blob: bda72174265c0bb1da918329348531a0597b1ca6 [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
Willy Tarreauac198b92019-11-25 15:55:45 +01007 2019/11/25
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
368 408 when the request timeout strikes before the request is complete
369 500 when haproxy encounters an unrecoverable internal error, such as a
370 memory allocation failure, which should never happen
371 502 when the server returns an empty, invalid or incomplete response, or
372 when an "rspdeny" filter blocks the response.
373 503 when no server was available to handle the request, or in response to
374 monitoring requests which match the "monitor fail" condition
375 504 when the response timeout strikes before the server responds
376
377The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3784.2).
379
380
3811.3.2. The response headers
382---------------------------
383
384Response headers work exactly like request headers, and as such, HAProxy uses
385the same parsing function for both. Please refer to paragraph 1.2.2 for more
386details.
387
388
3892. Configuring HAProxy
390----------------------
391
3922.1. Configuration file format
393------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200394
395HAProxy's configuration process involves 3 major sources of parameters :
396
397 - the arguments from the command-line, which always take precedence
398 - the "global" section, which sets process-wide parameters
399 - the proxies sections which can take form of "defaults", "listen",
400 "frontend" and "backend".
401
Willy Tarreau0ba27502007-12-24 16:55:16 +0100402The configuration file syntax consists in lines beginning with a keyword
403referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200404delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100405
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200406
William Lallemandf9873ba2015-05-05 17:37:14 +02004072.2. Quoting and escaping
408-------------------------
409
410HAProxy's configuration introduces a quoting and escaping system similar to
411many programming languages. The configuration file supports 3 types: escaping
412with a backslash, weak quoting with double quotes, and strong quoting with
413single quotes.
414
415If spaces have to be entered in strings, then they must be escaped by preceding
416them by a backslash ('\') or by quoting them. Backslashes also have to be
417escaped by doubling or strong quoting them.
418
419Escaping is achieved by preceding a special character by a backslash ('\'):
420
421 \ to mark a space and differentiate it from a delimiter
422 \# to mark a hash and differentiate it from a comment
423 \\ to use a backslash
424 \' to use a single quote and differentiate it from strong quoting
425 \" to use a double quote and differentiate it from weak quoting
426
427Weak quoting is achieved by using double quotes (""). Weak quoting prevents
428the interpretation of:
429
430 space as a parameter separator
431 ' single quote as a strong quoting delimiter
432 # hash as a comment start
433
William Lallemandb2f07452015-05-12 14:27:13 +0200434Weak quoting permits the interpretation of variables, if you want to use a non
435-interpreted dollar within a double quoted string, you should escape it with a
436backslash ("\$"), it does not work outside weak quoting.
437
438Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200439quoting.
440
441Strong quoting is achieved by using single quotes (''). Inside single quotes,
442nothing is interpreted, it's the efficient way to quote regexes.
443
444Quoted and escaped strings are replaced in memory by their interpreted
445equivalent, it allows you to perform concatenation.
446
447 Example:
448 # those are equivalents:
449 log-format %{+Q}o\ %t\ %s\ %{-Q}r
450 log-format "%{+Q}o %t %s %{-Q}r"
451 log-format '%{+Q}o %t %s %{-Q}r'
452 log-format "%{+Q}o %t"' %s %{-Q}r'
453 log-format "%{+Q}o %t"' %s'\ %{-Q}r
454
455 # those are equivalents:
456 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
457 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
458 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
459 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
460
461
William Lallemandb2f07452015-05-12 14:27:13 +02004622.3. Environment variables
463--------------------------
464
465HAProxy's configuration supports environment variables. Those variables are
466interpreted only within double quotes. Variables are expanded during the
467configuration parsing. Variable names must be preceded by a dollar ("$") and
468optionally enclosed with braces ("{}") similarly to what is done in Bourne
469shell. Variable names can contain alphanumerical characters or the character
470underscore ("_") but should not start with a digit.
471
472 Example:
473
474 bind "fd@${FD_APP1}"
475
476 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
477
478 user "$HAPROXY_USER"
479
William Lallemand4d03e432019-06-14 15:35:37 +0200480Some variables are defined by HAProxy, they can be used in the configuration
481file, or could be inherited by a program (See 3.7. Programs):
William Lallemanddaf4cd22018-04-17 16:46:13 +0200482
William Lallemand4d03e432019-06-14 15:35:37 +0200483* HAPROXY_LOCALPEER: defined at the startup of the process which contains the
484 name of the local peer. (See "-L" in the management guide.)
485
486* HAPROXY_CFGFILES: list of the configuration files loaded by HAProxy,
487 separated by semicolons. Can be useful in the case you specified a
488 directory.
489
490* HAPROXY_MWORKER: In master-worker mode, this variable is set to 1.
491
John Roesler27c754c2019-07-10 15:45:51 -0500492* HAPROXY_CLI: configured listeners addresses of the stats socket for every
William Lallemand4d03e432019-06-14 15:35:37 +0200493 processes, separated by semicolons.
494
John Roesler27c754c2019-07-10 15:45:51 -0500495* HAPROXY_MASTER_CLI: In master-worker mode, listeners addresses of the master
William Lallemand4d03e432019-06-14 15:35:37 +0200496 CLI, separated by semicolons.
497
498See also "external-check command" for other variables.
William Lallemandb2f07452015-05-12 14:27:13 +0200499
5002.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200501----------------
502
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100503Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100504values are generally expressed in milliseconds (unless explicitly stated
505otherwise) but may be expressed in any other unit by suffixing the unit to the
506numeric value. It is important to consider this because it will not be repeated
507for every keyword. Supported units are :
508
509 - us : microseconds. 1 microsecond = 1/1000000 second
510 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
511 - s : seconds. 1s = 1000ms
512 - m : minutes. 1m = 60s = 60000ms
513 - h : hours. 1h = 60m = 3600s = 3600000ms
514 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
515
516
Lukas Tribusaa83a312017-03-21 09:25:09 +00005172.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200518-------------
519
520 # Simple configuration for an HTTP proxy listening on port 80 on all
521 # interfaces and forwarding requests to a single backend "servers" with a
522 # single server "server1" listening on 127.0.0.1:8000
523 global
524 daemon
525 maxconn 256
526
527 defaults
528 mode http
529 timeout connect 5000ms
530 timeout client 50000ms
531 timeout server 50000ms
532
533 frontend http-in
534 bind *:80
535 default_backend servers
536
537 backend servers
538 server server1 127.0.0.1:8000 maxconn 32
539
540
541 # The same configuration defined with a single listen block. Shorter but
542 # less expressive, especially in HTTP mode.
543 global
544 daemon
545 maxconn 256
546
547 defaults
548 mode http
549 timeout connect 5000ms
550 timeout client 50000ms
551 timeout server 50000ms
552
553 listen http-in
554 bind *:80
555 server server1 127.0.0.1:8000 maxconn 32
556
557
558Assuming haproxy is in $PATH, test these configurations in a shell with:
559
Willy Tarreauccb289d2010-12-11 20:19:38 +0100560 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200561
562
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005633. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200564--------------------
565
566Parameters in the "global" section are process-wide and often OS-specific. They
567are generally set once for all and do not need being changed once correct. Some
568of them have command-line equivalents.
569
570The following keywords are supported in the "global" section :
571
572 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200573 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200574 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200575 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200576 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200577 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200578 - description
579 - deviceatlas-json-file
580 - deviceatlas-log-level
581 - deviceatlas-separator
582 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900583 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200584 - gid
585 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100586 - hard-stop-after
Christopher Fauleta99ff4d2019-07-22 16:18:24 +0200587 - h1-case-adjust
588 - h1-case-adjust-file
Willy Tarreau6a06a402007-07-15 20:15:28 +0200589 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200590 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100591 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200592 - lua-load
William Lallemand27edc4b2019-05-07 17:49:33 +0200593 - mworker-max-reloads
Willy Tarreau6a06a402007-07-15 20:15:28 +0200594 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200595 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200596 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200597 - pidfile
Willy Tarreau1d549722016-02-16 12:41:57 +0100598 - presetenv
599 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200600 - uid
601 - ulimit-n
602 - user
Willy Tarreau636848a2019-04-15 19:38:50 +0200603 - set-dumpable
Willy Tarreau1d549722016-02-16 12:41:57 +0100604 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200605 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200606 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200607 - ssl-default-bind-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200608 - ssl-default-bind-options
609 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200610 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200611 - ssl-default-server-options
612 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100613 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100614 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100615 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100616 - 51degrees-data-file
617 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200618 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200619 - 51degrees-cache-size
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200620 - wurfl-data-file
621 - wurfl-information-list
622 - wurfl-information-list-separator
Willy Tarreaub3cc9f22019-04-19 16:03:32 +0200623 - wurfl-cache-size
Willy Tarreaud72758d2010-01-12 10:42:19 +0100624
Willy Tarreau6a06a402007-07-15 20:15:28 +0200625 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200626 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200627 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200628 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100629 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100630 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100631 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200632 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200633 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200634 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200635 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200636 - noepoll
637 - nokqueue
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +0000638 - noevports
Willy Tarreau6a06a402007-07-15 20:15:28 +0200639 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100640 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300641 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000642 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100643 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200644 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200645 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200646 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000647 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000648 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200649 - tune.buffers.limit
650 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200651 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200652 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100653 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200654 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200655 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200656 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100657 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200658 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200659 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100660 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100661 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100662 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100663 - tune.lua.session-timeout
664 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200665 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100666 - tune.maxaccept
667 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200668 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200669 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200670 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100671 - tune.rcvbuf.client
672 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100673 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200674 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100675 - tune.sndbuf.client
676 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100677 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100678 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200679 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100680 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200681 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200682 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100683 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200684 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100685 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200686 - tune.vars.reqres-max-size
687 - tune.vars.sess-max-size
688 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100689 - tune.zlib.memlevel
690 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100691
Willy Tarreau6a06a402007-07-15 20:15:28 +0200692 * Debugging
693 - debug
694 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200695
696
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006973.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200698------------------------------------
699
Emeric Brunc8e8d122012-10-02 18:42:10 +0200700ca-base <dir>
701 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200702 relative path is used with "ca-file" or "crl-file" directives. Absolute
703 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200704
Willy Tarreau6a06a402007-07-15 20:15:28 +0200705chroot <jail dir>
706 Changes current directory to <jail dir> and performs a chroot() there before
707 dropping privileges. This increases the security level in case an unknown
708 vulnerability would be exploited, since it would make it very hard for the
709 attacker to exploit the system. This only works when the process is started
710 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100711 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100712
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100713cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
714 On Linux 2.6 and above, it is possible to bind a process or a thread to a
715 specific CPU set. This means that the process or the thread will never run on
716 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
717 sets. The first argument is a process set, eventually followed by a thread
718 set. These sets have the format
719
720 all | odd | even | number[-[number]]
721
722 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100723 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100724 ignored. It is possible to specify a range with two such number delimited by
725 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100726 "all", only odd numbers using "odd" or even numbers using "even", just like
727 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100728 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 +0100729 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100730 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100731 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100732 specified. Each "cpu-map" directive will replace the previous ones when they
733 overlap. A thread will be bound on the intersection of its mapping and the
734 one of the process on which it is attached. If the intersection is null, no
735 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100736
Christopher Fauletff4121f2017-11-22 16:38:49 +0100737 Ranges can be partially defined. The higher bound can be omitted. In such
738 case, it is replaced by the corresponding maximum value, 32 or 64 depending
739 on the machine's word size.
740
Christopher Faulet26028f62017-11-22 15:01:51 +0100741 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100742 automatically bind a process or a thread to a CPU by incrementing
743 process/thread and CPU sets. To be valid, both sets must have the same
744 size. No matter the declaration order of the CPU sets, it will be bound from
745 the lowest to the highest bound. Having a process and a thread range with the
746 "auto:" prefix is not supported. Only one range is supported, the other one
747 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100748
749 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100750 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
751
752 cpu-map 1/all 0-3 # bind all threads of the first process on the
753 # first 4 CPUs
754
755 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
756 # or "cpu-map 1-32 0-31" depending on the machine's
757 # word size.
758
Christopher Faulet26028f62017-11-22 15:01:51 +0100759 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100760 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100761 cpu-map auto:1-4 0-3
762 cpu-map auto:1-4 0-1 2-3
763 cpu-map auto:1-4 3 2 1 0
764
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100765 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
766 # and so on.
767 cpu-map auto:1/1-4 0-3
768 cpu-map auto:1/1-4 0-1 2-3
769 cpu-map auto:1/1-4 3 2 1 0
770
Davor Ocelice9ed2812017-12-25 17:49:28 +0100771 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100772 cpu-map auto:all 0-63
773 cpu-map auto:even 0-31
774 cpu-map auto:odd 32-63
775
776 # invalid cpu-map because process and CPU sets have different sizes.
777 cpu-map auto:1-4 0 # invalid
778 cpu-map auto:1 0-3 # invalid
779
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100780 # invalid cpu-map because automatic binding is used with a process range
781 # and a thread range.
782 cpu-map auto:all/all 0 # invalid
783 cpu-map auto:all/1-4 0 # invalid
784 cpu-map auto:1-4/all 0 # invalid
785
Emeric Brunc8e8d122012-10-02 18:42:10 +0200786crt-base <dir>
787 Assigns a default directory to fetch SSL certificates from when a relative
788 path is used with "crtfile" directives. Absolute locations specified after
789 "crtfile" prevail and ignore "crt-base".
790
Willy Tarreau6a06a402007-07-15 20:15:28 +0200791daemon
792 Makes the process fork into background. This is the recommended mode of
793 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100794 disabled by the command line "-db" argument. This option is ignored in
795 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200796
David Carlier8167f302015-06-01 13:50:06 +0200797deviceatlas-json-file <path>
798 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100799 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200800
801deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100802 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200803 optional and set to 0 by default if not set.
804
805deviceatlas-separator <char>
806 Sets the character separator for the API properties results. This directive
807 is optional and set to | by default if not set.
808
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100809deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200810 Sets the client cookie's name used for the detection if the DeviceAtlas
811 Client-side component was used during the request. This directive is optional
812 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100813
Simon Horman98637e52014-06-20 12:30:16 +0900814external-check
815 Allows the use of an external agent to perform health checks.
816 This is disabled by default as a security precaution.
817 See "option external-check".
818
Willy Tarreau6a06a402007-07-15 20:15:28 +0200819gid <number>
820 Changes the process' group ID to <number>. It is recommended that the group
821 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
822 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100823 Note that if haproxy is started from a user having supplementary groups, it
824 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200825 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100826
Willy Tarreau8b852462019-12-03 08:29:22 +0100827group <group name>
828 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
829 See also "gid" and "user".
830
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100831hard-stop-after <time>
832 Defines the maximum time allowed to perform a clean soft-stop.
833
834 Arguments :
835 <time> is the maximum time (by default in milliseconds) for which the
836 instance will remain alive when a soft-stop is received via the
837 SIGUSR1 signal.
838
839 This may be used to ensure that the instance will quit even if connections
840 remain opened during a soft-stop (for example with long timeouts for a proxy
841 in tcp mode). It applies both in TCP and HTTP mode.
842
843 Example:
844 global
845 hard-stop-after 30s
846
Christopher Fauleta99ff4d2019-07-22 16:18:24 +0200847h1-case-adjust <from> <to>
848 Defines the case adjustment to apply, when enabled, to the header name
849 <from>, to change it to <to> before sending it to HTTP/1 clients or
850 servers. <from> must be in lower case, and <from> and <to> must not differ
851 except for their case. It may be repeated if several header names need to be
852 ajusted. Duplicate entries are not allowed. If a lot of header names have to
853 be adjusted, it might be more convenient to use "h1-case-adjust-file".
854 Please note that no transformation will be applied unless "option
855 h1-case-adjust-bogus-client" or "option h1-case-adjust-bogus-server" is
856 specified in a proxy.
857
858 There is no standard case for header names because, as stated in RFC7230,
859 they are case-insensitive. So applications must handle them in a case-
860 insensitive manner. But some bogus applications violate the standards and
861 erroneously rely on the cases most commonly used by browsers. This problem
862 becomes critical with HTTP/2 because all header names must be exchanged in
863 lower case, and HAProxy follows the same convention. All header names are
864 sent in lower case to clients and servers, regardless of the HTTP version.
865
866 Applications which fail to properly process requests or responses may require
867 to temporarily use such workarounds to adjust header names sent to them for
868 the time it takes the application to be fixed. Please note that an
869 application which requires such workarounds might be vulnerable to content
870 smuggling attacks and must absolutely be fixed.
871
872 Example:
873 global
874 h1-case-adjust content-length Content-Length
875
876 See "h1-case-adjust-file", "option h1-case-adjust-bogus-client" and
877 "option h1-case-adjust-bogus-server".
878
879h1-case-adjust-file <hdrs-file>
880 Defines a file containing a list of key/value pairs used to adjust the case
881 of some header names before sending them to HTTP/1 clients or servers. The
882 file <hdrs-file> must contain 2 header names per line. The first one must be
883 in lower case and both must not differ except for their case. Lines which
884 start with '#' are ignored, just like empty lines. Leading and trailing tabs
885 and spaces are stripped. Duplicate entries are not allowed. Please note that
886 no transformation will be applied unless "option h1-case-adjust-bogus-client"
887 or "option h1-case-adjust-bogus-server" is specified in a proxy.
888
889 If this directive is repeated, only the last one will be processed. It is an
890 alternative to the directive "h1-case-adjust" if a lot of header names need
891 to be adjusted. Please read the risks associated with using this.
892
893 See "h1-case-adjust", "option h1-case-adjust-bogus-client" and
894 "option h1-case-adjust-bogus-server".
895
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200896log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
897 <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100898 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100899 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100900 configured with "log global".
901
902 <address> can be one of:
903
Willy Tarreau2769aa02007-12-27 18:26:09 +0100904 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100905 no port is specified, 514 is used by default (the standard syslog
906 port).
907
David du Colombier24bb5f52011-03-17 10:40:23 +0100908 - An IPv6 address followed by a colon and optionally a UDP port. If
909 no port is specified, 514 is used by default (the standard syslog
910 port).
911
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100912 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100913 considerations for chroot (be sure the path is accessible inside
914 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100915 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100916
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100917 - A file descriptor number in the form "fd@<number>", which may point
918 to a pipe, terminal, or socket. In this case unbuffered logs are used
919 and one writev() call per log is performed. This is a bit expensive
920 but acceptable for most workloads. Messages sent this way will not be
921 truncated but may be dropped, in which case the DroppedLogs counter
922 will be incremented. The writev() call is atomic even on pipes for
923 messages up to PIPE_BUF size, which POSIX recommends to be at least
924 512 and which is 4096 bytes on most modern operating systems. Any
925 larger message may be interleaved with messages from other processes.
926 Exceptionally for debugging purposes the file descriptor may also be
927 directed to a file, but doing so will significantly slow haproxy down
928 as non-blocking calls will be ignored. Also there will be no way to
929 purge nor rotate this file without restarting the process. Note that
930 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +0100931 for use with a TCP syslog server. See also the "short" and "raw"
932 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100933
934 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
935 "fd@2", see above.
936
William Lallemandb2f07452015-05-12 14:27:13 +0200937 You may want to reference some environment variables in the address
938 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100939
Willy Tarreau18324f52014-06-27 18:10:07 +0200940 <length> is an optional maximum line length. Log lines larger than this value
941 will be truncated before being sent. The reason is that syslog
942 servers act differently on log line length. All servers support the
943 default value of 1024, but some servers simply drop larger lines
944 while others do log them. If a server supports long lines, it may
945 make sense to set this value here in order to avoid truncating long
946 lines. Similarly, if a server drops long lines, it is preferable to
947 truncate them before sending them. Accepted values are 80 to 65535
948 inclusive. The default value of 1024 is generally fine for all
949 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100950 JSON-formatted logs may require larger values. You may also need to
951 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200952
Dragan Dosen7ad31542015-09-28 17:16:47 +0200953 <format> is the log format used when generating syslog messages. It may be
954 one of the following :
955
956 rfc3164 The RFC3164 syslog message format. This is the default.
957 (https://tools.ietf.org/html/rfc3164)
958
959 rfc5424 The RFC5424 syslog message format.
960 (https://tools.ietf.org/html/rfc5424)
961
Willy Tarreaue8746a02018-11-12 08:45:00 +0100962 short A message containing only a level between angle brackets such as
963 '<3>', followed by the text. The PID, date, time, process name
964 and system name are omitted. This is designed to be used with a
965 local log server. This format is compatible with what the systemd
966 logger consumes.
967
Willy Tarreauc1b06452018-11-12 11:57:56 +0100968 raw A message containing only the text. The level, PID, date, time,
969 process name and system name are omitted. This is designed to be
970 used in containers or during development, where the severity only
971 depends on the file descriptor used (stdout/stderr).
