetch function or after a transformation keyword returning a string
14157 type. The result is of type string.
14158
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014159ltime(<format>[,<offset>])
14160 Converts an integer supposed to contain a date since epoch to a string
14161 representing this date in local time using a format defined by the <format>
14162 string using strftime(3). The purpose is to allow any date format to be used
14163 in logs. An optional <offset> in seconds may be applied to the input date
14164 (positive or negative). See the strftime() man page for the format supported
14165 by your operating system. See also the utime converter.
14166
14167 Example :
14168
14169 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010014170 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014171 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
14172
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014173map(<map_file>[,<default_value>])
14174map_<match_type>(<map_file>[,<default_value>])
14175map_<match_type>_<output_type>(<map_file>[,<default_value>])
14176 Search the input value from <map_file> using the <match_type> matching method,
14177 and return the associated value converted to the type <output_type>. If the
14178 input value cannot be found in the <map_file>, the converter returns the
14179 <default_value>. If the <default_value> is not set, the converter fails and
14180 acts as if no input value could be fetched. If the <match_type> is not set, it
14181 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
14182 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
14183 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014184
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014185 It is important to avoid overlapping between the keys : IP addresses and
14186 strings are stored in trees, so the first of the finest match will be used.
14187 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014188
Tim Düsterhus4896c442016-11-29 02:15:19 +010014189 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014190 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014191
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014192 input type | match method | output type str | output type int | output type ip
14193 -----------+--------------+-----------------+-----------------+---------------
14194 str | str | map_str | map_str_int | map_str_ip
14195 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020014196 str | beg | map_beg | map_beg_int | map_end_ip
14197 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014198 str | sub | map_sub | map_sub_int | map_sub_ip
14199 -----------+--------------+-----------------+-----------------+---------------
14200 str | dir | map_dir | map_dir_int | map_dir_ip
14201 -----------+--------------+-----------------+-----------------+---------------
14202 str | dom | map_dom | map_dom_int | map_dom_ip
14203 -----------+--------------+-----------------+-----------------+---------------
14204 str | end | map_end | map_end_int | map_end_ip
14205 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080014206 str | reg | map_reg | map_reg_int | map_reg_ip
14207 -----------+--------------+-----------------+-----------------+---------------
14208 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014209 -----------+--------------+-----------------+-----------------+---------------
14210 int | int | map_int | map_int_int | map_int_ip
14211 -----------+--------------+-----------------+-----------------+---------------
14212 ip | ip | map_ip | map_ip_int | map_ip_ip
14213 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014214
Thierry Fournier8feaa662016-02-10 22:55:20 +010014215 The special map called "map_regm" expect matching zone in the regular
14216 expression and modify the output replacing back reference (like "\1") by
14217 the corresponding match text.
14218
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014219 The file contains one key + value per line. Lines which start with '#' are
14220 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
14221 is then the first "word" (series of non-space/tabs characters), and the value
14222 is what follows this series of space/tab till the end of the line excluding
14223 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014224
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014225 Example :
14226
14227 # this is a comment and is ignored
14228 2.22.246.0/23 United Kingdom \n
14229 <-><-----------><--><------------><---->
14230 | | | | `- trailing spaces ignored
14231 | | | `---------- value
14232 | | `-------------------- middle spaces ignored
14233 | `---------------------------- key
14234 `------------------------------------ leading spaces ignored
14235
Willy Tarreau97707872015-01-27 15:12:13 +010014236mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014237 Divides the input value of type signed integer by <value>, and returns the
14238 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014239 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014240 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014241 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014242 "sess" : the variable is shared with the whole session
14243 "txn" : the variable is shared with the transaction (request and response)
14244 "req" : the variable is shared only during request processing
14245 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014246 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014247 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014248
14249mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014250 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020014251 the product as an signed integer. In case of overflow, the largest possible
14252 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014253 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014254 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014255 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014256 "sess" : the variable is shared with the whole session
14257 "txn" : the variable is shared with the transaction (request and response)
14258 "req" : the variable is shared only during request processing
14259 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014260 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014261 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014262
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010014263nbsrv
14264 Takes an input value of type string, interprets it as a backend name and
14265 returns the number of usable servers in that backend. Can be used in places
14266 where we want to look up a backend from a dynamic name, like a result of a
14267 map lookup.
14268
Willy Tarreau97707872015-01-27 15:12:13 +010014269neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014270 Takes the input value of type signed integer, computes the opposite value,
14271 and returns the remainder as an signed integer. 0 is identity. This operator
14272 is provided for reversed subtracts : in order to subtract the input from a
14273 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010014274
14275not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014276 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010014277 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010014278 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010014279 absence of a flag).
14280
14281odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014282 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010014283 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
14284
14285or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014286 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014287 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010014288 numeric value or a variable name. The name of the variable starts with an
14289 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014290 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014291 "sess" : the variable is shared with the whole session
14292 "txn" : the variable is shared with the transaction (request and response)
14293 "req" : the variable is shared only during request processing
14294 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014295 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014296 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014297
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014298protobuf(<field_number>,[<field_type>])
14299 This extracts the protocol buffers message field in raw mode of an input binary
14300 sample representation of a protocol buffer message with <field_number> as field
14301 number (dotted notation) if <field_type> is not present, or as an integer sample
14302 if this field is present (see also "ungrpc" below).
14303 The list of the authorized types is the following one: "int32", "int64", "uint32",
14304 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
14305 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
14306 "float" for the wire type 5. Note that "string" is considered as a length-delimited
14307 type, so it does not require any <field_type> argument to be extracted.
14308 More information may be found here about the protocol buffers message field types:
14309 https://developers.google.com/protocol-buffers/docs/encoding
14310
Willy Tarreauc4dc3502015-01-23 20:39:28 +010014311regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010014312 Applies a regex-based substitution to the input string. It does the same
14313 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
14314 default it will replace in the input string the first occurrence of the
14315 largest part matching the regular expression <regex> with the substitution
14316 string <subst>. It is possible to replace all occurrences instead by adding
14317 the flag "g" in the third argument <flags>. It is also possible to make the
14318 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
14319 string, it is made up from the concatenation of all desired flags. Thus if
14320 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
14321 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010014322 configuration parser, some characters such as closing parenthesis, closing
14323 square brackets or comma are not possible to use in the arguments. The first
14324 use of this converter is to replace certain characters or sequence of
14325 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010014326
14327 Example :
14328
14329 # de-duplicate "/" in header "x-path".
14330 # input: x-path: /////a///b/c/xzxyz/
14331 # output: x-path: /a/b/c/xzxyz/
14332 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
14333
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020014334capture-req(<id>)
14335 Capture the string entry in the request slot <id> and returns the entry as
14336 is. If the slot doesn't exist, the capture fails silently.
14337
14338 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020014339 "http-response capture", "capture.req.hdr" and
14340 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020014341
14342capture-res(<id>)
14343 Capture the string entry in the response slot <id> and returns the entry as
14344 is. If the slot doesn't exist, the capture fails silently.
14345
14346 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020014347 "http-response capture", "capture.req.hdr" and
14348 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020014349
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014350sdbm([<avalanche>])
14351 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
14352 hash function. Optionally, it is possible to apply a full avalanche hash
14353 function to the output if the optional <avalanche> argument equals 1. This
14354 converter uses the same functions as used by the various hash-based load
14355 balancing algorithms, so it will provide exactly the same results. It is
14356 mostly intended for debugging, but can be used as a stick-table entry to
14357 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010014358 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
14359 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014360
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014361set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010014362 Sets a variable with the input content and returns the content on the output
14363 as-is. The variable keeps the value and the associated input type. The name of
14364 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014365 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014366 "sess" : the variable is shared with the whole session
14367 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014368 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014369 "req" : the variable is shared only during request processing,
14370 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014371 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014372 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014373
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020014374sha1
14375 Converts a binary input sample to a SHA1 digest. The result is a binary
14376 sample with length of 20 bytes.
14377
Tim Duesterhusca097c12018-04-27 21:18:45 +020014378strcmp(<var>)
14379 Compares the contents of <var> with the input value of type string. Returns
14380 the result as a signed integer compatible with strcmp(3): 0 if both strings
14381 are identical. A value less than 0 if the left string is lexicographically
14382 smaller than the right string or if the left string is shorter. A value greater
14383 than 0 otherwise (right string greater than left string or the right string is
14384 shorter).
14385
14386 Example :
14387
14388 http-request set-var(txn.host) hdr(host)
14389 # Check whether the client is attempting domain fronting.
14390 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
14391
14392
Willy Tarreau97707872015-01-27 15:12:13 +010014393sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014394 Subtracts <value> from the input value of type signed integer, and returns
14395 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014396 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010014397 or a variable name. The name of the variable starts with an indication about
14398 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014399 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014400 "sess" : the variable is shared with the whole session
14401 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014402 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014403 "req" : the variable is shared only during request processing,
14404 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014405 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014406 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014407
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014408table_bytes_in_rate(<table>)
14409 Uses the string representation of the input sample to perform a look up in
14410 the specified table. If the key is not found in the table, integer value zero
14411 is returned. Otherwise the converter returns the average client-to-server
14412 bytes rate associated with the input sample in the designated table, measured
14413 in amount of bytes over the period configured in the table. See also the
14414 sc_bytes_in_rate sample fetch keyword.
14415
14416
14417table_bytes_out_rate(<table>)
14418 Uses the string representation of the input sample to perform a look up in
14419 the specified table. If the key is not found in the table, integer value zero
14420 is returned. Otherwise the converter returns the average server-to-client
14421 bytes rate associated with the input sample in the designated table, measured
14422 in amount of bytes over the period configured in the table. See also the
14423 sc_bytes_out_rate sample fetch keyword.
14424
14425table_conn_cnt(<table>)
14426 Uses the string representation of the input sample to perform a look up in
14427 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014428 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014429 connections associated with the input sample in the designated table. See
14430 also the sc_conn_cnt sample fetch keyword.
14431
14432table_conn_cur(<table>)
14433 Uses the string representation of the input sample to perform a look up in
14434 the specified table. If the key is not found in the table, integer value zero
14435 is returned. Otherwise the converter returns the current amount of concurrent
14436 tracked connections associated with the input sample in the designated table.
14437 See also the sc_conn_cur sample fetch keyword.
14438
14439table_conn_rate(<table>)
14440 Uses the string representation of the input sample to perform a look up in
14441 the specified table. If the key is not found in the table, integer value zero
14442 is returned. Otherwise the converter returns the average incoming connection
14443 rate associated with the input sample in the designated table. See also the
14444 sc_conn_rate sample fetch keyword.
14445
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014446table_gpt0(<table>)
14447 Uses the string representation of the input sample to perform a look up in
14448 the specified table. If the key is not found in the table, boolean value zero
14449 is returned. Otherwise the converter returns the current value of the first
14450 general purpose tag associated with the input sample in the designated table.
14451 See also the sc_get_gpt0 sample fetch keyword.
14452
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014453table_gpc0(<table>)
14454 Uses the string representation of the input sample to perform a look up in
14455 the specified table. If the key is not found in the table, integer value zero
14456 is returned. Otherwise the converter returns the current value of the first
14457 general purpose counter associated with the input sample in the designated
14458 table. See also the sc_get_gpc0 sample fetch keyword.
14459
14460table_gpc0_rate(<table>)
14461 Uses the string representation of the input sample to perform a look up in
14462 the specified table. If the key is not found in the table, integer value zero
14463 is returned. Otherwise the converter returns the frequency which the gpc0
14464 counter was incremented over the configured period in the table, associated
14465 with the input sample in the designated table. See also the sc_get_gpc0_rate
14466 sample fetch keyword.
14467
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014468table_gpc1(<table>)
14469 Uses the string representation of the input sample to perform a look up in
14470 the specified table. If the key is not found in the table, integer value zero
14471 is returned. Otherwise the converter returns the current value of the second
14472 general purpose counter associated with the input sample in the designated
14473 table. See also the sc_get_gpc1 sample fetch keyword.
14474
14475table_gpc1_rate(<table>)
14476 Uses the string representation of the input sample to perform a look up in
14477 the specified table. If the key is not found in the table, integer value zero
14478 is returned. Otherwise the converter returns the frequency which the gpc1
14479 counter was incremented over the configured period in the table, associated
14480 with the input sample in the designated table. See also the sc_get_gpc1_rate
14481 sample fetch keyword.
14482
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014483table_http_err_cnt(<table>)
14484 Uses the string representation of the input sample to perform a look up in
14485 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014486 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014487 errors associated with the input sample in the designated table. See also the
14488 sc_http_err_cnt sample fetch keyword.
14489
14490table_http_err_rate(<table>)
14491 Uses the string representation of the input sample to perform a look up in
14492 the specified table. If the key is not found in the table, integer value zero
14493 is returned. Otherwise the average rate of HTTP errors associated with the
14494 input sample in the designated table, measured in amount of errors over the
14495 period configured in the table. See also the sc_http_err_rate sample fetch
14496 keyword.
14497
14498table_http_req_cnt(<table>)
14499 Uses the string representation of the input sample to perform a look up in
14500 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014501 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014502 requests associated with the input sample in the designated table. See also
14503 the sc_http_req_cnt sample fetch keyword.
14504
14505table_http_req_rate(<table>)
14506 Uses the string representation of the input sample to perform a look up in
14507 the specified table. If the key is not found in the table, integer value zero
14508 is returned. Otherwise the average rate of HTTP requests associated with the
14509 input sample in the designated table, measured in amount of requests over the
14510 period configured in the table. See also the sc_http_req_rate sample fetch
14511 keyword.
14512
14513table_kbytes_in(<table>)
14514 Uses the string representation of the input sample to perform a look up in
14515 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014516 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014517 to-server data associated with the input sample in the designated table,
14518 measured in kilobytes. The test is currently performed on 32-bit integers,
14519 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
14520 keyword.
14521
14522table_kbytes_out(<table>)
14523 Uses the string representation of the input sample to perform a look up in
14524 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014525 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014526 to-client data associated with the input sample in the designated table,
14527 measured in kilobytes. The test is currently performed on 32-bit integers,
14528 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
14529 keyword.
14530
14531table_server_id(<table>)
14532 Uses the string representation of the input sample to perform a look up in
14533 the specified table. If the key is not found in the table, integer value zero
14534 is returned. Otherwise the converter returns the server ID associated with
14535 the input sample in the designated table. A server ID is associated to a
14536 sample by a "stick" rule when a connection to a server succeeds. A server ID
14537 zero means that no server is associated with this key.
14538
14539table_sess_cnt(<table>)
14540 Uses the string representation of the input sample to perform a look up in
14541 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014542 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014543 sessions associated with the input sample in the designated table. Note that
14544 a session here refers to an incoming connection being accepted by the
14545 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
14546 keyword.
14547
14548table_sess_rate(<table>)
14549 Uses the string representation of the input sample to perform a look up in
14550 the specified table. If the key is not found in the table, integer value zero
14551 is returned. Otherwise the converter returns the average incoming session
14552 rate associated with the input sample in the designated table. Note that a
14553 session here refers to an incoming connection being accepted by the
14554 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
14555 keyword.
14556
14557table_trackers(<table>)
14558 Uses the string representation of the input sample to perform a look up in
14559 the specified table. If the key is not found in the table, integer value zero
14560 is returned. Otherwise the converter returns the current amount of concurrent
14561 connections tracking the same key as the input sample in the designated
14562 table. It differs from table_conn_cur in that it does not rely on any stored
14563 information but on the table's reference count (the "use" value which is
14564 returned by "show table" on the CLI). This may sometimes be more suited for
14565 layer7 tracking. It can be used to tell a server how many concurrent
14566 connections there are from a given address for example. See also the
14567 sc_trackers sample fetch keyword.
14568
Willy Tarreauffcb2e42014-07-10 16:29:08 +020014569upper
14570 Convert a string sample to upper case. This can only be placed after a string
14571 sample fetch function or after a transformation keyword returning a string
14572 type. The result is of type string.
14573
Willy Tarreau7e913cb2020-04-23 17:54:47 +020014574url_dec([<in_form>])
14575 Takes an url-encoded string provided as input and returns the decoded version
14576 as output. The input and the output are of type string. If the <in_form>
14577 argument is set to a non-zero integer value, the input string is assumed to
14578 be part of a form or query string and the '+' character will be turned into a
14579 space (' '). Otherwise this will only happen after a question mark indicating
14580 a query string ('?').
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020014581
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014582ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014583 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014584 sample representation of a gRPC message with <field_number> as field number
14585 (dotted notation) if <field_type> is not present, or as an integer sample if this
14586 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014587 The list of the authorized types is the following one: "int32", "int64", "uint32",
14588 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
14589 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
14590 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014591 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014592 More information may be found here about the protocol buffers message field types:
14593 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014594
14595 Example:
14596 // with such a protocol buffer .proto file content adapted from
14597 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
14598
14599 message Point {
14600 int32 latitude = 1;
14601 int32 longitude = 2;
14602 }
14603
14604 message PPoint {
14605 Point point = 59;
14606 }
14607
14608 message Rectangle {
14609 // One corner of the rectangle.
14610 PPoint lo = 48;
14611 // The other corner of the rectangle.
14612 PPoint hi = 49;
14613 }
14614
14615 let's say a body request is made of a "Rectangle" object value (two PPoint
14616 protocol buffers messages), the four protocol buffers fields could be
14617 extracted with these "ungrpc" directives:
14618
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014619 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
14620 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roesler27c754c2019-07-10 15:45:51 -050014621 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014622 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
14623
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014624 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014625
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014626 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014627
John Roesler27c754c2019-07-10 15:45:51 -050014628 As a gRPC message is always made of a gRPC header followed by protocol buffers
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014629 messages, in the previous example the "latitude" of "lo" first PPoint
14630 could be extracted with these equivalent directives:
14631
14632 req.body,ungrpc(48.59),protobuf(1,int32)
14633 req.body,ungrpc(48),protobuf(59.1,int32)
14634 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
14635
14636 Note that the first convert must be "ungrpc", the remaining ones must be
14637 "protobuf" and only the last one may have or not a second argument to
14638 interpret the previous binary sample.
14639
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014640
Christopher Faulet85d79c92016-11-09 16:54:56 +010014641unset-var(<var name>)
14642 Unsets a variable if the input content is defined. The name of the variable
14643 starts with an indication about its scope. The scopes allowed are:
14644 "proc" : the variable is shared with the whole process
14645 "sess" : the variable is shared with the whole session
14646 "txn" : the variable is shared with the transaction (request and
14647 response),
14648 "req" : the variable is shared only during request processing,
14649 "res" : the variable is shared only during response processing.
14650 This prefix is followed by a name. The separator is a '.'. The name may only
14651 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
14652
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014653utime(<format>[,<offset>])
14654 Converts an integer supposed to contain a date since epoch to a string
14655 representing this date in UTC time using a format defined by the <format>
14656 string using strftime(3). The purpose is to allow any date format to be used
14657 in logs. An optional <offset> in seconds may be applied to the input date
14658 (positive or negative). See the strftime() man page for the format supported
14659 by your operating system. See also the ltime converter.
14660
14661 Example :
14662
14663 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010014664 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014665 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
14666
Marcin Deranek9631a282018-04-16 14:30:46 +020014667word(<index>,<delimiters>[,<count>])
14668 Extracts the nth word counting from the beginning (positive index) or from
14669 the end (negative index) considering given delimiters from an input string.
14670 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
Jerome Magnind1fa5fa2020-01-28 13:33:44 +010014671 Delimiters at the beginning or end of the input string are ignored.
Marcin Deranek9631a282018-04-16 14:30:46 +020014672 Optionally you can specify <count> of words to extract (default: 1).
14673 Value of 0 indicates extraction of all remaining words.
14674
14675 Example :
14676 str(f1_f2_f3__f5),word(4,_) # f5
14677 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
14678 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
14679 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
14680 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Jerome Magnind1fa5fa2020-01-28 13:33:44 +010014681 str(/f1/f2/f3/f4),word(1,/) # f1
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010014682
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014683wt6([<avalanche>])
14684 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
14685 hash function. Optionally, it is possible to apply a full avalanche hash
14686 function to the output if the optional <avalanche> argument equals 1. This
14687 converter uses the same functions as used by the various hash-based load
14688 balancing algorithms, so it will provide exactly the same results. It is
14689 mostly intended for debugging, but can be used as a stick-table entry to
14690 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010014691 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
14692 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014693
Willy Tarreau97707872015-01-27 15:12:13 +010014694xor(<value>)
14695 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014696 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014697 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014698 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014699 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014700 "sess" : the variable is shared with the whole session
14701 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014702 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014703 "req" : the variable is shared only during request processing,
14704 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014705 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014706 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014707
Thierry FOURNIER01e09742016-12-26 11:46:11 +010014708xxh32([<seed>])
14709 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
14710 variant of the XXHash hash function. This hash supports a seed which defaults
14711 to zero but a different value maybe passed as the <seed> argument. This hash
14712 is known to be very good and very fast so it can be used to hash URLs and/or
14713 URL parameters for use as stick-table keys to collect statistics with a low
14714 collision rate, though care must be taken as the algorithm is not considered
14715 as cryptographically secure.
14716
14717xxh64([<seed>])
14718 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
14719 variant of the XXHash hash function. This hash supports a seed which defaults
14720 to zero but a different value maybe passed as the <seed> argument. This hash
14721 is known to be very good and very fast so it can be used to hash URLs and/or
14722 URL parameters for use as stick-table keys to collect statistics with a low
14723 collision rate, though care must be taken as the algorithm is not considered
14724 as cryptographically secure.
14725
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014726
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200147277.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020014728--------------------------------------------
14729
14730A first set of sample fetch methods applies to internal information which does
14731not even relate to any client information. These ones are sometimes used with
14732"monitor-fail" directives to report an internal status to external watchers.
14733The sample fetch methods described in this section are usable anywhere.
14734
14735always_false : boolean
14736 Always returns the boolean "false" value. It may be used with ACLs as a
14737 temporary replacement for another one when adjusting configurations.
14738
14739always_true : boolean
14740 Always returns the boolean "true" value. It may be used with ACLs as a
14741 temporary replacement for another one when adjusting configurations.
14742
14743avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014744 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020014745 divided by the number of active servers. The current backend is used if no
14746 backend is specified. This is very similar to "queue" except that the size of
14747 the farm is considered, in order to give a more accurate measurement of the
14748 time it may take for a new connection to be processed. The main usage is with
14749 ACL to return a sorry page to new users when it becomes certain they will get
14750 a degraded service, or to pass to the backend servers in a header so that
14751 they decide to work in degraded mode or to disable some functions to speed up
14752 the processing a bit. Note that in the event there would not be any active
14753 server anymore, twice the number of queued connections would be considered as
14754 the measured value. This is a fair estimate, as we expect one server to get
14755 back soon anyway, but we still prefer to send new traffic to another backend
14756 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
14757 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010014758
Willy Tarreau74ca5042013-06-11 23:12:07 +020014759be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020014760 Applies to the number of currently established connections on the backend,
14761 possibly including the connection being evaluated. If no backend name is
14762 specified, the current one is used. But it is also possible to check another
14763 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014764 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
14765
14766be_conn_free([<backend>]) : integer
14767 Returns an integer value corresponding to the number of available connections
14768 across available servers in the backend. Queue slots are not included. Backup
14769 servers are also not included, unless all other servers are down. If no
14770 backend name is specified, the current one is used. But it is also possible
14771 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014772 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
14773 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014774
14775 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
14776 (meaning unlimited), then this fetch clearly does not make sense, in which
14777 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014778
Willy Tarreau74ca5042013-06-11 23:12:07 +020014779be_sess_rate([<backend>]) : integer
14780 Returns an integer value corresponding to the sessions creation rate on the
14781 backend, in number of new sessions per second. This is used with ACLs to
14782 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010014783 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020014784 online dictionary). It can also be useful to add this element to logs using a
14785 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014786
14787 Example :
14788 # Redirect to an error page if the dictionary is requested too often
14789 backend dynamic
14790 mode http
14791 acl being_scanned be_sess_rate gt 100
14792 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010014793
Davor Ocelice9ed2812017-12-25 17:49:28 +010014794bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014795 Returns a binary chain. The input is the hexadecimal representation
14796 of the string.