972
Frédéric Lécailled690dfa2019-04-25 10:52:17 +0200973 <ranges> A list of comma-separated ranges to identify the logs to sample.
974 This is used to balance the load of the logs to send to the log
975 server. The limits of the ranges cannot be null. They are numbered
976 from 1. The size or period (in number of logs) of the sample must be
977 set with <sample_size> parameter.
978
979 <sample_size>
980 The size of the sample in number of logs to consider when balancing
981 their logging loads. It is used to balance the load of the logs to
982 send to the syslog server. This size must be greater or equal to the
983 maximum of the high limits of the ranges.
984 (see also <ranges> parameter).
985
Robert Tsai81ae1952007-12-05 10:47:29 +0100986 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200987
Willy Tarreaue8746a02018-11-12 08:45:00 +0100988 kern user mail daemon auth syslog lpr news
989 uucp cron auth2 ftp ntp audit alert cron2
990 local0 local1 local2 local3 local4 local5 local6 local7
991
Willy Tarreauc1b06452018-11-12 11:57:56 +0100992 Note that the facility is ignored for the "short" and "raw"
993 formats, but still required as a positional field. It is
994 recommended to use "daemon" in this case to make it clear that
995 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200996
997 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200998 all messages are sent. If a maximum level is specified, only messages with a
999 severity at least as important as this level will be sent. An optional minimum
1000 level can be specified. If it is set, logs emitted with a more severe level
1001 than this one will be capped to this level. This is used to avoid sending
1002 "emerg" messages on all terminals on some default syslog configurations.
1003 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +02001004
Cyril Bontédc4d9032012-04-08 21:57:39 +02001005 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +02001006
Joe Williamsdf5b38f2010-12-29 17:05:48 +01001007log-send-hostname [<string>]
1008 Sets the hostname field in the syslog header. If optional "string" parameter
1009 is set the header is set to the string contents, otherwise uses the hostname
1010 of the system. Generally used if one is not relaying logs through an
1011 intermediate syslog server or for simply customizing the hostname printed in
1012 the logs.
1013
Kevinm48936af2010-12-22 16:08:21 +00001014log-tag <string>
1015 Sets the tag field in the syslog header to this string. It defaults to the
1016 program name as launched from the command line, which usually is "haproxy".
1017 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +01001018 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +00001019
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001020lua-load <file>
1021 This global directive loads and executes a Lua file. This directive can be
1022 used multiple times.
1023
William Lallemand4cfede82017-11-24 22:02:34 +01001024master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +02001025 Master-worker mode. It is equivalent to the command line "-W" argument.
1026 This mode will launch a "master" which will monitor the "workers". Using
1027 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001028 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +02001029 or daemon mode. It is recommended to use this mode with multiprocess and
1030 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +01001031 By default, if a worker exits with a bad return code, in the case of a
1032 segfault for example, all workers will be killed, and the master will leave.
1033 It is convenient to combine this behavior with Restart=on-failure in a
1034 systemd unit file in order to relaunch the whole process. If you don't want
1035 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +02001036
William Lallemand4cfede82017-11-24 22:02:34 +01001037 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +02001038
William Lallemand27edc4b2019-05-07 17:49:33 +02001039mworker-max-reloads <number>
1040 In master-worker mode, this option limits the number of time a worker can
John Roesler27c754c2019-07-10 15:45:51 -05001041 survive to a reload. If the worker did not leave after a reload, once its
William Lallemand27edc4b2019-05-07 17:49:33 +02001042 number of reloads is greater than this number, the worker will receive a
1043 SIGTERM. This option helps to keep under control the number of workers.
1044 See also "show proc" in the Management Guide.
1045
Willy Tarreau6a06a402007-07-15 20:15:28 +02001046nbproc <number>
1047 Creates <number> processes when going daemon. This requires the "daemon"
1048 mode. By default, only one process is created, which is the recommended mode
1049 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +01001050 process, it may be needed to fork multiple daemons. When set to a value
1051 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreau1f672a82019-01-26 14:20:55 +01001052 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon" and
1053 "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001054
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001055nbthread <number>
1056 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +01001057 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
1058 "nbproc" historically used to be the only way to use multiple processors, it
1059 also involved a number of shortcomings related to the lack of synchronization
1060 between processes (health-checks, peers, stick-tables, stats, ...) which do
1061 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +01001062 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
1063 HAProxy is started in foreground. On some platforms supporting CPU affinity,
1064 when nbproc is not used, the default "nbthread" value is automatically set to
1065 the number of CPUs the process is bound to upon startup. This means that the
1066 thread count can easily be adjusted from the calling process using commands
1067 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
1068 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +02001069
Willy Tarreau6a06a402007-07-15 20:15:28 +02001070pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001071 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +02001072 the "-p" command line argument. The file must be accessible to the user
1073 starting the process. See also "daemon".
1074
Willy Tarreau1d549722016-02-16 12:41:57 +01001075presetenv <name> <value>
1076 Sets environment variable <name> to value <value>. If the variable exists, it
1077 is NOT overwritten. The changes immediately take effect so that the next line
1078 in the configuration file sees the new value. See also "setenv", "resetenv",
1079 and "unsetenv".
1080
1081resetenv [<name> ...]
1082 Removes all environment variables except the ones specified in argument. It
1083 allows to use a clean controlled environment before setting new values with
1084 setenv or unsetenv. Please note that some internal functions may make use of
1085 some environment variables, such as time manipulation functions, but also
1086 OpenSSL or even external checks. This must be used with extreme care and only
1087 after complete validation. The changes immediately take effect so that the
1088 next line in the configuration file sees the new environment. See also
1089 "setenv", "presetenv", and "unsetenv".
1090
Christopher Fauletff4121f2017-11-22 16:38:49 +01001091stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +02001092 Limits the stats socket to a certain set of processes numbers. By default the
1093 stats socket is bound to all processes, causing a warning to be emitted when
1094 nbproc is greater than 1 because there is no way to select the target process
1095 when connecting. However, by using this setting, it becomes possible to pin
1096 the stats socket to a specific set of processes, typically the first one. The
1097 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001098 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001099 word size (32 or 64). Ranges can be partially defined. The higher bound can
1100 be omitted. In such case, it is replaced by the corresponding maximum
1101 value. A better option consists in using the "process" setting of the "stats
1102 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001103
Baptiste Assmann5626f482015-08-23 10:00:10 +02001104server-state-base <directory>
1105 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001106 file names which do not start with a '/'. See also "server-state-file",
1107 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001108
1109server-state-file <file>
1110 Specifies the path to the file containing state of servers. If the path starts
1111 with a slash ('/'), it is considered absolute, otherwise it is considered
1112 relative to the directory specified using "server-state-base" (if set) or to
1113 the current directory. Before reloading HAProxy, it is possible to save the
1114 servers' current state using the stats command "show servers state". The
1115 output of this command must be written in the file pointed by <file>. When
1116 starting up, before handling traffic, HAProxy will read, load and apply state
1117 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001118 configuration. See also "server-state-base" and "show servers state",
1119 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001120
Willy Tarreau1d549722016-02-16 12:41:57 +01001121setenv <name> <value>
1122 Sets environment variable <name> to value <value>. If the variable exists, it
1123 is overwritten. The changes immediately take effect so that the next line in
1124 the configuration file sees the new value. See also "presetenv", "resetenv",
1125 and "unsetenv".
1126
Willy Tarreau636848a2019-04-15 19:38:50 +02001127set-dumpable
1128 This option is better left disabled by default and enabled only upon a
1129 developer's request. It has no impact on performance nor stability but will
1130 try hard to re-enable core dumps that were possibly disabled by file size
1131 limitations (ulimit -f), core size limitations (ulimit -c), or "dumpability"
1132 of a process after changing its UID/GID (such as /proc/sys/fs/suid_dumpable
1133 on Linux). Core dumps might still be limited by the current directory's
1134 permissions (check what directory the file is started from), the chroot
1135 directory's permission (it may be needed to temporarily disable the chroot
1136 directive or to move it to a dedicated writable location), or any other
1137 system-specific constraint. For example, some Linux flavours are notorious
1138 for replacing the default core file with a path to an executable not even
1139 installed on the system (check /proc/sys/kernel/core_pattern). Often, simply
1140 writing "core", "core.%p" or "/var/log/core/core.%p" addresses the issue.
1141 When trying to enable this option waiting for a rare issue to re-appear, it's
1142 often a good idea to first try to obtain such a dump by issuing, for example,
1143 "kill -11" to the haproxy process and verify that it leaves a core where
1144 expected when dying.
1145
Willy Tarreau610f04b2014-02-13 11:36:41 +01001146ssl-default-bind-ciphers <ciphers>
1147 This setting is only available when support for OpenSSL was built in. It sets
1148 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001149 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001150 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001151 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1152 information and recommendations see e.g.
1153 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1154 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1155 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1156 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001157
1158ssl-default-bind-ciphersuites <ciphersuites>
1159 This setting is only available when support for OpenSSL was built in and
1160 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1161 describing the list of cipher algorithms ("cipher suite") that are negotiated
1162 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1163 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001164 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1165 cipher configuration for TLSv1.2 and earlier, please check the
1166 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001167 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001168
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001169ssl-default-bind-options [<option>]...
1170 This setting is only available when support for OpenSSL was built in. It sets
1171 default ssl-options to force on all "bind" lines. Please check the "bind"
1172 keyword to see available options.
1173
1174 Example:
1175 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001176 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001177
Willy Tarreau610f04b2014-02-13 11:36:41 +01001178ssl-default-server-ciphers <ciphers>
1179 This setting is only available when support for OpenSSL was built in. It
1180 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001181 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001182 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001183 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1184 information and recommendations see e.g.
1185 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1186 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1187 For TLSv1.3 cipher configuration, please check the
1188 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1189 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001190
1191ssl-default-server-ciphersuites <ciphersuites>
1192 This setting is only available when support for OpenSSL was built in and
1193 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1194 string describing the list of cipher algorithms that are negotiated during
1195 the TLSv1.3 handshake with the server, for all "server" lines which do not
1196 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001197 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1198 cipher configuration for TLSv1.2 and earlier, please check the
1199 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1200 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001201
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001202ssl-default-server-options [<option>]...
1203 This setting is only available when support for OpenSSL was built in. It sets
1204 default ssl-options to force on all "server" lines. Please check the "server"
1205 keyword to see available options.
1206
Remi Gacogne47783ef2015-05-29 15:53:22 +02001207ssl-dh-param-file <file>
1208 This setting is only available when support for OpenSSL was built in. It sets
1209 the default DH parameters that are used during the SSL/TLS handshake when
1210 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001211 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001212 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001213 are not specified either by using ssl-dh-param-file or by setting them
1214 directly in the certificate file, pre-generated DH parameters of the size
1215 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1216 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001217 Custom DH parameters may be generated by using the OpenSSL command
1218 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1219 parameters should not be considered secure anymore.
1220
Emeric Brun850efd52014-01-29 12:24:34 +01001221ssl-server-verify [none|required]
1222 The default behavior for SSL verify on servers side. If specified to 'none',
1223 servers certificates are not verified. The default is 'required' except if
1224 forced using cmdline option '-dV'.
1225
Willy Tarreauabb175f2012-09-24 12:43:26 +02001226stats socket [<address:port>|<path>] [param*]
1227 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1228 Connections to this socket will return various statistics outputs and even
1229 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001230 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001231 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001232
Willy Tarreauabb175f2012-09-24 12:43:26 +02001233 All parameters supported by "bind" lines are supported, for instance to
1234 restrict access to some users or their access rights. Please consult
1235 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001236
1237stats timeout <timeout, in milliseconds>
1238 The default timeout on the stats socket is set to 10 seconds. It is possible
1239 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001240 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001241
1242stats maxconn <connections>
1243 By default, the stats socket is limited to 10 concurrent connections. It is
1244 possible to change this value with "stats maxconn".
1245
Willy Tarreau6a06a402007-07-15 20:15:28 +02001246uid <number>
1247 Changes the process' user ID to <number>. It is recommended that the user ID
1248 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1249 be started with superuser privileges in order to be able to switch to another
1250 one. See also "gid" and "user".
1251
1252ulimit-n <number>
1253 Sets the maximum number of per-process file-descriptors to <number>. By
1254 default, it is automatically computed, so it is recommended not to use this
1255 option.
1256
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001257unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1258 [ group <group> ] [ gid <gid> ]
1259
1260 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1261 This is mainly used to simplify declaration of those UNIX sockets and reduce
1262 the risk of errors, since those settings are most commonly required but are
1263 also process-specific. The <prefix> setting can be used to force all socket
1264 path to be relative to that directory. This might be needed to access another
1265 component's chroot. Note that those paths are resolved before haproxy chroots
1266 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1267 all have the same meaning as their homonyms used by the "bind" statement. If
1268 both are specified, the "bind" statement has priority, meaning that the
1269 "unix-bind" settings may be seen as process-wide default settings.
1270
Willy Tarreau1d549722016-02-16 12:41:57 +01001271unsetenv [<name> ...]
1272 Removes environment variables specified in arguments. This can be useful to
1273 hide some sensitive information that are occasionally inherited from the
1274 user's environment during some operations. Variables which did not exist are
1275 silently ignored so that after the operation, it is certain that none of
1276 these variables remain. The changes immediately take effect so that the next
1277 line in the configuration file will not see these variables. See also
1278 "setenv", "presetenv", and "resetenv".
1279
Willy Tarreau6a06a402007-07-15 20:15:28 +02001280user <user name>
1281 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1282 See also "uid" and "group".
1283
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001284node <name>
1285 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1286
1287 This statement is useful in HA configurations where two or more processes or
1288 servers share the same IP address. By setting a different node-name on all
1289 nodes, it becomes easy to immediately spot what server is handling the
1290 traffic.
1291
1292description <text>
1293 Add a text that describes the instance.
1294
1295 Please note that it is required to escape certain characters (# for example)
1296 and this text is inserted into a html page so you should avoid using
1297 "<" and ">" characters.
1298
Thomas Holmesdb04f192015-05-18 13:21:39 +0100129951degrees-data-file <file path>
1300 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001301 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001302
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001303 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001304 compiled with USE_51DEGREES.
1305
Ben Shillitof25e8e52016-12-02 14:25:37 +0000130651degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001307 A list of 51Degrees property names to be load from the dataset. A full list
1308 of names is available on the 51Degrees website:
1309 https://51degrees.com/resources/property-dictionary
1310
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001311 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001312 compiled with USE_51DEGREES.
1313
Dragan Dosen93b38d92015-06-29 16:43:25 +0200131451degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001315 A char that will be appended to every property value in a response header
1316 containing 51Degrees results. If not set that will be set as ','.
1317
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001318 Please note that this option is only available when haproxy has been
1319 compiled with USE_51DEGREES.
1320
132151degrees-cache-size <number>
1322 Sets the size of the 51Degrees converter cache to <number> entries. This
1323 is an LRU cache which reminds previous device detections and their results.
1324 By default, this cache is disabled.
1325
1326 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001327 compiled with USE_51DEGREES.
1328
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001329wurfl-data-file <file path>
1330 The path of the WURFL data file to provide device detection services. The
1331 file should be accessible by HAProxy with relevant permissions.
1332
1333 Please note that this option is only available when haproxy has been compiled
1334 with USE_WURFL=1.
1335
1336wurfl-information-list [<capability>]*
1337 A space-delimited list of WURFL capabilities, virtual capabilities, property
1338 names we plan to use in injected headers. A full list of capability and
1339 virtual capability names is available on the Scientiamobile website :
1340
1341 https://www.scientiamobile.com/wurflCapability
1342
1343 Valid WURFL properties are:
1344 - wurfl_id Contains the device ID of the matched device.
1345
1346 - wurfl_root_id Contains the device root ID of the matched
1347 device.
1348
1349 - wurfl_isdevroot Tells if the matched device is a root device.
1350 Possible values are "TRUE" or "FALSE".
1351
1352 - wurfl_useragent The original useragent coming with this
1353 particular web request.
1354
1355 - wurfl_api_version Contains a string representing the currently
1356 used Libwurfl API version.
1357
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001358 - wurfl_info A string containing information on the parsed
1359 wurfl.xml and its full path.
1360
1361 - wurfl_last_load_time Contains the UNIX timestamp of the last time
1362 WURFL has been loaded successfully.
1363
1364 - wurfl_normalized_useragent The normalized useragent.
1365
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001366 Please note that this option is only available when haproxy has been compiled
1367 with USE_WURFL=1.
1368
1369wurfl-information-list-separator <char>
1370 A char that will be used to separate values in a response header containing
1371 WURFL results. If not set that a comma (',') will be used by default.
1372
1373 Please note that this option is only available when haproxy has been compiled
1374 with USE_WURFL=1.
1375
1376wurfl-patch-file [<file path>]
1377 A list of WURFL patch file paths. Note that patches are loaded during startup
1378 thus before the chroot.
1379
1380 Please note that this option is only available when haproxy has been compiled
1381 with USE_WURFL=1.
1382
paulborilebad132c2019-04-18 11:57:04 +02001383wurfl-cache-size <size>
1384 Sets the WURFL Useragent cache size. For faster lookups, already processed user
1385 agents are kept in a LRU cache :
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001386 - "0" : no cache is used.
paulborilebad132c2019-04-18 11:57:04 +02001387 - <size> : size of lru cache in elements.
Willy Tarreaub3cc9f22019-04-19 16:03:32 +02001388
1389 Please note that this option is only available when haproxy has been compiled
1390 with USE_WURFL=1.
1391
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013923.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001393-----------------------
1394
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001395busy-polling
1396 In some situations, especially when dealing with low latency on processors
1397 supporting a variable frequency or when running inside virtual machines, each
1398 time the process waits for an I/O using the poller, the processor goes back
1399 to sleep or is offered to another VM for a long time, and it causes
1400 excessively high latencies. This option provides a solution preventing the
1401 processor from sleeping by always using a null timeout on the pollers. This
1402 results in a significant latency reduction (30 to 100 microseconds observed)
1403 at the expense of a risk to overheat the processor. It may even be used with
1404 threads, in which case improperly bound threads may heavily conflict,
1405 resulting in a worse performance and high values for the CPU stolen fields
1406 in "show info" output, indicating which threads are misconfigured. It is
1407 important not to let the process run on the same processor as the network
1408 interrupts when this option is used. It is also better to avoid using it on
1409 multiple CPU threads sharing the same core. This option is disabled by
1410 default. If it has been enabled, it may still be forcibly disabled by
1411 prefixing it with the "no" keyword. It is ignored by the "select" and
1412 "poll" pollers.
1413
Willy Tarreau1746eec2014-04-25 10:46:47 +02001414max-spread-checks <delay in milliseconds>
1415 By default, haproxy tries to spread the start of health checks across the
1416 smallest health check interval of all the servers in a farm. The principle is
1417 to avoid hammering services running on the same server. But when using large
1418 check intervals (10 seconds or more), the last servers in the farm take some
1419 time before starting to be tested, which can be a problem. This parameter is
1420 used to enforce an upper bound on delay between the first and the last check,
1421 even if the servers' check intervals are larger. When servers run with
1422 shorter intervals, their intervals will be respected though.
1423
Willy Tarreau6a06a402007-07-15 20:15:28 +02001424maxconn <number>
1425 Sets the maximum per-process number of concurrent connections to <number>. It
1426 is equivalent to the command-line argument "-n". Proxies will stop accepting
1427 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001428 automatically adjusted according to this value. See also "ulimit-n". Note:
1429 the "select" poller cannot reliably use more than 1024 file descriptors on
1430 some platforms. If your platform only supports select and reports "select
1431 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001432 below 500 in general). If this value is not set, it will automatically be
1433 calculated based on the current file descriptors limit reported by the
1434 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1435 is enforced, based on the buffer size, memory allocated to compression, SSL
1436 cache size, and use or not of SSL and the associated maxsslconn (which can
1437 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001438
Willy Tarreau81c25d02011-09-07 15:17:21 +02001439maxconnrate <number>
1440 Sets the maximum per-process number of connections per second to <number>.
1441 Proxies will stop accepting connections when this limit is reached. It can be
1442 used to limit the global capacity regardless of each frontend capacity. It is
1443 important to note that this can only be used as a service protection measure,
1444 as there will not necessarily be a fair share between frontends when the
1445 limit is reached, so it's a good idea to also limit each frontend to some
1446 value close to its expected share. Also, lowering tune.maxaccept can improve
1447 fairness.
1448
William Lallemandd85f9172012-11-09 17:05:39 +01001449maxcomprate <number>
1450 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001451 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001452 level will be decreased during the session. If the maximum is reached at the
1453 beginning of a session, the session will not compress at all. If the maximum
1454 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001455 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001456 default value.
1457
William Lallemand072a2bf2012-11-20 17:01:01 +01001458maxcompcpuusage <number>
1459 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1460 for new requests or decreasing the compression level of current requests.