14797
14798bool(<bool>) : bool
14799 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
14800 'false' and '0' are the same. 'true' and '1' are the same.
14801
Willy Tarreau74ca5042013-06-11 23:12:07 +020014802connslots([<backend>]) : integer
14803 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014804 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020014805 connections on all servers and the maximum queue size. This is probably only
14806 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050014807
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014808 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020014809 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014810 usage; see "use_backend" keyword) can be redirected to a different backend.
14811
Willy Tarreau55165fe2009-05-10 12:02:55 +020014812 'connslots' = number of available server connection slots, + number of
14813 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014814
Willy Tarreaua36af912009-10-10 12:02:45 +020014815 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020014816 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020014817 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020014818 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010014819 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020014820 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020014821 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014822
Willy Tarreau55165fe2009-05-10 12:02:55 +020014823 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
14824 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020014825 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020014826 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014827
Willy Tarreau70fe9442018-11-22 16:07:39 +010014828cpu_calls : integer
14829 Returns the number of calls to the task processing the stream or current
14830 request since it was allocated. This number is reset for each new request on
14831 the same connections in case of HTTP keep-alive. This value should usually be
14832 low and stable (around 2 calls for a typically simple request) but may become
14833 high if some processing (compression, caching or analysis) is performed. This
14834 is purely for performance monitoring purposes.
14835
14836cpu_ns_avg : integer
14837 Returns the average number of nanoseconds spent in each call to the task
14838 processing the stream or current request. This number is reset for each new
14839 request on the same connections in case of HTTP keep-alive. This value
14840 indicates the overall cost of processing the request or the connection for
14841 each call. There is no good nor bad value but the time spent in a call
14842 automatically causes latency for other processing (see lat_ns_avg below),
14843 and may affect other connection's apparent response time. Certain operations
14844 like compression, complex regex matching or heavy Lua operations may directly
14845 affect this value, and having it in the logs will make it easier to spot the
14846 faulty processing that needs to be fixed to recover decent performance.
14847 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
14848
14849cpu_ns_tot : integer
14850 Returns the total number of nanoseconds spent in each call to the task
14851 processing the stream or current request. This number is reset for each new
14852 request on the same connections in case of HTTP keep-alive. This value
14853 indicates the overall cost of processing the request or the connection for
14854 each call. There is no good nor bad value but the time spent in a call
14855 automatically causes latency for other processing (see lat_ns_avg below),
14856 induces CPU costs on the machine, and may affect other connection's apparent
14857 response time. Certain operations like compression, complex regex matching or
14858 heavy Lua operations may directly affect this value, and having it in the
14859 logs will make it easier to spot the faulty processing that needs to be fixed
14860 to recover decent performance. The value may be artificially high due to a
14861 high cpu_calls count, for example when processing many HTTP chunks, and for
14862 this reason it is often preferred to log cpu_ns_avg instead.
14863
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014864date([<offset>]) : integer
14865 Returns the current date as the epoch (number of seconds since 01/01/1970).
14866 If an offset value is specified, then it is a number of seconds that is added
14867 to the current date before returning the value. This is particularly useful
14868 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020014869 It is useful combined with the http_date converter.
14870
14871 Example :
14872
14873 # set an expires header to now+1 hour in every response
14874 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014875
Etienne Carrierea792a0a2018-01-17 13:43:24 +010014876date_us : integer
14877 Return the microseconds part of the date (the "second" part is returned by
14878 date sample). This sample is coherent with the date sample as it is comes
14879 from the same timeval structure.
14880
Willy Tarreaud716f9b2017-10-13 11:03:15 +020014881distcc_body(<token>[,<occ>]) : binary
14882 Parses a distcc message and returns the body associated to occurrence #<occ>
14883 of the token <token>. Occurrences start at 1, and when unspecified, any may
14884 match though in practice only the first one is checked for now. This can be
14885 used to extract file names or arguments in files built using distcc through
14886 haproxy. Please refer to distcc's protocol documentation for the complete
14887 list of supported tokens.
14888
14889distcc_param(<token>[,<occ>]) : integer
14890 Parses a distcc message and returns the parameter associated to occurrence
14891 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
14892 any may match though in practice only the first one is checked for now. This
14893 can be used to extract certain information such as the protocol version, the
14894 file size or the argument in files built using distcc through haproxy.
14895 Another use case consists in waiting for the start of the preprocessed file
14896 contents before connecting to the server to avoid keeping idle connections.
14897 Please refer to distcc's protocol documentation for the complete list of
14898 supported tokens.
14899
14900 Example :
14901 # wait up to 20s for the pre-processed file to be uploaded
14902 tcp-request inspect-delay 20s
14903 tcp-request content accept if { distcc_param(DOTI) -m found }
14904 # send large files to the big farm
14905 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14906
Willy Tarreau595ec542013-06-12 21:34:28 +020014907env(<name>) : string
14908 Returns a string containing the value of environment variable <name>. As a
14909 reminder, environment variables are per-process and are sampled when the
14910 process starts. This can be useful to pass some information to a next hop
14911 server, or with ACLs to take specific action when the process is started a
14912 certain way.
14913
14914 Examples :
14915 # Pass the Via header to next hop with the local hostname in it
14916 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14917
14918 # reject cookie-less requests when the STOP environment variable is set
14919 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14920
Willy Tarreau74ca5042013-06-11 23:12:07 +020014921fe_conn([<frontend>]) : integer
14922 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014923 possibly including the connection being evaluated. If no frontend name is
14924 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014925 frontend. It can be used to return a sorry page before hard-blocking, or to
14926 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014927 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014928 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14929 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014930
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014931fe_req_rate([<frontend>]) : integer
14932 Returns an integer value corresponding to the number of HTTP requests per
14933 second sent to a frontend. This number can differ from "fe_sess_rate" in
14934 situations where client-side keep-alive is enabled.
14935
Willy Tarreau74ca5042013-06-11 23:12:07 +020014936fe_sess_rate([<frontend>]) : integer
14937 Returns an integer value corresponding to the sessions creation rate on the
14938 frontend, in number of new sessions per second. This is used with ACLs to
14939 limit the incoming session rate to an acceptable range in order to prevent
14940 abuse of service at the earliest moment, for example when combined with other
14941 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14942 down below the limit. It can also be useful to add this element to logs using
14943 a log-format directive. See also the "rate-limit sessions" directive for use
14944 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014945
14946 Example :
14947 # This frontend limits incoming mails to 10/s with a max of 100
14948 # concurrent connections. We accept any connection below 10/s, and
14949 # force excess clients to wait for 100 ms. Since clients are limited to
14950 # 100 max, there cannot be more than 10 incoming mails per second.
14951 frontend mail
14952 bind :25
14953 mode tcp
14954 maxconn 100
14955 acl too_fast fe_sess_rate ge 10
14956 tcp-request inspect-delay 100ms
14957 tcp-request content accept if ! too_fast
14958 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014959
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014960hostname : string
14961 Returns the system hostname.
14962
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014963int(<integer>) : signed integer
14964 Returns a signed integer.
14965
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014966ipv4(<ipv4>) : ipv4
14967 Returns an ipv4.
14968
14969ipv6(<ipv6>) : ipv6
14970 Returns an ipv6.
14971
Willy Tarreau70fe9442018-11-22 16:07:39 +010014972lat_ns_avg : integer
14973 Returns the average number of nanoseconds spent between the moment the task
14974 handling the stream is woken up and the moment it is effectively called. This
14975 number is reset for each new request on the same connections in case of HTTP
14976 keep-alive. This value indicates the overall latency inflicted to the current
14977 request by all other requests being processed in parallel, and is a direct
14978 indicator of perceived performance due to noisy neighbours. In order to keep
14979 the value low, it is possible to reduce the scheduler's run queue depth using
14980 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14981 once using "tune.maxpollevents", to decrease the stream's nice value using
14982 the "nice" option on the "bind" lines or in the frontend, or to look for
14983 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14984 whose processing needs to be adjusted or fixed. Compression of large buffers
14985 could be a culprit, like heavy regex or long lists of regex.
14986 Note: this value is exactly lat_ns_tot divided by cpu_calls.
14987
14988lat_ns_tot : integer
14989 Returns the total number of nanoseconds spent between the moment the task
14990 handling the stream is woken up and the moment it is effectively called. This
14991 number is reset for each new request on the same connections in case of HTTP
14992 keep-alive. This value indicates the overall latency inflicted to the current
14993 request by all other requests being processed in parallel, and is a direct
14994 indicator of perceived performance due to noisy neighbours. In order to keep
14995 the value low, it is possible to reduce the scheduler's run queue depth using
14996 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14997 once using "tune.maxpollevents", to decrease the stream's nice value using
14998 the "nice" option on the "bind" lines or in the frontend, or to look for
14999 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
15000 whose processing needs to be adjusted or fixed. Compression of large buffers
15001 could be a culprit, like heavy regex or long lists of regex. Note: while it
15002 may intuitively seem that the total latency adds to a transfer time, it is
15003 almost never true because while a task waits for the CPU, network buffers
15004 continue to fill up and the next call will process more at once. The value
15005 may be artificially high due to a high cpu_calls count, for example when
15006 processing many HTTP chunks, and for this reason it is often preferred to log
15007 lat_ns_avg instead, which is a more relevant performance indicator.
15008
Thierry FOURNIERcc103292015-06-06 19:30:17 +020015009meth(<method>) : method
15010 Returns a method.
15011
Willy Tarreau0f30d262014-11-24 16:02:05 +010015012nbproc : integer
15013 Returns an integer value corresponding to the number of processes that were
15014 started (it equals the global "nbproc" setting). This is useful for logging
15015 and debugging purposes.
15016
Willy Tarreau74ca5042013-06-11 23:12:07 +020015017nbsrv([<backend>]) : integer
15018 Returns an integer value corresponding to the number of usable servers of
15019 either the current backend or the named backend. This is mostly used with
15020 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010015021 switch to an alternate backend when the number of servers is too low to
15022 to handle some load. It is useful to report a failure when combined with
15023 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010015024
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040015025prio_class : integer
15026 Returns the priority class of the current session for http mode or connection
15027 for tcp mode. The value will be that set by the last call to "http-request
15028 set-priority-class" or "tcp-request content set-priority-class".
15029
15030prio_offset : integer
15031 Returns the priority offset of the current session for http mode or
15032 connection for tcp mode. The value will be that set by the last call to
15033 "http-request set-priority-offset" or "tcp-request content
15034 set-priority-offset".
15035
Willy Tarreau0f30d262014-11-24 16:02:05 +010015036proc : integer
15037 Returns an integer value corresponding to the position of the process calling
15038 the function, between 1 and global.nbproc. This is useful for logging and
15039 debugging purposes.
15040
Willy Tarreau74ca5042013-06-11 23:12:07 +020015041queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010015042 Returns the total number of queued connections of the designated backend,
15043 including all the connections in server queues. If no backend name is
15044 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020015045 one. This is useful with ACLs or to pass statistics to backend servers. This
15046 can be used to take actions when queuing goes above a known level, generally
15047 indicating a surge of traffic or a massive slowdown on the servers. One
15048 possible action could be to reject new users but still accept old ones. See
15049 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
15050
Willy Tarreau84310e22014-02-14 11:59:04 +010015051rand([<range>]) : integer
15052 Returns a random integer value within a range of <range> possible values,
15053 starting at zero. If the range is not specified, it defaults to 2^32, which
15054 gives numbers between 0 and 4294967295. It can be useful to pass some values
15055 needed to take some routing decisions for example, or just for debugging
15056 purposes. This random must not be used for security purposes.
15057
Luca Schimweg77306662019-09-10 15:42:52 +020015058uuid([<version>]) : string
15059 Returns a UUID following the RFC4122 standard. If the version is not
15060 specified, a UUID version 4 (fully random) is returned.
15061 Currently, only version 4 is supported.
15062
Willy Tarreau74ca5042013-06-11 23:12:07 +020015063srv_conn([<backend>/]<server>) : integer
15064 Returns an integer value corresponding to the number of currently established
15065 connections on the designated server, possibly including the connection being
15066 evaluated. If <backend> is omitted, then the server is looked up in the
15067 current backend. It can be used to use a specific farm when one server is
15068 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040015069 connections with it. See also the "fe_conn", "be_conn", "queue", and
15070 "srv_conn_free" fetch methods.
15071
15072srv_conn_free([<backend>/]<server>) : integer
15073 Returns an integer value corresponding to the number of available connections
15074 on the designated server, possibly including the connection being evaluated.
15075 The value does not include queue slots. If <backend> is omitted, then the
15076 server is looked up in the current backend. It can be used to use a specific
15077 farm when one server is full, or to inform the server about our view of the
15078 number of active connections with it. See also the "be_conn_free" and
15079 "srv_conn" fetch methods.
15080
15081 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
15082 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015083
15084srv_is_up([<backend>/]<server>) : boolean
15085 Returns true when the designated server is UP, and false when it is either
15086 DOWN or in maintenance mode. If <backend> is omitted, then the server is
15087 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010015088 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020015089 availability). Another possible use which is more of a hack consists in
15090 using dummy servers as boolean variables that can be enabled or disabled from
15091 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
15092
Willy Tarreauff2b7af2017-10-13 11:46:26 +020015093srv_queue([<backend>/]<server>) : integer
15094 Returns an integer value corresponding to the number of connections currently
15095 pending in the designated server's queue. If <backend> is omitted, then the
15096 server is looked up in the current backend. It can sometimes be used together
15097 with the "use-server" directive to force to use a known faster server when it
15098 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
15099 fetch methods.
15100
Willy Tarreau74ca5042013-06-11 23:12:07 +020015101srv_sess_rate([<backend>/]<server>) : integer
15102 Returns an integer corresponding to the sessions creation rate on the
15103 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015104 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020015105 used with ACLs but can make sense with logs too. This is used to switch to an
15106 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010015107 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020015108 overloading servers).
15109
15110 Example :
15111 # Redirect to a separate back
15112 acl srv1_full srv_sess_rate(be1/srv1) gt 50
15113 acl srv2_full srv_sess_rate(be1/srv2) gt 50
15114 use_backend be2 if srv1_full or srv2_full
15115
Willy Tarreau0f30d262014-11-24 16:02:05 +010015116stopping : boolean
15117 Returns TRUE if the process calling the function is currently stopping. This
15118 can be useful for logging, or for relaxing certain checks or helping close
15119 certain connections upon graceful shutdown.
15120
Thierry FOURNIERcc103292015-06-06 19:30:17 +020015121str(<string>) : string
15122 Returns a string.
15123
Willy Tarreau74ca5042013-06-11 23:12:07 +020015124table_avl([<table>]) : integer
15125 Returns the total number of available entries in the current proxy's
15126 stick-table or in the designated stick-table. See also table_cnt.
15127
15128table_cnt([<table>]) : integer
15129 Returns the total number of entries currently in use in the current proxy's
15130 stick-table or in the designated stick-table. See also src_conn_cnt and
15131 table_avl for other entry counting methods.
15132
Christopher Faulet34adb2a2017-11-21 21:45:38 +010015133thread : integer
15134 Returns an integer value corresponding to the position of the thread calling
15135 the function, between 0 and (global.nbthread-1). This is useful for logging
15136 and debugging purposes.
15137
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015138var(<var-name>) : undefined
15139 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010015140 sample fetch fails. The name of the variable starts with an indication
15141 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015142 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015143 "sess" : the variable is shared with the whole session
15144 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015145 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010015146 "req" : the variable is shared only during request processing,
15147 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015148 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015149 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015150
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200151517.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020015152----------------------------------
15153
15154The layer 4 usually describes just the transport layer which in haproxy is
15155closest to the connection, where no content is yet made available. The fetch
15156methods described here are usable as low as the "tcp-request connection" rule
15157sets unless they require some future information. Those generally include
15158TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015159the incoming connection. For retrieving a value from a sticky counters, the
15160counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020015161"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
15162used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
15163can be specified as the first integer argument when using the "sc_" prefix.
15164Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
15165table may be specified with the "sc*" form, in which case the currently
15166tracked key will be looked up into this alternate table instead of the table
15167currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015168
Jérôme Magnin35e53a62019-01-16 14:38:37 +010015169bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010015170 Returns the backend connection's HTTP major version encoding, which may be 1
15171 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
15172 encoding and not the version present in the request header.
15173
Willy Tarreau74ca5042013-06-11 23:12:07 +020015174be_id : integer
15175 Returns an integer containing the current backend's id. It can be used in
15176 frontends with responses to check which backend processed the request.
15177
Marcin Deranekd2471c22016-12-12 14:08:05 +010015178be_name : string
15179 Returns a string containing the current backend's name. It can be used in
15180 frontends with responses to check which backend processed the request.
15181
Willy Tarreau74ca5042013-06-11 23:12:07 +020015182dst : ip
15183 This is the destination IPv4 address of the connection on the client side,
15184 which is the address the client connected to. It can be useful when running
15185 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
15186 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010015187 RFC 4291. When the incoming connection passed through address translation or
15188 redirection involving connection tracking, the original destination address
15189 before the redirection will be reported. On Linux systems, the source and
15190 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
15191 is set, because a late response may reopen a timed out connection and switch
15192 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015193
15194dst_conn : integer
15195 Returns an integer value corresponding to the number of currently established
15196 connections on the same socket including the one being evaluated. It is
15197 normally used with ACLs but can as well be used to pass the information to
15198 servers in an HTTP header or in logs. It can be used to either return a sorry
15199 page before hard-blocking, or to use a specific backend to drain new requests
15200 when the socket is considered saturated. This offers the ability to assign
15201 different limits to different listening ports or addresses. See also the
15202 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015203
Willy Tarreau16e01562016-08-09 16:46:18 +020015204dst_is_local : boolean
15205 Returns true if the destination address of the incoming connection is local
15206 to the system, or false if the address doesn't exist on the system, meaning
15207 that it was intercepted in transparent mode. It can be useful to apply
15208 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010015209 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020015210 be delivered only on this address, or SSH access could be locally redirected.
15211 Please note that the check involves a few system calls, so it's better to do
15212 it only once per connection.
15213
Willy Tarreau74ca5042013-06-11 23:12:07 +020015214dst_port : integer
15215 Returns an integer value corresponding to the destination TCP port of the
15216 connection on the client side, which is the port the client connected to.
15217 This might be used when running in transparent mode, when assigning dynamic
15218 ports to some clients for a whole application session, to stick all users to
15219 a same server, or to pass the destination port information to a server using
15220 an HTTP header.
15221
Willy Tarreau60ca10a2017-08-18 15:26:54 +020015222fc_http_major : integer
15223 Reports the front connection's HTTP major version encoding, which may be 1
15224 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
15225 encoding and not on the version present in the request header.
15226
Emeric Brun4f603012017-01-05 15:11:44 +010015227fc_rcvd_proxy : boolean
15228 Returns true if the client initiated the connection with a PROXY protocol
15229 header.
15230
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020015231fc_rtt(<unit>) : integer
15232 Returns the Round Trip Time (RTT) measured by the kernel for the client
15233 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
15234 can be set to "ms" for milliseconds or "us" for microseconds. If the server
15235 connection is not established, if the connection is not TCP or if the
15236 operating system does not support TCP_INFO, for example Linux kernels before
15237 2.4, the sample fetch fails.
15238
15239fc_rttvar(<unit>) : integer
15240 Returns the Round Trip Time (RTT) variance measured by the kernel for the
15241 client connection. <unit> is facultative, by default the unit is milliseconds.
15242 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
15243 server connection is not established, if the connection is not TCP or if the
15244 operating system does not support TCP_INFO, for example Linux kernels before
15245 2.4, the sample fetch fails.
15246
Christopher Faulet297df182019-10-17 14:40:48 +020015247fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015248 Returns the unacked counter measured by the kernel for the client connection.
15249 If the server connection is not established, if the connection is not TCP or
15250 if the operating system does not support TCP_INFO, for example Linux kernels
15251 before 2.4, the sample fetch fails.
15252
Christopher Faulet297df182019-10-17 14:40:48 +020015253fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015254 Returns the sacked counter measured by the kernel for the client connection.
15255 If the server connection is not established, if the connection is not TCP or
15256 if the operating system does not support TCP_INFO, for example Linux kernels
15257 before 2.4, the sample fetch fails.
15258
Christopher Faulet297df182019-10-17 14:40:48 +020015259fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015260 Returns the retransmits counter measured by the kernel for the client
15261 connection. If the server connection is not established, if the connection is
15262 not TCP or if the operating system does not support TCP_INFO, for example
15263 Linux kernels before 2.4, the sample fetch fails.
15264
Christopher Faulet297df182019-10-17 14:40:48 +020015265fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015266 Returns the fack counter measured by the kernel for the client
15267 connection. If the server connection is not established, if the connection is
15268 not TCP or if the operating system does not support TCP_INFO, for example
15269 Linux kernels before 2.4, the sample fetch fails.
15270
Christopher Faulet297df182019-10-17 14:40:48 +020015271fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015272 Returns the lost counter measured by the kernel for the client
15273 connection. If the server connection is not established, if the connection is
15274 not TCP or if the operating system does not support TCP_INFO, for example
15275 Linux kernels before 2.4, the sample fetch fails.
15276
Christopher Faulet297df182019-10-17 14:40:48 +020015277fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015278 Returns the reordering counter measured by the kernel for the client
15279 connection. If the server connection is not established, if the connection is
15280 not TCP or if the operating system does not support TCP_INFO, for example
15281 Linux kernels before 2.4, the sample fetch fails.
15282
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020015283fe_defbe : string
15284 Returns a string containing the frontend's default backend name. It can be
15285 used in frontends to check which backend will handle requests by default.
15286
Willy Tarreau74ca5042013-06-11 23:12:07 +020015287fe_id : integer
15288 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010015289 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020015290 coming via a same frontend to the same server.
15291
Marcin Deranekd2471c22016-12-12 14:08:05 +010015292fe_name : string
15293 Returns a string containing the current frontend's name. It can be used in
15294 backends to check from which frontend it was called, or to stick all users
15295 coming via a same frontend to the same server.
15296
Cyril Bonté62ba8702014-04-22 23:52:25 +020015297sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015298sc0_bytes_in_rate([<table>]) : integer
15299sc1_bytes_in_rate([<table>]) : integer
15300sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015301 Returns the average client-to-server bytes rate from the currently tracked
15302 counters, measured in amount of bytes over the period configured in the
15303 table. See also src_bytes_in_rate.
15304
Cyril Bonté62ba8702014-04-22 23:52:25 +020015305sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015306sc0_bytes_out_rate([<table>]) : integer
15307sc1_bytes_out_rate([<table>]) : integer
15308sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015309 Returns the average server-to-client bytes rate from the currently tracked
15310 counters, measured in amount of bytes over the period configured in the
15311 table. See also src_bytes_out_rate.
15312
Cyril Bonté62ba8702014-04-22 23:52:25 +020015313sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015314sc0_clr_gpc0([<table>]) : integer
15315sc1_clr_gpc0([<table>]) : integer
15316sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020015317 Clears the first General Purpose Counter associated to the currently tracked
15318 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010015319 stored value is zero, so first invocation will always return zero. This is
15320 typically used as a second ACL in an expression in order to mark a connection
15321 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020015322
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015323 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020015324 # block if 5 consecutive requests continue to come faster than 10 sess
15325 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015326 acl abuse sc0_http_req_rate gt 10
15327 acl kill sc0_inc_gpc0 gt 5
15328 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020015329 tcp-request connection accept if !abuse save
15330 tcp-request connection reject if abuse kill
15331
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015332sc_clr_gpc1(<ctr>[,<table>]) : integer
15333sc0_clr_gpc1([<table>]) : integer
15334sc1_clr_gpc1([<table>]) : integer
15335sc2_clr_gpc1([<table>]) : integer
15336 Clears the second General Purpose Counter associated to the currently tracked
15337 counters, and returns its previous value. Before the first invocation, the
15338 stored value is zero, so first invocation will always return zero. This is
15339 typically used as a second ACL in an expression in order to mark a connection
15340 when a first ACL was verified.
15341
Cyril Bonté62ba8702014-04-22 23:52:25 +020015342sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015343sc0_conn_cnt([<table>]) : integer
15344sc1_conn_cnt([<table>]) : integer
15345sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015346 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020015347 counters. See also src_conn_cnt.
15348
Cyril Bonté62ba8702014-04-22 23:52:25 +020015349sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015350sc0_conn_cur([<table>]) : integer
15351sc1_conn_cur([<table>]) : integer
15352sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015353 Returns the current amount of concurrent connections tracking the same
15354 tracked counters. This number is automatically incremented when tracking
15355 begins and decremented when tracking stops. See also src_conn_cur.
15356
Cyril Bonté62ba8702014-04-22 23:52:25 +020015357sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015358sc0_conn_rate([<table>]) : integer
15359sc1_conn_rate([<table>]) : integer
15360sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015361 Returns the average connection rate from the currently tracked counters,
15362 measured in amount of connections over the period configured in the table.
15363 See also src_conn_rate.
15364
Cyril Bonté62ba8702014-04-22 23:52:25 +020015365sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015366sc0_get_gpc0([<table>]) : integer
15367sc1_get_gpc0([<table>]) : integer
15368sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015369 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015370 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015371
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015372sc_get_gpc1(<ctr>[,<table>]) : integer
15373sc0_get_gpc1([<table>]) : integer
15374sc1_get_gpc1([<table>]) : integer
15375sc2_get_gpc1([<table>]) : integer
15376 Returns the value of the second General Purpose Counter associated to the
15377 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
15378
Thierry FOURNIER236657b2015-08-19 08:25:14 +020015379sc_get_gpt0(<ctr>[,<table>]) : integer
15380sc0_get_gpt0([<table>]) : integer
15381sc1_get_gpt0([<table>]) : integer
15382sc2_get_gpt0([<table>]) : integer
15383 Returns the value of the first General Purpose Tag associated to the
15384 currently tracked counters. See also src_get_gpt0.
15385
Cyril Bonté62ba8702014-04-22 23:52:25 +020015386sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015387sc0_gpc0_rate([<table>]) : integer
15388sc1_gpc0_rate([<table>]) : integer
15389sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015390 Returns the average increment rate of the first General Purpose Counter
15391 associated to the currently tracked counters. It reports the frequency
15392 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015393 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
15394 that the "gpc0_rate" counter must be stored in the stick-table for a value to
15395 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020015396
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015397sc_gpc1_rate(<ctr>[,<table>]) : integer
15398sc0_gpc1_rate([<table>]) : integer
15399sc1_gpc1_rate([<table>]) : integer
15400sc2_gpc1_rate([<table>]) : integer
15401 Returns the average increment rate of the second General Purpose Counter
15402 associated to the currently tracked counters. It reports the frequency
15403 which the gpc1 counter was incremented over the configured period. See also
15404 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
15405 that the "gpc1_rate" counter must be stored in the stick-table for a value to
15406 be returned, as "gpc1" only holds the event count.
15407
Cyril Bonté62ba8702014-04-22 23:52:25 +020015408sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015409sc0_http_err_cnt([<table>]) : integer
15410sc1_http_err_cnt([<table>]) : integer
15411sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015412 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020015413 counters. This includes the both request errors and 4xx error responses.
15414 See also src_http_err_cnt.
15415
Cyril Bonté62ba8702014-04-22 23:52:25 +020015416sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015417sc0_http_err_rate([<table>]) : integer
15418sc1_http_err_rate([<table>]) : integer
15419sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015420 Returns the average rate of HTTP errors from the currently tracked counters,
15421 measured in amount of errors over the period configured in the table. This
15422 includes the both request errors and 4xx error responses. See also
15423 src_http_err_rate.
15424
Cyril Bonté62ba8702014-04-22 23:52:25 +020015425sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015426sc0_http_req_cnt([<table>]) : integer
15427sc1_http_req_cnt([<table>]) : integer
15428sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015429 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020015430 counters. This includes every started request, valid or not. See also
15431 src_http_req_cnt.
15432
Cyril Bonté62ba8702014-04-22 23:52:25 +020015433sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015434sc0_http_req_rate([<table>]) : integer
15435sc1_http_req_rate([<table>]) : integer
15436sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015437 Returns the average rate of HTTP requests from the currently tracked
15438 counters, measured in amount of requests over the period configured in
15439 the table. This includes every started request, valid or not. See also
15440 src_http_req_rate.
15441
Cyril Bonté62ba8702014-04-22 23:52:25 +020015442sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015443sc0_inc_gpc0([<table>]) : integer
15444sc1_inc_gpc0([<table>]) : integer
15445sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015446 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010015447 tracked counters, and returns its new value. Before the first invocation,
15448 the stored value is zero, so first invocation will increase it to 1 and will
15449 return 1. This is typically used as a second ACL in an expression in order
15450 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020015451
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015452 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015453 acl abuse sc0_http_req_rate gt 10
15454 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020015455 tcp-request connection reject if abuse kill
15456
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015457sc_inc_gpc1(<ctr>[,<table>]) : integer
15458sc0_inc_gpc1([<table>]) : integer
15459sc1_inc_gpc1([<table>]) : integer
15460sc2_inc_gpc1([<table>]) : integer
15461 Increments the second General Purpose Counter associated to the currently
15462 tracked counters, and returns its new value. Before the first invocation,
15463 the stored value is zero, so first invocation will increase it to 1 and will
15464 return 1. This is typically used as a second ACL in an expression in order
15465 to mark a connection when a first ACL was verified.
15466
Cyril Bonté62ba8702014-04-22 23:52:25 +020015467sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015468sc0_kbytes_in([<table>]) : integer
15469sc1_kbytes_in([<table>]) : integer
15470sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015471 Returns the total amount of client-to-server data from the currently tracked
15472 counters, measured in kilobytes. The test is currently performed on 32-bit
15473 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020015474
Cyril Bonté62ba8702014-04-22 23:52:25 +020015475sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015476sc0_kbytes_out([<table>]) : integer
15477sc1_kbytes_out([<table>]) : integer
15478sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015479 Returns the total amount of server-to-client data from the currently tracked
15480 counters, measured in kilobytes. The test is currently performed on 32-bit
15481 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020015482
Cyril Bonté62ba8702014-04-22 23:52:25 +020015483sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015484sc0_sess_cnt([<table>]) : integer
15485sc1_sess_cnt([<table>]) : integer
15486sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015487 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020015488 into sessions, which means that they were accepted by a "tcp-request
15489 connection" rule, from the currently tracked counters. A backend may count
15490 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040015491 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020015492 with the client. See also src_sess_cnt.
15493
Cyril Bonté62ba8702014-04-22 23:52:25 +020015494sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015495sc0_sess_rate([<table>]) : integer
15496sc1_sess_rate([<table>]) : integer
15497sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015498 Returns the average session rate from the currently tracked counters,
15499 measured in amount of sessions over the period configured in the table. A
15500 session is a connection that got past the early "tcp-request connection"
15501 rules. A backend may count more sessions than connections because each
15502 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040015503 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020015504
Cyril Bonté62ba8702014-04-22 23:52:25 +020015505sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020015506sc0_tracked([<table>]) : boolean
15507sc1_tracked([<table>]) : boolean
15508sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020015509 Returns true if the designated session counter is currently being tracked by
15510 the current session. This can be useful when deciding whether or not we want
15511 to set some values in a header passed to the server.
15512
Cyril Bonté62ba8702014-04-22 23:52:25 +020015513sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015514sc0_trackers([<table>]) : integer
15515sc1_trackers([<table>]) : integer
15516sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010015517 Returns the current amount of concurrent connections tracking the same
15518 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015519 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010015520 that it does not rely on any stored information but on the table's reference
15521 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015522 may sometimes be more suited for layer7 tracking. It can be used to tell a
15523 server how many concurrent connections there are from a given address for
15524 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010015525
Willy Tarreau74ca5042013-06-11 23:12:07 +020015526so_id : integer
15527 Returns an integer containing the current listening socket's id. It is useful
15528 in frontends involving many "bind" lines, or to stick all users coming via a
15529 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015530
Jerome Magnin28b90332020-03-27 22:08:40 +010015531so_name : string
15532 Returns a string containing the current listening socket's name, as defined
15533 with name on a "bind" line. It can serve the same purposes as so_id but with
15534 strings instead of integers.
15535
Willy Tarreau74ca5042013-06-11 23:12:07 +020015536src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010015537 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020015538 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
15539 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
15540 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010015541 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
15542 directive is used, it can be the address of a client behind another
15543 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010015544 "tcp-request connection" which sees the real address. When the incoming
15545 connection passed through address translation or redirection involving
15546 connection tracking, the original destination address before the redirection
15547 will be reported. On Linux systems, the source and destination may seldom
15548 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
15549 response may reopen a timed out connection and switch what is believed to be
15550 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015551
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015552 Example:
15553 # add an HTTP header in requests with the originating address' country
15554 http-request set-header X-Country %[src,map_ip(geoip.lst)]
15555
Willy Tarreau74ca5042013-06-11 23:12:07 +020015556src_bytes_in_rate([<table>]) : integer
15557 Returns the average bytes rate from the incoming connection's source address
15558 in the current proxy's stick-table or in the designated stick-table, measured
15559 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015560 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015561
Willy Tarreau74ca5042013-06-11 23:12:07 +020015562src_bytes_out_rate([<table>]) : integer
15563 Returns the average bytes rate to the incoming connection's source address in
15564 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015565 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015566 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015567
Willy Tarreau74ca5042013-06-11 23:12:07 +020015568src_clr_gpc0([<table>]) : integer
15569 Clears the first General Purpose Counter associated to the incoming
15570 connection's source address in the current proxy's stick-table or in the
15571 designated stick-table, and returns its previous value. If the address is not
15572 found, an entry is created and 0 is returned. This is typically used as a
15573 second ACL in an expression in order to mark a connection when a first ACL
15574 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020015575
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015576 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020015577 # block if 5 consecutive requests continue to come faster than 10 sess
15578 # per second, and reset the counter as soon as the traffic slows down.
15579 acl abuse src_http_req_rate gt 10
15580 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010015581 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020015582 tcp-request connection accept if !abuse save
15583 tcp-request connection reject if abuse kill
15584
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015585src_clr_gpc1([<table>]) : integer
15586 Clears the second General Purpose Counter associated to the incoming
15587 connection's source address in the current proxy's stick-table or in the
15588 designated stick-table, and returns its previous value. If the address is not
15589 found, an entry is created and 0 is returned. This is typically used as a
15590 second ACL in an expression in order to mark a connection when a first ACL
15591 was verified.
15592
Willy Tarreau74ca5042013-06-11 23:12:07 +020015593src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015594 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020015595 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015596 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015597 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015598
Willy Tarreau74ca5042013-06-11 23:12:07 +020015599src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020015600 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015601 current incoming connection's source address in the current proxy's
15602 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015603 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015604
Willy Tarreau74ca5042013-06-11 23:12:07 +020015605src_conn_rate([<table>]) : integer
15606 Returns the average connection rate from the incoming connection's source
15607 address in the current proxy's stick-table or in the designated stick-table,
15608 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015609 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015610
Willy Tarreau74ca5042013-06-11 23:12:07 +020015611src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020015612 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015613 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015614 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015615 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015616
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015617src_get_gpc1([<table>]) : integer
15618 Returns the value of the second General Purpose Counter associated to the
15619 incoming connection's source address in the current proxy's stick-table or in
15620 the designated stick-table. If the address is not found, zero is returned.
15621 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
15622
Thierry FOURNIER236657b2015-08-19 08:25:14 +020015623src_get_gpt0([<table>]) : integer
15624 Returns the value of the first General Purpose Tag associated to the
15625 incoming connection's source address in the current proxy's stick-table or in
15626 the designated stick-table. If the address is not found, zero is returned.
15627 See also sc/sc0/sc1/sc2_get_gpt0.
15628
Willy Tarreau74ca5042013-06-11 23:12:07 +020015629src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015630 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020015631 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015632 stick-table or in the designated stick-table. It reports the frequency
15633 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015634 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
15635 that the "gpc0_rate" counter must be stored in the stick-table for a value to
15636 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015637
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015638src_gpc1_rate([<table>]) : integer
15639 Returns the average increment rate of the second General Purpose Counter
15640 associated to the incoming connection's source address in the current proxy's
15641 stick-table or in the designated stick-table. It reports the frequency
15642 which the gpc1 counter was incremented over the configured period. See also
15643 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
15644 that the "gpc1_rate" counter must be stored in the stick-table for a value to
15645 be returned, as "gpc1" only holds the event count.
15646
Willy Tarreau74ca5042013-06-11 23:12:07 +020015647src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015648 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020015649 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020015650 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015651 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020015652 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015653
Willy Tarreau74ca5042013-06-11 23:12:07 +020015654src_http_err_rate([<table>]) : integer
15655 Returns the average rate of HTTP errors from the incoming connection's source
15656 address in the current proxy's stick-table or in the designated stick-table,
15657 measured in amount of errors over the period configured in the table. This
15658 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015659 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015660
Willy Tarreau74ca5042013-06-11 23:12:07 +020015661src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015662 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020015663 source address in the current proxy's stick-table or in the designated stick-
15664 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015665 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015666
Willy Tarreau74ca5042013-06-11 23:12:07 +020015667src_http_req_rate([<table>]) : integer
15668 Returns the average rate of HTTP requests from the incoming connection's
15669 source address in the current proxy's stick-table or in the designated stick-
15670 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020015671 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015672 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015673
Willy Tarreau74ca5042013-06-11 23:12:07 +020015674src_inc_gpc0([<table>]) : integer
15675 Increments the first General Purpose Counter associated to the incoming
15676 connection's source address in the current proxy's stick-table or in the
15677 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015678 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015679 This is typically used as a second ACL in an expression in order to mark a
15680 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020015681
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015682 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020015683 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010015684 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020015685 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020015686
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015687src_inc_gpc1([<table>]) : integer
15688 Increments the second General Purpose Counter associated to the incoming
15689 connection's source address in the current proxy's stick-table or in the
15690 designated stick-table, and returns its new value. If the address is not
15691 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
15692 This is typically used as a second ACL in an expression in order to mark a
15693 connection when a first ACL was verified.
15694
Willy Tarreau16e01562016-08-09 16:46:18 +020015695src_is_local : boolean
15696 Returns true if the source address of the incoming connection is local to the
15697 system, or false if the address doesn't exist on the system, meaning that it
15698 comes from a remote machine. Note that UNIX addresses are considered local.
15699 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015700 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020015701 note that the check involves a few system calls, so it's better to do it only
15702 once per connection.
15703
Willy Tarreau74ca5042013-06-11 23:12:07 +020015704src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015705 Returns the total amount of data received from the incoming connection's
15706 source address in the current proxy's stick-table or in the designated
15707 stick-table, measured in kilobytes. If the address is not found, zero is
15708 returned. The test is currently performed on 32-bit integers, which limits
15709 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015710
Willy Tarreau74ca5042013-06-11 23:12:07 +020015711src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015712 Returns the total amount of data sent to the incoming connection's source
15713 address in the current proxy's stick-table or in the designated stick-table,
15714 measured in kilobytes. If the address is not found, zero is returned. The
15715 test is currently performed on 32-bit integers, which limits values to 4
15716 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015717
Willy Tarreau74ca5042013-06-11 23:12:07 +020015718src_port : integer
15719 Returns an integer value corresponding to the TCP source port of the
15720 connection on the client side, which is the port the client connected from.
15721 Usage of this function is very limited as modern protocols do not care much
15722 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010015723
Willy Tarreau74ca5042013-06-11 23:12:07 +020015724src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015725 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020015726 connection's source IPv4 address in the current proxy's stick-table or in the
15727 designated stick-table, that were transformed into sessions, which means that
15728 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015729 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015730
Willy Tarreau74ca5042013-06-11 23:12:07 +020015731src_sess_rate([<table>]) : integer
15732 Returns the average session rate from the incoming connection's source
15733 address in the current proxy's stick-table or in the designated stick-table,
15734 measured in amount of sessions over the period configured in the table. A
15735 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015736 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015737
Willy Tarreau74ca5042013-06-11 23:12:07 +020015738src_updt_conn_cnt([<table>]) : integer
15739 Creates or updates the entry associated to the incoming connection's source
15740 address in the current proxy's stick-table or in the designated stick-table.
15741 This table must be configured to store the "conn_cnt" data type, otherwise
15742 the match will be ignored. The current count is incremented by one, and the
15743 expiration timer refreshed. The updated count is returned, so this match
15744 can't return zero. This was used to reject service abusers based on their
15745 source address. Note: it is recommended to use the more complete "track-sc*"
15746 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015747
15748 Example :
15749 # This frontend limits incoming SSH connections to 3 per 10 second for
15750 # each source address, and rejects excess connections until a 10 second
15751 # silence is observed. At most 20 addresses are tracked.
15752 listen ssh
15753 bind :22
15754 mode tcp
15755 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020015756 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020015757 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015758 server local 127.0.0.1:22
15759
Willy Tarreau74ca5042013-06-11 23:12:07 +020015760srv_id : integer
15761 Returns an integer containing the server's id when processing the response.
15762 While it's almost only used with ACLs, it may be used for logging or
15763 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020015764
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200157657.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020015766----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020015767
Willy Tarreau74ca5042013-06-11 23:12:07 +020015768The layer 5 usually describes just the session layer which in haproxy is
15769closest to the session once all the connection handshakes are finished, but
15770when no content is yet made available. The fetch methods described here are
15771usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015772future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020015773
Ben Shillitof25e8e52016-12-02 14:25:37 +00001577451d.all(<prop>[,<prop>*]) : string
15775 Returns values for the properties requested as a string, where values are
15776 separated by the delimiter specified with "51degrees-property-separator".
15777 The device is identified using all the important HTTP headers from the
15778 request. The function can be passed up to five property names, and if a
15779 property name can't be found, the value "NoData" is returned.
15780
15781 Example :
15782 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
15783 # containing the three properties requested using all relevant headers from
15784 # the request.
15785 frontend http-in
15786 bind *:8081
15787 default_backend servers
15788 http-request set-header X-51D-DeviceTypeMobileTablet \
15789 %[51d.all(DeviceType,IsMobile,IsTablet)]
15790
Emeric Brun645ae792014-04-30 14:21:06 +020015791ssl_bc : boolean
15792 Returns true when the back connection was made via an SSL/TLS transport
15793 layer and is locally deciphered. This means the outgoing connection was made
15794 other a server with the "ssl" option.
15795
15796ssl_bc_alg_keysize : integer
15797 Returns the symmetric cipher key size supported in bits when the outgoing
15798 connection was made over an SSL/TLS transport layer.
15799
Olivier Houchard6b77f492018-11-22 18:18:29 +010015800ssl_bc_alpn : string
15801 This extracts the Application Layer Protocol Negotiation field from an
15802 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020015803 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010015804 server. The SSL library must have been built with support for TLS
15805 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15806 not advertised unless the "alpn" keyword on the "server" line specifies a
15807 protocol list. Also, nothing forces the server to pick a protocol from this
15808 list, any other one may be requested. The TLS ALPN extension is meant to
15809 replace the TLS NPN extension. See also "ssl_bc_npn".
15810
Emeric Brun645ae792014-04-30 14:21:06 +020015811ssl_bc_cipher : string
15812 Returns the name of the used cipher when the outgoing connection was made
15813 over an SSL/TLS transport layer.
15814
Patrick Hemmer65674662019-06-04 08:13:03 -040015815ssl_bc_client_random : binary
15816 Returns the client random of the back connection when the incoming connection
15817 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15818 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15819
Emeric Brun74f7ffa2018-02-19 16:14:12 +010015820ssl_bc_is_resumed : boolean
15821 Returns true when the back connection was made over an SSL/TLS transport
15822 layer and the newly created SSL session was resumed using a cached
15823 session or a TLS ticket.
15824
Olivier Houchard6b77f492018-11-22 18:18:29 +010015825ssl_bc_npn : string
15826 This extracts the Next Protocol Negotiation field from an outgoing connection
15827 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020015828 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010015829 built with support for TLS extensions enabled (check haproxy -vv). Note that
15830 the TLS NPN extension is not advertised unless the "npn" keyword on the
15831 "server" line specifies a protocol list. Also, nothing forces the server to
15832 pick a protocol from this list, any other one may be used. Please note that
15833 the TLS NPN extension was replaced with ALPN.
15834
Emeric Brun645ae792014-04-30 14:21:06 +020015835ssl_bc_protocol : string
15836 Returns the name of the used protocol when the outgoing connection was made
15837 over an SSL/TLS transport layer.
15838
Emeric Brunb73a9b02014-04-30 18:49:19 +020015839ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020015840 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015841 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15842 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020015843
Patrick Hemmer65674662019-06-04 08:13:03 -040015844ssl_bc_server_random : binary
15845 Returns the server random of the back connection when the incoming connection
15846 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15847 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15848
Emeric Brun645ae792014-04-30 14:21:06 +020015849ssl_bc_session_id : binary
15850 Returns the SSL ID of the back connection when the outgoing connection was
15851 made over an SSL/TLS transport layer. It is useful to log if we want to know
15852 if session was reused or not.
15853
Patrick Hemmere0275472018-04-28 19:15:51 -040015854ssl_bc_session_key : binary
15855 Returns the SSL session master key of the back connection when the outgoing
15856 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15857 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15858 BoringSSL.
15859
Emeric Brun645ae792014-04-30 14:21:06 +020015860ssl_bc_use_keysize : integer
15861 Returns the symmetric cipher key size used in bits when the outgoing
15862 connection was made over an SSL/TLS transport layer.
15863
Willy Tarreau74ca5042013-06-11 23:12:07 +020015864ssl_c_ca_err : integer
15865 When the incoming connection was made over an SSL/TLS transport layer,
15866 returns the ID of the first error detected during verification of the client
15867 certificate at depth > 0, or 0 if no error was encountered during this
15868 verification process. Please refer to your SSL library's documentation to
15869 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020015870
Willy Tarreau74ca5042013-06-11 23:12:07 +020015871ssl_c_ca_err_depth : integer
15872 When the incoming connection was made over an SSL/TLS transport layer,
15873 returns the depth in the CA chain of the first error detected during the
15874 verification of the client certificate. If no error is encountered, 0 is
15875 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015876
Emeric Brun43e79582014-10-29 19:03:26 +010015877ssl_c_der : binary
15878 Returns the DER formatted certificate presented by the client when the
15879 incoming connection was made over an SSL/TLS transport layer. When used for
15880 an ACL, the value(s) to match against can be passed in hexadecimal form.