1461 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1462 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1463 case of multiple processes (nbproc > 1), each process manages its individual
1464 usage. A value of 100 disable the limit. The default value is 100. Setting
1465 a lower value will prevent the compression work from slowing the whole
1466 process down and from introducing high latencies.
1467
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001468maxpipes <number>
1469 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1470 are only used by kernel-based tcp splicing. Since a pipe contains two file
1471 descriptors, the "ulimit-n" value will be increased accordingly. The default
1472 value is maxconn/4, which seems to be more than enough for most heavy usages.
1473 The splice code dynamically allocates and releases pipes, and can fall back
1474 to standard copy, so setting this value too low may only impact performance.
1475
Willy Tarreau93e7c002013-10-07 18:51:07 +02001476maxsessrate <number>
1477 Sets the maximum per-process number of sessions per second to <number>.
1478 Proxies will stop accepting connections when this limit is reached. It can be
1479 used to limit the global capacity regardless of each frontend capacity. It is
1480 important to note that this can only be used as a service protection measure,
1481 as there will not necessarily be a fair share between frontends when the
1482 limit is reached, so it's a good idea to also limit each frontend to some
1483 value close to its expected share. Also, lowering tune.maxaccept can improve
1484 fairness.
1485
Willy Tarreau403edff2012-09-06 11:58:37 +02001486maxsslconn <number>
1487 Sets the maximum per-process number of concurrent SSL connections to
1488 <number>. By default there is no SSL-specific limit, which means that the
1489 global maxconn setting will apply to all connections. Setting this limit
1490 avoids having openssl use too much memory and crash when malloc returns NULL
1491 (since it unfortunately does not reliably check for such conditions). Note
1492 that the limit applies both to incoming and outgoing connections, so one
1493 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001494 If this value is not set, but a memory limit is enforced, this value will be
1495 automatically computed based on the memory limit, maxconn, the buffer size,
1496 memory allocated to compression, SSL cache size, and use of SSL in either
1497 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1498 when there is a memory limit, haproxy will automatically adjust these values
1499 so that 100% of the connections can be made over SSL with no risk, and will
1500 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001501
Willy Tarreaue43d5322013-10-07 20:01:52 +02001502maxsslrate <number>
1503 Sets the maximum per-process number of SSL sessions per second to <number>.
1504 SSL listeners will stop accepting connections when this limit is reached. It
1505 can be used to limit the global SSL CPU usage regardless of each frontend
1506 capacity. It is important to note that this can only be used as a service
1507 protection measure, as there will not necessarily be a fair share between
1508 frontends when the limit is reached, so it's a good idea to also limit each
1509 frontend to some value close to its expected share. It is also important to
1510 note that the sessions are accounted before they enter the SSL stack and not
1511 after, which also protects the stack against bad handshakes. Also, lowering
1512 tune.maxaccept can improve fairness.
1513
William Lallemand9d5f5482012-11-07 16:12:57 +01001514maxzlibmem <number>
1515 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1516 When the maximum amount is reached, future sessions will not compress as long
1517 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001518 The default value is 0. The value is available in bytes on the UNIX socket
1519 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1520 "ZlibMemUsage" in bytes.
1521
Willy Tarreau6a06a402007-07-15 20:15:28 +02001522noepoll
1523 Disables the use of the "epoll" event polling system on Linux. It is
1524 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001525 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001526
1527nokqueue
1528 Disables the use of the "kqueue" event polling system on BSD. It is
1529 equivalent to the command-line argument "-dk". The next polling system
1530 used will generally be "poll". See also "nopoll".
1531
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001532noevports
1533 Disables the use of the event ports event polling system on SunOS systems
1534 derived from Solaris 10 and later. It is equivalent to the command-line
1535 argument "-dv". The next polling system used will generally be "poll". See
1536 also "nopoll".
1537
Willy Tarreau6a06a402007-07-15 20:15:28 +02001538nopoll
1539 Disables the use of the "poll" event polling system. It is equivalent to the
1540 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001541 It should never be needed to disable "poll" since it's available on all
Emmanuel Hocdet0ba4f482019-04-08 16:53:32 +00001542 platforms supported by HAProxy. See also "nokqueue", "noepoll" and
1543 "noevports".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001544
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001545nosplice
1546 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001547 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001548 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001549 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001550 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1551 be used. This option makes it easier to globally disable kernel splicing in
1552 case of doubt. See also "option splice-auto", "option splice-request" and
1553 "option splice-response".
1554
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001555nogetaddrinfo
1556 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1557 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1558
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001559noreuseport
1560 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1561 command line argument "-dR".
1562
Willy Tarreaud2d33482019-04-25 17:09:07 +02001563profiling.tasks { auto | on | off }
1564 Enables ('on') or disables ('off') per-task CPU profiling. When set to 'auto'
1565 the profiling automatically turns on a thread when it starts to suffer from
1566 an average latency of 1000 microseconds or higher as reported in the
1567 "avg_loop_us" activity field, and automatically turns off when the latency
John Roesler27c754c2019-07-10 15:45:51 -05001568 returns below 990 microseconds (this value is an average over the last 1024
Willy Tarreaud2d33482019-04-25 17:09:07 +02001569 loops so it does not vary quickly and tends to significantly smooth short
1570 spikes). It may also spontaneously trigger from time to time on overloaded
1571 systems, containers, or virtual machines, or when the system swaps (which
1572 must absolutely never happen on a load balancer).
1573
1574 CPU profiling per task can be very convenient to report where the time is
1575 spent and which requests have what effect on which other request. Enabling
1576 it will typically affect the overall's performance by less than 1%, thus it
1577 is recommended to leave it to the default 'auto' value so that it only
1578 operates when a problem is identified. This feature requires a system
Willy Tarreau75c62c22018-11-22 11:02:09 +01001579 supporting the clock_gettime(2) syscall with clock identifiers
1580 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1581 be zero. This option may be changed at run time using "set profiling" on the
1582 CLI.
1583
Willy Tarreaufe255b72007-10-14 23:09:26 +02001584spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001585 Sometimes it is desirable to avoid sending agent and health checks to
1586 servers at exact intervals, for instance when many logical servers are
1587 located on the same physical server. With the help of this parameter, it
1588 becomes possible to add some randomness in the check interval between 0
1589 and +/- 50%. A value between 2 and 5 seems to show good results. The
1590 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001591
Davor Ocelice9ed2812017-12-25 17:49:28 +01001592ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001593 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001594 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001595 multiple times, it will simply enable multiple crypto engines. Referencing an
1596 unsupported engine will prevent haproxy from starting. Note that many engines
1597 will lead to lower HTTPS performance than pure software with recent
1598 processors. The optional command "algo" sets the default algorithms an ENGINE
1599 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001600 of "ALL" uses the engine for all cryptographic operations. If no list of
1601 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001602 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1603 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1604 openssl configuration file uses:
1605 https://www.openssl.org/docs/man1.0.2/apps/config.html
1606
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001607ssl-mode-async
1608 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001609 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001610 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1611 doesn't support moving read/write buffers and is not compliant with
1612 haproxy's buffer management. So the asynchronous mode is disabled on
John Roesler27c754c2019-07-10 15:45:51 -05001613 read/write operations (it is only enabled during initial and renegotiation
Emeric Brunb5e42a82017-06-06 12:35:14 +00001614 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001615
Willy Tarreau33cb0652014-12-23 22:52:37 +01001616tune.buffers.limit <number>
1617 Sets a hard limit on the number of buffers which may be allocated per process.
1618 The default value is zero which means unlimited. The minimum non-zero value
1619 will always be greater than "tune.buffers.reserve" and should ideally always
1620 be about twice as large. Forcing this value can be particularly useful to
1621 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001622 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001623 another one to be released by another session. Since buffers are dynamically
1624 allocated and released, the waiting time is very short and not perceptible
1625 provided that limits remain reasonable. In fact sometimes reducing the limit
1626 may even increase performance by increasing the CPU cache's efficiency. Tests
1627 have shown good results on average HTTP traffic with a limit to 1/10 of the
1628 expected global maxconn setting, which also significantly reduces memory
1629 usage. The memory savings come from the fact that a number of connections
1630 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1631 advised to do so by an haproxy core developer.
1632
Willy Tarreau1058ae72014-12-23 22:40:40 +01001633tune.buffers.reserve <number>
1634 Sets the number of buffers which are pre-allocated and reserved for use only
1635 during memory shortage conditions resulting in failed memory allocations. The
1636 minimum value is 2 and is also the default. There is no reason a user would
1637 want to change this value, it's mostly aimed at haproxy core developers.
1638
Willy Tarreau27a674e2009-08-17 07:23:33 +02001639tune.bufsize <number>
1640 Sets the buffer size to this size (in bytes). Lower values allow more
1641 sessions to coexist in the same amount of RAM, and higher values allow some
1642 applications with very large cookies to work. The default value is 16384 and
1643 can be changed at build time. It is strongly recommended not to change this
1644 from the default value, as very low values will break some services such as
1645 statistics, and values larger than default size will increase memory usage,
1646 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001647 parameter should be decreased by the same factor as this one is increased. In
1648 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1649 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001650 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001651 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1652 value set using this parameter will automatically be rounded up to the next
1653 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001654
Willy Tarreau43961d52010-10-04 20:39:20 +02001655tune.chksize <number>
1656 Sets the check buffer size to this size (in bytes). Higher values may help
1657 find string or regex patterns in very large pages, though doing so may imply
1658 more memory and CPU usage. The default value is 16384 and can be changed at
1659 build time. It is not recommended to change this value, but to use better
1660 checks whenever possible.
1661
William Lallemandf3747832012-11-09 12:33:10 +01001662tune.comp.maxlevel <number>
1663 Sets the maximum compression level. The compression level affects CPU
1664 usage during compression. This value affects CPU usage during compression.
1665 Each session using compression initializes the compression algorithm with
1666 this value. The default value is 1.
1667
Willy Tarreauc299e1e2019-02-27 11:35:12 +01001668tune.fail-alloc
1669 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
1670 allocation attempt fails. Must be between 0 (no failure) and 100 (no
1671 success). This is useful to debug and make sure memory failures are handled
1672 gracefully.
1673
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001674tune.h2.header-table-size <number>
1675 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1676 cannot be larger than 65536 bytes. A larger value may help certain clients
1677 send more compact requests, depending on their capabilities. This amount of
1678 memory is consumed for each HTTP/2 connection. It is recommended not to
1679 change it.
1680
Willy Tarreaue6baec02017-07-27 11:45:11 +02001681tune.h2.initial-window-size <number>
1682 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001683 can upload before waiting for an acknowledgment from haproxy. This setting
1684 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001685 The default value is 65535, which roughly allows up to 5 Mbps of upload
1686 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1687 over a 1-ms local network. It can make sense to increase this value to allow
1688 faster uploads, or to reduce it to increase fairness when dealing with many
1689 clients. It doesn't affect resource usage.
1690
Willy Tarreau5242ef82017-07-27 11:47:28 +02001691tune.h2.max-concurrent-streams <number>
1692 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1693 number of outstanding requests on a single connection). The default value is
1694 100. A larger one may slightly improve page load time for complex sites when
1695 visited over high latency networks, but increases the amount of resources a
1696 single client may allocate. A value of zero disables the limit so a single
1697 client may create as many streams as allocatable by haproxy. It is highly
1698 recommended not to change this value.
1699
Willy Tarreaua24b35c2019-02-21 13:24:36 +01001700tune.h2.max-frame-size <number>
1701 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
1702 receive to its peers. The default value is the largest between 16384 and the
1703 buffer size (tune.bufsize). In any case, haproxy will not announce support
1704 for frame sizes larger than buffers. The main purpose of this setting is to
1705 allow to limit the maximum frame size setting when using large buffers. Too
1706 large frame sizes might have performance impact or cause some peers to
1707 misbehave. It is highly recommended not to change this value.
1708
Willy Tarreau193b8c62012-11-22 00:17:38 +01001709tune.http.cookielen <number>
1710 Sets the maximum length of captured cookies. This is the maximum value that
1711 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1712 will automatically be truncated to this one. It is important not to set too
1713 high a value because all cookie captures still allocate this size whatever
1714 their configured value (they share a same pool). This value is per request
1715 per response, so the memory allocated is twice this value per connection.
1716 When not specified, the limit is set to 63 characters. It is recommended not
1717 to change this value.
1718
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001719tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001720 Sets the maximum length of request URI in logs. This prevents truncating long
1721 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001722 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001723 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001724 configuration directives too.
1725 The default value is 1024.
1726
Willy Tarreauac1932d2011-10-24 19:14:41 +02001727tune.http.maxhdr <number>
1728 Sets the maximum number of headers in a request. When a request comes with a
1729 number of headers greater than this value (including the first line), it is
1730 rejected with a "400 Bad Request" status code. Similarly, too large responses
1731 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1732 for all usages, considering that the widely deployed Apache server uses the
1733 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001734 a buggy application to work by the time it gets fixed. The accepted range is
1735 1..32767. Keep in mind that each new header consumes 32bits of memory for
1736 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001737
Willy Tarreau7e312732014-02-12 16:35:14 +01001738tune.idletimer <timeout>
1739 Sets the duration after which haproxy will consider that an empty buffer is
1740 probably associated with an idle stream. This is used to optimally adjust
1741 some packet sizes while forwarding large and small data alternatively. The
1742 decision to use splice() or to send large buffers in SSL is modulated by this
1743 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1744 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001745 which seems to correctly detect end user pauses (e.g. read a page before
John Roesler27c754c2019-07-10 15:45:51 -05001746 clicking). There should be no reason for changing this value. Please check
Willy Tarreau7e312732014-02-12 16:35:14 +01001747 tune.ssl.maxrecord below.
1748
Willy Tarreau7ac908b2019-02-27 12:02:18 +01001749tune.listener.multi-queue { on | off }
1750 Enables ('on') or disables ('off') the listener's multi-queue accept which
1751 spreads the incoming traffic to all threads a "bind" line is allowed to run
1752 on instead of taking them for itself. This provides a smoother traffic
1753 distribution and scales much better, especially in environments where threads
1754 may be unevenly loaded due to external activity (network interrupts colliding
1755 with one thread for example). This option is enabled by default, but it may
1756 be forcefully disabled for troubleshooting or for situations where it is
1757 estimated that the operating system already provides a good enough
1758 distribution and connections are extremely short-lived.
1759
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001760tune.lua.forced-yield <number>
1761 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001762 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001763 HAProxy scheduler to process other tasks like accepting connections or
1764 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001765 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001766 lowered. If the Lua code is quite long and its result is absolutely required
1767 to process the data, the <number> can be increased.
1768
Willy Tarreau32f61e22015-03-18 17:54:59 +01001769tune.lua.maxmem
1770 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1771 default it is zero which means unlimited. It is important to set a limit to
1772 ensure that a bug in a script will not result in the system running out of
1773 memory.
1774
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001775tune.lua.session-timeout <timeout>
1776 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001777 preventing infinite loops or spending too much time in Lua. This timeout
1778 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001779 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001780
1781tune.lua.task-timeout <timeout>
1782 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1783 dedicated to the tasks. By default, this timeout isn't set because a task may
1784 remain alive during of the lifetime of HAProxy. For example, a task used to
1785 check servers.
1786
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001787tune.lua.service-timeout <timeout>
1788 This is the execution timeout for the Lua services. This is useful for
1789 preventing infinite loops or spending too much time in Lua. This timeout
1790 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001791 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001792
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001793tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001794 Sets the maximum number of consecutive connections a process may accept in a
1795 row before switching to other work. In single process mode, higher numbers
1796 give better performance at high connection rates. However in multi-process
1797 modes, keeping a bit of fairness between processes generally is better to
1798 increase performance. This value applies individually to each listener, so
1799 that the number of processes a listener is bound to is taken into account.
1800 This value defaults to 64. In multi-process mode, it is divided by twice
1801 the number of processes the listener is bound to. Setting this value to -1
1802 completely disables the limitation. It should normally not be needed to tweak
1803 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001804
1805tune.maxpollevents <number>
1806 Sets the maximum amount of events that can be processed at once in a call to
1807 the polling system. The default value is adapted to the operating system. It
1808 has been noticed that reducing it below 200 tends to slightly decrease
1809 latency at the expense of network bandwidth, and increasing it above 200
1810 tends to trade latency for slightly increased bandwidth.
1811
Willy Tarreau27a674e2009-08-17 07:23:33 +02001812tune.maxrewrite <number>
1813 Sets the reserved buffer space to this size in bytes. The reserved space is
1814 used for header rewriting or appending. The first reads on sockets will never
1815 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1816 bufsize, though that does not make much sense since there are rarely large
1817 numbers of headers to add. Setting it too high prevents processing of large
1818 requests or responses. Setting it too low prevents addition of new headers
1819 to already large requests or to POST requests. It is generally wise to set it
1820 to about 1024. It is automatically readjusted to half of bufsize if it is
1821 larger than that. This means you don't have to worry about it when changing
1822 bufsize.
1823
Willy Tarreauf3045d22015-04-29 16:24:50 +02001824tune.pattern.cache-size <number>
1825 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1826 cache which reminds previous lookups and their results. It is used by ACLs
1827 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1828 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1829 strings. It applies to pattern expressions which means that it will be able
1830 to memorize the result of a lookup among all the patterns specified on a
1831 configuration line (including all those loaded from files). It automatically
1832 invalidates entries which are updated using HTTP actions or on the CLI. The
1833 default cache size is set to 10000 entries, which limits its footprint to
Willy Tarreau7fdd81c2019-10-23 06:59:31 +02001834 about 5 MB per process/thread on 32-bit systems and 8 MB per process/thread
1835 on 64-bit systems, as caches are thread/process local. There is a very low
Willy Tarreauf3045d22015-04-29 16:24:50 +02001836 risk of collision in this cache, which is in the order of the size of the
1837 cache divided by 2^64. Typically, at 10000 requests per second with the
1838 default cache size of 10000 entries, there's 1% chance that a brute force
1839 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1840 This is considered much lower than the risk of a memory corruption caused by
1841 aging components. If this is not acceptable, the cache can be disabled by
1842 setting this parameter to 0.
1843
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001844tune.pipesize <number>
1845 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1846 are the default size for the system. But sometimes when using TCP splicing,
1847 it can improve performance to increase pipe sizes, especially if it is
1848 suspected that pipes are not filled and that many calls to splice() are
1849 performed. This has an impact on the kernel's memory footprint, so this must
1850 not be changed if impacts are not understood.
1851
Olivier Houchard88698d92019-04-16 19:07:22 +02001852tune.pool-low-fd-ratio <number>
1853 This setting sets the max number of file descriptors (in percentage) used by
1854 haproxy globally against the maximum number of file descriptors haproxy can
1855 use before we stop putting connection into the idle pool for reuse. The
1856 default is 20.
1857
1858tune.pool-high-fd-ratio <number>
1859 This setting sets the max number of file descriptors (in percentage) used by
1860 haproxy globally against the maximum number of file descriptors haproxy can
1861 use before we start killing idle connections when we can't reuse a connection
1862 and we have to create a new one. The default is 25 (one quarter of the file
1863 descriptor will mean that roughly half of the maximum front connections can
1864 keep an idle connection behind, anything beyond this probably doesn't make
John Roesler27c754c2019-07-10 15:45:51 -05001865 much sense in the general case when targeting connection reuse).
Olivier Houchard88698d92019-04-16 19:07:22 +02001866
Willy Tarreaue803de22010-01-21 17:43:04 +01001867tune.rcvbuf.client <number>
1868tune.rcvbuf.server <number>
1869 Forces the kernel socket receive buffer size on the client or the server side
1870 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1871 and backends. It should normally never be set, and the default size (0) lets
John Roesler27c754c2019-07-10 15:45:51 -05001872 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001873 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001874 order to save kernel memory by preventing it from buffering too large amounts
1875 of received data. Lower values will significantly increase CPU usage though.
1876
Willy Tarreaub22fc302015-12-14 12:04:35 +01001877tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001878 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001879 socket buffers. One of them is that a read returns more than <recv_enough>
1880 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1881 may be changed by this setting to better deal with workloads involving lots
1882 of short messages such as telnet or SSH sessions.
1883
Olivier Houchard1599b802018-05-24 18:59:04 +02001884tune.runqueue-depth <number>
John Roesler27c754c2019-07-10 15:45:51 -05001885 Sets the maximum amount of task that can be processed at once when running
Olivier Houchard1599b802018-05-24 18:59:04 +02001886 tasks. The default value is 200. Increasing it may incur latency when
1887 dealing with I/Os, making it too small can incur extra overhead.