15881
Willy Tarreau74ca5042013-06-11 23:12:07 +020015882ssl_c_err : integer
15883 When the incoming connection was made over an SSL/TLS transport layer,
15884 returns the ID of the first error detected during verification at depth 0, or
15885 0 if no error was encountered during this verification process. Please refer
15886 to your SSL library's documentation to find the exhaustive list of error
15887 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020015888
Willy Tarreau74ca5042013-06-11 23:12:07 +020015889ssl_c_i_dn([<entry>[,<occ>]]) : string
15890 When the incoming connection was made over an SSL/TLS transport layer,
15891 returns the full distinguished name of the issuer of the certificate
15892 presented by the client when no <entry> is specified, or the value of the
15893 first given entry found from the beginning of the DN. If a positive/negative
15894 occurrence number is specified as the optional second argument, it returns
15895 the value of the nth given entry value from the beginning/end of the DN.
15896 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
15897 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020015898
Willy Tarreau74ca5042013-06-11 23:12:07 +020015899ssl_c_key_alg : string
15900 Returns the name of the algorithm used to generate the key of the certificate
15901 presented by the client when the incoming connection was made over an SSL/TLS
15902 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020015903
Willy Tarreau74ca5042013-06-11 23:12:07 +020015904ssl_c_notafter : string
15905 Returns the end date presented by the client as a formatted string
15906 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15907 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020015908
Willy Tarreau74ca5042013-06-11 23:12:07 +020015909ssl_c_notbefore : string
15910 Returns the start date presented by the client as a formatted string
15911 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15912 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015913
Willy Tarreau74ca5042013-06-11 23:12:07 +020015914ssl_c_s_dn([<entry>[,<occ>]]) : string
15915 When the incoming connection was made over an SSL/TLS transport layer,
15916 returns the full distinguished name of the subject of the certificate
15917 presented by the client when no <entry> is specified, or the value of the
15918 first given entry found from the beginning of the DN. If a positive/negative
15919 occurrence number is specified as the optional second argument, it returns
15920 the value of the nth given entry value from the beginning/end of the DN.
15921 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
15922 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015923
Willy Tarreau74ca5042013-06-11 23:12:07 +020015924ssl_c_serial : binary
15925 Returns the serial of the certificate presented by the client when the
15926 incoming connection was made over an SSL/TLS transport layer. When used for
15927 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015928
Willy Tarreau74ca5042013-06-11 23:12:07 +020015929ssl_c_sha1 : binary
15930 Returns the SHA-1 fingerprint of the certificate presented by the client when
15931 the incoming connection was made over an SSL/TLS transport layer. This can be
15932 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015933 Note that the output is binary, so if you want to pass that signature to the
15934 server, you need to encode it in hex or base64, such as in the example below:
15935
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015936 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015937 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020015938
Willy Tarreau74ca5042013-06-11 23:12:07 +020015939ssl_c_sig_alg : string
15940 Returns the name of the algorithm used to sign the certificate presented by
15941 the client when the incoming connection was made over an SSL/TLS transport
15942 layer.
Emeric Brun87855892012-10-17 17:39:35 +020015943
Willy Tarreau74ca5042013-06-11 23:12:07 +020015944ssl_c_used : boolean
15945 Returns true if current SSL session uses a client certificate even if current
15946 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020015947
Willy Tarreau74ca5042013-06-11 23:12:07 +020015948ssl_c_verify : integer
15949 Returns the verify result error ID when the incoming connection was made over
15950 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
15951 refer to your SSL library's documentation for an exhaustive list of error
15952 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020015953
Willy Tarreau74ca5042013-06-11 23:12:07 +020015954ssl_c_version : integer
15955 Returns the version of the certificate presented by the client when the
15956 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020015957
Emeric Brun43e79582014-10-29 19:03:26 +010015958ssl_f_der : binary
15959 Returns the DER formatted certificate presented by the frontend when the
15960 incoming connection was made over an SSL/TLS transport layer. When used for
15961 an ACL, the value(s) to match against can be passed in hexadecimal form.
15962
Willy Tarreau74ca5042013-06-11 23:12:07 +020015963ssl_f_i_dn([<entry>[,<occ>]]) : string
15964 When the incoming connection was made over an SSL/TLS transport layer,
15965 returns the full distinguished name of the issuer of the certificate
15966 presented by the frontend when no <entry> is specified, or the value of the
15967 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020015968 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020015969 the value of the nth given entry value from the beginning/end of the DN.
15970 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
15971 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020015972
Willy Tarreau74ca5042013-06-11 23:12:07 +020015973ssl_f_key_alg : string
15974 Returns the name of the algorithm used to generate the key of the certificate
15975 presented by the frontend when the incoming connection was made over an
15976 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015977
Willy Tarreau74ca5042013-06-11 23:12:07 +020015978ssl_f_notafter : string
15979 Returns the end date presented by the frontend as a formatted string
15980 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15981 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015982
Willy Tarreau74ca5042013-06-11 23:12:07 +020015983ssl_f_notbefore : string
15984 Returns the start date presented by the frontend as a formatted string
15985 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15986 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020015987
Willy Tarreau74ca5042013-06-11 23:12:07 +020015988ssl_f_s_dn([<entry>[,<occ>]]) : string
15989 When the incoming connection was made over an SSL/TLS transport layer,
15990 returns the full distinguished name of the subject of the certificate
15991 presented by the frontend when no <entry> is specified, or the value of the
15992 first given entry found from the beginning of the DN. If a positive/negative
15993 occurrence number is specified as the optional second argument, it returns
15994 the value of the nth given entry value from the beginning/end of the DN.
15995 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
15996 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020015997
Willy Tarreau74ca5042013-06-11 23:12:07 +020015998ssl_f_serial : binary
15999 Returns the serial of the certificate presented by the frontend when the
16000 incoming connection was made over an SSL/TLS transport layer. When used for
16001 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020016002
Emeric Brun55f4fa82014-04-30 17:11:25 +020016003ssl_f_sha1 : binary
16004 Returns the SHA-1 fingerprint of the certificate presented by the frontend
16005 when the incoming connection was made over an SSL/TLS transport layer. This
16006 can be used to know which certificate was chosen using SNI.
16007
Willy Tarreau74ca5042013-06-11 23:12:07 +020016008ssl_f_sig_alg : string
16009 Returns the name of the algorithm used to sign the certificate presented by
16010 the frontend when the incoming connection was made over an SSL/TLS transport
16011 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020016012
Willy Tarreau74ca5042013-06-11 23:12:07 +020016013ssl_f_version : integer
16014 Returns the version of the certificate presented by the frontend when the
16015 incoming connection was made over an SSL/TLS transport layer.
16016
16017ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020016018 Returns true when the front connection was made via an SSL/TLS transport
16019 layer and is locally deciphered. This means it has matched a socket declared
16020 with a "bind" line having the "ssl" option.
16021
Willy Tarreau74ca5042013-06-11 23:12:07 +020016022 Example :
16023 # This passes "X-Proto: https" to servers when client connects over SSL
16024 listen http-https
16025 bind :80
16026 bind :443 ssl crt /etc/haproxy.pem
16027 http-request add-header X-Proto https if { ssl_fc }
16028
16029ssl_fc_alg_keysize : integer
16030 Returns the symmetric cipher key size supported in bits when the incoming
16031 connection was made over an SSL/TLS transport layer.
16032
16033ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016034 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020016035 incoming connection made via a TLS transport layer and locally deciphered by
16036 haproxy. The result is a string containing the protocol name advertised by
16037 the client. The SSL library must have been built with support for TLS
16038 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
16039 not advertised unless the "alpn" keyword on the "bind" line specifies a
16040 protocol list. Also, nothing forces the client to pick a protocol from this
16041 list, any other one may be requested. The TLS ALPN extension is meant to
16042 replace the TLS NPN extension. See also "ssl_fc_npn".
16043
Willy Tarreau74ca5042013-06-11 23:12:07 +020016044ssl_fc_cipher : string
16045 Returns the name of the used cipher when the incoming connection was made
16046 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020016047
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010016048ssl_fc_cipherlist_bin : binary
16049 Returns the binary form of the client hello cipher list. The maximum returned
16050 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010016051 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010016052
16053ssl_fc_cipherlist_hex : string
16054 Returns the binary form of the client hello cipher list encoded as
16055 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010016056 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010016057
16058ssl_fc_cipherlist_str : string
16059 Returns the decoded text form of the client hello cipher list. The maximum
16060 number of ciphers returned is according with the value of
16061 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010016062 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020016063 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010016064
16065ssl_fc_cipherlist_xxh : integer
16066 Returns a xxh64 of the cipher list. This hash can be return only is the value
16067 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010016068 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010016069
Patrick Hemmer65674662019-06-04 08:13:03 -040016070ssl_fc_client_random : binary
16071 Returns the client random of the front connection when the incoming connection
16072 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
16073 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
16074
Willy Tarreau74ca5042013-06-11 23:12:07 +020016075ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020016076 Returns true if a client certificate is present in an incoming connection over
16077 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010016078 Note: on SSL session resumption with Session ID or TLS ticket, client
16079 certificate is not present in the current connection but may be retrieved
16080 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
16081 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020016082
Olivier Houchardccaa7de2017-10-02 11:51:03 +020016083ssl_fc_has_early : boolean
16084 Returns true if early data were sent, and the handshake didn't happen yet. As
16085 it has security implications, it is useful to be able to refuse those, or
16086 wait until the handshake happened.
16087
Willy Tarreau74ca5042013-06-11 23:12:07 +020016088ssl_fc_has_sni : boolean
16089 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020016090 in an incoming connection was made over an SSL/TLS transport layer. Returns
16091 true when the incoming connection presents a TLS SNI field. This requires
John Roesler27c754c2019-07-10 15:45:51 -050016092 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020016093 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020016094
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020016095ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020016096 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010016097 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
16098 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020016099
Willy Tarreau74ca5042013-06-11 23:12:07 +020016100ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016101 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020016102 made via a TLS transport layer and locally deciphered by haproxy. The result
16103 is a string containing the protocol name advertised by the client. The SSL
16104 library must have been built with support for TLS extensions enabled (check
16105 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
16106 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
16107 forces the client to pick a protocol from this list, any other one may be
16108 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020016109
Willy Tarreau74ca5042013-06-11 23:12:07 +020016110ssl_fc_protocol : string
16111 Returns the name of the used protocol when the incoming connection was made
16112 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020016113
Emeric Brunb73a9b02014-04-30 18:49:19 +020016114ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040016115 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020016116 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
16117 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040016118
Patrick Hemmer65674662019-06-04 08:13:03 -040016119ssl_fc_server_random : binary
16120 Returns the server random of the front connection when the incoming connection
16121 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
16122 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
16123
Willy Tarreau74ca5042013-06-11 23:12:07 +020016124ssl_fc_session_id : binary
16125 Returns the SSL ID of the front connection when the incoming connection was
16126 made over an SSL/TLS transport layer. It is useful to stick a given client to
16127 a server. It is important to note that some browsers refresh their session ID
16128 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020016129
Patrick Hemmere0275472018-04-28 19:15:51 -040016130ssl_fc_session_key : binary
16131 Returns the SSL session master key of the front connection when the incoming
16132 connection was made over an SSL/TLS transport layer. It is useful to decrypt
16133 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
16134 BoringSSL.
16135
16136
Willy Tarreau74ca5042013-06-11 23:12:07 +020016137ssl_fc_sni : string
16138 This extracts the Server Name Indication TLS extension (SNI) field from an
16139 incoming connection made via an SSL/TLS transport layer and locally
16140 deciphered by haproxy. The result (when present) typically is a string
16141 matching the HTTPS host name (253 chars or less). The SSL library must have
16142 been built with support for TLS extensions enabled (check haproxy -vv).
16143
16144 This fetch is different from "req_ssl_sni" above in that it applies to the
16145 connection being deciphered by haproxy and not to SSL contents being blindly
16146 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roesler27c754c2019-07-10 15:45:51 -050016147 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020016148 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020016149
Willy Tarreau74ca5042013-06-11 23:12:07 +020016150 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020016151 ssl_fc_sni_end : suffix match
16152 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020016153
Willy Tarreau74ca5042013-06-11 23:12:07 +020016154ssl_fc_use_keysize : integer
16155 Returns the symmetric cipher key size used in bits when the incoming
16156 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020016157
Willy Tarreaub6fb4202008-07-20 11:18:28 +020016158
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200161597.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020016160------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020016161
Willy Tarreau74ca5042013-06-11 23:12:07 +020016162Fetching samples from buffer contents is a bit different from the previous
16163sample fetches above because the sampled data are ephemeral. These data can
16164only be used when they're available and will be lost when they're forwarded.
16165For this reason, samples fetched from buffer contents during a request cannot
16166be used in a response for example. Even while the data are being fetched, they
16167can change. Sometimes it is necessary to set some delays or combine multiple
16168sample fetch methods to ensure that the expected data are complete and usable,
16169for example through TCP request content inspection. Please see the "tcp-request
16170content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020016171
Willy Tarreau74ca5042013-06-11 23:12:07 +020016172payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010016173 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016174 "stick on", "stick match"), and for "res.payload" when used in the context of
16175 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010016176
Willy Tarreau74ca5042013-06-11 23:12:07 +020016177payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
16178 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010016179 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020016180 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010016181
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020016182req.hdrs : string
16183 Returns the current request headers as string including the last empty line
16184 separating headers from the request body. The last empty line can be used to
16185 detect a truncated header block. This sample fetch is useful for some SPOE
16186 headers analyzers and for advanced logging.
16187
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020016188req.hdrs_bin : binary
16189 Returns the current request headers contained in preparsed binary form. This
16190 is useful for offloading some processing with SPOE. Each string is described
16191 by a length followed by the number of bytes indicated in the length. The
16192 length is represented using the variable integer encoding detailed in the
16193 SPOE documentation. The end of the list is marked by a couple of empty header
16194 names and values (length of 0 for both).
16195
16196 *(<str:header-name><str:header-value>)<empty string><empty string>
16197
16198 int: refer to the SPOE documentation for the encoding
16199 str: <int:length><bytes>
16200
Willy Tarreau74ca5042013-06-11 23:12:07 +020016201req.len : integer
16202req_len : integer (deprecated)
16203 Returns an integer value corresponding to the number of bytes present in the
16204 request buffer. This is mostly used in ACL. It is important to understand
16205 that this test does not return false as long as the buffer is changing. This
16206 means that a check with equality to zero will almost always immediately match
16207 at the beginning of the session, while a test for more data will wait for
16208 that data to come in and return false only when haproxy is certain that no
16209 more data will come in. This test was designed to be used with TCP request
16210 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016211
Willy Tarreau74ca5042013-06-11 23:12:07 +020016212req.payload(<offset>,<length>) : binary
16213 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020016214 in the request buffer. As a special case, if the <length> argument is zero,
16215 the the whole buffer from <offset> to the end is extracted. This can be used
16216 with ACLs in order to check for the presence of some content in a buffer at
16217 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016218
Willy Tarreau74ca5042013-06-11 23:12:07 +020016219 ACL alternatives :
16220 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016221
Willy Tarreau74ca5042013-06-11 23:12:07 +020016222req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
16223 This extracts a binary block whose size is specified at <offset1> for <length>
16224 bytes, and which starts at <offset2> if specified or just after the length in
16225 the request buffer. The <offset2> parameter also supports relative offsets if
16226 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016227
Willy Tarreau74ca5042013-06-11 23:12:07 +020016228 ACL alternatives :
16229 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016230
Willy Tarreau74ca5042013-06-11 23:12:07 +020016231 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016232
Willy Tarreau74ca5042013-06-11 23:12:07 +020016233req.proto_http : boolean
16234req_proto_http : boolean (deprecated)
16235 Returns true when data in the request buffer look like HTTP and correctly
16236 parses as such. It is the same parser as the common HTTP request parser which
16237 is used so there should be no surprises. The test does not match until the
16238 request is complete, failed or timed out. This test may be used to report the
16239 protocol in TCP logs, but the biggest use is to block TCP request analysis
16240 until a complete HTTP request is present in the buffer, for example to track
16241 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016242
Willy Tarreau74ca5042013-06-11 23:12:07 +020016243 Example:
16244 # track request counts per "base" (concatenation of Host+URL)
16245 tcp-request inspect-delay 10s
16246 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016247 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016248
Willy Tarreau74ca5042013-06-11 23:12:07 +020016249req.rdp_cookie([<name>]) : string
16250rdp_cookie([<name>]) : string (deprecated)
16251 When the request buffer looks like the RDP protocol, extracts the RDP cookie
16252 <name>, or any cookie if unspecified. The parser only checks for the first
16253 cookie, as illustrated in the RDP protocol specification. The cookie name is
16254 case insensitive. Generally the "MSTS" cookie name will be used, as it can
16255 contain the user name of the client connecting to the server if properly
16256 configured on the client. The "MSTSHASH" cookie is often used as well for
16257 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016258
Willy Tarreau74ca5042013-06-11 23:12:07 +020016259 This differs from "balance rdp-cookie" in that any balancing algorithm may be
16260 used and thus the distribution of clients to backend servers is not linked to
16261 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
16262 such as "balance roundrobin" or "balance leastconn" will lead to a more even
16263 distribution of clients to backend servers than the hash used by "balance
16264 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016265
Willy Tarreau74ca5042013-06-11 23:12:07 +020016266 ACL derivatives :
16267 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016268
Willy Tarreau74ca5042013-06-11 23:12:07 +020016269 Example :
16270 listen tse-farm
16271 bind 0.0.0.0:3389
16272 # wait up to 5s for an RDP cookie in the request
16273 tcp-request inspect-delay 5s
16274 tcp-request content accept if RDP_COOKIE
16275 # apply RDP cookie persistence
16276 persist rdp-cookie
16277 # Persist based on the mstshash cookie
16278 # This is only useful makes sense if
16279 # balance rdp-cookie is not used
16280 stick-table type string size 204800
16281 stick on req.rdp_cookie(mstshash)
16282 server srv1 1.1.1.1:3389
16283 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016284
Willy Tarreau74ca5042013-06-11 23:12:07 +020016285 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
16286 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016287
Willy Tarreau74ca5042013-06-11 23:12:07 +020016288req.rdp_cookie_cnt([name]) : integer
16289rdp_cookie_cnt([name]) : integer (deprecated)
16290 Tries to parse the request buffer as RDP protocol, then returns an integer
16291 corresponding to the number of RDP cookies found. If an optional cookie name
16292 is passed, only cookies matching this name are considered. This is mostly
16293 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016294
Willy Tarreau74ca5042013-06-11 23:12:07 +020016295 ACL derivatives :
16296 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016297
Alex Zorin4afdd132018-12-30 13:56:28 +110016298req.ssl_alpn : string
16299 Returns a string containing the values of the Application-Layer Protocol
16300 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
16301 ClientHello message. Note that this only applies to raw contents found in the
16302 request buffer and not to the contents deciphered via an SSL data layer, so
16303 this will not work with "bind" lines having the "ssl" option. This is useful
16304 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020016305 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110016306
16307 Examples :
16308 # Wait for a client hello for at most 5 seconds
16309 tcp-request inspect-delay 5s
16310 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020016311 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110016312 default_backend bk_default
16313
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020016314req.ssl_ec_ext : boolean
16315 Returns a boolean identifying if client sent the Supported Elliptic Curves
16316 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020016317 message. This can be used to present ECC compatible clients with EC
16318 certificate and to use RSA for all others, on the same IP address. Note that
16319 this only applies to raw contents found in the request buffer and not to
16320 contents deciphered via an SSL data layer, so this will not work with "bind"
16321 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020016322
Willy Tarreau74ca5042013-06-11 23:12:07 +020016323req.ssl_hello_type : integer
16324req_ssl_hello_type : integer (deprecated)
16325 Returns an integer value containing the type of the SSL hello message found
16326 in the request buffer if the buffer contains data that parse as a complete
16327 SSL (v3 or superior) client hello message. Note that this only applies to raw
16328 contents found in the request buffer and not to contents deciphered via an
16329 SSL data layer, so this will not work with "bind" lines having the "ssl"
16330 option. This is mostly used in ACL to detect presence of an SSL hello message
16331 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016332
Willy Tarreau74ca5042013-06-11 23:12:07 +020016333req.ssl_sni : string
16334req_ssl_sni : string (deprecated)
16335 Returns a string containing the value of the Server Name TLS extension sent
16336 by a client in a TLS stream passing through the request buffer if the buffer
16337 contains data that parse as a complete SSL (v3 or superior) client hello
16338 message. Note that this only applies to raw contents found in the request
16339 buffer and not to contents deciphered via an SSL data layer, so this will not
16340 work with "bind" lines having the "ssl" option. SNI normally contains the
16341 name of the host the client tries to connect to (for recent browsers). SNI is
16342 useful for allowing or denying access to certain hosts when SSL/TLS is used
16343 by the client. This test was designed to be used with TCP request content
16344 inspection. If content switching is needed, it is recommended to first wait
16345 for a complete client hello (type 1), like in the example below. See also
16346 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016347
Willy Tarreau74ca5042013-06-11 23:12:07 +020016348 ACL derivatives :
16349 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016350
Willy Tarreau74ca5042013-06-11 23:12:07 +020016351 Examples :
16352 # Wait for a client hello for at most 5 seconds
16353 tcp-request inspect-delay 5s
16354 tcp-request content accept if { req_ssl_hello_type 1 }
16355 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
16356 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016357
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053016358req.ssl_st_ext : integer
16359 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
16360 Returns 1 if the client sent SessionTicket TLS Extension
16361 Returns 2 if the client also sent non-zero length TLS SessionTicket
16362 Note that this only applies to raw contents found in the request buffer and
16363 not to contents deciphered via an SSL data layer, so this will not work with
16364 "bind" lines having the "ssl" option. This can for example be used to detect
16365 whether the client sent a SessionTicket or not and stick it accordingly, if
16366 no SessionTicket then stick on SessionID or don't stick as there's no server
16367 side state is there when SessionTickets are in use.
16368
Willy Tarreau74ca5042013-06-11 23:12:07 +020016369req.ssl_ver : integer
16370req_ssl_ver : integer (deprecated)
16371 Returns an integer value containing the version of the SSL/TLS protocol of a
16372 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
16373 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
16374 composed of the major version multiplied by 65536, added to the minor
16375 version. Note that this only applies to raw contents found in the request
16376 buffer and not to contents deciphered via an SSL data layer, so this will not
16377 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010016378 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020016379 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016380
Willy Tarreau74ca5042013-06-11 23:12:07 +020016381 ACL derivatives :
16382 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010016383
Willy Tarreau47e8eba2013-09-11 23:28:46 +020016384res.len : integer
16385 Returns an integer value corresponding to the number of bytes present in the
16386 response buffer. This is mostly used in ACL. It is important to understand
16387 that this test does not return false as long as the buffer is changing. This
16388 means that a check with equality to zero will almost always immediately match
16389 at the beginning of the session, while a test for more data will wait for
16390 that data to come in and return false only when haproxy is certain that no
16391 more data will come in. This test was designed to be used with TCP response
16392 content inspection.