1888
Willy Tarreaue803de22010-01-21 17:43:04 +01001889tune.sndbuf.client <number>
1890tune.sndbuf.server <number>
1891 Forces the kernel socket send buffer size on the client or the server side to
1892 the specified value in bytes. This value applies to all TCP/HTTP frontends
1893 and backends. It should normally never be set, and the default size (0) lets
John Roesler27c754c2019-07-10 15:45:51 -05001894 the kernel auto-tune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001895 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001896 order to save kernel memory by preventing it from buffering too large amounts
1897 of received data. Lower values will significantly increase CPU usage though.
1898 Another use case is to prevent write timeouts with extremely slow clients due
1899 to the kernel waiting for a large part of the buffer to be read before
1900 notifying haproxy again.
1901
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001902tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001903 Sets the size of the global SSL session cache, in a number of blocks. A block
1904 is large enough to contain an encoded session without peer certificate.
1905 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001906 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001907 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001908 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001909 and reassigned. Higher values reduce the occurrence of such a purge, hence
1910 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1911 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001912 and are shared between all processes if "nbproc" is greater than 1. Setting
1913 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001914
Emeric Brun8dc60392014-05-09 13:52:00 +02001915tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001916 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001917 should normally not be used since it will force many renegotiations due to
1918 clients hitting a random process. But it may be required on some operating
1919 systems where none of the SSL cache synchronization method may be used. In
1920 this case, adding a first layer of hash-based load balancing before the SSL
1921 layer might limit the impact of the lack of session sharing.
1922
Emeric Brun4f65bff2012-11-16 15:11:00 +01001923tune.ssl.lifetime <timeout>
1924 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001925 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001926 does not guarantee that sessions will last that long, because if the cache is
1927 full, the longest idle sessions will be purged despite their configured
1928 lifetime. The real usefulness of this setting is to prevent sessions from
1929 being used for too long.
1930
Willy Tarreaubfd59462013-02-21 07:46:09 +01001931tune.ssl.maxrecord <number>
1932 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1933 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1934 data only once it has received a full record. With large records, it means
1935 that clients might have to download up to 16kB of data before starting to
1936 process them. Limiting the value can improve page load times on browsers
1937 located over high latency or low bandwidth networks. It is suggested to find
1938 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1939 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1940 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1941 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001942 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001943 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001944
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001945tune.ssl.default-dh-param <number>
1946 Sets the maximum size of the Diffie-Hellman parameters used for generating
1947 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1948 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1949 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1950 this maximum value. Default value if 1024. Only 1024 or higher values are
1951 allowed. Higher values will increase the CPU load, and values greater than
1952 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001953 used if static Diffie-Hellman parameters are supplied either directly
1954 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001955
Christopher Faulet31af49d2015-06-09 17:29:50 +02001956tune.ssl.ssl-ctx-cache-size <number>
1957 Sets the size of the cache used to store generated certificates to <number>
1958 entries. This is a LRU cache. Because generating a SSL certificate
1959 dynamically is expensive, they are cached. The default cache size is set to
1960 1000 entries.
1961
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001962tune.ssl.capture-cipherlist-size <number>
1963 Sets the maximum size of the buffer used for capturing client-hello cipher
1964 list. If the value is 0 (default value) the capture is disabled, otherwise
1965 a buffer is allocated for each SSL/TLS connection.
1966
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001967tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001968tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001969tune.vars.reqres-max-size <size>
1970tune.vars.sess-max-size <size>
1971tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001972 These five tunes help to manage the maximum amount of memory used by the
1973 variables system. "global" limits the overall amount of memory available for
1974 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1975 memory for the session scope, "txn" for the transaction scope, and "reqres"
1976 limits the memory for each request or response processing.
1977 Memory accounting is hierarchical, meaning more coarse grained limits include
1978 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1979 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001980
Daniel Schneller0b547052016-03-21 20:46:57 +01001981 For example, when "tune.vars.sess-max-size" is limited to 100,
1982 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1983 100 either. If we create a variable "txn.var" that contains 100 bytes,
1984 all available space is consumed.
1985 Notice that exceeding the limits at runtime will not result in an error
1986 message, but values might be cut off or corrupted. So make sure to accurately
1987 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001988
William Lallemanda509e4c2012-11-07 16:54:34 +01001989tune.zlib.memlevel <number>
1990 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001991 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001992 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01001993 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01001994 between 1 and 9. The default value is 8.
1995
1996tune.zlib.windowsize <number>
1997 Sets the window size (the size of the history buffer) as a parameter of the
1998 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01001999 in better compression at the expense of memory usage. Can be a value between
2000 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002001
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020023.3. Debugging
2003--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02002004
2005debug
2006 Enables debug mode which dumps to stdout all exchanges, and disables forking
2007 into background. It is the equivalent of the command-line argument "-d". It
2008 should never be used in a production configuration since it may prevent full
2009 system startup.
2010
2011quiet
2012 Do not display any message during startup. It is equivalent to the command-
2013 line argument "-q".
2014
Emeric Brunf099e792010-09-27 12:05:28 +02002015
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010020163.4. Userlists
2017--------------
2018It is possible to control access to frontend/backend/listen sections or to
2019http stats by allowing only authenticated and authorized users. To do this,
2020it is required to create at least one userlist and to define users.
2021
2022userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01002023 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002024 used to store authentication & authorization data for independent customers.
2025
2026group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01002027 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002028 attach users to this group by using a comma separated list of names
2029 proceeded by "users" keyword.
2030
Cyril Bontéf0c60612010-02-06 14:44:47 +01002031user <username> [password|insecure-password <password>]
2032 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002033 Adds user <username> to the current userlist. Both secure (encrypted) and
2034 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002035 evaluated using the crypt(3) function, so depending on the system's
2036 capabilities, different algorithms are supported. For example, modern Glibc
2037 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
2038 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002039
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01002040 Attention: Be aware that using encrypted passwords might cause significantly
2041 increased CPU usage, depending on the number of requests, and the algorithm
2042 used. For any of the hashed variants, the password for each request must
2043 be processed through the chosen algorithm, before it can be compared to the
2044 value specified in the config file. Most current algorithms are deliberately
2045 designed to be expensive to compute to achieve resistance against brute
2046 force attacks. They do not simply salt/hash the clear text password once,
2047 but thousands of times. This can quickly become a major factor in haproxy's
2048 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002049
2050 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01002051 userlist L1
2052 group G1 users tiger,scott
2053 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002054
Cyril Bontéf0c60612010-02-06 14:44:47 +01002055 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
2056 user scott insecure-password elgato
2057 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002058
Cyril Bontéf0c60612010-02-06 14:44:47 +01002059 userlist L2
2060 group G1
2061 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002062
Cyril Bontéf0c60612010-02-06 14:44:47 +01002063 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
2064 user scott insecure-password elgato groups G1,G2
2065 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01002066
2067 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002068
Emeric Brunf099e792010-09-27 12:05:28 +02002069
20703.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02002071----------
Emeric Brun94900952015-06-11 18:25:54 +02002072It is possible to propagate entries of any data-types in stick-tables between
2073several haproxy instances over TCP connections in a multi-master fashion. Each
2074instance pushes its local updates and insertions to remote peers. The pushed
2075values overwrite remote ones without aggregation. Interrupted exchanges are
2076automatically detected and recovered from the last known point.
2077In addition, during a soft restart, the old process connects to the new one
2078using such a TCP connection to push all its entries before the new process
2079tries to connect to other peers. That ensures very fast replication during a
2080reload, it typically takes a fraction of a second even for large tables.
2081Note that Server IDs are used to identify servers remotely, so it is important
2082that configurations look similar or at least that the same IDs are forced on
2083each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02002084
2085peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002086 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02002087 which is referenced by one or more stick-tables.
2088
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002089bind [<address>]:<port_range> [, ...] [param*]
2090 Defines the binding parameters of the local peer of this "peers" section.
2091 Such lines are not supported with "peer" line in the same "peers" section.
2092
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002093disabled
2094 Disables a peers section. It disables both listening and any synchronization
2095 related to this section. This is provided to disable synchronization of stick
2096 tables without having to comment out all "peers" references.
2097
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002098default-bind [param*]
2099 Defines the binding parameters for the local peer, excepted its address.
2100
2101default-server [param*]
2102 Change default options for a server in a "peers" section.
2103
2104 Arguments:
2105 <param*> is a list of parameters for this server. The "default-server"
2106 keyword accepts an important number of options and has a complete
2107 section dedicated to it. Please refer to section 5 for more
2108 details.
2109
2110
2111 See also: "server" and section 5 about server options
2112
Willy Tarreau77e4bd12015-05-01 20:02:17 +02002113enable
2114 This re-enables a disabled peers section which was previously disabled.
2115
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002116peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02002117 Defines a peer inside a peers section.
2118 If <peername> is set to the local peer name (by default hostname, or forced
2119 using "-L" command line option), haproxy will listen for incoming remote peer
2120 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
2121 to join the remote peer, and <peername> is used at the protocol level to
2122 identify and validate the remote peer on the server side.
2123
2124 During a soft restart, local peer <ip>:<port> is used by the old instance to
2125 connect the new one and initiate a complete replication (teaching process).
2126
2127 It is strongly recommended to have the exact same peers declaration on all
2128 peers and to only rely on the "-L" command line argument to change the local
2129 peer name. This makes it easier to maintain coherent configuration files
2130 across all peers.
2131
William Lallemandb2f07452015-05-12 14:27:13 +02002132 You may want to reference some environment variables in the address
2133 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01002134
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002135 Note: "peer" keyword may transparently be replaced by "server" keyword (see
2136 "server" keyword explanation below).
2137
2138server <peername> [<ip>:<port>] [param*]
Michael Prokop4438c602019-05-24 10:25:45 +02002139 As previously mentioned, "peer" keyword may be replaced by "server" keyword
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002140 with a support for all "server" parameters found in 5.2 paragraph.
2141 If the underlying peer is local, <ip>:<port> parameters must not be present.
2142 These parameters must be provided on a "bind" line (see "bind" keyword
2143 of this "peers" section).
2144 Some of these parameters are irrelevant for "peers" sections.
2145
2146
Cyril Bontédc4d9032012-04-08 21:57:39 +02002147 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002148 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02002149 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002150 peer haproxy1 192.168.0.1:1024
2151 peer haproxy2 192.168.0.2:1024
2152 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02002153
2154 backend mybackend
2155 mode tcp
2156 balance roundrobin
2157 stick-table type ip size 20k peers mypeers
2158 stick on src
2159
Willy Tarreauf7b30a92010-12-06 22:59:17 +01002160 server srv1 192.168.0.30:80
2161 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02002162
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01002163 Example:
2164 peers mypeers
2165 bind 127.0.0.11:10001 ssl crt mycerts/pem
2166 default-server ssl verify none
2167 server hostA 127.0.0.10:10000
2168 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02002169
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002170
2171table <tablename> type {ip | integer | string [len <length>] | binary [len <length>]}
2172 size <size> [expire <expire>] [nopurge] [store <data_type>]*
2173
2174 Configure a stickiness table for the current section. This line is parsed
2175 exactly the same way as the "stick-table" keyword in others section, except
John Roesler27c754c2019-07-10 15:45:51 -05002176 for the "peers" argument which is not required here and with an additional
Frédéric Lécaille4f5b77c2019-03-18 14:05:58 +01002177 mandatory first parameter to designate the stick-table. Contrary to others
2178 sections, there may be several "table" lines in "peers" sections (see also
2179 "stick-table" keyword).
2180
2181 Also be aware of the fact that "peers" sections have their own stick-table
2182 namespaces to avoid collisions between stick-table names identical in
2183 different "peers" section. This is internally handled prepending the "peers"
2184 sections names to the name of the stick-tables followed by a '/' character.
2185 If somewhere else in the configuration file you have to refer to such
2186 stick-tables declared in "peers" sections you must use the prefixed version
2187 of the stick-table name as follows:
2188
2189 peers mypeers
2190 peer A ...
2191 peer B ...
2192 table t1 ...
2193
2194 frontend fe1
2195 tcp-request content track-sc0 src table mypeers/t1
2196
2197 This is also this prefixed version of the stick-table names which must be
2198 used to refer to stick-tables through the CLI.
2199
2200 About "peers" protocol, as only "peers" belonging to the same section may
2201 communicate with each others, there is no need to do such a distinction.
2202 Several "peers" sections may declare stick-tables with the same name.
2203 This is shorter version of the stick-table name which is sent over the network.
2204 There is only a '/' character as prefix to avoid stick-table name collisions between
2205 stick-tables declared as backends and stick-table declared in "peers" sections
2206 as follows in this weird but supported configuration:
2207
2208 peers mypeers
2209 peer A ...
2210 peer B ...
2211 table t1 type string size 10m store gpc0
2212
2213 backend t1
2214 stick-table type string size 10m store gpc0 peers mypeers
2215
2216 Here "t1" table declared in "mypeeers" section has "mypeers/t1" as global name.
2217 "t1" table declared as a backend as "t1" as global name. But at peer protocol
2218 level the former table is named "/t1", the latter is again named "t1".
2219
Simon Horman51a1cf62015-02-03 13:00:44 +090022203.6. Mailers
2221------------
2222It is possible to send email alerts when the state of servers changes.
2223If configured email alerts are sent to each mailer that is configured
2224in a mailers section. Email is sent to mailers using SMTP.
2225
Pieter Baauw386a1272015-08-16 15:26:24 +02002226mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09002227 Creates a new mailer list with the name <mailersect>. It is an
2228 independent section which is referenced by one or more proxies.
2229
2230mailer <mailername> <ip>:<port>
2231 Defines a mailer inside a mailers section.
2232
2233 Example:
2234 mailers mymailers
2235 mailer smtp1 192.168.0.1:587
2236 mailer smtp2 192.168.0.2:587
2237
2238 backend mybackend
2239 mode tcp
2240 balance roundrobin
2241
2242 email-alert mailers mymailers
2243 email-alert from test1@horms.org
2244 email-alert to test2@horms.org
2245
2246 server srv1 192.168.0.30:80
2247 server srv2 192.168.0.31:80
2248
Pieter Baauw235fcfc2016-02-13 15:33:40 +01002249timeout mail <time>
2250 Defines the time available for a mail/connection to be made and send to
2251 the mail-server. If not defined the default value is 10 seconds. To allow
2252 for at least two SYN-ACK packets to be send during initial TCP handshake it
2253 is advised to keep this value above 4 seconds.
2254
2255 Example:
2256 mailers mymailers
2257 timeout mail 20s
2258 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09002259
William Lallemandc9515522019-06-12 16:32:11 +020022603.7. Programs
2261-------------
2262In master-worker mode, it is possible to launch external binaries with the
2263master, these processes are called programs. These programs are launched and
2264managed the same way as the workers.
2265
2266During a reload of HAProxy, those processes are dealing with the same
2267sequence as a worker:
2268
2269 - the master is re-executed
2270 - the master sends a SIGUSR1 signal to the program
2271 - if "option start-on-reload" is not disabled, the master launches a new
2272 instance of the program
2273
2274During a stop, or restart, a SIGTERM is sent to the programs.
2275
2276program <name>
2277 This is a new program section, this section will create an instance <name>
2278 which is visible in "show proc" on the master CLI. (See "9.4. Master CLI" in
2279 the management guide).
2280
2281command <command> [arguments*]
2282 Define the command to start with optional arguments. The command is looked
2283 up in the current PATH if it does not include an absolute path. This is a
2284 mandatory option of the program section. Arguments containing spaces must
2285 be enclosed in quotes or double quotes or be prefixed by a backslash.
2286
2287option start-on-reload
2288no option start-on-reload
2289 Start (or not) a new instance of the program upon a reload of the master.
2290 The default is to start a new instance. This option may only be used in a
2291 program section.
2292
2293
Willy Tarreauc57f0e22009-05-10 13:12:33 +020022944. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002295----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002296
Willy Tarreau6a06a402007-07-15 20:15:28 +02002297Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002298 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002299 - frontend <name>
2300 - backend <name>
2301 - listen <name>
2302
2303A "defaults" section sets default parameters for all other sections following
2304its declaration. Those default parameters are reset by the next "defaults"
2305section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002306section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002307
2308A "frontend" section describes a set of listening sockets accepting client
2309connections.
2310
2311A "backend" section describes a set of servers to which the proxy will connect
2312to forward incoming connections.
2313
2314A "listen" section defines a complete proxy with its frontend and backend
2315parts combined in one section. It is generally useful for TCP-only traffic.
2316
Willy Tarreau0ba27502007-12-24 16:55:16 +01002317All proxy names must be formed from upper and lower case letters, digits,
2318'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2319case-sensitive, which means that "www" and "WWW" are two different proxies.
2320
2321Historically, all proxy names could overlap, it just caused troubles in the
2322logs. Since the introduction of content switching, it is mandatory that two
2323proxies with overlapping capabilities (frontend/backend) have different names.
2324However, it is still permitted that a frontend and a backend share the same
2325name, as this configuration seems to be commonly encountered.
2326
2327Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2328and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002329bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002330protocol, and can interact with it by allowing, blocking, switching, adding,
2331modifying, or removing arbitrary contents in requests or responses, based on
2332arbitrary criteria.
2333
Willy Tarreau70dffda2014-01-30 03:07:23 +01002334In HTTP mode, the processing applied to requests and responses flowing over
2335a connection depends in the combination of the frontend's HTTP options and
Christopher Faulet315b39c2018-09-21 16:26:19 +02002336the backend's. HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002337
2338 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2339 requests and responses are processed, and connections remain open but idle
2340 between responses and new requests.
2341
2342 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
2343 1.0 to 1.5-dev21 : only the first request and response are processed, and
2344 everything else is forwarded with no analysis at all. This mode should not
Christopher Faulet6c9bbb22019-03-26 21:37:23 +01002345 be used as it creates lots of trouble with logging and HTTP processing.
2346 And because it cannot work in HTTP/2, this option is deprecated and it is
2347 only supported on legacy HTTP frontends. In HTX, it is ignored and a
2348 warning is emitted during HAProxy startup.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002349
Willy Tarreau70dffda2014-01-30 03:07:23 +01002350 - SCL: server close ("option http-server-close") : the server-facing
2351 connection is closed after the end of the response is received, but the
2352 client-facing connection remains open.
2353
Christopher Faulet315b39c2018-09-21 16:26:19 +02002354 - CLO: close ("option httpclose"): the connection is closed after the end of
2355 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002356
2357The effective mode that will be applied to a connection passing through a
2358frontend and a backend can be determined by both proxy modes according to the
2359following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002360weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002361
Christopher Faulet315b39c2018-09-21 16:26:19 +02002362 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002363
Christopher Faulet315b39c2018-09-21 16:26:19 +02002364 | KAL | SCL | CLO
2365 ----+-----+-----+----
2366 KAL | KAL | SCL | CLO
2367 ----+-----+-----+----
2368 TUN | TUN | SCL | CLO
2369 Frontend ----+-----+-----+----
2370 mode SCL | SCL | SCL | CLO
2371 ----+-----+-----+----
2372 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002373
Willy Tarreau0ba27502007-12-24 16:55:16 +01002374
Willy Tarreau70dffda2014-01-30 03:07:23 +01002375
Willy Tarreauc57f0e22009-05-10 13:12:33 +020023764.1. Proxy keywords matrix
2377--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002378
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002379The following list of keywords is supported. Most of them may only be used in a
2380limited set of section types. Some of them are marked as "deprecated" because
2381they are inherited from an old syntax which may be confusing or functionally
2382limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002383marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002384option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002385and must be disabled for a specific instance. Such options may also be prefixed
2386with "default" in order to restore default settings regardless of what has been
2387specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002388
Willy Tarreau6a06a402007-07-15 20:15:28 +02002389
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002390 keyword defaults frontend listen backend
2391------------------------------------+----------+----------+---------+---------
2392acl - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002393backlog X X X -
2394balance X - X X
2395bind - X X -
2396bind-process X X X X
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002397block (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002398capture cookie - X X -
2399capture request header - X X -
2400capture response header - X X -
2401clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002402compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002403contimeout (deprecated) X - X X
2404cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002405declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002406default-server X - X X
2407default_backend X X X -
2408description - X X X
2409disabled X X X X
2410dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002411email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002412email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002413email-alert mailers X X X X
2414email-alert myhostname X X X X
2415email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002416enabled X X X X
2417errorfile X X X X
2418errorloc X X X X
2419errorloc302 X X X X
2420-- keyword -------------------------- defaults - frontend - listen -- backend -
2421errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002422force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002423filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002424fullconn X - X X
2425grace X X X X
2426hash-type X - X X
2427http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002428http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002429http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002430http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002431http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002432http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002433http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002434id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002435ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002436load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002437log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002438log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002439log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002440log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002441max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002442maxconn X X X -
2443mode X X X X
2444monitor fail - X X -
2445monitor-net X X X -
2446monitor-uri X X X -
2447option abortonclose (*) X - X X
2448option accept-invalid-http-request (*) X X X -
2449option accept-invalid-http-response (*) X - X X
2450option allbackups (*) X - X X
2451option checkcache (*) X - X X
2452option clitcpka (*) X X X -
2453option contstats (*) X X X -
2454option dontlog-normal (*) X X X -
2455option dontlognull (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002456-- keyword -------------------------- defaults - frontend - listen -- backend -
2457option forwardfor X X X X
Christopher Fauleta99ff4d2019-07-22 16:18:24 +02002458option h1-case-adjust-bogus-client (*) X X X -
2459option h1-case-adjust-bogus-server (*) X - X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002460option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002461option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002462option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002463option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002464option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002465option http-server-close (*) X X X X
Christopher Faulet6c9bbb22019-03-26 21:37:23 +01002466option http-tunnel (deprecated) (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002467option http-use-proxy-header (*) X X X -
Willy Tarreau68ad3a42018-10-22 11:49:15 +02002468option http-use-htx (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002469option httpchk X - X X
2470option httpclose (*) X X X X
Freddy Spierenburge88b7732019-03-25 14:35:17 +01002471option httplog X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002472option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002473option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002474option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002475option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002476option log-health-checks (*) X - X X
2477option log-separate-errors (*) X X X -
2478option logasap (*) X X X -
2479option mysql-check X - X X
2480option nolinger (*) X X X X
2481option originalto X X X X
2482option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002483option pgsql-check X - X X
2484option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002485option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002486option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002487option smtpchk X - X X
2488option socket-stats (*) X X X -
2489option splice-auto (*) X X X X
2490option splice-request (*) X X X X
2491option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002492option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002493option srvtcpka (*) X - X X
2494option ssl-hello-chk X - X X
2495-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002496option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002497option tcp-smart-accept (*) X X X -
2498option tcp-smart-connect (*) X - X X
2499option tcpka X X X X
2500option tcplog X X X X
2501option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002502external-check command X - X X
2503external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002504persist rdp-cookie X - X X
2505rate-limit sessions X X X -
2506redirect - X X X
2507redisp (deprecated) X - X X
2508redispatch (deprecated) X - X X
Willy Tarreau96d51952019-05-22 20:34:35 +02002509reqadd (deprecated) - X X X
2510reqallow (deprecated) - X X X
2511reqdel (deprecated) - X X X
2512reqdeny (deprecated) - X X X
2513reqiallow (deprecated) - X X X
2514reqidel (deprecated) - X X X
2515reqideny (deprecated) - X X X
2516reqipass (deprecated) - X X X
2517reqirep (deprecated) - X X X
2518reqitarpit (deprecated) - X X X
2519reqpass (deprecated) - X X X
2520reqrep (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002521-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau96d51952019-05-22 20:34:35 +02002522reqtarpit (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002523retries X - X X
Olivier Houcharda254a372019-04-05 15:30:12 +02002524retry-on X - X X
Willy Tarreau96d51952019-05-22 20:34:35 +02002525rspadd (deprecated) - X X X
2526rspdel (deprecated) - X X X
2527rspdeny (deprecated) - X X X
2528rspidel (deprecated) - X X X
2529rspideny (deprecated) - X X X
2530rspirep (deprecated) - X X X
2531rsprep (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002532server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002533server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002534server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002535source X - X X
2536srvtimeout (deprecated) X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002537stats admin - X X X
2538stats auth X X X X
2539stats enable X X X X
2540stats hide-version X X X X
2541stats http-request - X X X
2542stats realm X X X X
2543stats refresh X X X X
2544stats scope X X X X
2545stats show-desc X X X X
2546stats show-legends X X X X
2547stats show-node X X X X
2548stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002549-- keyword -------------------------- defaults - frontend - listen -- backend -
2550stick match - - X X
2551stick on - - X X
2552stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002553stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002554stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002555tcp-check connect - - X X
2556tcp-check expect - - X X
2557tcp-check send - - X X
2558tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002559tcp-request connection - X X -
2560tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002561tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002562tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002563tcp-response content - - X X
2564tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002565timeout check X - X X
2566timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002567timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002568timeout clitimeout (deprecated) X X X -
2569timeout connect X - X X
2570timeout contimeout (deprecated) X - X X
2571timeout http-keep-alive X X X X
2572timeout http-request X X X X
2573timeout queue X - X X
2574timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002575timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002576timeout srvtimeout (deprecated) X - X X
2577timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002578timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002579transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002580unique-id-format X X X -
2581unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002582use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002583use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002584------------------------------------+----------+----------+---------+---------
2585 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002586
Willy Tarreau0ba27502007-12-24 16:55:16 +01002587
Willy Tarreauc57f0e22009-05-10 13:12:33 +020025884.2. Alphabetically sorted keywords reference
2589---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002590
2591This section provides a description of each keyword and its usage.
2592
2593
2594acl <aclname> <criterion> [flags] [operator] <value> ...
2595 Declare or complete an access list.
2596 May be used in sections : defaults | frontend | listen | backend
2597 no | yes | yes | yes
2598 Example:
2599 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2600 acl invalid_src src_port 0:1023
2601 acl local_dst hdr(host) -i localhost
2602
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002603 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002604
2605
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002606backlog <conns>
2607 Give hints to the system about the approximate listen backlog desired size
2608 May be used in sections : defaults | frontend | listen | backend
2609 yes | yes | yes | no
2610 Arguments :
2611 <conns> is the number of pending connections. Depending on the operating
2612 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002613 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002614
2615 In order to protect against SYN flood attacks, one solution is to increase
2616 the system's SYN backlog size. Depending on the system, sometimes it is just
2617 tunable via a system parameter, sometimes it is not adjustable at all, and
2618 sometimes the system relies on hints given by the application at the time of
2619 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2620 to the listen() syscall. On systems which can make use of this value, it can
2621 sometimes be useful to be able to specify a different value, hence this
2622 backlog parameter.
2623
2624 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2625 used as a hint and the system accepts up to the smallest greater power of
2626 two, and never more than some limits (usually 32768).
2627
2628 See also : "maxconn" and the target operating system's tuning guide.
2629
2630
Willy Tarreau0ba27502007-12-24 16:55:16 +01002631balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002632balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002633 Define the load balancing algorithm to be used in a backend.
2634 May be used in sections : defaults | frontend | listen | backend
2635 yes | no | yes | yes
2636 Arguments :
2637 <algorithm> is the algorithm used to select a server when doing load
2638 balancing. This only applies when no persistence information
2639 is available, or when a connection is redispatched to another
2640 server. <algorithm> may be one of the following :
2641
2642 roundrobin Each server is used in turns, according to their weights.
2643 This is the smoothest and fairest algorithm when the server's
2644 processing time remains equally distributed. This algorithm
2645 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002646 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002647 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002648 large farms, when a server becomes up after having been down
2649 for a very short time, it may sometimes take a few hundreds
2650 requests for it to be re-integrated into the farm and start
2651 receiving traffic. This is normal, though very rare. It is
2652 indicated here in case you would have the chance to observe
2653 it, so that you don't worry.
2654
2655 static-rr Each server is used in turns, according to their weights.
2656 This algorithm is as similar to roundrobin except that it is
2657 static, which means that changing a server's weight on the
2658 fly will have no effect. On the other hand, it has no design
2659 limitation on the number of servers, and when a server goes
2660 up, it is always immediately reintroduced into the farm, once
2661 the full map is recomputed. It also uses slightly less CPU to
2662 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002663
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002664 leastconn The server with the lowest number of connections receives the
2665 connection. Round-robin is performed within groups of servers
2666 of the same load to ensure that all servers will be used. Use
2667 of this algorithm is recommended where very long sessions are
2668 expected, such as LDAP, SQL, TSE, etc... but is not very well
2669 suited for protocols using short sessions such as HTTP. This
2670 algorithm is dynamic, which means that server weights may be
2671 adjusted on the fly for slow starts for instance.
2672
Willy Tarreauf09c6602012-02-13 17:12:08 +01002673 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002674 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002675 identifier to the highest (see server parameter "id"), which
2676 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002677 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002678 not make sense to use this algorithm without setting maxconn.
2679 The purpose of this algorithm is to always use the smallest
2680 number of servers so that extra servers can be powered off
2681 during non-intensive hours. This algorithm ignores the server
2682 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002683 or IMAP than HTTP, though it can be useful there too. In
2684 order to use this algorithm efficiently, it is recommended
2685 that a cloud controller regularly checks server usage to turn
2686 them off when unused, and regularly checks backend queue to
2687 turn new servers on when the queue inflates. Alternatively,
2688 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002689
Willy Tarreau0ba27502007-12-24 16:55:16 +01002690 source The source IP address is hashed and divided by the total
2691 weight of the running servers to designate which server will
2692 receive the request. This ensures that the same client IP
2693 address will always reach the same server as long as no
2694 server goes down or up. If the hash result changes due to the
2695 number of running servers changing, many clients will be
2696 directed to a different server. This algorithm is generally
2697 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002698 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002699 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002700 static by default, which means that changing a server's
2701 weight on the fly will have no effect, but this can be
2702 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002703
Oskar Stolc8dc41842012-05-19 10:19:54 +01002704 uri This algorithm hashes either the left part of the URI (before
2705 the question mark) or the whole URI (if the "whole" parameter
2706 is present) and divides the hash value by the total weight of
2707 the running servers. The result designates which server will
2708 receive the request. This ensures that the same URI will
2709 always be directed to the same server as long as no server
2710 goes up or down. This is used with proxy caches and
2711 anti-virus proxies in order to maximize the cache hit rate.
2712 Note that this algorithm may only be used in an HTTP backend.
2713 This algorithm is static by default, which means that
2714 changing a server's weight on the fly will have no effect,
2715 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002716
Oskar Stolc8dc41842012-05-19 10:19:54 +01002717 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002718 "depth", both followed by a positive integer number. These
2719 options may be helpful when it is needed to balance servers
2720 based on the beginning of the URI only. The "len" parameter
2721 indicates that the algorithm should only consider that many
2722 characters at the beginning of the URI to compute the hash.
2723 Note that having "len" set to 1 rarely makes sense since most
2724 URIs start with a leading "/".
2725
2726 The "depth" parameter indicates the maximum directory depth
2727 to be used to compute the hash. One level is counted for each
2728 slash in the request. If both parameters are specified, the
2729 evaluation stops when either is reached.
2730
Willy Tarreau0ba27502007-12-24 16:55:16 +01002731 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002732 the query string of each HTTP GET request.
2733
2734 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002735 request entity will be searched for the parameter argument,
2736 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002737 ('?') in the URL. The message body will only start to be
2738 analyzed once either the advertised amount of data has been
2739 received or the request buffer is full. In the unlikely event
2740 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002741 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002742 be randomly balanced if at all. This keyword used to support
2743 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002744
2745 If the parameter is found followed by an equal sign ('=') and
2746 a value, then the value is hashed and divided by the total
2747 weight of the running servers. The result designates which
2748 server will receive the request.
2749
2750 This is used to track user identifiers in requests and ensure
2751 that a same user ID will always be sent to the same server as
2752 long as no server goes up or down. If no value is found or if
2753 the parameter is not found, then a round robin algorithm is
2754 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002755 backend. This algorithm is static by default, which means
2756 that changing a server's weight on the fly will have no
2757 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002758
Cyril Bontédc4d9032012-04-08 21:57:39 +02002759 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2760 request. Just as with the equivalent ACL 'hdr()' function,
2761 the header name in parenthesis is not case sensitive. If the
2762 header is absent or if it does not contain any value, the
2763 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002764
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002765 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002766 reducing the hash algorithm to the main domain part with some
2767 specific headers such as 'Host'. For instance, in the Host
2768 value "haproxy.1wt.eu", only "1wt" will be considered.
2769
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002770 This algorithm is static by default, which means that
2771 changing a server's weight on the fly will have no effect,
2772 but this can be changed using "hash-type".
2773
Willy Tarreau21c741a2019-01-14 18:14:27 +01002774 random
2775 random(<draws>)
2776 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02002777 hashing function. This means that the servers' weights are
2778 respected, dynamic weight changes immediately take effect, as
2779 well as new server additions. Random load balancing can be
2780 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01002781 or removed as it may avoid the hammering effect that could
2782 result from roundrobin or leastconn in this situation. The
2783 hash-balance-factor directive can be used to further improve
2784 fairness of the load balancing, especially in situations
2785 where servers show highly variable response times. When an
2786 argument <draws> is present, it must be an integer value one
2787 or greater, indicating the number of draws before selecting
2788 the least loaded of these servers. It was indeed demonstrated
2789 that picking the least loaded of two servers is enough to
2790 significantly improve the fairness of the algorithm, by
2791 always avoiding to pick the most loaded server within a farm
2792 and getting rid of any bias that could be induced by the
2793 unfair distribution of the consistent list. Higher values N
2794 will take away N-1 of the highest loaded servers at the
2795 expense of performance. With very high values, the algorithm
2796 will converge towards the leastconn's result but much slower.
2797 The default value is 2, which generally shows very good
2798 distribution and performance. This algorithm is also known as
2799 the Power of Two Random Choices and is described here :
2800 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02002801
Emeric Brun736aa232009-06-30 17:56:00 +02002802 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002803 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002804 The RDP cookie <name> (or "mstshash" if omitted) will be
2805 looked up and hashed for each incoming TCP request. Just as
2806 with the equivalent ACL 'req_rdp_cookie()' function, the name
2807 is not case-sensitive. This mechanism is useful as a degraded
2808 persistence mode, as it makes it possible to always send the
2809 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002810 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002811 used instead.
2812
2813 Note that for this to work, the frontend must ensure that an
2814 RDP cookie is already present in the request buffer. For this
2815 you must use 'tcp-request content accept' rule combined with
2816 a 'req_rdp_cookie_cnt' ACL.
2817
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002818 This algorithm is static by default, which means that
2819 changing a server's weight on the fly will have no effect,
2820 but this can be changed using "hash-type".
2821
Cyril Bontédc4d9032012-04-08 21:57:39 +02002822 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002823
Willy Tarreau0ba27502007-12-24 16:55:16 +01002824 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002825 algorithms. Right now, only "url_param" and "uri" support an
2826 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002827
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002828 The load balancing algorithm of a backend is set to roundrobin when no other
2829 algorithm, mode nor option have been set. The algorithm may only be set once
2830 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002831
Lukas Tribus80512b12018-10-27 20:07:40 +02002832 With authentication schemes that require the same connection like NTLM, URI
John Roesler27c754c2019-07-10 15:45:51 -05002833 based algorithms must not be used, as they would cause subsequent requests
Lukas Tribus80512b12018-10-27 20:07:40 +02002834 to be routed to different backend servers, breaking the invalid assumptions
2835 NTLM relies on.
2836
Willy Tarreau0ba27502007-12-24 16:55:16 +01002837 Examples :
2838 balance roundrobin
2839 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002840 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002841 balance hdr(User-Agent)
2842 balance hdr(host)
2843 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002844
2845 Note: the following caveats and limitations on using the "check_post"
2846 extension with "url_param" must be considered :
2847
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002848 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002849 to determine if the parameters will be found in the body or entity which
2850 may contain binary data. Therefore another method may be required to
2851 restrict consideration of POST requests that have no URL parameters in
2852 the body. (see acl reqideny http_end)
2853
2854 - using a <max_wait> value larger than the request buffer size does not
2855 make sense and is useless. The buffer size is set at build time, and
2856 defaults to 16 kB.
2857
2858 - Content-Encoding is not supported, the parameter search will probably
2859 fail; and load balancing will fall back to Round Robin.
2860
2861 - Expect: 100-continue is not supported, load balancing will fall back to
2862 Round Robin.
2863
Lukas Tribus23953682017-04-28 13:24:30 +00002864 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002865 If the entire parameter value is not present in the first chunk, the
2866 selection of server is undefined (actually, defined by how little
2867 actually appeared in the first chunk).
2868
2869 - This feature does not support generation of a 100, 411 or 501 response.
2870
2871 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002872 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002873 white space or control characters are found, indicating the end of what
2874 might be a URL parameter list. This is probably not a concern with SGML
2875 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002876
Willy Tarreau294d0f02015-08-10 19:40:12 +02002877 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002878
2879
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002880bind [<address>]:<port_range> [, ...] [param*]
2881bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002882 Define one or several listening addresses and/or ports in a frontend.
2883 May be used in sections : defaults | frontend | listen | backend
2884 no | yes | yes | no
2885 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002886 <address> is optional and can be a host name, an IPv4 address, an IPv6
2887 address, or '*'. It designates the address the frontend will
2888 listen on. If unset, all IPv4 addresses of the system will be
2889 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002890 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002891 Optionally, an address family prefix may be used before the
2892 address to force the family regardless of the address format,
2893 which can be useful to specify a path to a unix socket with
2894 no slash ('/'). Currently supported prefixes are :
2895 - 'ipv4@' -> address is always IPv4
2896 - 'ipv6@' -> address is always IPv6
2897 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002898 - 'abns@' -> address is in abstract namespace (Linux only).
2899 Note: since abstract sockets are not "rebindable", they
2900 do not cope well with multi-process mode during
2901 soft-restart, so it is better to avoid them if
2902 nbproc is greater than 1. The effect is that if the
2903 new process fails to start, only one of the old ones
2904 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002905 - 'fd@<n>' -> use file descriptor <n> inherited from the
2906 parent. The fd must be bound and may or may not already
2907 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002908 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2909 connected unix socket or of a socketpair. The bind waits
2910 to receive a FD over the unix socket and uses it as if it
2911 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02002912 You may want to reference some environment variables in the
2913 address parameter, see section 2.3 about environment
2914 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002915
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002916 <port_range> is either a unique TCP port, or a port range for which the
2917 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002918 above. The port is mandatory for TCP listeners. Note that in
2919 the case of an IPv6 address, the port is always the number
2920 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002921 - a numerical port (ex: '80')
2922 - a dash-delimited ports range explicitly stating the lower
2923 and upper bounds (ex: '2000-2100') which are included in
2924 the range.
2925
2926 Particular care must be taken against port ranges, because
2927 every <address:port> couple consumes one socket (= a file
2928 descriptor), so it's easy to consume lots of descriptors
2929 with a simple range, and to run out of sockets. Also, each
2930 <address:port> couple must be used only once among all
2931 instances running on a same system. Please note that binding
2932 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002933 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002934 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002935
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002936 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002937 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002938 receive UNIX connections on the socket located at this place.
2939 The path must begin with a slash and by default is absolute.
2940 It can be relative to the prefix defined by "unix-bind" in
2941 the global section. Note that the total length of the prefix
2942 followed by the socket path cannot exceed some system limits
2943 for UNIX sockets, which commonly are set to 107 characters.
2944
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002945 <param*> is a list of parameters common to all sockets declared on the
2946 same line. These numerous parameters depend on OS and build
2947 options and have a complete section dedicated to them. Please
2948 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002949
Willy Tarreau0ba27502007-12-24 16:55:16 +01002950 It is possible to specify a list of address:port combinations delimited by
2951 commas. The frontend will then listen on all of these addresses. There is no
2952 fixed limit to the number of addresses and ports which can be listened on in
2953 a frontend, as well as there is no limit to the number of "bind" statements
2954 in a frontend.
2955
2956 Example :
2957 listen http_proxy
2958 bind :80,:443
2959 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002960 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002961
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002962 listen http_https_proxy
2963 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002964 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002965
Willy Tarreau24709282013-03-10 21:32:12 +01002966 listen http_https_proxy_explicit
2967 bind ipv6@:80
2968 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2969 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2970
Willy Tarreaudad36a32013-03-11 01:20:04 +01002971 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002972 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002973
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002974 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2975 sun_path length is used for the address length. Some other programs
2976 such as socat use the string length only by default. Pass the option
2977 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2978 make it compatible with HAProxy's.
2979
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002980 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002981 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002982
2983
Christopher Fauletff4121f2017-11-22 16:38:49 +01002984bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002985 Limit visibility of an instance to a certain set of processes numbers.
2986 May be used in sections : defaults | frontend | listen | backend
2987 yes | yes | yes | yes
2988 Arguments :
2989 all All process will see this instance. This is the default. It
2990 may be used to override a default value.
2991
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002992 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002993 option may be combined with other numbers.
2994
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002995 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002996 option may be combined with other numbers. Do not use it
2997 with less than 2 processes otherwise some instances might be
2998 missing from all processes.
2999
Christopher Fauletff4121f2017-11-22 16:38:49 +01003000 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003001 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01003002 the machine's word size. Ranges can be partially defined. The
3003 higher bound can be omitted. In such case, it is replaced by
3004 the corresponding maximum value. If a proxy is bound to
3005 process numbers greater than the configured global.nbproc, it
3006 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02003007 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003008
3009 This keyword limits binding of certain instances to certain processes. This
3010 is useful in order not to have too many processes listening to the same
3011 ports. For instance, on a dual-core machine, it might make sense to set
3012 'nbproc 2' in the global section, then distributes the listeners among 'odd'
3013 and 'even' instances.