16393
Willy Tarreau74ca5042013-06-11 23:12:07 +020016394res.payload(<offset>,<length>) : binary
16395 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020016396 in the response buffer. As a special case, if the <length> argument is zero,
16397 the the whole buffer from <offset> to the end is extracted. This can be used
16398 with ACLs in order to check for the presence of some content in a buffer at
16399 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016400
Willy Tarreau74ca5042013-06-11 23:12:07 +020016401res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
16402 This extracts a binary block whose size is specified at <offset1> for <length>
16403 bytes, and which starts at <offset2> if specified or just after the length in
16404 the response buffer. The <offset2> parameter also supports relative offsets
16405 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016406
Willy Tarreau74ca5042013-06-11 23:12:07 +020016407 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016408
Willy Tarreau971f7b62015-09-29 14:06:59 +020016409res.ssl_hello_type : integer
16410rep_ssl_hello_type : integer (deprecated)
16411 Returns an integer value containing the type of the SSL hello message found
16412 in the response buffer if the buffer contains data that parses as a complete
16413 SSL (v3 or superior) hello message. Note that this only applies to raw
16414 contents found in the response buffer and not to contents deciphered via an
16415 SSL data layer, so this will not work with "server" lines having the "ssl"
16416 option. This is mostly used in ACL to detect presence of an SSL hello message
16417 that is supposed to contain an SSL session ID usable for stickiness.
16418
Willy Tarreau74ca5042013-06-11 23:12:07 +020016419wait_end : boolean
16420 This fetch either returns true when the inspection period is over, or does
16421 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010016422 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020016423 actions, such as a delayed reject for some special addresses. Since it either
16424 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010016425 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020016426 "WAIT_END" is always usable without prior declaration. This test was designed
16427 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016428
Willy Tarreau74ca5042013-06-11 23:12:07 +020016429 Examples :
16430 # delay every incoming request by 2 seconds
16431 tcp-request inspect-delay 2s
16432 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010016433
Willy Tarreau74ca5042013-06-11 23:12:07 +020016434 # don't immediately tell bad guys they are rejected
16435 tcp-request inspect-delay 10s
16436 acl goodguys src 10.0.0.0/24
16437 acl badguys src 10.0.1.0/24
16438 tcp-request content accept if goodguys
16439 tcp-request content reject if badguys WAIT_END
16440 tcp-request content reject
16441
16442
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200164437.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020016444--------------------------------------
16445
16446It is possible to fetch samples from HTTP contents, requests and responses.
16447This application layer is also called layer 7. It is only possible to fetch the
16448data in this section when a full HTTP request or response has been parsed from
16449its respective request or response buffer. This is always the case with all
16450HTTP specific rules and for sections running with "mode http". When using TCP
16451content inspection, it may be necessary to support an inspection delay in order
16452to let the request or response come in first. These fetches may require a bit
16453more CPU resources than the layer 4 ones, but not much since the request and
16454response are indexed.
16455
16456base : string
16457 This returns the concatenation of the first Host header and the path part of
16458 the request, which starts at the first slash and ends before the question
16459 mark. It can be useful in virtual hosted environments to detect URL abuses as
16460 well as to improve shared caches efficiency. Using this with a limited size
16461 stick table also allows one to collect statistics about most commonly
16462 requested objects by host/path. With ACLs it can allow simple content
16463 switching rules involving the host and the path at the same time, such as
16464 "www.example.com/favicon.ico". See also "path" and "uri".
16465
16466 ACL derivatives :
16467 base : exact string match
16468 base_beg : prefix match
16469 base_dir : subdir match
16470 base_dom : domain match
16471 base_end : suffix match
16472 base_len : length match
16473 base_reg : regex match
16474 base_sub : substring match
16475
16476base32 : integer
16477 This returns a 32-bit hash of the value returned by the "base" fetch method
16478 above. This is useful to track per-URL activity on high traffic sites without
16479 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016480 memory. The output type is an unsigned integer. The hash function used is
16481 SDBM with full avalanche on the output. Technically, base32 is exactly equal
16482 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020016483
16484base32+src : binary
16485 This returns the concatenation of the base32 fetch above and the src fetch
16486 below. The resulting type is of type binary, with a size of 8 or 20 bytes
16487 depending on the source address family. This can be used to track per-IP,
16488 per-URL counters.
16489
William Lallemand65ad6e12014-01-31 15:08:02 +010016490capture.req.hdr(<idx>) : string
16491 This extracts the content of the header captured by the "capture request
16492 header", idx is the position of the capture keyword in the configuration.
16493 The first entry is an index of 0. See also: "capture request header".
16494
16495capture.req.method : string
16496 This extracts the METHOD of an HTTP request. It can be used in both request
16497 and response. Unlike "method", it can be used in both request and response
16498 because it's allocated.
16499
16500capture.req.uri : string
16501 This extracts the request's URI, which starts at the first slash and ends
16502 before the first space in the request (without the host part). Unlike "path"
16503 and "url", it can be used in both request and response because it's
16504 allocated.
16505
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020016506capture.req.ver : string
16507 This extracts the request's HTTP version and returns either "HTTP/1.0" or
16508 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
16509 logs because it relies on a persistent flag.
16510
William Lallemand65ad6e12014-01-31 15:08:02 +010016511capture.res.hdr(<idx>) : string
16512 This extracts the content of the header captured by the "capture response
16513 header", idx is the position of the capture keyword in the configuration.
16514 The first entry is an index of 0.
16515 See also: "capture response header"
16516
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020016517capture.res.ver : string
16518 This extracts the response's HTTP version and returns either "HTTP/1.0" or
16519 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
16520 persistent flag.
16521
Willy Tarreaua5910cc2015-05-02 00:46:08 +020016522req.body : binary
16523 This returns the HTTP request's available body as a block of data. It
16524 requires that the request body has been buffered made available using
16525 "option http-buffer-request". In case of chunked-encoded body, currently only
16526 the first chunk is analyzed.
16527
Thierry FOURNIER9826c772015-05-20 15:50:54 +020016528req.body_param([<name>) : string
16529 This fetch assumes that the body of the POST request is url-encoded. The user
16530 can check if the "content-type" contains the value
16531 "application/x-www-form-urlencoded". This extracts the first occurrence of the
16532 parameter <name> in the body, which ends before '&'. The parameter name is
16533 case-sensitive. If no name is given, any parameter will match, and the first
16534 one will be returned. The result is a string corresponding to the value of the
16535 parameter <name> as presented in the request body (no URL decoding is
16536 performed). Note that the ACL version of this fetch iterates over multiple
16537 parameters and will iteratively report all parameters values if no name is
16538 given.
16539
Willy Tarreaua5910cc2015-05-02 00:46:08 +020016540req.body_len : integer
16541 This returns the length of the HTTP request's available body in bytes. It may
16542 be lower than the advertised length if the body is larger than the buffer. It
16543 requires that the request body has been buffered made available using
16544 "option http-buffer-request".
16545
16546req.body_size : integer
16547 This returns the advertised length of the HTTP request's body in bytes. It
16548 will represent the advertised Content-Length header, or the size of the first
16549 chunk in case of chunked encoding. In order to parse the chunks, it requires
16550 that the request body has been buffered made available using
16551 "option http-buffer-request".
16552
Willy Tarreau74ca5042013-06-11 23:12:07 +020016553req.cook([<name>]) : string
16554cook([<name>]) : string (deprecated)
16555 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16556 header line from the request, and returns its value as string. If no name is
16557 specified, the first cookie value is returned. When used with ACLs, all
16558 matching cookies are evaluated. Spaces around the name and the value are
16559 ignored as requested by the Cookie header specification (RFC6265). The cookie
16560 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
16561 well return an empty value if it is present. Use the "found" match to detect
16562 presence. Use the res.cook() variant for response cookies sent by the server.
16563
16564 ACL derivatives :
16565 cook([<name>]) : exact string match
16566 cook_beg([<name>]) : prefix match
16567 cook_dir([<name>]) : subdir match
16568 cook_dom([<name>]) : domain match
16569 cook_end([<name>]) : suffix match
16570 cook_len([<name>]) : length match
16571 cook_reg([<name>]) : regex match
16572 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010016573
Willy Tarreau74ca5042013-06-11 23:12:07 +020016574req.cook_cnt([<name>]) : integer
16575cook_cnt([<name>]) : integer (deprecated)
16576 Returns an integer value representing the number of occurrences of the cookie
16577 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016578
Willy Tarreau74ca5042013-06-11 23:12:07 +020016579req.cook_val([<name>]) : integer
16580cook_val([<name>]) : integer (deprecated)
16581 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16582 header line from the request, and converts its value to an integer which is
16583 returned. If no name is specified, the first cookie value is returned. When
16584 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020016585
Willy Tarreau74ca5042013-06-11 23:12:07 +020016586cookie([<name>]) : string (deprecated)
16587 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16588 header line from the request, or a "Set-Cookie" header from the response, and
16589 returns its value as a string. A typical use is to get multiple clients
16590 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020016591 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020016592 multi-peer synchronization and state keeping across restarts. If no name is
16593 specified, the first cookie value is returned. This fetch should not be used
16594 anymore and should be replaced by req.cook() or res.cook() instead as it
16595 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016596
Willy Tarreau74ca5042013-06-11 23:12:07 +020016597hdr([<name>[,<occ>]]) : string
16598 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
16599 used on responses. Please refer to these respective fetches for more details.
16600 In case of doubt about the fetch direction, please use the explicit ones.
16601 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016602 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016603
Willy Tarreau74ca5042013-06-11 23:12:07 +020016604req.fhdr(<name>[,<occ>]) : string
16605 This extracts the last occurrence of header <name> in an HTTP request. When
16606 used from an ACL, all occurrences are iterated over until a match is found.
16607 Optionally, a specific occurrence might be specified as a position number.
16608 Positive values indicate a position from the first occurrence, with 1 being
16609 the first one. Negative values indicate positions relative to the last one,
16610 with -1 being the last one. It differs from req.hdr() in that any commas
16611 present in the value are returned and are not used as delimiters. This is
16612 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016613
Willy Tarreau74ca5042013-06-11 23:12:07 +020016614req.fhdr_cnt([<name>]) : integer
16615 Returns an integer value representing the number of occurrences of request
16616 header field name <name>, or the total number of header fields if <name> is
16617 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
16618 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016619
Willy Tarreau74ca5042013-06-11 23:12:07 +020016620req.hdr([<name>[,<occ>]]) : string
16621 This extracts the last occurrence of header <name> in an HTTP request. When
16622 used from an ACL, all occurrences are iterated over until a match is found.
16623 Optionally, a specific occurrence might be specified as a position number.
16624 Positive values indicate a position from the first occurrence, with 1 being
16625 the first one. Negative values indicate positions relative to the last one,
16626 with -1 being the last one. A typical use is with the X-Forwarded-For header
16627 once converted to IP, associated with an IP stick-table. The function
16628 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000016629 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020016630 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010016631 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010016632
Willy Tarreau74ca5042013-06-11 23:12:07 +020016633 ACL derivatives :
16634 hdr([<name>[,<occ>]]) : exact string match
16635 hdr_beg([<name>[,<occ>]]) : prefix match
16636 hdr_dir([<name>[,<occ>]]) : subdir match
16637 hdr_dom([<name>[,<occ>]]) : domain match
16638 hdr_end([<name>[,<occ>]]) : suffix match
16639 hdr_len([<name>[,<occ>]]) : length match
16640 hdr_reg([<name>[,<occ>]]) : regex match
16641 hdr_sub([<name>[,<occ>]]) : substring match
16642
16643req.hdr_cnt([<name>]) : integer
16644hdr_cnt([<header>]) : integer (deprecated)
16645 Returns an integer value representing the number of occurrences of request
16646 header field name <name>, or the total number of header field values if
16647 <name> is not specified. It is important to remember that one header line may
16648 count as several headers if it has several values. The function considers any
16649 comma as a delimiter for distinct values. If full-line headers are desired
16650 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
16651 detect presence, absence or abuse of a specific header, as well as to block
16652 request smuggling attacks by rejecting requests which contain more than one
16653 of certain headers. See "req.hdr" for more information on header matching.
16654
16655req.hdr_ip([<name>[,<occ>]]) : ip
16656hdr_ip([<name>[,<occ>]]) : ip (deprecated)
16657 This extracts the last occurrence of header <name> in an HTTP request,
16658 converts it to an IPv4 or IPv6 address and returns this address. When used
16659 with ACLs, all occurrences are checked, and if <name> is omitted, every value
16660 of every header is checked. Optionally, a specific occurrence might be
16661 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010016662 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020016663 positions relative to the last one, with -1 being the last one. A typical use
16664 is with the X-Forwarded-For and X-Client-IP headers.
16665
16666req.hdr_val([<name>[,<occ>]]) : integer
16667hdr_val([<name>[,<occ>]]) : integer (deprecated)
16668 This extracts the last occurrence of header <name> in an HTTP request, and
16669 converts it to an integer value. When used with ACLs, all occurrences are
16670 checked, and if <name> is omitted, every value of every header is checked.
16671 Optionally, a specific occurrence might be specified as a position number.
16672 Positive values indicate a position from the first occurrence, with 1 being
16673 the first one. Negative values indicate positions relative to the last one,
16674 with -1 being the last one. A typical use is with the X-Forwarded-For header.
16675
Frédéric Lécailleec891192019-02-26 15:02:35 +010016676
16677
Willy Tarreau74ca5042013-06-11 23:12:07 +020016678http_auth(<userlist>) : boolean
16679 Returns a boolean indicating whether the authentication data received from
16680 the client match a username & password stored in the specified userlist. This
16681 fetch function is not really useful outside of ACLs. Currently only http
16682 basic auth is supported.
16683
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010016684http_auth_group(<userlist>) : string
16685 Returns a string corresponding to the user name found in the authentication
16686 data received from the client if both the user name and password are valid
16687 according to the specified userlist. The main purpose is to use it in ACLs
16688 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016689 This fetch function is not really useful outside of ACLs. Currently only http
16690 basic auth is supported.
16691
16692 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010016693 http_auth_group(<userlist>) : group ...
16694 Returns true when the user extracted from the request and whose password is
16695 valid according to the specified userlist belongs to at least one of the
16696 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016697
16698http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020016699 Returns true when the request being processed is the first one of the
16700 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020016701 from some requests when a request is not the first one, or to help grouping
16702 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020016703
Willy Tarreau74ca5042013-06-11 23:12:07 +020016704method : integer + string
16705 Returns an integer value corresponding to the method in the HTTP request. For
16706 example, "GET" equals 1 (check sources to establish the matching). Value 9
16707 means "other method" and may be converted to a string extracted from the
16708 stream. This should not be used directly as a sample, this is only meant to
16709 be used from ACLs, which transparently convert methods from patterns to these
16710 integer + string values. Some predefined ACL already check for most common
16711 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016712
Willy Tarreau74ca5042013-06-11 23:12:07 +020016713 ACL derivatives :
16714 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016715
Willy Tarreau74ca5042013-06-11 23:12:07 +020016716 Example :
16717 # only accept GET and HEAD requests
16718 acl valid_method method GET HEAD
16719 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020016720
Willy Tarreau74ca5042013-06-11 23:12:07 +020016721path : string
16722 This extracts the request's URL path, which starts at the first slash and
16723 ends before the question mark (without the host part). A typical use is with
16724 prefetch-capable caches, and with portals which need to aggregate multiple
16725 information from databases and keep them in caches. Note that with outgoing
16726 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010016727 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020016728 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016729
Willy Tarreau74ca5042013-06-11 23:12:07 +020016730 ACL derivatives :
16731 path : exact string match
16732 path_beg : prefix match
16733 path_dir : subdir match
16734 path_dom : domain match
16735 path_end : suffix match
16736 path_len : length match
16737 path_reg : regex match
16738 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016739
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016740query : string
16741 This extracts the request's query string, which starts after the first
16742 question mark. If no question mark is present, this fetch returns nothing. If
16743 a question mark is present but nothing follows, it returns an empty string.
16744 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010016745 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016746 which stops before the question mark.
16747
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016748req.hdr_names([<delim>]) : string
16749 This builds a string made from the concatenation of all header names as they
16750 appear in the request when the rule is evaluated. The default delimiter is
16751 the comma (',') but it may be overridden as an optional argument <delim>. In
16752 this case, only the first character of <delim> is considered.
16753
Willy Tarreau74ca5042013-06-11 23:12:07 +020016754req.ver : string
16755req_ver : string (deprecated)
16756 Returns the version string from the HTTP request, for example "1.1". This can
16757 be useful for logs, but is mostly there for ACL. Some predefined ACL already
16758 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016759
Willy Tarreau74ca5042013-06-11 23:12:07 +020016760 ACL derivatives :
16761 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020016762
Willy Tarreau74ca5042013-06-11 23:12:07 +020016763res.comp : boolean
16764 Returns the boolean "true" value if the response has been compressed by
16765 HAProxy, otherwise returns boolean "false". This may be used to add
16766 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016767
Willy Tarreau74ca5042013-06-11 23:12:07 +020016768res.comp_algo : string
16769 Returns a string containing the name of the algorithm used if the response
16770 was compressed by HAProxy, for example : "deflate". This may be used to add
16771 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016772
Willy Tarreau74ca5042013-06-11 23:12:07 +020016773res.cook([<name>]) : string
16774scook([<name>]) : string (deprecated)
16775 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16776 header line from the response, and returns its value as string. If no name is
16777 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016778
Willy Tarreau74ca5042013-06-11 23:12:07 +020016779 ACL derivatives :
16780 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016781
Willy Tarreau74ca5042013-06-11 23:12:07 +020016782res.cook_cnt([<name>]) : integer
16783scook_cnt([<name>]) : integer (deprecated)
16784 Returns an integer value representing the number of occurrences of the cookie
16785 <name> in the response, or all cookies if <name> is not specified. This is
16786 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016787
Willy Tarreau74ca5042013-06-11 23:12:07 +020016788res.cook_val([<name>]) : integer
16789scook_val([<name>]) : integer (deprecated)
16790 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16791 header line from the response, and converts its value to an integer which is
16792 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016793
Willy Tarreau74ca5042013-06-11 23:12:07 +020016794res.fhdr([<name>[,<occ>]]) : string
16795 This extracts the last occurrence of header <name> in an HTTP response, or of
16796 the last header if no <name> is specified. Optionally, a specific occurrence
16797 might be specified as a position number. Positive values indicate a position
16798 from the first occurrence, with 1 being the first one. Negative values
16799 indicate positions relative to the last one, with -1 being the last one. It
16800 differs from res.hdr() in that any commas present in the value are returned
16801 and are not used as delimiters. If this is not desired, the res.hdr() fetch
16802 should be used instead. This is sometimes useful with headers such as Date or
16803 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016804
Willy Tarreau74ca5042013-06-11 23:12:07 +020016805res.fhdr_cnt([<name>]) : integer
16806 Returns an integer value representing the number of occurrences of response
16807 header field name <name>, or the total number of header fields if <name> is
16808 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
16809 the number of full line headers and does not stop on commas. If this is not
16810 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016811
Willy Tarreau74ca5042013-06-11 23:12:07 +020016812res.hdr([<name>[,<occ>]]) : string
16813shdr([<name>[,<occ>]]) : string (deprecated)
16814 This extracts the last occurrence of header <name> in an HTTP response, or of
16815 the last header if no <name> is specified. Optionally, a specific occurrence
16816 might be specified as a position number. Positive values indicate a position
16817 from the first occurrence, with 1 being the first one. Negative values
16818 indicate positions relative to the last one, with -1 being the last one. This
16819 can be useful to learn some data into a stick-table. The function considers
16820 any comma as a delimiter for distinct values. If this is not desired, the
16821 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016822
Willy Tarreau74ca5042013-06-11 23:12:07 +020016823 ACL derivatives :
16824 shdr([<name>[,<occ>]]) : exact string match
16825 shdr_beg([<name>[,<occ>]]) : prefix match
16826 shdr_dir([<name>[,<occ>]]) : subdir match
16827 shdr_dom([<name>[,<occ>]]) : domain match
16828 shdr_end([<name>[,<occ>]]) : suffix match
16829 shdr_len([<name>[,<occ>]]) : length match
16830 shdr_reg([<name>[,<occ>]]) : regex match
16831 shdr_sub([<name>[,<occ>]]) : substring match
16832
16833res.hdr_cnt([<name>]) : integer
16834shdr_cnt([<name>]) : integer (deprecated)
16835 Returns an integer value representing the number of occurrences of response
16836 header field name <name>, or the total number of header fields if <name> is
16837 not specified. The function considers any comma as a delimiter for distinct
16838 values. If this is not desired, the res.fhdr_cnt() fetch should be used
16839 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016840
Willy Tarreau74ca5042013-06-11 23:12:07 +020016841res.hdr_ip([<name>[,<occ>]]) : ip
16842shdr_ip([<name>[,<occ>]]) : ip (deprecated)
16843 This extracts the last occurrence of header <name> in an HTTP response,
16844 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
16845 specific occurrence might be specified as a position number. Positive values
16846 indicate a position from the first occurrence, with 1 being the first one.
16847 Negative values indicate positions relative to the last one, with -1 being
16848 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016849
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016850res.hdr_names([<delim>]) : string
16851 This builds a string made from the concatenation of all header names as they
16852 appear in the response when the rule is evaluated. The default delimiter is
16853 the comma (',') but it may be overridden as an optional argument <delim>. In
16854 this case, only the first character of <delim> is considered.
16855
Willy Tarreau74ca5042013-06-11 23:12:07 +020016856res.hdr_val([<name>[,<occ>]]) : integer
16857shdr_val([<name>[,<occ>]]) : integer (deprecated)
16858 This extracts the last occurrence of header <name> in an HTTP response, and
16859 converts it to an integer value. Optionally, a specific occurrence might be
16860 specified as a position number. Positive values indicate a position from the
16861 first occurrence, with 1 being the first one. Negative values indicate
16862 positions relative to the last one, with -1 being the last one. This can be
16863 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016864
Willy Tarreau74ca5042013-06-11 23:12:07 +020016865res.ver : string
16866resp_ver : string (deprecated)
16867 Returns the version string from the HTTP response, for example "1.1". This
16868 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020016869
Willy Tarreau74ca5042013-06-11 23:12:07 +020016870 ACL derivatives :
16871 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016872
Willy Tarreau74ca5042013-06-11 23:12:07 +020016873set-cookie([<name>]) : string (deprecated)
16874 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16875 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020016876 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020016877 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016878
Willy Tarreau74ca5042013-06-11 23:12:07 +020016879 This fetch function is deprecated and has been superseded by the "res.cook"
16880 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016881
Willy Tarreau74ca5042013-06-11 23:12:07 +020016882status : integer
16883 Returns an integer containing the HTTP status code in the HTTP response, for
16884 example, 302. It is mostly used within ACLs and integer ranges, for example,
16885 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016886
Thierry Fournier0e00dca2016-04-07 15:47:40 +020016887unique-id : string
16888 Returns the unique-id attached to the request. The directive
16889 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
16890 fails. Note that the unique-id is usually used with HTTP requests, however this
16891 sample fetch can be used with other protocols. Obviously, if it is used with
16892 other protocols than HTTP, the unique-id-format directive must not contain
16893 HTTP parts. See: unique-id-format and unique-id-header
16894
Willy Tarreau74ca5042013-06-11 23:12:07 +020016895url : string
16896 This extracts the request's URL as presented in the request. A typical use is
16897 with prefetch-capable caches, and with portals which need to aggregate
16898 multiple information from databases and keep them in caches. With ACLs, using
16899 "path" is preferred over using "url", because clients may send a full URL as
16900 is normally done with proxies. The only real use is to match "*" which does
16901 not match in "path", and for which there is already a predefined ACL. See
16902 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016903
Willy Tarreau74ca5042013-06-11 23:12:07 +020016904 ACL derivatives :
16905 url : exact string match
16906 url_beg : prefix match
16907 url_dir : subdir match
16908 url_dom : domain match
16909 url_end : suffix match
16910 url_len : length match
16911 url_reg : regex match
16912 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016913
Willy Tarreau74ca5042013-06-11 23:12:07 +020016914url_ip : ip
16915 This extracts the IP address from the request's URL when the host part is
16916 presented as an IP address. Its use is very limited. For instance, a
16917 monitoring system might use this field as an alternative for the source IP in
16918 order to test what path a given source address would follow, or to force an
16919 entry in a table for a given source address. With ACLs it can be used to
16920 restrict access to certain systems through a proxy, for example when combined
16921 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016922
Willy Tarreau74ca5042013-06-11 23:12:07 +020016923url_port : integer
16924 This extracts the port part from the request's URL. Note that if the port is
16925 not specified in the request, port 80 is assumed. With ACLs it can be used to
16926 restrict access to certain systems through a proxy, for example when combined
16927 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016928
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016929urlp([<name>[,<delim>]]) : string
16930url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020016931 This extracts the first occurrence of the parameter <name> in the query
16932 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016933 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
16934 any parameter will match, and the first one will be returned. The result is
16935 a string corresponding to the value of the parameter <name> as presented in
16936 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020016937 stickiness based on a client ID, to extract an application cookie passed as a
16938 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016939 this fetch iterates over multiple parameters and will iteratively report all
16940 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016941
Willy Tarreau74ca5042013-06-11 23:12:07 +020016942 ACL derivatives :
16943 urlp(<name>[,<delim>]) : exact string match
16944 urlp_beg(<name>[,<delim>]) : prefix match
16945 urlp_dir(<name>[,<delim>]) : subdir match
16946 urlp_dom(<name>[,<delim>]) : domain match
16947 urlp_end(<name>[,<delim>]) : suffix match
16948 urlp_len(<name>[,<delim>]) : length match
16949 urlp_reg(<name>[,<delim>]) : regex match
16950 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016951
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016952
Willy Tarreau74ca5042013-06-11 23:12:07 +020016953 Example :
16954 # match http://example.com/foo?PHPSESSIONID=some_id
16955 stick on urlp(PHPSESSIONID)
16956 # match http://example.com/foo;JSESSIONID=some_id
16957 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016958
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016959urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020016960 See "urlp" above. This one extracts the URL parameter <name> in the request
16961 and converts it to an integer value. This can be used for session stickiness
16962 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020016963
Dragan Dosen0070cd52016-06-16 12:19:49 +020016964url32 : integer
16965 This returns a 32-bit hash of the value obtained by concatenating the first
16966 Host header and the whole URL including parameters (not only the path part of
16967 the request, as in the "base32" fetch above). This is useful to track per-URL
16968 activity. A shorter hash is stored, saving a lot of memory. The output type
16969 is an unsigned integer.
16970
16971url32+src : binary
16972 This returns the concatenation of the "url32" fetch and the "src" fetch. The
16973 resulting type is of type binary, with a size of 8 or 20 bytes depending on
16974 the source address family. This can be used to track per-IP, per-URL counters.
16975
Willy Tarreau198a7442008-01-17 12:05:32 +010016976
Willy Tarreau74ca5042013-06-11 23:12:07 +0200169777.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016978---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016979
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016980Some predefined ACLs are hard-coded so that they do not have to be declared in
16981every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020016982order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010016983
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016984ACL name Equivalent to Usage
16985---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016986FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020016987HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016988HTTP_1.0 req_ver 1.0 match HTTP version 1.0
16989HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010016990HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
16991HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
16992HTTP_URL_SLASH url_beg / match URL beginning with "/"
16993HTTP_URL_STAR url * match URL equal to "*"
16994LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016995METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016996METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016997METH_GET method GET HEAD match HTTP GET or HEAD method
16998METH_HEAD method HEAD match HTTP HEAD method
16999METH_OPTIONS method OPTIONS match HTTP OPTIONS method
17000METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020017001METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017002METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020017003RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017004REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010017005TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017006WAIT_END wait_end wait for end of content analysis
17007---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010017008
Willy Tarreaub937b7e2010-01-12 15:27:54 +010017009
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170108. Logging
17011----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010017012
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017013One of HAProxy's strong points certainly lies is its precise logs. It probably
17014provides the finest level of information available for such a product, which is
17015very important for troubleshooting complex environments. Standard information
17016provided in logs include client ports, TCP/HTTP state timers, precise session
17017state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017018to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017019headers.
17020
17021In order to improve administrators reactivity, it offers a great transparency
17022about encountered problems, both internal and external, and it is possible to
17023send logs to different sources at the same time with different level filters :
17024
17025 - global process-level logs (system errors, start/stop, etc..)
17026 - per-instance system and internal errors (lack of resource, bugs, ...)
17027 - per-instance external troubles (servers up/down, max connections)
17028 - per-instance activity (client connections), either at the establishment or
17029 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010017030 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060017031 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017032
17033The ability to distribute different levels of logs to different log servers
17034allow several production teams to interact and to fix their problems as soon
17035as possible. For example, the system team might monitor system-wide errors,
17036while the application team might be monitoring the up/down for their servers in
17037real time, and the security team might analyze the activity logs with one hour
17038delay.
17039
17040
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170418.1. Log levels
17042---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017043
Simon Hormandf791f52011-05-29 15:01:10 +090017044TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017045source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090017046HTTP request, HTTP return code, number of bytes transmitted, conditions
17047in which the session ended, and even exchanged cookies values. For example
17048track a particular user's problems. All messages may be sent to up to two
17049syslog servers. Check the "log" keyword in section 4.2 for more information
17050about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017051
17052
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170538.2. Log formats
17054----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017055
William Lallemand48940402012-01-30 16:47:22 +010017056HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090017057and will be detailed in the following sections. A few of them may vary
17058slightly with the configuration, due to indicators specific to certain
17059options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017060
17061 - the default format, which is very basic and very rarely used. It only
17062 provides very basic information about the incoming connection at the moment
17063 it is accepted : source IP:port, destination IP:port, and frontend-name.
17064 This mode will eventually disappear so it will not be described to great
17065 extents.
17066
17067 - the TCP format, which is more advanced. This format is enabled when "option
17068 tcplog" is set on the frontend. HAProxy will then usually wait for the
17069 connection to terminate before logging. This format provides much richer
17070 information, such as timers, connection counts, queue size, etc... This
17071 format is recommended for pure TCP proxies.
17072
17073 - the HTTP format, which is the most advanced for HTTP proxying. This format
17074 is enabled when "option httplog" is set on the frontend. It provides the
17075 same information as the TCP format with some HTTP-specific fields such as
17076 the request, the status code, and captures of headers and cookies. This
17077 format is recommended for HTTP proxies.
17078
Emeric Brun3a058f32009-06-30 18:26:00 +020017079 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
17080 fields arranged in the same order as the CLF format. In this mode, all
17081 timers, captures, flags, etc... appear one per field after the end of the
17082 common fields, in the same order they appear in the standard HTTP format.
17083
William Lallemand48940402012-01-30 16:47:22 +010017084 - the custom log format, allows you to make your own log line.
17085
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017086Next sections will go deeper into details for each of these formats. Format
17087specification will be performed on a "field" basis. Unless stated otherwise, a
17088field is a portion of text delimited by any number of spaces. Since syslog
17089servers are susceptible of inserting fields at the beginning of a line, it is
17090always assumed that the first field is the one containing the process name and
17091identifier.
17092
17093Note : Since log lines may be quite long, the log examples in sections below
17094 might be broken into multiple lines. The example log lines will be
17095 prefixed with 3 closing angle brackets ('>>>') and each time a log is
17096 broken into multiple lines, each non-final line will end with a
17097 backslash ('\') and the next line will start indented by two characters.
17098
17099
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171008.2.1. Default log format
17101-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017102
17103This format is used when no specific option is set. The log is emitted as soon
17104as the connection is accepted. One should note that this currently is the only
17105format which logs the request's destination IP and ports.
17106
17107 Example :
17108 listen www
17109 mode http
17110 log global
17111 server srv1 127.0.0.1:8000
17112
17113 >>> Feb 6 12:12:09 localhost \
17114 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
17115 (www/HTTP)
17116
17117 Field Format Extract from the example above
17118 1 process_name '[' pid ']:' haproxy[14385]:
17119 2 'Connect from' Connect from
17120 3 source_ip ':' source_port 10.0.1.2:33312
17121 4 'to' to
17122 5 destination_ip ':' destination_port 10.0.3.31:8012
17123 6 '(' frontend_name '/' mode ')' (www/HTTP)
17124
17125Detailed fields description :
17126 - "source_ip" is the IP address of the client which initiated the connection.
17127 - "source_port" is the TCP port of the client which initiated the connection.
17128 - "destination_ip" is the IP address the client connected to.
17129 - "destination_port" is the TCP port the client connected to.
17130 - "frontend_name" is the name of the frontend (or listener) which received
17131 and processed the connection.
17132 - "mode is the mode the frontend is operating (TCP or HTTP).
17133
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017134In case of a UNIX socket, the source and destination addresses are marked as
17135"unix:" and the ports reflect the internal ID of the socket which accepted the
17136connection (the same ID as reported in the stats).
17137
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017138It is advised not to use this deprecated format for newer installations as it
17139will eventually disappear.
17140
17141
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171428.2.2. TCP log format
17143---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017144
17145The TCP format is used when "option tcplog" is specified in the frontend, and
17146is the recommended format for pure TCP proxies. It provides a lot of precious
17147information for troubleshooting. Since this format includes timers and byte
17148counts, the log is normally emitted at the end of the session. It can be
17149emitted earlier if "option logasap" is specified, which makes sense in most
17150environments with long sessions such as remote terminals. Sessions which match
17151the "monitor" rules are never logged. It is also possible not to emit logs for
17152sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017153specifying "option dontlognull" in the frontend. Successful connections will
17154not be logged if "option dontlog-normal" is specified in the frontend. A few
17155fields may slightly vary depending on some configuration options, those are
17156marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017157
17158 Example :
17159 frontend fnt
17160 mode tcp
17161 option tcplog
17162 log global
17163 default_backend bck
17164
17165 backend bck
17166 server srv1 127.0.0.1:8000
17167
17168 >>> Feb 6 12:12:56 localhost \
17169 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
17170 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
17171
17172 Field Format Extract from the example above
17173 1 process_name '[' pid ']:' haproxy[14387]:
17174 2 client_ip ':' client_port 10.0.1.2:33313
17175 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
17176 4 frontend_name fnt
17177 5 backend_name '/' server_name bck/srv1
17178 6 Tw '/' Tc '/' Tt* 0/0/5007
17179 7 bytes_read* 212
17180 8 termination_state --
17181 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
17182 10 srv_queue '/' backend_queue 0/0
17183
17184Detailed fields description :
17185 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017186 connection to haproxy. If the connection was accepted on a UNIX socket
17187 instead, the IP address would be replaced with the word "unix". Note that
17188 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017189 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010017190 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017191 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017192
17193 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017194 If the connection was accepted on a UNIX socket instead, the port would be
17195 replaced with the ID of the accepting socket, which is also reported in the
17196 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017197
17198 - "accept_date" is the exact date when the connection was received by haproxy
17199 (which might be very slightly different from the date observed on the
17200 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020017201 the same date which may appear in any upstream firewall's log. When used in
17202 HTTP mode, the accept_date field will be reset to the first moment the
17203 connection is ready to receive a new request (end of previous response for
17204 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017205
17206 - "frontend_name" is the name of the frontend (or listener) which received
17207 and processed the connection.
17208
17209 - "backend_name" is the name of the backend (or listener) which was selected
17210 to manage the connection to the server. This will be the same as the
17211 frontend if no switching rule has been applied, which is common for TCP
17212 applications.
17213
17214 - "server_name" is the name of the last server to which the connection was
17215 sent, which might differ from the first one if there were connection errors
17216 and a redispatch occurred. Note that this server belongs to the backend
17217 which processed the request. If the connection was aborted before reaching
17218 a server, "<NOSRV>" is indicated instead of a server name.
17219
17220 - "Tw" is the total time in milliseconds spent waiting in the various queues.
17221 It can be "-1" if the connection was aborted before reaching the queue.
17222 See "Timers" below for more details.
17223
17224 - "Tc" is the total time in milliseconds spent waiting for the connection to
17225 establish to the final server, including retries. It can be "-1" if the
17226 connection was aborted before a connection could be established. See
17227 "Timers" below for more details.
17228
17229 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017230 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017231 "option logasap" was specified, then the time counting stops at the moment
17232 the log is emitted. In this case, a '+' sign is prepended before the value,
17233 indicating that the final one will be larger. See "Timers" below for more
17234 details.
17235
17236 - "bytes_read" is the total number of bytes transmitted from the server to
17237 the client when the log is emitted. If "option logasap" is specified, the
17238 this value will be prefixed with a '+' sign indicating that the final one
17239 may be larger. Please note that this value is a 64-bit counter, so log
17240 analysis tools must be able to handle it without overflowing.
17241
17242 - "termination_state" is the condition the session was in when the session
17243 ended. This indicates the session state, which side caused the end of
17244 session to happen, and for what reason (timeout, error, ...). The normal
17245 flags should be "--", indicating the session was closed by either end with
17246 no data remaining in buffers. See below "Session state at disconnection"
17247 for more details.
17248
17249 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017250 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017251 limits have been reached. For instance, if actconn is close to 512 when
17252 multiple connection errors occur, chances are high that the system limits
17253 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017254 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017255
17256 - "feconn" is the total number of concurrent connections on the frontend when
17257 the session was logged. It is useful to estimate the amount of resource
17258 required to sustain high loads, and to detect when the frontend's "maxconn"
17259 has been reached. Most often when this value increases by huge jumps, it is
17260 because there is congestion on the backend servers, but sometimes it can be
17261 caused by a denial of service attack.
17262
17263 - "beconn" is the total number of concurrent connections handled by the
17264 backend when the session was logged. It includes the total number of
17265 concurrent connections active on servers as well as the number of
17266 connections pending in queues. It is useful to estimate the amount of
17267 additional servers needed to support high loads for a given application.
17268 Most often when this value increases by huge jumps, it is because there is
17269 congestion on the backend servers, but sometimes it can be caused by a
17270 denial of service attack.
17271
17272 - "srv_conn" is the total number of concurrent connections still active on
17273 the server when the session was logged. It can never exceed the server's
17274 configured "maxconn" parameter. If this value is very often close or equal
17275 to the server's "maxconn", it means that traffic regulation is involved a
17276 lot, meaning that either the server's maxconn value is too low, or that
17277 there aren't enough servers to process the load with an optimal response
17278 time. When only one of the server's "srv_conn" is high, it usually means
17279 that this server has some trouble causing the connections to take longer to
17280 be processed than on other servers.
17281
17282 - "retries" is the number of connection retries experienced by this session
17283 when trying to connect to the server. It must normally be zero, unless a
17284 server is being stopped at the same moment the connection was attempted.
17285 Frequent retries generally indicate either a network problem between
17286 haproxy and the server, or a misconfigured system backlog on the server
17287 preventing new connections from being queued. This field may optionally be
17288 prefixed with a '+' sign, indicating that the session has experienced a
17289 redispatch after the maximal retry count has been reached on the initial
17290 server. In this case, the server name appearing in the log is the one the
17291 connection was redispatched to, and not the first one, though both may
17292 sometimes be the same in case of hashing for instance. So as a general rule
17293 of thumb, when a '+' is present in front of the retry count, this count
17294 should not be attributed to the logged server.
17295
17296 - "srv_queue" is the total number of requests which were processed before
17297 this one in the server queue. It is zero when the request has not gone
17298 through the server queue. It makes it possible to estimate the approximate
17299 server's response time by dividing the time spent in queue by the number of
17300 requests in the queue. It is worth noting that if a session experiences a
17301 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010017302 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017303 backend queue unless a redispatch occurs.
17304
17305 - "backend_queue" is the total number of requests which were processed before
17306 this one in the backend's global queue. It is zero when the request has not
17307 gone through the global queue. It makes it possible to estimate the average
17308 queue length, which easily translates into a number of missing servers when
17309 divided by a server's "maxconn" parameter. It is worth noting that if a
17310 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010017311 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017312 through both the server queue and the backend queue unless a redispatch
17313 occurs.
17314
17315
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200173168.2.3. HTTP log format
17317----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017318
17319The HTTP format is the most complete and the best suited for HTTP proxies. It
17320is enabled by when "option httplog" is specified in the frontend. It provides
17321the same level of information as the TCP format with additional features which
17322are specific to the HTTP protocol. Just like the TCP format, the log is usually
17323emitted at the end of the session, unless "option logasap" is specified, which
17324generally only makes sense for download sites. A session which matches the
17325"monitor" rules will never logged. It is also possible not to log sessions for
17326which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017327frontend. Successful connections will not be logged if "option dontlog-normal"
17328is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017329
17330Most fields are shared with the TCP log, some being different. A few fields may
17331slightly vary depending on some configuration options. Those ones are marked
17332with a star ('*') after the field name below.
17333
17334 Example :
17335 frontend http-in
17336 mode http
17337 option httplog
17338 log global
17339 default_backend bck
17340
17341 backend static
17342 server srv1 127.0.0.1:8000
17343
17344 >>> Feb 6 12:14:14 localhost \
17345 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
17346 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 +010017347 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017348
17349 Field Format Extract from the example above
17350 1 process_name '[' pid ']:' haproxy[14389]:
17351 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017352 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017353 4 frontend_name http-in
17354 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017355 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017356 7 status_code 200
17357 8 bytes_read* 2750
17358 9 captured_request_cookie -
17359 10 captured_response_cookie -
17360 11 termination_state ----
17361 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
17362 13 srv_queue '/' backend_queue 0/0
17363 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
17364 15 '{' captured_response_headers* '}' {}
17365 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010017366
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017367Detailed fields description :
17368 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017369 connection to haproxy. If the connection was accepted on a UNIX socket
17370 instead, the IP address would be replaced with the word "unix". Note that
17371 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017372 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010017373 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017374 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017375
17376 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017377 If the connection was accepted on a UNIX socket instead, the port would be
17378 replaced with the ID of the accepting socket, which is also reported in the
17379 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017380
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017381 - "request_date" is the exact date when the first byte of the HTTP request
17382 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017383
17384 - "frontend_name" is the name of the frontend (or listener) which received
17385 and processed the connection.
17386
17387 - "backend_name" is the name of the backend (or listener) which was selected
17388 to manage the connection to the server. This will be the same as the
17389 frontend if no switching rule has been applied.
17390
17391 - "server_name" is the name of the last server to which the connection was
17392 sent, which might differ from the first one if there were connection errors
17393 and a redispatch occurred. Note that this server belongs to the backend
17394 which processed the request. If the request was aborted before reaching a
17395 server, "<NOSRV>" is indicated instead of a server name. If the request was
17396 intercepted by the stats subsystem, "<STATS>" is indicated instead.
17397
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017398 - "TR" is the total time in milliseconds spent waiting for a full HTTP
17399 request from the client (not counting body) after the first byte was
17400 received. It can be "-1" if the connection was aborted before a complete
John Roesler27c754c2019-07-10 15:45:51 -050017401 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017402 always be very small because a request generally fits in one single packet.
17403 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020017404 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
17405 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017406
17407 - "Tw" is the total time in milliseconds spent waiting in the various queues.
17408 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020017409 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017410
17411 - "Tc" is the total time in milliseconds spent waiting for the connection to
17412 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020017413 request was aborted before a connection could be established. See section
17414 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017415
17416 - "Tr" is the total time in milliseconds spent waiting for the server to send
17417 a full HTTP response, not counting data. It can be "-1" if the request was
17418 aborted before a complete response could be received. It generally matches
17419 the server's processing time for the request, though it may be altered by
17420 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020017421 "GET" requests generally indicate an overloaded server. See section 8.4
17422 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017423
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017424 - "Ta" is the time the request remained active in haproxy, which is the total
17425 time in milliseconds elapsed between the first byte of the request was
17426 received and the last byte of response was sent. It covers all possible
17427 processing except the handshake (see Th) and idle time (see Ti). There is
17428 one exception, if "option logasap" was specified, then the time counting
17429 stops at the moment the log is emitted. In this case, a '+' sign is
17430 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020017431 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017432
17433 - "status_code" is the HTTP status code returned to the client. This status
17434 is generally set by the server, but it might also be set by haproxy when
17435 the server cannot be reached or when its response is blocked by haproxy.
17436
17437 - "bytes_read" is the total number of bytes transmitted to the client when
17438 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roesler27c754c2019-07-10 15:45:51 -050017439 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017440 the final one may be larger. Please note that this value is a 64-bit
17441 counter, so log analysis tools must be able to handle it without
17442 overflowing.
17443
17444 - "captured_request_cookie" is an optional "name=value" entry indicating that
17445 the client had this cookie in the request. The cookie name and its maximum
17446 length are defined by the "capture cookie" statement in the frontend
17447 configuration. The field is a single dash ('-') when the option is not
17448 set. Only one cookie may be captured, it is generally used to track session
17449 ID exchanges between a client and a server to detect session crossing
17450 between clients due to application bugs. For more details, please consult
17451 the section "Capturing HTTP headers and cookies" below.
17452
17453 - "captured_response_cookie" is an optional "name=value" entry indicating
17454 that the server has returned a cookie with its response. The cookie name
17455 and its maximum length are defined by the "capture cookie" statement in the
17456 frontend configuration. The field is a single dash ('-') when the option is
17457 not set. Only one cookie may be captured, it is generally used to track
17458 session ID exchanges between a client and a server to detect session
17459 crossing between clients due to application bugs. For more details, please
17460 consult the section "Capturing HTTP headers and cookies" below.
17461
17462 - "termination_state" is the condition the session was in when the session
17463 ended. This indicates the session state, which side caused the end of
17464 session to happen, for what reason (timeout, error, ...), just like in TCP
17465 logs, and information about persistence operations on cookies in the last
17466 two characters. The normal flags should begin with "--", indicating the
17467 session was closed by either end with no data remaining in buffers. See
17468 below "Session state at disconnection" for more details.
17469
17470 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017471 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017472 limits have been reached. For instance, if actconn is close to 512 or 1024
17473 when multiple connection errors occur, chances are high that the system
17474 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017475 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017476 system.
17477
17478 - "feconn" is the total number of concurrent connections on the frontend when
17479 the session was logged. It is useful to estimate the amount of resource
17480 required to sustain high loads, and to detect when the frontend's "maxconn"
17481 has been reached. Most often when this value increases by huge jumps, it is
17482 because there is congestion on the backend servers, but sometimes it can be
17483 caused by a denial of service attack.
17484
17485 - "beconn" is the total number of concurrent connections handled by the
17486 backend when the session was logged. It includes the total number of
17487 concurrent connections active on servers as well as the number of
17488 connections pending in queues. It is useful to estimate the amount of
17489 additional servers needed to support high loads for a given application.
17490 Most often when this value increases by huge jumps, it is because there is
17491 congestion on the backend servers, but sometimes it can be caused by a
17492 denial of service attack.