3014
Willy Tarreaua9db57e2013-01-18 11:29:29 +01003015 At the moment, it is not possible to reference more than 32 or 64 processes
3016 using this keyword, but this should be more than enough for most setups.
3017 Please note that 'all' really means all processes regardless of the machine's
3018 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003019
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003020 Each "bind" line may further be limited to a subset of the proxy's processes,
3021 please consult the "process" bind keyword in section 5.1.
3022
Willy Tarreaub369a042014-09-16 13:21:03 +02003023 When a frontend has no explicit "bind-process" line, it tries to bind to all
3024 the processes referenced by its "bind" lines. That means that frontends can
3025 easily adapt to their listeners' processes.
3026
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003027 If some backends are referenced by frontends bound to other processes, the
3028 backend automatically inherits the frontend's processes.
3029
3030 Example :
3031 listen app_ip1
3032 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003033 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003034
3035 listen app_ip2
3036 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003037 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003038
3039 listen management
3040 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02003041 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003042
Willy Tarreau110ecc12012-11-15 17:50:01 +01003043 listen management
3044 bind 10.0.0.4:80
3045 bind-process 1-4
3046
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02003047 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01003048
3049
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02003050block { if | unless } <condition> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003051 Block a layer 7 request if/unless a condition is matched
3052 May be used in sections : defaults | frontend | listen | backend
3053 no | yes | yes | yes
3054
3055 The HTTP request will be blocked very early in the layer 7 processing
3056 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003057 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02003058 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01003059 conditions are met or not met. There is no fixed limit to the number of
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03003060 "block" statements per instance. To block connections at layer 4 (without
3061 sending a 403 error) see "tcp-request connection reject" and
3062 "tcp-request content reject" rules.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003063
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02003064 This form is deprecated, do not use it in any new configuration, use the new
3065 "http-request deny" instead.
3066
Willy Tarreau0ba27502007-12-24 16:55:16 +01003067 Example:
3068 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
3069 acl invalid_src src_port 0:1023
3070 acl local_dst hdr(host) -i localhost
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +03003071 # block is deprecated. Use http-request deny instead:
3072 #block if invalid_src || local_dst
3073 http-request deny if invalid_src || local_dst
Willy Tarreau0ba27502007-12-24 16:55:16 +01003074
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03003075 See also : section 7 about ACL usage, "http-request deny",
3076 "http-response deny", "tcp-request connection reject" and
3077 "tcp-request content reject".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003078
3079capture cookie <name> len <length>
3080 Capture and log a cookie in the request and in the response.
3081 May be used in sections : defaults | frontend | listen | backend
3082 no | yes | yes | no
3083 Arguments :
3084 <name> is the beginning of the name of the cookie to capture. In order
3085 to match the exact name, simply suffix the name with an equal
3086 sign ('='). The full name will appear in the logs, which is
3087 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01003088 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003089
3090 <length> is the maximum number of characters to report in the logs, which
3091 include the cookie name, the equal sign and the value, all in the
3092 standard "name=value" form. The string will be truncated on the
3093 right if it exceeds <length>.
3094
3095 Only the first cookie is captured. Both the "cookie" request headers and the
3096 "set-cookie" response headers are monitored. This is particularly useful to
3097 check for application bugs causing session crossing or stealing between
3098 users, because generally the user's cookies can only change on a login page.
3099
3100 When the cookie was not presented by the client, the associated log column
3101 will report "-". When a request does not cause a cookie to be assigned by the
3102 server, a "-" is reported in the response column.
3103
3104 The capture is performed in the frontend only because it is necessary that
3105 the log format does not change for a given frontend depending on the
3106 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01003107 "capture cookie" statement in a frontend. The maximum capture length is set
3108 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
3109 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003110
3111 Example:
3112 capture cookie ASPSESSION len 32
3113
3114 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003115 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003116
3117
3118capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003119 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003120 May be used in sections : defaults | frontend | listen | backend
3121 no | yes | yes | no
3122 Arguments :
3123 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003124 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003125 appear in the requests, with the first letter of each word in
3126 upper case. The header name will not appear in the logs, only the
3127 value is reported, but the position in the logs is respected.
3128
3129 <length> is the maximum number of characters to extract from the value and
3130 report in the logs. The string will be truncated on the right if
3131 it exceeds <length>.
3132
Willy Tarreau4460d032012-11-21 23:37:37 +01003133 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003134 value will be added to the logs between braces ('{}'). If multiple headers
3135 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003136 in the same order they were declared in the configuration. Non-existent
3137 headers will be logged just as an empty string. Common uses for request
3138 header captures include the "Host" field in virtual hosting environments, the
3139 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003140 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003141 environments to find where the request came from.
3142
3143 Note that when capturing headers such as "User-agent", some spaces may be
3144 logged, making the log analysis more difficult. Thus be careful about what
3145 you log if you know your log parser is not smart enough to rely on the
3146 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003147
Willy Tarreau0900abb2012-11-22 00:21:46 +01003148 There is no limit to the number of captured request headers nor to their
3149 length, though it is wise to keep them low to limit memory usage per session.
3150 In order to keep log format consistent for a same frontend, header captures
3151 can only be declared in a frontend. It is not possible to specify a capture
3152 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003153
3154 Example:
3155 capture request header Host len 15
3156 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01003157 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01003158
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003159 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003160 about logging.
3161
3162
3163capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01003164 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003165 May be used in sections : defaults | frontend | listen | backend
3166 no | yes | yes | no
3167 Arguments :
3168 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003169 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01003170 appear in the response, with the first letter of each word in
3171 upper case. The header name will not appear in the logs, only the
3172 value is reported, but the position in the logs is respected.
3173
3174 <length> is the maximum number of characters to extract from the value and
3175 report in the logs. The string will be truncated on the right if
3176 it exceeds <length>.
3177
Willy Tarreau4460d032012-11-21 23:37:37 +01003178 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01003179 result will be added to the logs between braces ('{}') after the captured
3180 request headers. If multiple headers are captured, they will be delimited by
3181 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01003182 the configuration. Non-existent headers will be logged just as an empty
3183 string. Common uses for response header captures include the "Content-length"
3184 header which indicates how many bytes are expected to be returned, the
3185 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003186
Willy Tarreau0900abb2012-11-22 00:21:46 +01003187 There is no limit to the number of captured response headers nor to their
3188 length, though it is wise to keep them low to limit memory usage per session.
3189 In order to keep log format consistent for a same frontend, header captures
3190 can only be declared in a frontend. It is not possible to specify a capture
3191 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003192
3193 Example:
3194 capture response header Content-length len 9
3195 capture response header Location len 15
3196
Willy Tarreauc57f0e22009-05-10 13:12:33 +02003197 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01003198 about logging.
3199
3200
Cyril Bontéf0c60612010-02-06 14:44:47 +01003201clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003202 Set the maximum inactivity time on the client side.
3203 May be used in sections : defaults | frontend | listen | backend
3204 yes | yes | yes | no
3205 Arguments :
3206 <timeout> is the timeout value is specified in milliseconds by default, but
3207 can be in any other unit if the number is suffixed by the unit,
3208 as explained at the top of this document.
3209
3210 The inactivity timeout applies when the client is expected to acknowledge or
3211 send data. In HTTP mode, this timeout is particularly important to consider
3212 during the first phase, when the client sends the request, and during the
3213 response while it is reading data sent by the server. The value is specified
3214 in milliseconds by default, but can be in any other unit if the number is
3215 suffixed by the unit, as specified at the top of this document. In TCP mode
3216 (and to a lesser extent, in HTTP mode), it is highly recommended that the
3217 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003218 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01003219 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +01003220 (e.g. 4 or 5 seconds).
Willy Tarreau0ba27502007-12-24 16:55:16 +01003221
3222 This parameter is specific to frontends, but can be specified once for all in
3223 "defaults" sections. This is in fact one of the easiest solutions not to
3224 forget about it. An unspecified timeout results in an infinite timeout, which
3225 is not recommended. Such a usage is accepted and works but reports a warning
3226 during startup because it may results in accumulation of expired sessions in
3227 the system if the system's timeouts are not configured either.
3228
3229 This parameter is provided for compatibility but is currently deprecated.
3230 Please use "timeout client" instead.
3231
Willy Tarreau036fae02008-01-06 13:24:40 +01003232 See also : "timeout client", "timeout http-request", "timeout server", and
3233 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003234
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003235compression algo <algorithm> ...
3236compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02003237compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02003238 Enable HTTP compression.
3239 May be used in sections : defaults | frontend | listen | backend
3240 yes | yes | yes | yes
3241 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003242 algo is followed by the list of supported compression algorithms.
3243 type is followed by the list of MIME types that will be compressed.
3244 offload makes haproxy work as a compression offloader only (see notes).
3245
3246 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01003247 identity this is mostly for debugging, and it was useful for developing
3248 the compression feature. Identity does not apply any change on
3249 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003250
Willy Tarreauc91840a2015-03-28 17:00:39 +01003251 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003252 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003253
3254 deflate same as "gzip", but with deflate algorithm and zlib format.
3255 Note that this algorithm has ambiguous support on many
3256 browsers and no support at all from recent ones. It is
3257 strongly recommended not to use it for anything else than
3258 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003259 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003260
Willy Tarreauc91840a2015-03-28 17:00:39 +01003261 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3262 alternative when the browser wants "deflate". All major
3263 browsers understand it and despite violating the standards,
3264 it is known to work better than "deflate", at least on MSIE
3265 and some versions of Safari. Do not use it in conjunction
3266 with "deflate", use either one or the other since both react
3267 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003268 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003269
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003270 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003271 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003272 If backend servers support HTTP compression, these directives
3273 will be no-op: haproxy will see the compressed response and will not
3274 compress again. If backend servers do not support HTTP compression and
3275 there is Accept-Encoding header in request, haproxy will compress the
3276 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003277
3278 The "offload" setting makes haproxy remove the Accept-Encoding header to
3279 prevent backend servers from compressing responses. It is strongly
3280 recommended not to do this because this means that all the compression work
3281 will be done on the single point where haproxy is located. However in some
3282 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003283 with broken HTTP compression implementation which can't be turned off.
3284 In that case haproxy can be used to prevent that gateway from emitting
3285 invalid payloads. In this case, simply removing the header in the
3286 configuration does not work because it applies before the header is parsed,
3287 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003288 then be used for such scenarios. Note: for now, the "offload" setting is
3289 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003290
William Lallemand05097442012-11-20 12:14:28 +01003291 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003292 * the request does not advertise a supported compression algorithm in the
3293 "Accept-Encoding" header
3294 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003295 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003296 * response contain neither a "Content-Length" header nor a
3297 "Transfer-Encoding" whose last value is "chunked"
3298 * response contains a "Content-Type" header whose first value starts with
3299 "multipart"
3300 * the response contains the "no-transform" value in the "Cache-control"
3301 header
3302 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3303 and later
3304 * The response contains a "Content-Encoding" header, indicating that the
3305 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003306 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01003307
Tim Duesterhusb229f012019-01-29 16:38:56 +01003308 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003309
William Lallemand82fe75c2012-10-23 10:25:10 +02003310 Examples :
3311 compression algo gzip
3312 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003313
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003314
Cyril Bontéf0c60612010-02-06 14:44:47 +01003315contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003316 Set the maximum time to wait for a connection attempt to a server to succeed.
3317 May be used in sections : defaults | frontend | listen | backend
3318 yes | no | yes | yes
3319 Arguments :
3320 <timeout> is the timeout value is specified in milliseconds by default, but
3321 can be in any other unit if the number is suffixed by the unit,
3322 as explained at the top of this document.
3323
3324 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003325 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01003326 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +01003327 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003328 connect timeout also presets the queue timeout to the same value if this one
3329 has not been specified. Historically, the contimeout was also used to set the
3330 tarpit timeout in a listen section, which is not possible in a pure frontend.
3331
3332 This parameter is specific to backends, but can be specified once for all in
3333 "defaults" sections. This is in fact one of the easiest solutions not to
3334 forget about it. An unspecified timeout results in an infinite timeout, which
3335 is not recommended. Such a usage is accepted and works but reports a warning
3336 during startup because it may results in accumulation of failed sessions in
3337 the system if the system's timeouts are not configured either.
3338
3339 This parameter is provided for backwards compatibility but is currently
3340 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
3341 instead.
3342
3343 See also : "timeout connect", "timeout queue", "timeout tarpit",
3344 "timeout server", "contimeout".
3345
3346
Willy Tarreau55165fe2009-05-10 12:02:55 +02003347cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003348 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3349 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01003350 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003351 Enable cookie-based persistence in a backend.
3352 May be used in sections : defaults | frontend | listen | backend
3353 yes | no | yes | yes
3354 Arguments :
3355 <name> is the name of the cookie which will be monitored, modified or
3356 inserted in order to bring persistence. This cookie is sent to
3357 the client via a "Set-Cookie" header in the response, and is
3358 brought back by the client in a "Cookie" header in all requests.
3359 Special care should be taken to choose a name which does not
3360 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003361 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003362 HTTP/HTTPS), care should be taken to use different cookie names
3363 between all backends if persistence between them is not desired.
3364
3365 rewrite This keyword indicates that the cookie will be provided by the
3366 server and that haproxy will have to modify its value to set the
3367 server's identifier in it. This mode is handy when the management
3368 of complex combinations of "Set-cookie" and "Cache-control"
3369 headers is left to the application. The application can then
3370 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003371 cookie. Since all responses should be monitored, this mode
3372 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003373 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003374 start with this mode for new deployments. This keyword is
3375 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003376
3377 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003378 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003379
Willy Tarreaua79094d2010-08-31 22:54:15 +02003380 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003381 server. When used without the "preserve" option, if the server
Michael Prokop4438c602019-05-24 10:25:45 +02003382 emits a cookie with the same name, it will be removed before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003383 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003384 existing configurations running in the "rewrite" mode. The cookie
3385 will only be a session cookie and will not be stored on the
3386 client's disk. By default, unless the "indirect" option is added,
3387 the server will see the cookies emitted by the client. Due to
3388 caching effects, it is generally wise to add the "nocache" or
3389 "postonly" keywords (see below). The "insert" keyword is not
3390 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003391
3392 prefix This keyword indicates that instead of relying on a dedicated
3393 cookie for the persistence, an existing one will be completed.
3394 This may be needed in some specific environments where the client
3395 does not support more than one single cookie and the application
3396 already needs it. In this case, whenever the server sets a cookie
3397 named <name>, it will be prefixed with the server's identifier
3398 and a delimiter. The prefix will be removed from all client
3399 requests so that the server still finds the cookie it emitted.
3400 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003401 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003402 not compatible with "rewrite" and "insert". Note: it is highly
3403 recommended not to use "indirect" with "prefix", otherwise server
3404 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003405
Willy Tarreaua79094d2010-08-31 22:54:15 +02003406 indirect When this option is specified, no cookie will be emitted to a
3407 client which already has a valid one for the server which has
3408 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003409 it will be removed, unless the "preserve" option is also set. In
3410 "insert" mode, this will additionally remove cookies from the
3411 requests transmitted to the server, making the persistence
3412 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003413 Note: it is highly recommended not to use "indirect" with
3414 "prefix", otherwise server cookie updates would not be sent to
3415 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003416
3417 nocache This option is recommended in conjunction with the insert mode
3418 when there is a cache between the client and HAProxy, as it
3419 ensures that a cacheable response will be tagged non-cacheable if
3420 a cookie needs to be inserted. This is important because if all
3421 persistence cookies are added on a cacheable home page for
3422 instance, then all customers will then fetch the page from an
3423 outer cache and will all share the same persistence cookie,
3424 leading to one server receiving much more traffic than others.
3425 See also the "insert" and "postonly" options.
3426
3427 postonly This option ensures that cookie insertion will only be performed
3428 on responses to POST requests. It is an alternative to the
3429 "nocache" option, because POST responses are not cacheable, so
3430 this ensures that the persistence cookie will never get cached.
3431 Since most sites do not need any sort of persistence before the
3432 first POST which generally is a login request, this is a very
3433 efficient method to optimize caching without risking to find a
3434 persistence cookie in the cache.
3435 See also the "insert" and "nocache" options.
3436
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003437 preserve This option may only be used with "insert" and/or "indirect". It
3438 allows the server to emit the persistence cookie itself. In this
3439 case, if a cookie is found in the response, haproxy will leave it
3440 untouched. This is useful in order to end persistence after a
3441 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003442 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003443 the past. By combining this mechanism with the "disable-on-404"
3444 check option, it is possible to perform a completely graceful
3445 shutdown because users will definitely leave the server after
3446 they logout.
3447
Willy Tarreau4992dd22012-05-31 21:02:17 +02003448 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3449 when a cookie is inserted. This attribute is used so that a
3450 user agent doesn't share the cookie with non-HTTP components.
3451 Please check RFC6265 for more information on this attribute.
3452
3453 secure This option tells haproxy to add a "Secure" cookie attribute when
3454 a cookie is inserted. This attribute is used so that a user agent
3455 never emits this cookie over non-secure channels, which means
3456 that a cookie learned with this flag will be presented only over
3457 SSL/TLS connections. Please check RFC6265 for more information on
3458 this attribute.
3459
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003460 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003461 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003462 name. If the domain begins with a dot, the browser is allowed to
3463 use it for any host ending with that name. It is also possible to
3464 specify several domain names by invoking this option multiple
3465 times. Some browsers might have small limits on the number of
3466 domains, so be careful when doing that. For the record, sending
3467 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003468
Willy Tarreau996a92c2010-10-13 19:30:47 +02003469 maxidle This option allows inserted cookies to be ignored after some idle
3470 time. It only works with insert-mode cookies. When a cookie is
3471 sent to the client, the date this cookie was emitted is sent too.
3472 Upon further presentations of this cookie, if the date is older
3473 than the delay indicated by the parameter (in seconds), it will
3474 be ignored. Otherwise, it will be refreshed if needed when the
3475 response is sent to the client. This is particularly useful to
3476 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003477 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003478 this option is set and a cookie has no date, it is always
3479 accepted, but gets refreshed in the response. This maintains the
3480 ability for admins to access their sites. Cookies that have a
3481 date in the future further than 24 hours are ignored. Doing so
3482 lets admins fix timezone issues without risking kicking users off
3483 the site.
3484
3485 maxlife This option allows inserted cookies to be ignored after some life
3486 time, whether they're in use or not. It only works with insert
3487 mode cookies. When a cookie is first sent to the client, the date
3488 this cookie was emitted is sent too. Upon further presentations
3489 of this cookie, if the date is older than the delay indicated by
3490 the parameter (in seconds), it will be ignored. If the cookie in
3491 the request has no date, it is accepted and a date will be set.
3492 Cookies that have a date in the future further than 24 hours are
3493 ignored. Doing so lets admins fix timezone issues without risking
3494 kicking users off the site. Contrary to maxidle, this value is
3495 not refreshed, only the first visit date counts. Both maxidle and
3496 maxlife may be used at the time. This is particularly useful to
3497 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003498 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003499 is stronger than the maxidle method in that it forces a
3500 redispatch after some absolute delay.
3501
Olivier Houchard4e694042017-03-14 20:01:29 +01003502 dynamic Activate dynamic cookies. When used, a session cookie is
3503 dynamically created for each server, based on the IP and port
3504 of the server, and a secret key, specified in the
3505 "dynamic-cookie-key" backend directive.
3506 The cookie will be regenerated each time the IP address change,
3507 and is only generated for IPv4/IPv6.
3508
Willy Tarreau0ba27502007-12-24 16:55:16 +01003509 There can be only one persistence cookie per HTTP backend, and it can be
3510 declared in a defaults section. The value of the cookie will be the value
3511 indicated after the "cookie" keyword in a "server" statement. If no cookie
3512 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003513
Willy Tarreau0ba27502007-12-24 16:55:16 +01003514 Examples :
3515 cookie JSESSIONID prefix
3516 cookie SRV insert indirect nocache
3517 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003518 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003519
Willy Tarreau294d0f02015-08-10 19:40:12 +02003520 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003521
Willy Tarreau983e01e2010-01-11 18:42:06 +01003522
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003523declare capture [ request | response ] len <length>
3524 Declares a capture slot.
3525 May be used in sections : defaults | frontend | listen | backend
3526 no | yes | yes | no
3527 Arguments:
3528 <length> is the length allowed for the capture.
3529
3530 This declaration is only available in the frontend or listen section, but the
3531 reserved slot can be used in the backends. The "request" keyword allocates a
3532 capture slot for use in the request, and "response" allocates a capture slot
3533 for use in the response.
3534
3535 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003536 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003537 "http-request capture" and "http-response capture".