17493
17494 - "srv_conn" is the total number of concurrent connections still active on
17495 the server when the session was logged. It can never exceed the server's
17496 configured "maxconn" parameter. If this value is very often close or equal
17497 to the server's "maxconn", it means that traffic regulation is involved a
17498 lot, meaning that either the server's maxconn value is too low, or that
17499 there aren't enough servers to process the load with an optimal response
17500 time. When only one of the server's "srv_conn" is high, it usually means
17501 that this server has some trouble causing the requests to take longer to be
17502 processed than on other servers.
17503
17504 - "retries" is the number of connection retries experienced by this session
17505 when trying to connect to the server. It must normally be zero, unless a
17506 server is being stopped at the same moment the connection was attempted.
17507 Frequent retries generally indicate either a network problem between
17508 haproxy and the server, or a misconfigured system backlog on the server
17509 preventing new connections from being queued. This field may optionally be
17510 prefixed with a '+' sign, indicating that the session has experienced a
17511 redispatch after the maximal retry count has been reached on the initial
17512 server. In this case, the server name appearing in the log is the one the
17513 connection was redispatched to, and not the first one, though both may
17514 sometimes be the same in case of hashing for instance. So as a general rule
17515 of thumb, when a '+' is present in front of the retry count, this count
17516 should not be attributed to the logged server.
17517
17518 - "srv_queue" is the total number of requests which were processed before
17519 this one in the server queue. It is zero when the request has not gone
17520 through the server queue. It makes it possible to estimate the approximate
17521 server's response time by dividing the time spent in queue by the number of
17522 requests in the queue. It is worth noting that if a session experiences a
17523 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010017524 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017525 backend queue unless a redispatch occurs.
17526
17527 - "backend_queue" is the total number of requests which were processed before
17528 this one in the backend's global queue. It is zero when the request has not
17529 gone through the global queue. It makes it possible to estimate the average
17530 queue length, which easily translates into a number of missing servers when
17531 divided by a server's "maxconn" parameter. It is worth noting that if a
17532 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010017533 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017534 through both the server queue and the backend queue unless a redispatch
17535 occurs.
17536
17537 - "captured_request_headers" is a list of headers captured in the request due
17538 to the presence of the "capture request header" statement in the frontend.
17539 Multiple headers can be captured, they will be delimited by a vertical bar
17540 ('|'). When no capture is enabled, the braces do not appear, causing a
17541 shift of remaining fields. It is important to note that this field may
17542 contain spaces, and that using it requires a smarter log parser than when
17543 it's not used. Please consult the section "Capturing HTTP headers and
17544 cookies" below for more details.
17545
17546 - "captured_response_headers" is a list of headers captured in the response
17547 due to the presence of the "capture response header" statement in the
17548 frontend. Multiple headers can be captured, they will be delimited by a
17549 vertical bar ('|'). When no capture is enabled, the braces do not appear,
17550 causing a shift of remaining fields. It is important to note that this
17551 field may contain spaces, and that using it requires a smarter log parser
17552 than when it's not used. Please consult the section "Capturing HTTP headers
17553 and cookies" below for more details.
17554
17555 - "http_request" is the complete HTTP request line, including the method,
17556 request and HTTP version string. Non-printable characters are encoded (see
17557 below the section "Non-printable characters"). This is always the last
17558 field, and it is always delimited by quotes and is the only one which can
17559 contain quotes. If new fields are added to the log format, they will be
17560 added before this field. This field might be truncated if the request is
17561 huge and does not fit in the standard syslog buffer (1024 characters). This
17562 is the reason why this field must always remain the last one.
17563
17564
Cyril Bontédc4d9032012-04-08 21:57:39 +0200175658.2.4. Custom log format
17566------------------------
William Lallemand48940402012-01-30 16:47:22 +010017567
Willy Tarreau2beef582012-12-20 17:22:52 +010017568The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010017569mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010017570
Davor Ocelice9ed2812017-12-25 17:49:28 +010017571HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010017572Variables can take arguments using braces ('{}'), and multiple arguments are
17573separated by commas within the braces. Flags may be added or removed by
17574prefixing them with a '+' or '-' sign.
17575
17576Special variable "%o" may be used to propagate its flags to all other
17577variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010017578("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010017579
Willy Tarreauc8368452012-12-21 00:09:23 +010017580If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020017581as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010017582less common information such as the client's SSL certificate's DN, or to log
17583the key that would be used to store an entry into a stick table.
17584
William Lallemand48940402012-01-30 16:47:22 +010017585Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017586In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010017587in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010017588
Dragan Dosen835b9212016-02-12 13:23:03 +010017589Note: when using the RFC5424 syslog message format, the characters '"',
17590'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
17591https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
17592such cases, the use of the flag "E" should be considered.
17593
William Lallemand48940402012-01-30 16:47:22 +010017594Flags are :
17595 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017596 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010017597 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
17598 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010017599
17600 Example:
17601
17602 log-format %T\ %t\ Some\ Text
17603 log-format %{+Q}o\ %t\ %s\ %{-Q}r
17604
Dragan Dosen835b9212016-02-12 13:23:03 +010017605 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
17606
William Lallemand48940402012-01-30 16:47:22 +010017607At the moment, the default HTTP format is defined this way :
17608
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017609 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
17610 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010017611
William Lallemandbddd4fd2012-02-27 11:23:10 +010017612the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010017613
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017614 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
17615 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
17616 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010017617
William Lallemandbddd4fd2012-02-27 11:23:10 +010017618and the default TCP format is defined this way :
17619
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017620 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
17621 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010017622
William Lallemand48940402012-01-30 16:47:22 +010017623Please refer to the table below for currently defined variables :
17624
William Lallemandbddd4fd2012-02-27 11:23:10 +010017625 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017626 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017627 +---+------+-----------------------------------------------+-------------+
17628 | | %o | special variable, apply flags on all next var | |
17629 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010017630 | | %B | bytes_read (from server to client) | numeric |
17631 | H | %CC | captured_request_cookie | string |
17632 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020017633 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000017634 | H | %HM | HTTP method (ex: POST) | string |
17635 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000017636 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000017637 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
17638 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010017639 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020017640 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017641 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017642 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017643 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020017644 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080017645 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017646 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
17647 | H | %Ti | idle time before the HTTP request | numeric |
17648 | H | %Tq | Th + Ti + TR | numeric |
17649 | H | %TR | time to receive the full request from 1st byte| numeric |
17650 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017651 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017652 | | %Tt | Tt | numeric |
17653 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010017654 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017655 | | %ac | actconn | numeric |
17656 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017657 | | %bc | beconn (backend concurrent connections) | numeric |
17658 | | %bi | backend_source_ip (connecting address) | IP |
17659 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017660 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010017661 | | %ci | client_ip (accepted address) | IP |
17662 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017663 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017664 | | %fc | feconn (frontend concurrent connections) | numeric |
17665 | | %fi | frontend_ip (accepting address) | IP |
17666 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020017667 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020017668 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017669 | | %hr | captured_request_headers default style | string |
17670 | | %hrl | captured_request_headers CLF style | string list |
17671 | | %hs | captured_response_headers default style | string |
17672 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020017673 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017674 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017675 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017676 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010017677 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017678 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017679 | | %sc | srv_conn (server concurrent connections) | numeric |
17680 | | %si | server_IP (target address) | IP |
17681 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017682 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017683 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
17684 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017685 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017686 | H | %tr | date_time of HTTP request | date |
17687 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020017688 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017689 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017690 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017691 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010017692
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017693 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010017694
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010017695
176968.2.5. Error log format
17697-----------------------
17698
17699When an incoming connection fails due to an SSL handshake or an invalid PROXY
17700protocol header, haproxy will log the event using a shorter, fixed line format.
17701By default, logs are emitted at the LOG_INFO level, unless the option
17702"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017703will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010017704logged if the "dontlognull" option is set.
17705
17706The format looks like this :
17707
17708 >>> Dec 3 18:27:14 localhost \
17709 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
17710 Connection error during SSL handshake
17711
17712 Field Format Extract from the example above
17713 1 process_name '[' pid ']:' haproxy[6103]:
17714 2 client_ip ':' client_port 127.0.0.1:56059
17715 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
17716 4 frontend_name "/" bind_name ":" frt/f1:
17717 5 message Connection error during SSL handshake
17718
17719These fields just provide minimal information to help debugging connection
17720failures.
17721
17722
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200177238.3. Advanced logging options
17724-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017725
17726Some advanced logging options are often looked for but are not easy to find out
17727just by looking at the various options. Here is an entry point for the few
17728options which can enable better logging. Please refer to the keywords reference
17729for more information about their usage.
17730
17731
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200177328.3.1. Disabling logging of external tests
17733------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017734
17735It is quite common to have some monitoring tools perform health checks on
17736haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
17737commercial load-balancer, and sometimes it will simply be a more complete
17738monitoring system such as Nagios. When the tests are very frequent, users often
17739ask how to disable logging for those checks. There are three possibilities :
17740
17741 - if connections come from everywhere and are just TCP probes, it is often
17742 desired to simply disable logging of connections without data exchange, by
17743 setting "option dontlognull" in the frontend. It also disables logging of
17744 port scans, which may or may not be desired.
17745
17746 - if the connection come from a known source network, use "monitor-net" to
17747 declare this network as monitoring only. Any host in this network will then
17748 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017749 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017750 such as other load-balancers.
17751
17752 - if the tests are performed on a known URI, use "monitor-uri" to declare
17753 this URI as dedicated to monitoring. Any host sending this request will
17754 only get the result of a health-check, and the request will not be logged.
17755
17756
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200177578.3.2. Logging before waiting for the session to terminate
17758----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017759
17760The problem with logging at end of connection is that you have no clue about
17761what is happening during very long sessions, such as remote terminal sessions
17762or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010017763"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017764just before data transfer begins. This means that in case of TCP, it will still
17765log the connection status to the server, and in case of HTTP, it will log just
17766after processing the server headers. In this case, the number of bytes reported
17767is the number of header bytes sent to the client. In order to avoid confusion
17768with normal logs, the total time field and the number of bytes are prefixed
17769with a '+' sign which means that real numbers are certainly larger.
17770
17771
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200177728.3.3. Raising log level upon errors
17773------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017774
17775Sometimes it is more convenient to separate normal traffic from errors logs,
17776for instance in order to ease error monitoring from log files. When the option
17777"log-separate-errors" is used, connections which experience errors, timeouts,
17778retries, redispatches or HTTP status codes 5xx will see their syslog level
17779raised from "info" to "err". This will help a syslog daemon store the log in
17780a separate file. It is very important to keep the errors in the normal traffic
17781file too, so that log ordering is not altered. You should also be careful if
17782you already have configured your syslog daemon to store all logs higher than
17783"notice" in an "admin" file, because the "err" level is higher than "notice".
17784
17785
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200177868.3.4. Disabling logging of successful connections
17787--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017788
17789Although this may sound strange at first, some large sites have to deal with
17790multiple thousands of logs per second and are experiencing difficulties keeping
17791them intact for a long time or detecting errors within them. If the option
17792"dontlog-normal" is set on the frontend, all normal connections will not be
17793logged. In this regard, a normal connection is defined as one without any
17794error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
17795and a response with a status 5xx is not considered normal and will be logged
17796too. Of course, doing is is really discouraged as it will remove most of the
17797useful information from the logs. Do this only if you have no other
17798alternative.
17799
17800
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200178018.4. Timing events
17802------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017803
17804Timers provide a great help in troubleshooting network problems. All values are
17805reported in milliseconds (ms). These timers should be used in conjunction with
17806the session termination flags. In TCP mode with "option tcplog" set on the
17807frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017808mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
17809addition, three other measures are provided, "Th", "Ti", and "Tq".
17810
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010017811Timings events in HTTP mode:
17812
17813 first request 2nd request
17814 |<-------------------------------->|<-------------- ...
17815 t tr t tr ...
17816 ---|----|----|----|----|----|----|----|----|--
17817 : Th Ti TR Tw Tc Tr Td : Ti ...
17818 :<---- Tq ---->: :
17819 :<-------------- Tt -------------->:
17820 :<--------- Ta --------->:
17821
17822Timings events in TCP mode:
17823
17824 TCP session
17825 |<----------------->|
17826 t t
17827 ---|----|----|----|----|---
17828 | Th Tw Tc Td |
17829 |<------ Tt ------->|
17830
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017831 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017832 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017833 only happen once during the whole connection's lifetime. A large time here
17834 may indicate that the client only pre-established the connection without
17835 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010017836 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020017837 SSL handshake was very expensive to compute. Please note that this time is
17838 reported only before the first request, so it is safe to average it over
17839 all request to calculate the amortized value. The second and subsequent
17840 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017841
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017842 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
17843 counts between the end of the handshakes and the first byte of the HTTP
17844 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020017845 to count after the end of the transmission the previous response. When a
17846 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
17847 after the previous request. Some browsers pre-establish connections to a
17848 server in order to reduce the latency of a future request, and keep them
17849 pending until they need it. This delay will be reported as the idle time. A
17850 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017851
17852 - TR: total time to get the client request (HTTP mode only). It's the time
17853 elapsed between the first bytes received and the moment the proxy received
17854 the empty line marking the end of the HTTP headers. The value "-1"
17855 indicates that the end of headers has never been seen. This happens when
17856 the client closes prematurely or times out. This time is usually very short
17857 since most requests fit in a single packet. A large time may indicate a
17858 request typed by hand during a test.
17859
17860 - Tq: total time to get the client request from the accept date or since the
17861 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010017862 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 +020017863 returns -1 as well. This timer used to be very useful before the arrival of
17864 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
17865 it in favor of TR nowadays, as the idle time adds a lot of noise to the
17866 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017867
17868 - Tw: total time spent in the queues waiting for a connection slot. It
17869 accounts for backend queue as well as the server queues, and depends on the
17870 queue size, and the time needed for the server to complete previous
17871 requests. The value "-1" means that the request was killed before reaching
17872 the queue, which is generally what happens with invalid or denied requests.
17873
17874 - Tc: total time to establish the TCP connection to the server. It's the time
17875 elapsed between the moment the proxy sent the connection request, and the
17876 moment it was acknowledged by the server, or between the TCP SYN packet and
17877 the matching SYN/ACK packet in return. The value "-1" means that the
17878 connection never established.
17879
17880 - Tr: server response time (HTTP mode only). It's the time elapsed between
17881 the moment the TCP connection was established to the server and the moment
17882 the server sent its complete response headers. It purely shows its request
17883 processing time, without the network overhead due to the data transmission.
17884 It is worth noting that when the client has data to send to the server, for
17885 instance during a POST request, the time already runs, and this can distort
17886 apparent response time. For this reason, it's generally wise not to trust
17887 too much this field for POST requests initiated from clients behind an
17888 untrusted network. A value of "-1" here means that the last the response
17889 header (empty line) was never seen, most likely because the server timeout
17890 stroke before the server managed to process the request.
17891
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017892 - Ta: total active time for the HTTP request, between the moment the proxy
17893 received the first byte of the request header and the emission of the last
17894 byte of the response body. The exception is when the "logasap" option is
17895 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
17896 a '+' sign. From this field, we can deduce "Td", the data transmission time,
17897 by subtracting other timers when valid :
17898
17899 Td = Ta - (TR + Tw + Tc + Tr)
17900
17901 Timers with "-1" values have to be excluded from this equation. Note that
17902 "Ta" can never be negative.
17903
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017904 - Tt: total session duration time, between the moment the proxy accepted it
17905 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017906 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
17907 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017908 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017909
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017910 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017911
17912 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017913 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
17914 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017915
17916These timers provide precious indications on trouble causes. Since the TCP
17917protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
17918that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017919due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
17920"Tt" is close to a timeout value specified in the configuration, it often means
17921that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017922
17923Most common cases :
17924
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017925 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
17926 the client and the proxy. This is very rare on local networks but might
17927 happen when clients are on far remote networks and send large requests. It
17928 may happen that values larger than usual appear here without any network
17929 cause. Sometimes, during an attack or just after a resource starvation has
17930 ended, haproxy may accept thousands of connections in a few milliseconds.
17931 The time spent accepting these connections will inevitably slightly delay
17932 processing of other connections, and it can happen that request times in the
17933 order of a few tens of milliseconds are measured after a few thousands of
17934 new connections have been accepted at once. Using one of the keep-alive
17935 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020017936 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017937
17938 - If "Tc" is close to 3000, a packet has probably been lost between the
17939 server and the proxy during the server connection phase. This value should
17940 always be very low, such as 1 ms on local networks and less than a few tens
17941 of ms on remote networks.
17942
Willy Tarreau55165fe2009-05-10 12:02:55 +020017943 - If "Tr" is nearly always lower than 3000 except some rare values which seem
17944 to be the average majored by 3000, there are probably some packets lost
17945 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017946
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017947 - If "Ta" is large even for small byte counts, it generally is because
17948 neither the client nor the server decides to close the connection while
17949 haproxy is running in tunnel mode and both have agreed on a keep-alive
17950 connection mode. In order to solve this issue, it will be needed to specify
17951 one of the HTTP options to manipulate keep-alive or close options on either
17952 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
17953 important when connection regulation is used with the "maxconn" option on
17954 the servers, since no new connection will be sent to the server until
17955 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017956
17957Other noticeable HTTP log cases ('xx' means any value to be ignored) :
17958
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017959 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017960 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017961 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017962
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017963 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017964 or it aborted too early. Check the session termination flags
17965 then "timeout http-request" and "timeout client" settings.
17966
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017967 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017968 servers were out of order, because the request was invalid
17969 or forbidden by ACL rules. Check the session termination
17970 flags.
17971
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017972 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
17973 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017974 Check the session termination flags, then check the
17975 "timeout connect" setting. Note that the tarpit action might
17976 return similar-looking patterns, with "Tw" equal to the time
17977 the client connection was maintained open.
17978
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017979 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017980 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017981 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017982 termination flags, then check the "timeout server" setting.
17983
17984
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200179858.5. Session state at disconnection
17986-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017987
17988TCP and HTTP logs provide a session termination indicator in the
17989"termination_state" field, just before the number of active connections. It is
179902-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
17991each of which has a special meaning :
17992
17993 - On the first character, a code reporting the first event which caused the
17994 session to terminate :
17995
17996 C : the TCP session was unexpectedly aborted by the client.
17997
17998 S : the TCP session was unexpectedly aborted by the server, or the
17999 server explicitly refused it.
18000
18001 P : the session was prematurely aborted by the proxy, because of a
18002 connection limit enforcement, because a DENY filter was matched,
18003 because of a security check which detected and blocked a dangerous
18004 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010018005 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020018006
18007 L : the session was locally processed by haproxy and was not passed to
18008 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018009
18010 R : a resource on the proxy has been exhausted (memory, sockets, source
18011 ports, ...). Usually, this appears during the connection phase, and
18012 system logs should contain a copy of the precise error. If this
18013 happens, it must be considered as a very serious anomaly which
18014 should be fixed as soon as possible by any means.
18015
18016 I : an internal error was identified by the proxy during a self-check.
18017 This should NEVER happen, and you are encouraged to report any log
18018 containing this, because this would almost certainly be a bug. It
18019 would be wise to preventively restart the process after such an
18020 event too, in case it would be caused by memory corruption.
18021
Simon Horman752dc4a2011-06-21 14:34:59 +090018022 D : the session was killed by haproxy because the server was detected
18023 as down and was configured to kill all connections when going down.
18024
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070018025 U : the session was killed by haproxy on this backup server because an
18026 active server was detected as up and was configured to kill all
18027 backup connections when going up.
18028
Willy Tarreaua2a64e92011-09-07 23:01:56 +020018029 K : the session was actively killed by an admin operating on haproxy.
18030
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018031 c : the client-side timeout expired while waiting for the client to
18032 send or receive data.
18033
18034 s : the server-side timeout expired while waiting for the server to
18035 send or receive data.
18036
18037 - : normal session completion, both the client and the server closed
18038 with nothing left in the buffers.
18039
18040 - on the second character, the TCP or HTTP session state when it was closed :
18041
Willy Tarreauf7b30a92010-12-06 22:59:17 +010018042 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018043 (HTTP mode only). Nothing was sent to any server.
18044
18045 Q : the proxy was waiting in the QUEUE for a connection slot. This can
18046 only happen when servers have a 'maxconn' parameter set. It can
18047 also happen in the global queue after a redispatch consecutive to
18048 a failed attempt to connect to a dying server. If no redispatch is
18049 reported, then no connection attempt was made to any server.
18050
18051 C : the proxy was waiting for the CONNECTION to establish on the
18052 server. The server might at most have noticed a connection attempt.
18053
18054 H : the proxy was waiting for complete, valid response HEADERS from the
18055 server (HTTP only).
18056
18057 D : the session was in the DATA phase.
18058
18059 L : the proxy was still transmitting LAST data to the client while the
18060 server had already finished. This one is very rare as it can only
18061 happen when the client dies while receiving the last packets.
18062
18063 T : the request was tarpitted. It has been held open with the client
18064 during the whole "timeout tarpit" duration or until the client
18065 closed, both of which will be reported in the "Tw" timer.
18066
18067 - : normal session completion after end of data transfer.
18068
18069 - the third character tells whether the persistence cookie was provided by
18070 the client (only in HTTP mode) :
18071
18072 N : the client provided NO cookie. This is usually the case for new
18073 visitors, so counting the number of occurrences of this flag in the
18074 logs generally indicate a valid trend for the site frequentation.
18075
18076 I : the client provided an INVALID cookie matching no known server.
18077 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020018078 cookies between HTTP/HTTPS sites, persistence conditionally
18079 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018080
18081 D : the client provided a cookie designating a server which was DOWN,
18082 so either "option persist" was used and the client was sent to
18083 this server, or it was not set and the client was redispatched to
18084 another server.
18085
Willy Tarreau996a92c2010-10-13 19:30:47 +020018086 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018087 server.
18088
Willy Tarreau996a92c2010-10-13 19:30:47 +020018089 E : the client provided a valid cookie, but with a last date which was
18090 older than what is allowed by the "maxidle" cookie parameter, so
18091 the cookie is consider EXPIRED and is ignored. The request will be
18092 redispatched just as if there was no cookie.
18093
18094 O : the client provided a valid cookie, but with a first date which was
18095 older than what is allowed by the "maxlife" cookie parameter, so
18096 the cookie is consider too OLD and is ignored. The request will be
18097 redispatched just as if there was no cookie.
18098
Willy Tarreauc89ccb62012-04-05 21:18:22 +020018099 U : a cookie was present but was not used to select the server because
18100 some other server selection mechanism was used instead (typically a
18101 "use-server" rule).
18102
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018103 - : does not apply (no cookie set in configuration).
18104
18105 - the last character reports what operations were performed on the persistence
18106 cookie returned by the server (only in HTTP mode) :
18107
18108 N : NO cookie was provided by the server, and none was inserted either.
18109
18110 I : no cookie was provided by the server, and the proxy INSERTED one.
18111 Note that in "cookie insert" mode, if the server provides a cookie,
18112 it will still be overwritten and reported as "I" here.
18113
Willy Tarreau996a92c2010-10-13 19:30:47 +020018114 U : the proxy UPDATED the last date in the cookie that was presented by
18115 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030018116 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020018117 date indicated in the cookie. If any other change happens, such as
18118 a redispatch, then the cookie will be marked as inserted instead.
18119
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018120 P : a cookie was PROVIDED by the server and transmitted as-is.
18121
18122 R : the cookie provided by the server was REWRITTEN by the proxy, which
18123 happens in "cookie rewrite" or "cookie prefix" modes.
18124
18125 D : the cookie provided by the server was DELETED by the proxy.
18126
18127 - : does not apply (no cookie set in configuration).