3538
3539
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003540default-server [param*]
3541 Change default options for a server in a backend
3542 May be used in sections : defaults | frontend | listen | backend
3543 yes | no | yes | yes
3544 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003545 <param*> is a list of parameters for this server. The "default-server"
3546 keyword accepts an important number of options and has a complete
3547 section dedicated to it. Please refer to section 5 for more
3548 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003549
Willy Tarreau983e01e2010-01-11 18:42:06 +01003550 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003551 default-server inter 1000 weight 13
3552
3553 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003554
Willy Tarreau983e01e2010-01-11 18:42:06 +01003555
Willy Tarreau0ba27502007-12-24 16:55:16 +01003556default_backend <backend>
3557 Specify the backend to use when no "use_backend" rule has been matched.
3558 May be used in sections : defaults | frontend | listen | backend
3559 yes | yes | yes | no
3560 Arguments :
3561 <backend> is the name of the backend to use.
3562
3563 When doing content-switching between frontend and backends using the
3564 "use_backend" keyword, it is often useful to indicate which backend will be
3565 used when no rule has matched. It generally is the dynamic backend which
3566 will catch all undetermined requests.
3567
Willy Tarreau0ba27502007-12-24 16:55:16 +01003568 Example :
3569
3570 use_backend dynamic if url_dyn
3571 use_backend static if url_css url_img extension_img
3572 default_backend dynamic
3573
Willy Tarreau98d04852015-05-26 12:18:29 +02003574 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003575
Willy Tarreau0ba27502007-12-24 16:55:16 +01003576
Baptiste Assmann27f51342013-10-09 06:51:49 +02003577description <string>
3578 Describe a listen, frontend or backend.
3579 May be used in sections : defaults | frontend | listen | backend
3580 no | yes | yes | yes
3581 Arguments : string
3582
3583 Allows to add a sentence to describe the related object in the HAProxy HTML
3584 stats page. The description will be printed on the right of the object name
3585 it describes.
3586 No need to backslash spaces in the <string> arguments.
3587
3588
Willy Tarreau0ba27502007-12-24 16:55:16 +01003589disabled
3590 Disable a proxy, frontend or backend.
3591 May be used in sections : defaults | frontend | listen | backend
3592 yes | yes | yes | yes
3593 Arguments : none
3594
3595 The "disabled" keyword is used to disable an instance, mainly in order to
3596 liberate a listening port or to temporarily disable a service. The instance
3597 will still be created and its configuration will be checked, but it will be
3598 created in the "stopped" state and will appear as such in the statistics. It
3599 will not receive any traffic nor will it send any health-checks or logs. It
3600 is possible to disable many instances at once by adding the "disabled"
3601 keyword in a "defaults" section.
3602
3603 See also : "enabled"
3604
3605
Willy Tarreau5ce94572010-06-07 14:35:41 +02003606dispatch <address>:<port>
3607 Set a default server address
3608 May be used in sections : defaults | frontend | listen | backend
3609 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003610 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003611
3612 <address> is the IPv4 address of the default server. Alternatively, a
3613 resolvable hostname is supported, but this name will be resolved
3614 during start-up.
3615
3616 <ports> is a mandatory port specification. All connections will be sent
3617 to this port, and it is not permitted to use port offsets as is
3618 possible with normal servers.
3619
Willy Tarreau787aed52011-04-15 06:45:37 +02003620 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003621 server can take the connection. In the past it was used to forward non
3622 persistent connections to an auxiliary load balancer. Due to its simple
3623 syntax, it has also been used for simple TCP relays. It is recommended not to
3624 use it for more clarity, and to use the "server" directive instead.
3625
3626 See also : "server"
3627
Olivier Houchard4e694042017-03-14 20:01:29 +01003628
3629dynamic-cookie-key <string>
3630 Set the dynamic cookie secret key for a backend.
3631 May be used in sections : defaults | frontend | listen | backend
3632 yes | no | yes | yes
3633 Arguments : The secret key to be used.
3634
3635 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003636 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003637 specified on the "server" line), using a hash of the IP address of the
3638 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003639 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003640 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003641
Willy Tarreau0ba27502007-12-24 16:55:16 +01003642enabled
3643 Enable a proxy, frontend or backend.
3644 May be used in sections : defaults | frontend | listen | backend
3645 yes | yes | yes | yes
3646 Arguments : none
3647
3648 The "enabled" keyword is used to explicitly enable an instance, when the
3649 defaults has been set to "disabled". This is very rarely used.
3650
3651 See also : "disabled"
3652
3653
3654errorfile <code> <file>
3655 Return a file contents instead of errors generated by HAProxy
3656 May be used in sections : defaults | frontend | listen | backend
3657 yes | yes | yes | yes
3658 Arguments :
3659 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003660 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3661 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003662
3663 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003664 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003665 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003666 error pages, and to use absolute paths, since files are read
3667 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003668
3669 It is important to understand that this keyword is not meant to rewrite
3670 errors returned by the server, but errors detected and returned by HAProxy.
3671 This is why the list of supported errors is limited to a small set.
3672
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003673 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3674
Willy Tarreau0ba27502007-12-24 16:55:16 +01003675 The files are returned verbatim on the TCP socket. This allows any trick such
3676 as redirections to another URL or site, as well as tricks to clean cookies,
3677 force enable or disable caching, etc... The package provides default error
3678 files returning the same contents as default errors.
3679
Willy Tarreau59140a22009-02-22 12:02:30 +01003680 The files should not exceed the configured buffer size (BUFSIZE), which
3681 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003682 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003683 loops between the client and HAProxy when all servers are down, causing an
3684 error to be returned instead of an image. For better HTTP compliance, it is
3685 recommended that all header lines end with CR-LF and not LF alone.
3686
Willy Tarreau0ba27502007-12-24 16:55:16 +01003687 The files are read at the same time as the configuration and kept in memory.
3688 For this reason, the errors continue to be returned even when the process is
3689 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003690 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003691 403 status code and interrogating a blocked URL.
3692
3693 See also : "errorloc", "errorloc302", "errorloc303"
3694
Willy Tarreau59140a22009-02-22 12:02:30 +01003695 Example :
3696 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003697 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003698 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3699 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3700
Willy Tarreau2769aa02007-12-27 18:26:09 +01003701
3702errorloc <code> <url>
3703errorloc302 <code> <url>
3704 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3705 May be used in sections : defaults | frontend | listen | backend
3706 yes | yes | yes | yes
3707 Arguments :
3708 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003709 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3710 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003711
3712 <url> it is the exact contents of the "Location" header. It may contain
3713 either a relative URI to an error page hosted on the same site,
3714 or an absolute URI designating an error page on another site.
3715 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003716 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003717
3718 It is important to understand that this keyword is not meant to rewrite
3719 errors returned by the server, but errors detected and returned by HAProxy.
3720 This is why the list of supported errors is limited to a small set.
3721
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003722 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3723
Willy Tarreau2769aa02007-12-27 18:26:09 +01003724 Note that both keyword return the HTTP 302 status code, which tells the
3725 client to fetch the designated URL using the same HTTP method. This can be
3726 quite problematic in case of non-GET methods such as POST, because the URL
3727 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003728 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003729 status code, indicating to the client that the URL must be fetched with a GET
3730 request.
3731
3732 See also : "errorfile", "errorloc303"
3733
3734
3735errorloc303 <code> <url>
3736 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3737 May be used in sections : defaults | frontend | listen | backend
3738 yes | yes | yes | yes
3739 Arguments :
3740 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003741 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3742 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003743
3744 <url> it is the exact contents of the "Location" header. It may contain
3745 either a relative URI to an error page hosted on the same site,
3746 or an absolute URI designating an error page on another site.
3747 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003748 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003749
3750 It is important to understand that this keyword is not meant to rewrite
3751 errors returned by the server, but errors detected and returned by HAProxy.
3752 This is why the list of supported errors is limited to a small set.
3753
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003754 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3755
Willy Tarreau2769aa02007-12-27 18:26:09 +01003756 Note that both keyword return the HTTP 303 status code, which tells the
3757 client to fetch the designated URL using the same HTTP GET method. This
3758 solves the usual problems associated with "errorloc" and the 302 code. It is
3759 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003760 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003761
3762 See also : "errorfile", "errorloc", "errorloc302"
3763
3764
Simon Horman51a1cf62015-02-03 13:00:44 +09003765email-alert from <emailaddr>
3766 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003767 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003768 May be used in sections: defaults | frontend | listen | backend
3769 yes | yes | yes | yes
3770
3771 Arguments :
3772
3773 <emailaddr> is the from email address to use when sending email alerts
3774
3775 Also requires "email-alert mailers" and "email-alert to" to be set
3776 and if so sending email alerts is enabled for the proxy.
3777
Simon Horman64e34162015-02-06 11:11:57 +09003778 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003779 "email-alert myhostname", "email-alert to", section 3.6 about
3780 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003781
3782
3783email-alert level <level>
3784 Declare the maximum log level of messages for which email alerts will be
3785 sent. This acts as a filter on the sending of email alerts.
3786 May be used in sections: defaults | frontend | listen | backend
3787 yes | yes | yes | yes
3788
3789 Arguments :
3790
3791 <level> One of the 8 syslog levels:
3792 emerg alert crit err warning notice info debug
3793 The above syslog levels are ordered from lowest to highest.
3794
3795 By default level is alert
3796
3797 Also requires "email-alert from", "email-alert mailers" and
3798 "email-alert to" to be set and if so sending email alerts is enabled
3799 for the proxy.
3800
Simon Horman1421e212015-04-30 13:10:35 +09003801 Alerts are sent when :
3802
3803 * An un-paused server is marked as down and <level> is alert or lower
3804 * A paused server is marked as down and <level> is notice or lower
3805 * A server is marked as up or enters the drain state and <level>
3806 is notice or lower
3807 * "option log-health-checks" is enabled, <level> is info or lower,
3808 and a health check status update occurs
3809
Simon Horman64e34162015-02-06 11:11:57 +09003810 See also : "email-alert from", "email-alert mailers",
3811 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003812 section 3.6 about mailers.
3813
3814
3815email-alert mailers <mailersect>
3816 Declare the mailers to be used when sending email alerts
3817 May be used in sections: defaults | frontend | listen | backend
3818 yes | yes | yes | yes
3819
3820 Arguments :
3821
3822 <mailersect> is the name of the mailers section to send email alerts.
3823
3824 Also requires "email-alert from" and "email-alert to" to be set
3825 and if so sending email alerts is enabled for the proxy.
3826
Simon Horman64e34162015-02-06 11:11:57 +09003827 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3828 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003829
3830
3831email-alert myhostname <hostname>
3832 Declare the to hostname address to be used when communicating with
3833 mailers.
3834 May be used in sections: defaults | frontend | listen | backend
3835 yes | yes | yes | yes
3836
3837 Arguments :
3838
Baptiste Assmann738bad92015-12-21 15:27:53 +01003839 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003840
3841 By default the systems hostname is used.
3842
3843 Also requires "email-alert from", "email-alert mailers" and
3844 "email-alert to" to be set and if so sending email alerts is enabled
3845 for the proxy.
3846
Simon Horman64e34162015-02-06 11:11:57 +09003847 See also : "email-alert from", "email-alert level", "email-alert mailers",
3848 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003849
3850
3851email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003852 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003853 header of email alerts. This is the address that email alerts are sent to.
3854 May be used in sections: defaults | frontend | listen | backend
3855 yes | yes | yes | yes
3856
3857 Arguments :
3858
3859 <emailaddr> is the to email address to use when sending email alerts
3860
3861 Also requires "email-alert mailers" and "email-alert to" to be set
3862 and if so sending email alerts is enabled for the proxy.
3863
Simon Horman64e34162015-02-06 11:11:57 +09003864 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003865 "email-alert myhostname", section 3.6 about mailers.
3866
3867
Willy Tarreau4de91492010-01-22 19:10:05 +01003868force-persist { if | unless } <condition>
3869 Declare a condition to force persistence on down servers
3870 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003871 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003872
3873 By default, requests are not dispatched to down servers. It is possible to
3874 force this using "option persist", but it is unconditional and redispatches
3875 to a valid server if "option redispatch" is set. That leaves with very little
3876 possibilities to force some requests to reach a server which is artificially
3877 marked down for maintenance operations.
3878
3879 The "force-persist" statement allows one to declare various ACL-based
3880 conditions which, when met, will cause a request to ignore the down status of
3881 a server and still try to connect to it. That makes it possible to start a
3882 server, still replying an error to the health checks, and run a specially
3883 configured browser to test the service. Among the handy methods, one could
3884 use a specific source IP address, or a specific cookie. The cookie also has
3885 the advantage that it can easily be added/removed on the browser from a test
3886 page. Once the service is validated, it is then possible to open the service
3887 to the world by returning a valid response to health checks.
3888
3889 The forced persistence is enabled when an "if" condition is met, or unless an
3890 "unless" condition is met. The final redispatch is always disabled when this
3891 is used.
3892
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003893 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003894 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003895
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003896
3897filter <name> [param*]
3898 Add the filter <name> in the filter list attached to the proxy.
3899 May be used in sections : defaults | frontend | listen | backend
3900 no | yes | yes | yes
3901 Arguments :
3902 <name> is the name of the filter. Officially supported filters are
3903 referenced in section 9.
3904
Tim Düsterhus4896c442016-11-29 02:15:19 +01003905 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003906 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003907 filter. Please refer to the documentation of the corresponding
3908 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003909
3910 Multiple occurrences of the filter line can be used for the same proxy. The
3911 same filter can be referenced many times if needed.
3912
3913 Example:
3914 listen
3915 bind *:80
3916
3917 filter trace name BEFORE-HTTP-COMP
3918 filter compression
3919 filter trace name AFTER-HTTP-COMP
3920
3921 compression algo gzip
3922 compression offload
3923
3924 server srv1 192.168.0.1:80
3925
3926 See also : section 9.
3927
Willy Tarreau4de91492010-01-22 19:10:05 +01003928
Willy Tarreau2769aa02007-12-27 18:26:09 +01003929fullconn <conns>
3930 Specify at what backend load the servers will reach their maxconn
3931 May be used in sections : defaults | frontend | listen | backend
3932 yes | no | yes | yes
3933 Arguments :
3934 <conns> is the number of connections on the backend which will make the
3935 servers use the maximal number of connections.
3936
Willy Tarreau198a7442008-01-17 12:05:32 +01003937 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003938 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003939 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003940 load. The server will then always accept at least <minconn> connections,
3941 never more than <maxconn>, and the limit will be on the ramp between both
3942 values when the backend has less than <conns> concurrent connections. This
3943 makes it possible to limit the load on the servers during normal loads, but
3944 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003945 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003946
Willy Tarreaufbb78422011-06-05 15:38:35 +02003947 Since it's hard to get this value right, haproxy automatically sets it to
3948 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003949 backend (based on "use_backend" and "default_backend" rules). That way it's
3950 safe to leave it unset. However, "use_backend" involving dynamic names are
3951 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003952
Willy Tarreau2769aa02007-12-27 18:26:09 +01003953 Example :
3954 # The servers will accept between 100 and 1000 concurrent connections each
3955 # and the maximum of 1000 will be reached when the backend reaches 10000
3956 # connections.
3957 backend dynamic
3958 fullconn 10000
3959 server srv1 dyn1:80 minconn 100 maxconn 1000
3960 server srv2 dyn2:80 minconn 100 maxconn 1000
3961
3962 See also : "maxconn", "server"
3963
3964
3965grace <time>
3966 Maintain a proxy operational for some time after a soft stop
3967 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003968 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003969 Arguments :
3970 <time> is the time (by default in milliseconds) for which the instance
3971 will remain operational with the frontend sockets still listening
3972 when a soft-stop is received via the SIGUSR1 signal.
3973
3974 This may be used to ensure that the services disappear in a certain order.
3975 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003976 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003977 needed by the equipment to detect the failure.
3978
3979 Note that currently, there is very little benefit in using this parameter,
3980 and it may in fact complicate the soft-reconfiguration process more than
3981 simplify it.
3982
Willy Tarreau0ba27502007-12-24 16:55:16 +01003983
Andrew Rodland17be45e2016-10-25 17:04:12 -04003984hash-balance-factor <factor>
3985 Specify the balancing factor for bounded-load consistent hashing
3986 May be used in sections : defaults | frontend | listen | backend
3987 yes | no | no | yes
3988 Arguments :
3989 <factor> is the control for the maximum number of concurrent requests to
3990 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01003991 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04003992
3993 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3994 enables an algorithm that prevents any one server from getting too many
3995 requests at once, even if some hash buckets receive many more requests than
3996 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3997 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3998 then no server will be allowed to have a load more than 1.5 times the average.
3999 If server weights are used, they will be respected.
4000
4001 If the first-choice server is disqualified, the algorithm will choose another
4002 server based on the request hash, until a server with additional capacity is
4003 found. A higher <factor> allows more imbalance between the servers, while a
4004 lower <factor> means that more servers will be checked on average, affecting
4005 performance. Reasonable values are from 125 to 200.
4006
Willy Tarreau760e81d2018-05-03 07:20:40 +02004007 This setting is also used by "balance random" which internally relies on the
4008 consistent hashing mechanism.
4009
Andrew Rodland17be45e2016-10-25 17:04:12 -04004010 See also : "balance" and "hash-type".
4011
4012
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004013hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004014 Specify a method to use for mapping hashes to servers
4015 May be used in sections : defaults | frontend | listen | backend
4016 yes | no | yes | yes
4017 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04004018 <method> is the method used to select a server from the hash computed by
4019 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004020
Bhaskar98634f02013-10-29 23:30:51 -04004021 map-based the hash table is a static array containing all alive servers.
4022 The hashes will be very smooth, will consider weights, but
4023 will be static in that weight changes while a server is up
4024 will be ignored. This means that there will be no slow start.
4025 Also, since a server is selected by its position in the array,
4026 most mappings are changed when the server count changes. This
4027 means that when a server goes up or down, or when a server is
4028 added to a farm, most connections will be redistributed to
4029 different servers. This can be inconvenient with caches for
4030 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01004031
Bhaskar98634f02013-10-29 23:30:51 -04004032 consistent the hash table is a tree filled with many occurrences of each
4033 server. The hash key is looked up in the tree and the closest
4034 server is chosen. This hash is dynamic, it supports changing
4035 weights while the servers are up, so it is compatible with the
4036 slow start feature. It has the advantage that when a server
4037 goes up or down, only its associations are moved. When a
4038 server is added to the farm, only a few part of the mappings
4039 are redistributed, making it an ideal method for caches.
4040 However, due to its principle, the distribution will never be
4041 very smooth and it may sometimes be necessary to adjust a
4042 server's weight or its ID to get a more balanced distribution.
4043 In order to get the same distribution on multiple load
4044 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004045 same IDs. Note: consistent hash uses sdbm and avalanche if no
4046 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04004047
4048 <function> is the hash function to be used :
4049
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03004050 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04004051 reimplementation of ndbm) database library. It was found to do
4052 well in scrambling bits, causing better distribution of the keys
4053 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004054 function with good distribution, unless the total server weight
4055 is a multiple of 64, in which case applying the avalanche
4056 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04004057
4058 djb2 this function was first proposed by Dan Bernstein many years ago
4059 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004060 function provides a better distribution than sdbm. It generally
4061 works well with text-based inputs though it can perform extremely
4062 poorly with numeric-only input or when the total server weight is
4063 a multiple of 33, unless the avalanche modifier is also used.
4064
Willy Tarreaua0f42712013-11-14 14:30:35 +01004065 wt6 this function was designed for haproxy while testing other
4066 functions in the past. It is not as smooth as the other ones, but
4067 is much less sensible to the input data set or to the number of
4068 servers. It can make sense as an alternative to sdbm+avalanche or
4069 djb2+avalanche for consistent hashing or when hashing on numeric
4070 data such as a source IP address or a visitor identifier in a URL
4071 parameter.
4072
Willy Tarreau324f07f2015-01-20 19:44:50 +01004073 crc32 this is the most common CRC32 implementation as used in Ethernet,
4074 gzip, PNG, etc. It is slower than the other ones but may provide
4075 a better distribution or less predictable results especially when
4076 used on strings.
4077
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05004078 <modifier> indicates an optional method applied after hashing the key :
4079
4080 avalanche This directive indicates that the result from the hash
4081 function above should not be used in its raw form but that
4082 a 4-byte full avalanche hash must be applied first. The
4083 purpose of this step is to mix the resulting bits from the
4084 previous hash in order to avoid any undesired effect when
4085 the input contains some limited values or when the number of
4086 servers is a multiple of one of the hash's components (64
4087 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
4088 result less predictable, but it's also not as smooth as when
4089 using the original function. Some testing might be needed
4090 with some workloads. This hash is one of the many proposed
4091 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004092
Bhaskar98634f02013-10-29 23:30:51 -04004093 The default hash type is "map-based" and is recommended for most usages. The
4094 default function is "sdbm", the selection of a function should be based on
4095 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004096
Andrew Rodland17be45e2016-10-25 17:04:12 -04004097 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02004098
4099
Willy Tarreau0ba27502007-12-24 16:55:16 +01004100http-check disable-on-404
4101 Enable a maintenance mode upon HTTP/404 response to health-checks
4102 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01004103 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01004104 Arguments : none
4105
4106 When this option is set, a server which returns an HTTP code 404 will be
4107 excluded from further load-balancing, but will still receive persistent
4108 connections. This provides a very convenient method for Web administrators
4109 to perform a graceful shutdown of their servers. It is also important to note
4110 that a server which is detected as failed while it was in this mode will not
4111 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
4112 will immediately be reinserted into the farm. The status on the stats page
4113 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01004114 option only works in conjunction with the "httpchk" option. If this option
4115 is used with "http-check expect", then it has precedence over it so that 404
4116 responses will still be considered as soft-stop.