18128
Willy Tarreau996a92c2010-10-13 19:30:47 +020018129The combination of the two first flags gives a lot of information about what
18130was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018131helpful to detect server saturation, network troubles, local system resource
18132starvation, attacks, etc...
18133
18134The most common termination flags combinations are indicated below. They are
18135alphabetically sorted, with the lowercase set just after the upper case for
18136easier finding and understanding.
18137
18138 Flags Reason
18139
18140 -- Normal termination.
18141
18142 CC The client aborted before the connection could be established to the
18143 server. This can happen when haproxy tries to connect to a recently
18144 dead (or unchecked) server, and the client aborts while haproxy is
18145 waiting for the server to respond or for "timeout connect" to expire.
18146
18147 CD The client unexpectedly aborted during data transfer. This can be
18148 caused by a browser crash, by an intermediate equipment between the
18149 client and haproxy which decided to actively break the connection,
18150 by network routing issues between the client and haproxy, or by a
18151 keep-alive session between the server and the client terminated first
18152 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010018153
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018154 cD The client did not send nor acknowledge any data for as long as the
18155 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020018156 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018157
18158 CH The client aborted while waiting for the server to start responding.
18159 It might be the server taking too long to respond or the client
18160 clicking the 'Stop' button too fast.
18161
18162 cH The "timeout client" stroke while waiting for client data during a
18163 POST request. This is sometimes caused by too large TCP MSS values
18164 for PPPoE networks which cannot transport full-sized packets. It can
18165 also happen when client timeout is smaller than server timeout and
18166 the server takes too long to respond.
18167
18168 CQ The client aborted while its session was queued, waiting for a server
18169 with enough empty slots to accept it. It might be that either all the
18170 servers were saturated or that the assigned server was taking too
18171 long a time to respond.
18172
18173 CR The client aborted before sending a full HTTP request. Most likely
18174 the request was typed by hand using a telnet client, and aborted
18175 too early. The HTTP status code is likely a 400 here. Sometimes this
18176 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020018177 and the client. "option http-ignore-probes" can be used to ignore
18178 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018179
18180 cR The "timeout http-request" stroke before the client sent a full HTTP
18181 request. This is sometimes caused by too large TCP MSS values on the
18182 client side for PPPoE networks which cannot transport full-sized
18183 packets, or by clients sending requests by hand and not typing fast
18184 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020018185 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020018186 some browsers started to implement a "pre-connect" feature consisting
18187 in speculatively connecting to some recently visited web sites just
18188 in case the user would like to visit them. This results in many
18189 connections being established to web sites, which end up in 408
18190 Request Timeout if the timeout strikes first, or 400 Bad Request when
18191 the browser decides to close them first. These ones pollute the log
18192 and feed the error counters. Some versions of some browsers have even
18193 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010018194 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020018195 http-ignore-probes" in the frontend, resulting in connections with
18196 zero data transfer to be totally ignored. This will definitely hide
18197 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018198
18199 CT The client aborted while its session was tarpitted. It is important to
18200 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020018201 wrong tarpit rules have been written. If a lot of them happen, it
18202 might make sense to lower the "timeout tarpit" value to something
18203 closer to the average reported "Tw" timer, in order not to consume
18204 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018205
Willy Tarreau570f2212013-06-10 16:42:09 +020018206 LR The request was intercepted and locally handled by haproxy. Generally
18207 it means that this was a redirect or a stats request.
18208
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018209 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018210 the TCP connection (the proxy received a TCP RST or an ICMP message
18211 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010018212 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018213 or no ARP response on local network). When this happens in HTTP mode,
18214 the status code is likely a 502 or 503 here.
18215
18216 sC The "timeout connect" stroke before a connection to the server could
18217 complete. When this happens in HTTP mode, the status code is likely a
18218 503 or 504 here.
18219
18220 SD The connection to the server died with an error during the data
18221 transfer. This usually means that haproxy has received an RST from
18222 the server or an ICMP message from an intermediate equipment while
18223 exchanging data with the server. This can be caused by a server crash
18224 or by a network issue on an intermediate equipment.
18225
18226 sD The server did not send nor acknowledge any data for as long as the
18227 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010018228 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018229 load-balancers, ...), as well as keep-alive sessions maintained
18230 between the client and the server expiring first on haproxy.
18231
18232 SH The server aborted before sending its full HTTP response headers, or
18233 it crashed while processing the request. Since a server aborting at
18234 this moment is very rare, it would be wise to inspect its logs to
18235 control whether it crashed and why. The logged request may indicate a
18236 small set of faulty requests, demonstrating bugs in the application.
18237 Sometimes this might also be caused by an IDS killing the connection
18238 between haproxy and the server.
18239
18240 sH The "timeout server" stroke before the server could return its
18241 response headers. This is the most common anomaly, indicating too
18242 long transactions, probably caused by server or database saturation.
18243 The immediate workaround consists in increasing the "timeout server"
18244 setting, but it is important to keep in mind that the user experience
18245 will suffer from these long response times. The only long term
18246 solution is to fix the application.
18247
18248 sQ The session spent too much time in queue and has been expired. See
18249 the "timeout queue" and "timeout connect" settings to find out how to
18250 fix this if it happens too often. If it often happens massively in
18251 short periods, it may indicate general problems on the affected
18252 servers due to I/O or database congestion, or saturation caused by
18253 external attacks.
18254
18255 PC The proxy refused to establish a connection to the server because the
18256 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020018257 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018258 so that it does not happen anymore. This status is very rare and
18259 might happen when the global "ulimit-n" parameter is forced by hand.
18260
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010018261 PD The proxy blocked an incorrectly formatted chunked encoded message in
18262 a request or a response, after the server has emitted its headers. In
18263 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018264 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020018265 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010018266
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018267 PH The proxy blocked the server's response, because it was invalid,
18268 incomplete, dangerous (cache control), or matched a security filter.
18269 In any case, an HTTP 502 error is sent to the client. One possible
18270 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010018271 containing unauthorized characters. It is also possible but quite
18272 rare, that the proxy blocked a chunked-encoding request from the
18273 client due to an invalid syntax, before the server responded. In this
18274 case, an HTTP 400 error is sent to the client and reported in the
18275 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018276
18277 PR The proxy blocked the client's HTTP request, either because of an
18278 invalid HTTP syntax, in which case it returned an HTTP 400 error to
18279 the client, or because a deny filter matched, in which case it
18280 returned an HTTP 403 error.
18281
18282 PT The proxy blocked the client's request and has tarpitted its
18283 connection before returning it a 500 server error. Nothing was sent
18284 to the server. The connection was maintained open for as long as
18285 reported by the "Tw" timer field.
18286
18287 RC A local resource has been exhausted (memory, sockets, source ports)
18288 preventing the connection to the server from establishing. The error
18289 logs will tell precisely what was missing. This is very rare and can
18290 only be solved by proper system tuning.
18291
Willy Tarreau996a92c2010-10-13 19:30:47 +020018292The combination of the two last flags gives a lot of information about how
18293persistence was handled by the client, the server and by haproxy. This is very
18294important to troubleshoot disconnections, when users complain they have to
18295re-authenticate. The commonly encountered flags are :
18296
18297 -- Persistence cookie is not enabled.
18298
18299 NN No cookie was provided by the client, none was inserted in the
18300 response. For instance, this can be in insert mode with "postonly"
18301 set on a GET request.
18302
18303 II A cookie designating an invalid server was provided by the client,
18304 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040018305 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020018306 value can be presented by a client when no other server knows it.
18307
18308 NI No cookie was provided by the client, one was inserted in the
18309 response. This typically happens for first requests from every user
18310 in "insert" mode, which makes it an easy way to count real users.
18311
18312 VN A cookie was provided by the client, none was inserted in the
18313 response. This happens for most responses for which the client has
18314 already got a cookie.
18315
18316 VU A cookie was provided by the client, with a last visit date which is
18317 not completely up-to-date, so an updated cookie was provided in
18318 response. This can also happen if there was no date at all, or if
18319 there was a date but the "maxidle" parameter was not set, so that the
18320 cookie can be switched to unlimited time.
18321
18322 EI A cookie was provided by the client, with a last visit date which is
18323 too old for the "maxidle" parameter, so the cookie was ignored and a
18324 new cookie was inserted in the response.
18325
18326 OI A cookie was provided by the client, with a first visit date which is
18327 too old for the "maxlife" parameter, so the cookie was ignored and a
18328 new cookie was inserted in the response.
18329
18330 DI The server designated by the cookie was down, a new server was
18331 selected and a new cookie was emitted in the response.
18332
18333 VI The server designated by the cookie was not marked dead but could not
18334 be reached. A redispatch happened and selected another one, which was
18335 then advertised in the response.
18336
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018337
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200183388.6. Non-printable characters
18339-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018340
18341In order not to cause trouble to log analysis tools or terminals during log
18342consulting, non-printable characters are not sent as-is into log files, but are
18343converted to the two-digits hexadecimal representation of their ASCII code,
18344prefixed by the character '#'. The only characters that can be logged without
18345being escaped are comprised between 32 and 126 (inclusive). Obviously, the
18346escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
18347is the same for the character '"' which becomes "#22", as well as '{', '|' and
18348'}' when logging headers.
18349
18350Note that the space character (' ') is not encoded in headers, which can cause
18351issues for tools relying on space count to locate fields. A typical header
18352containing spaces is "User-Agent".
18353
18354Last, it has been observed that some syslog daemons such as syslog-ng escape
18355the quote ('"') with a backslash ('\'). The reverse operation can safely be
18356performed since no quote may appear anywhere else in the logs.
18357
18358
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200183598.7. Capturing HTTP cookies
18360---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018361
18362Cookie capture simplifies the tracking a complete user session. This can be
18363achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018364section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018365cookie will simultaneously be checked in the request ("Cookie:" header) and in
18366the response ("Set-Cookie:" header). The respective values will be reported in
18367the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018368locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018369not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
18370user switches to a new session for example, because the server will reassign it
18371a new cookie. It is also possible to detect if a server unexpectedly sets a
18372wrong cookie to a client, leading to session crossing.
18373
18374 Examples :
18375 # capture the first cookie whose name starts with "ASPSESSION"
18376 capture cookie ASPSESSION len 32
18377
18378 # capture the first cookie whose name is exactly "vgnvisitor"
18379 capture cookie vgnvisitor= len 32
18380
18381
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200183828.8. Capturing HTTP headers
18383---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018384
18385Header captures are useful to track unique request identifiers set by an upper
18386proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
18387the response, one can search for information about the response length, how the
18388server asked the cache to behave, or an object location during a redirection.
18389
18390Header captures are performed using the "capture request header" and "capture
18391response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018392section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018393
18394It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018395time. Non-existent headers are logged as empty strings, and if one header
18396appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018397are grouped within braces '{' and '}' in the same order as they were declared,
18398and delimited with a vertical bar '|' without any space. Response headers
18399follow the same representation, but are displayed after a space following the
18400request headers block. These blocks are displayed just before the HTTP request
18401in the logs.
18402
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020018403As a special case, it is possible to specify an HTTP header capture in a TCP
18404frontend. The purpose is to enable logging of headers which will be parsed in
18405an HTTP backend if the request is then switched to this HTTP backend.
18406
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018407 Example :
18408 # This instance chains to the outgoing proxy
18409 listen proxy-out
18410 mode http
18411 option httplog
18412 option logasap
18413 log global
18414 server cache1 192.168.1.1:3128
18415
18416 # log the name of the virtual server
18417 capture request header Host len 20
18418
18419 # log the amount of data uploaded during a POST
18420 capture request header Content-Length len 10
18421
18422 # log the beginning of the referrer
18423 capture request header Referer len 20
18424
18425 # server name (useful for outgoing proxies only)
18426 capture response header Server len 20
18427
18428 # logging the content-length is useful with "option logasap"
18429 capture response header Content-Length len 10
18430
Davor Ocelice9ed2812017-12-25 17:49:28 +010018431 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018432 capture response header Cache-Control len 8
18433
18434 # the Via header will report the next proxy's name
18435 capture response header Via len 20
18436
18437 # log the URL location during a redirection
18438 capture response header Location len 20
18439
18440 >>> Aug 9 20:26:09 localhost \
18441 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
18442 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
18443 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
18444 "GET http://fr.adserver.yahoo.com/"
18445
18446 >>> Aug 9 20:30:46 localhost \
18447 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
18448 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
18449 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018450 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018451
18452 >>> Aug 9 20:30:46 localhost \
18453 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
18454 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
18455 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
18456 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018457 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018458
18459
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200184608.9. Examples of logs
18461---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018462
18463These are real-world examples of logs accompanied with an explanation. Some of
18464them have been made up by hand. The syslog part has been removed for better
18465reading. Their sole purpose is to explain how to decipher them.
18466
18467 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
18468 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
18469 "HEAD / HTTP/1.0"
18470
18471 => long request (6.5s) entered by hand through 'telnet'. The server replied
18472 in 147 ms, and the session ended normally ('----')
18473
18474 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
18475 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
18476 0/9 "HEAD / HTTP/1.0"
18477
18478 => Idem, but the request was queued in the global queue behind 9 other
18479 requests, and waited there for 1230 ms.
18480
18481 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
18482 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
18483 "GET /image.iso HTTP/1.0"
18484
18485 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018486 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018487 14 ms, 243 bytes of headers were sent to the client, and total time from
18488 accept to first data byte is 30 ms.
18489
18490 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
18491 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
18492 "GET /cgi-bin/bug.cgi? HTTP/1.0"
18493
18494 => the proxy blocked a server response either because of an "rspdeny" or
18495 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020018496 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018497 risked being cached. In this case, the response is replaced with a "502
18498 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
18499 to return the 502 and not the server.
18500
18501 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018502 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 +010018503
18504 => the client never completed its request and aborted itself ("C---") after
18505 8.5s, while the proxy was waiting for the request headers ("-R--").
18506 Nothing was sent to any server.
18507
18508 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
18509 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
18510
18511 => The client never completed its request, which was aborted by the
18512 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010018513 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018514 send a 408 return code to the client.
18515
18516 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
18517 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
18518
18519 => This log was produced with "option tcplog". The client timed out after
18520 5 seconds ("c----").
18521
18522 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
18523 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018524 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018525
18526 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018527 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018528 (config says 'retries 3'), and no redispatch (otherwise we would have
18529 seen "/+3"). Status code 503 was returned to the client. There were 115
18530 connections on this server, 202 connections on this proxy, and 205 on
18531 the global process. It is possible that the server refused the
18532 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010018533
Willy Tarreau52b2d222011-09-07 23:48:48 +020018534
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200185359. Supported filters
18536--------------------
18537
18538Here are listed officially supported filters with the list of parameters they
18539accept. Depending on compile options, some of these filters might be
18540unavailable. The list of available filters is reported in haproxy -vv.
18541
18542See also : "filter"
18543
185449.1. Trace
18545----------
18546
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010018547filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018548
18549 Arguments:
18550 <name> is an arbitrary name that will be reported in
18551 messages. If no name is provided, "TRACE" is used.
18552
18553 <random-parsing> enables the random parsing of data exchanged between
18554 the client and the server. By default, this filter
18555 parses all available data. With this parameter, it
18556 only parses a random amount of the available data.
18557
Davor Ocelice9ed2812017-12-25 17:49:28 +010018558 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018559 default, this filter forwards all previously parsed
18560 data. With this parameter, it only forwards a random
18561 amount of the parsed data.
18562
Davor Ocelice9ed2812017-12-25 17:49:28 +010018563 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010018564
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018565This filter can be used as a base to develop new filters. It defines all
18566callbacks and print a message on the standard error stream (stderr) with useful
18567information for all of them. It may be useful to debug the activity of other
18568filters or, quite simply, HAProxy's activity.
18569
18570Using <random-parsing> and/or <random-forwarding> parameters is a good way to
18571tests the behavior of a filter that parses data exchanged between a client and
18572a server by adding some latencies in the processing.
18573
18574
185759.2. HTTP compression
18576---------------------
18577
18578filter compression
18579
18580The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
18581keyword must still be used to enable and configure the HTTP compression. And
Christopher Faulet27d93c32018-12-15 22:32:02 +010018582when no other filter is used, it is enough. When used with the cache enabled,
18583it is also enough. In this case, the compression is always done after the
18584response is stored in the cache. But it is mandatory to explicitly use a filter
18585line to enable the HTTP compression when at least one filter other than the
18586cache is used for the same listener/frontend/backend. This is important to know
18587the filters evaluation order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018588
Christopher Faulet27d93c32018-12-15 22:32:02 +010018589See also : "compression" and section 9.4 about the cache filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018590
18591
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200185929.3. Stream Processing Offload Engine (SPOE)
18593--------------------------------------------
18594
18595filter spoe [engine <name>] config <file>
18596
18597 Arguments :
18598
18599 <name> is the engine name that will be used to find the right scope in
18600 the configuration file. If not provided, all the file will be
18601 parsed.
18602
18603 <file> is the path of the engine configuration file. This file can
18604 contain configuration of several engines. In this case, each
18605 part must be placed in its own scope.
18606
18607The Stream Processing Offload Engine (SPOE) is a filter communicating with
18608external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010018609streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020018610exchanged with them are configured in dedicated files, for the main part. It
18611also requires dedicated backends, defined in HAProxy configuration.
18612
18613SPOE communicates with external components using an in-house binary protocol,
18614the Stream Processing Offload Protocol (SPOP).
18615
Tim Düsterhus4896c442016-11-29 02:15:19 +010018616For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020018617"doc/SPOE.txt".
18618
18619Important note:
18620 The SPOE filter is highly experimental for now and was not heavily
18621 tested. It is really not production ready. So use it carefully.
18622
Christopher Faulet99a17a22018-12-11 09:18:27 +0100186239.4. Cache
18624----------
18625
18626filter cache <name>
18627
18628 Arguments :
18629
18630 <name> is name of the cache section this filter will use.
18631
18632The cache uses a filter to store cacheable responses. The HTTP rules
18633"cache-store" and "cache-use" must be used to define how and when to use a
John Roesler27c754c2019-07-10 15:45:51 -050018634cache. By default the corresponding filter is implicitly defined. And when no
Christopher Faulet27d93c32018-12-15 22:32:02 +010018635other filters than cache or compression are used, it is enough. In such case,
18636the compression filter is always evaluated after the cache filter. But it is
18637mandatory to explicitly use a filter line to use a cache when at least one
18638filter other than the compression is used for the same
18639listener/frontend/backend. This is important to know the filters evaluation
18640order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010018641
Christopher Faulet27d93c32018-12-15 22:32:02 +010018642See also : section 9.2 about the compression filter and section 10 about cache.
Christopher Faulet99a17a22018-12-11 09:18:27 +010018643
William Lallemand86d0df02017-11-24 21:36:45 +01001864410. Cache
18645---------
18646
18647HAProxy provides a cache, which was designed to perform cache on small objects
18648(favicon, css...). This is a minimalist low-maintenance cache which runs in
18649RAM.
18650
18651The cache is based on a memory which is shared between processes and threads,
Cyril Bonté7b888f12017-11-26 22:24:31 +010018652this memory is split in blocks of 1k.
William Lallemand86d0df02017-11-24 21:36:45 +010018653
18654If an object is not used anymore, it can be deleted to store a new object
18655independently of its expiration date. The oldest objects are deleted first
18656when we try to allocate a new one.
18657
Cyril Bonté7b888f12017-11-26 22:24:31 +010018658The cache uses a hash of the host header and the URI as the key.
William Lallemand86d0df02017-11-24 21:36:45 +010018659
18660It's possible to view the status of a cache using the Unix socket command
18661"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
18662for more details.
18663
18664When an object is delivered from the cache, the server name in the log is
18665replaced by "<CACHE>".
18666
Cyril Bonté7b888f12017-11-26 22:24:31 +01001866710.1. Limitation
18668----------------
William Lallemand86d0df02017-11-24 21:36:45 +010018669
18670The cache won't store and won't deliver objects in these cases:
18671
18672- If the response is not a 200
18673- If the response contains a Vary header
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020018674- If the Content-Length + the headers size is greater than "max-object-size"
William Lallemand86d0df02017-11-24 21:36:45 +010018675- If the response is not cacheable
18676
18677- If the request is not a GET
18678- If the HTTP version of the request is smaller than 1.1
William Lallemand8a16fe02018-05-22 11:04:33 +020018679- If the request contains an Authorization header
William Lallemand86d0df02017-11-24 21:36:45 +010018680
Christopher Faulet99a17a22018-12-11 09:18:27 +010018681Caution!: For HAProxy version prior to 1.9, due to the limitation of the
18682filters, it is not recommended to use the cache with other filters. Using them
18683can cause undefined behavior if they modify the response (compression for
18684example). For HAProxy 1.9 and greater, it is safe, for HTX proxies only (see
18685"option http-use-htx" for details).
William Lallemand86d0df02017-11-24 21:36:45 +010018686
Cyril Bonté7b888f12017-11-26 22:24:31 +01001868710.2. Setup
18688-----------
William Lallemand86d0df02017-11-24 21:36:45 +010018689
18690To setup a cache, you must define a cache section and use it in a proxy with
18691the corresponding http-request and response actions.
18692
Cyril Bonté7b888f12017-11-26 22:24:31 +01001869310.2.1. Cache section
18694---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010018695
18696cache <name>
18697 Declare a cache section, allocate a shared cache memory named <name>, the
18698 size of cache is mandatory.
18699
18700total-max-size <megabytes>
Davor Ocelice9ed2812017-12-25 17:49:28 +010018701 Define the size in RAM of the cache in megabytes. This size is split in
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020018702 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
William Lallemand86d0df02017-11-24 21:36:45 +010018703
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020018704max-object-size <bytes>
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020018705 Define the maximum size of the objects to be cached. Must not be greater than
18706 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
18707 All objects with sizes larger than "max-object-size" will not be cached.
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020018708
William Lallemand86d0df02017-11-24 21:36:45 +010018709max-age <seconds>
18710 Define the maximum expiration duration. The expiration is set has the lowest
18711 value between the s-maxage or max-age (in this order) directive in the
18712 Cache-Control response header and this value. The default value is 60
18713 seconds, which means that you can't cache an object more than 60 seconds by
18714 default.
18715
Cyril Bonté7b888f12017-11-26 22:24:31 +01001871610.2.2. Proxy section
18717---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010018718
Jarno Huuskonen251a6b72019-01-04 14:05:02 +020018719http-request cache-use <name> [ { if | unless } <condition> ]
William Lallemand86d0df02017-11-24 21:36:45 +010018720 Try to deliver a cached object from the cache <name>. This directive is also
18721 mandatory to store the cache as it calculates the cache hash. If you want to
18722 use a condition for both storage and delivering that's a good idea to put it
18723 after this one.
18724
Jarno Huuskonen251a6b72019-01-04 14:05:02 +020018725http-response cache-store <name> [ { if | unless } <condition> ]
William Lallemand86d0df02017-11-24 21:36:45 +010018726 Store an http-response within the cache. The storage of the response headers
18727 is done at this step, which means you can use others http-response actions
18728 to modify headers before or after the storage of the response. This action
18729 is responsible for the setup of the cache storage filter.
18730
18731
18732Example:
18733
18734 backend bck1
18735 mode http
18736
18737 http-request cache-use foobar
18738 http-response cache-store foobar
18739 server srv1 127.0.0.1:80
18740
18741 cache foobar
18742 total-max-size 4
18743 max-age 240
18744
Willy Tarreau0ba27502007-12-24 16:55:16 +010018745/*
18746 * Local variables:
18747 * fill-column: 79
18748 * End:
18749 */