4117
4118 See also : "option httpchk", "http-check expect"
4119
4120
4121http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004122 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01004123 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02004124 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01004125 Arguments :
4126 <match> is a keyword indicating how to look for a specific pattern in the
4127 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004128 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01004129 exclamation mark ("!") to negate the match. Spaces are allowed
4130 between the exclamation mark and the keyword. See below for more
4131 details on the supported keywords.
4132
4133 <pattern> is the pattern to look for. It may be a string or a regular
4134 expression. If the pattern contains spaces, they must be escaped
4135 with the usual backslash ('\').
4136
4137 By default, "option httpchk" considers that response statuses 2xx and 3xx
4138 are valid, and that others are invalid. When "http-check expect" is used,
4139 it defines what is considered valid or invalid. Only one "http-check"
4140 statement is supported in a backend. If a server fails to respond or times
4141 out, the check obviously fails. The available matches are :
4142
4143 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004144 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004145 response's status code is exactly this string. If the
4146 "status" keyword is prefixed with "!", then the response
4147 will be considered invalid if the status code matches.
4148
4149 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004150 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004151 response's status code matches the expression. If the
4152 "rstatus" keyword is prefixed with "!", then the response
4153 will be considered invalid if the status code matches.
4154 This is mostly used to check for multiple codes.
4155
4156 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004157 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004158 response's body contains this exact string. If the
4159 "string" keyword is prefixed with "!", then the response
4160 will be considered invalid if the body contains this
4161 string. This can be used to look for a mandatory word at
4162 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004163 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01004164 trace).
4165
4166 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04004167 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01004168 response's body matches this expression. If the "rstring"
4169 keyword is prefixed with "!", then the response will be
4170 considered invalid if the body matches the expression.
4171 This can be used to look for a mandatory word at the end
4172 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01004173 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01004174
4175 It is important to note that the responses will be limited to a certain size
4176 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
4177 Thus, too large responses may not contain the mandatory pattern when using
4178 "string" or "rstring". If a large response is absolutely required, it is
4179 possible to change the default max size by setting the global variable.
4180 However, it is worth keeping in mind that parsing very large responses can
4181 waste some CPU cycles, especially when regular expressions are used, and that
4182 it is always better to focus the checks on smaller resources.
4183
Cyril Bonté32602d22015-01-30 00:07:07 +01004184 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
4185 will automatically append a "Connection: close" header, meaning that this
4186 header should not be present in the request provided by "option httpchk".
4187
Willy Tarreaubd741542010-03-16 18:46:54 +01004188 Last, if "http-check expect" is combined with "http-check disable-on-404",
4189 then this last one has precedence when the server responds with 404.
4190
4191 Examples :
4192 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004193 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01004194
4195 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004196 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01004197
4198 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01004199 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01004200
4201 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03004202 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01004203
Willy Tarreaubd741542010-03-16 18:46:54 +01004204 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01004205
4206
Willy Tarreauef781042010-01-27 11:53:01 +01004207http-check send-state
4208 Enable emission of a state header with HTTP health checks
4209 May be used in sections : defaults | frontend | listen | backend
4210 yes | no | yes | yes
4211 Arguments : none
4212
4213 When this option is set, haproxy will systematically send a special header
4214 "X-Haproxy-Server-State" with a list of parameters indicating to each server
4215 how they are seen by haproxy. This can be used for instance when a server is
4216 manipulated without access to haproxy and the operator needs to know whether
4217 haproxy still sees it up or not, or if the server is the last one in a farm.
4218
4219 The header is composed of fields delimited by semi-colons, the first of which
4220 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
4221 checks on the total number before transition, just as appears in the stats
4222 interface. Next headers are in the form "<variable>=<value>", indicating in
4223 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08004224 - a variable "address", containing the address of the backend server.
4225 This corresponds to the <address> field in the server declaration. For
4226 unix domain sockets, it will read "unix".
4227
4228 - a variable "port", containing the port of the backend server. This
4229 corresponds to the <port> field in the server declaration. For unix
4230 domain sockets, it will read "unix".
4231
Willy Tarreauef781042010-01-27 11:53:01 +01004232 - a variable "name", containing the name of the backend followed by a slash
4233 ("/") then the name of the server. This can be used when a server is
4234 checked in multiple backends.
4235
4236 - a variable "node" containing the name of the haproxy node, as set in the
4237 global "node" variable, otherwise the system's hostname if unspecified.
4238
4239 - a variable "weight" indicating the weight of the server, a slash ("/")
4240 and the total weight of the farm (just counting usable servers). This
4241 helps to know if other servers are available to handle the load when this
4242 one fails.
4243
4244 - a variable "scur" indicating the current number of concurrent connections
4245 on the server, followed by a slash ("/") then the total number of
4246 connections on all servers of the same backend.
4247
4248 - a variable "qcur" indicating the current number of requests in the
4249 server's queue.
4250
4251 Example of a header received by the application server :
4252 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
4253 scur=13/22; qcur=0
4254
4255 See also : "option httpchk", "http-check disable-on-404"
4256
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004257
4258http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004259 Access control for Layer 7 requests
4260
4261 May be used in sections: defaults | frontend | listen | backend
4262 no | yes | yes | yes
4263
Willy Tarreau20b0de52012-12-24 15:45:22 +01004264 The http-request statement defines a set of rules which apply to layer 7
4265 processing. The rules are evaluated in their declaration order when they are
4266 met in a frontend, listen or backend section. Any rule may optionally be
4267 followed by an ACL-based condition, in which case it will only be evaluated
4268 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004269
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004270 The first keyword is the rule's action. The supported actions are described
4271 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004272
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004273 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004274
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004275 It is important to know that http-request rules are processed very early in
4276 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
4277 or "reqadd" rules. That way, headers added by "add-header"/"set-header" are
4278 visible by almost all further ACL rules.
Willy Tarreau53275e82017-11-24 07:52:01 +01004279
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004280 Using "reqadd"/"reqdel"/"reqrep" to manipulate request headers is discouraged
4281 in newer versions (>= 1.5). But if you need to use regular expression to
4282 delete headers, you can still use "reqdel". Also please use
4283 "http-request deny/allow/tarpit" instead of "reqdeny"/"reqpass"/"reqtarpit".
Willy Tarreauccbcc372012-12-27 12:37:57 +01004284
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004285 Example:
4286 acl nagios src 192.168.129.3
4287 acl local_net src 192.168.0.0/16
4288 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004289
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004290 http-request allow if nagios
4291 http-request allow if local_net auth_ok
4292 http-request auth realm Gimme if local_net auth_ok
4293 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004294
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004295 Example:
4296 acl key req.hdr(X-Add-Acl-Key) -m found
4297 acl add path /addacl
4298 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004299
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004300 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004301
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004302 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4303 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004304
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004305 Example:
4306 acl value req.hdr(X-Value) -m found
4307 acl setmap path /setmap
4308 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004309
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004310 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004311
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004312 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4313 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004314
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004315 See also : "stats http-request", section 3.4 about userlists and section 7
4316 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004317
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004318http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004319
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004320 This is used to add a new entry into an ACL. The ACL must be loaded from a
4321 file (even a dummy empty file). The file name of the ACL to be updated is
4322 passed between parentheses. It takes one argument: <key fmt>, which follows
4323 log-format rules, to collect content of the new entry. It performs a lookup
4324 in the ACL before insertion, to avoid duplicated (or more) values. This
4325 lookup is done by a linear search and can be expensive with large lists!
4326 It is the equivalent of the "add acl" command from the stats socket, but can
4327 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004328
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004329http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004330
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004331 This appends an HTTP header field whose name is specified in <name> and
4332 whose value is defined by <fmt> which follows the log-format rules (see
4333 Custom Log Format in section 8.2.4). This is particularly useful to pass
4334 connection-specific information to the server (e.g. the client's SSL
4335 certificate), or to combine several headers into one. This rule is not
4336 final, so it is possible to add other similar rules. Note that header
4337 addition is performed immediately, so one rule might reuse the resulting
4338 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004339
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004340http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004341
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004342 This stops the evaluation of the rules and lets the request pass the check.
4343 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004344
Sasha Pachev218f0642014-06-16 12:05:59 -06004345
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004346http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004347
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004348 This stops the evaluation of the rules and immediately responds with an
4349 HTTP 401 or 407 error code to invite the user to present a valid user name
4350 and password. No further "http-request" rules are evaluated. An optional
4351 "realm" parameter is supported, it sets the authentication realm that is
4352 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004353
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004354 Example:
4355 acl auth_ok http_auth_group(L1) G1
4356 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004357
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004358http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004359
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004360 See section 10.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004361
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004362http-request capture <sample> [ len <length> | id <id> ]
4363 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004364
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004365 This captures sample expression <sample> from the request buffer, and
4366 converts it to a string of at most <len> characters. The resulting string is
4367 stored into the next request "capture" slot, so it will possibly appear next
4368 to some captured HTTP headers. It will then automatically appear in the logs,
4369 and it will be possible to extract it using sample fetch rules to feed it
4370 into headers or anything. The length should be limited given that this size
4371 will be allocated for each capture during the whole session life.
4372 Please check section 7.3 (Fetching samples) and "capture request header" for
4373 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004374
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004375 If the keyword "id" is used instead of "len", the action tries to store the
4376 captured string in a previously declared capture slot. This is useful to run
4377 captures in backends. The slot id can be declared by a previous directive
4378 "http-request capture" or with the "declare capture" keyword. If the slot
4379 <id> doesn't exist, then HAProxy fails parsing the configuration to prevent
4380 unexpected behavior at run time.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004381
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004382http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004383
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004384 This is used to delete an entry from an ACL. The ACL must be loaded from a
4385 file (even a dummy empty file). The file name of the ACL to be updated is
4386 passed between parentheses. It takes one argument: <key fmt>, which follows
4387 log-format rules, to collect content of the entry to delete.
4388 It is the equivalent of the "del acl" command from the stats socket, but can
4389 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004390
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004391http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004392
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004393 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004394
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004395http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004396
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004397 This is used to delete an entry from a MAP. The MAP must be loaded from a
4398 file (even a dummy empty file). The file name of the MAP to be updated is
4399 passed between parentheses. It takes one argument: <key fmt>, which follows
4400 log-format rules, to collect content of the entry to delete.
4401 It takes one argument: "file name" It is the equivalent of the "del map"
4402 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004403
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004404http-request deny [deny_status <status>] [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004405
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004406 This stops the evaluation of the rules and immediately rejects the request
4407 and emits an HTTP 403 error, or optionally the status code specified as an
4408 argument to "deny_status". The list of permitted status codes is limited to
4409 those that can be overridden by the "errorfile" directive.
4410 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004411
Olivier Houchard602bf7d2019-05-10 13:59:15 +02004412http-request disable-l7-retry [ { if | unless } <condition> ]
4413 This disables any attempt to retry the request if it fails for any other
4414 reason than a connection failure. This can be useful for example to make
4415 sure POST requests aren't retried on failure.
4416
Baptiste Assmann333939c2019-01-21 08:34:50 +01004417http-request do-resolve(<var>,<resolvers>,[ipv4,ipv6]) <expr> :
4418
4419 This action performs a DNS resolution of the output of <expr> and stores
4420 the result in the variable <var>. It uses the DNS resolvers section
4421 pointed by <resolvers>.
4422 It is possible to choose a resolution preference using the optional
4423 arguments 'ipv4' or 'ipv6'.
4424 When performing the DNS resolution, the client side connection is on
4425 pause waiting till the end of the resolution.
4426 If an IP address can be found, it is stored into <var>. If any kind of
4427 error occurs, then <var> is not set.
4428 One can use this action to discover a server IP address at run time and
4429 based on information found in the request (IE a Host header).
4430 If this action is used to find the server's IP address (using the
4431 "set-dst" action), then the server IP address in the backend must be set
4432 to 0.0.0.0.
4433
4434 Example:
4435 resolvers mydns
4436 nameserver local 127.0.0.53:53
4437 nameserver google 8.8.8.8:53
4438 timeout retry 1s
4439 hold valid 10s
4440 hold nx 3s
4441 hold other 3s
4442 hold obsolete 0s
4443 accepted_payload_size 8192
4444
4445 frontend fe
4446 bind 10.42.0.1:80
4447 http-request do-resolve(txn.myip,mydns,ipv4) hdr(Host),lower
4448 http-request capture var(txn.myip) len 40
4449
4450 # return 503 when the variable is not set,
4451 # which mean DNS resolution error
4452 use_backend b_503 unless { var(txn.myip) -m found }
4453
4454 default_backend be
4455
4456 backend b_503
4457 # dummy backend used to return 503.
4458 # one can use the errorfile directive to send a nice
4459 # 503 error page to end users
4460
4461 backend be
4462 # rule to prevent HAProxy from reconnecting to services
4463 # on the local network (forged DNS name used to scan the network)
4464 http-request deny if { var(txn.myip) -m ip 127.0.0.0/8 10.0.0.0/8 }
4465 http-request set-dst var(txn.myip)
4466 server clear 0.0.0.0:0
4467
4468 NOTE: Don't forget to set the "protection" rules to ensure HAProxy won't
4469 be used to scan the network or worst won't loop over itself...
4470
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004471http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4472
4473 This is used to build an HTTP 103 Early Hints response prior to any other one.
4474 This appends an HTTP header field to this response whose name is specified in
4475 <name> and whose value is defined by <fmt> which follows the log-format rules
4476 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004477 to the client some Link headers to preload resources required to render the
4478 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004479
4480 See RFC 8297 for more information.
4481
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004482http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004483
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004484 This performs an HTTP redirection based on a redirect rule. This is exactly
4485 the same as the "redirect" statement except that it inserts a redirect rule
4486 which can be processed in the middle of other "http-request" rules and that
4487 these rules use the "log-format" strings. See the "redirect" keyword for the
4488 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004489
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004490http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004491
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004492 This stops the evaluation of the rules and immediately closes the connection
4493 without sending any response. It acts similarly to the
4494 "tcp-request content reject" rules. It can be useful to force an immediate
4495 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004496
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004497http-request replace-header <name> <match-regex> <replace-fmt>
4498 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004499
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004500 This matches the regular expression in all occurrences of header field
4501 <name> according to <match-regex>, and replaces them with the
4502 <replace-fmt> argument. Format characters are allowed in replace-fmt and
4503 work like in <fmt> arguments in "http-request add-header". The match is
4504 only case-sensitive. It is important to understand that this action only
4505 considers whole header lines, regardless of the number of values they may
4506 contain. This usage is suited to headers naturally containing commas in
4507 their value, such as If-Modified-Since and so on.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004508
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004509 Example:
4510 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
William Lallemand86d0df02017-11-24 21:36:45 +01004511
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004512 # applied to:
4513 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004514
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004515 # outputs:
4516 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004517
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004518 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004519
Willy Tarreau33810222019-06-12 17:44:02 +02004520http-request replace-uri <match-regex> <replace-fmt>
4521 [ { if | unless } <condition> ]
4522
4523 This matches the regular expression in the URI part of the request
4524 according to <match-regex>, and replaces it with the <replace-fmt>
4525 argument. Standard back-references using the backslash ('\') followed by a
4526 number are supported. The <fmt> field is interpreted as a log-format string
4527 so it may contain special expressions just like the <fmt> argument passed
4528 to "http-request set-uri". The match is exclusively case-sensitive. Any
4529 optional scheme, authority or query string are considered in the matching
4530 part of the URI. It is worth noting that regular expressions may be more
4531 expensive to evaluate than certain ACLs, so rare replacements may benefit
4532 from a condition to avoid performing the evaluation at all if it does not
4533 match.
4534
4535 Example:
4536 # prefix /foo : turn /bar?q=1 into /foo/bar?q=1 :
4537 http-request replace-uri (.*) /foo\1
4538
4539 # suffix /foo : turn /bar?q=1 into /bar/foo?q=1 :
4540 http-request replace-uri ([^?]*)(\?(.*))? \1/foo\2
4541
4542 # strip /foo : turn /foo/bar?q=1 into /bar?q=1
4543 http-request replace-uri /foo/(.*) /\1
4544 # or more efficient if only some requests match :
4545 http-request replace-uri /foo/(.*) /\1 if { url_beg /foo/ }
4546
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004547http-request replace-value <name> <match-regex> <replace-fmt>
4548 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004549
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004550 This works like "replace-header" except that it matches the regex against
4551 every comma-delimited value of the header field <name> instead of the
4552 entire header. This is suited for all headers which are allowed to carry
4553 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004554
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004555 Example:
4556 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004557
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004558 # applied to:
4559 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004560
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004561 # outputs:
4562 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004563
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004564http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4565http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004566
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004567 This actions increments the GPC0 or GPC1 counter according with the sticky
4568 counter designated by <sc-id>. If an error occurs, this action silently fails
4569 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004570
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004571http-request sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004572
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004573 This action sets the GPT0 tag according to the sticky counter designated by
4574 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4575 occurs, this action silently fails and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004576
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004577http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004578
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004579 This is used to set the destination IP address to the value of specified
4580 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4581 but provides the correct IP in a HTTP header; or you want to mask the IP for
4582 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4583 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004584
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004585 Arguments:
4586 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4587 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004588
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004589 Example:
4590 http-request set-dst hdr(x-dst)
4591 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004592
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004593 When possible, set-dst preserves the original destination port as long as the
4594 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004595
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004596http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004597
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004598 This is used to set the destination port address to the value of specified
4599 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4600 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004601
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004602 Arguments:
4603 <expr> Is a standard HAProxy expression formed by a sample-fetch
4604 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004605
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004606 Example:
4607 http-request set-dst-port hdr(x-port)
4608 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004609
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004610 When possible, set-dst-port preserves the original destination address as
4611 long as the address family supports a port, otherwise it forces the
4612 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004613
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004614http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004615
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004616 This does the same as "http-request add-header" except that the header name
4617 is first removed if it existed. This is useful when passing security
4618 information to the server, where the header must not be manipulated by
4619 external users. Note that the new value is computed before the removal so it
4620 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004621
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004622 Example:
4623 http-request set-header X-Haproxy-Current-Date %T
4624 http-request set-header X-SSL %[ssl_fc]
4625 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4626 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4627 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4628 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4629 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4630 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4631 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02004632
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004633http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004634
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004635 This is used to change the log level of the current request when a certain
4636 condition is met. Valid levels are the 8 syslog levels (see the "log"
4637 keyword) plus the special level "silent" which disables logging for this
4638 request. This rule is not final so the last matching rule wins. This rule
4639 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004640
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004641http-request set-map(<file-name>) <key fmt> <value fmt>
4642 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004643
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004644 This is used to add a new entry into a MAP. The MAP must be loaded from a
4645 file (even a dummy empty file). The file name of the MAP to be updated is
4646 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4647 log-format rules, used to collect MAP key, and <value fmt>, which follows
4648 log-format rules, used to collect content for the new entry.
4649 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4650 more) values. This lookup is done by a linear search and can be expensive
4651 with large lists! It is the equivalent of the "set map" command from the
4652 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004653
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004654http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004655
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004656 This is used to set the Netfilter MARK on all packets sent to the client to
4657 the value passed in <mark> on platforms which support it. This value is an
4658 unsigned 32 bit value which can be matched by netfilter and by the routing
4659 table. It can be expressed both in decimal or hexadecimal format (prefixed by
4660 "0x"). This can be useful to force certain packets to take a different route
4661 (for example a cheaper network path for bulk downloads). This works on Linux
4662 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02004663
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004664http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004665
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004666 This rewrites the request method with the result of the evaluation of format
4667 string <fmt>. There should be very few valid reasons for having to do so as
4668 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004669
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004670http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004671
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004672 This sets the "nice" factor of the current request being processed. It only
4673 has effect against the other requests being processed at the same time.
4674 The default value is 0, unless altered by the "nice" setting on the "bind"
4675 line. The accepted range is -1024..1024. The higher the value, the nicest
4676 the request will be. Lower values will make the request more important than
4677 other ones. This can be useful to improve the speed of some requests, or
4678 lower the priority of non-important requests. Using this setting without
4679 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004680
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004681http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02004682
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004683 This rewrites the request path with the result of the evaluation of format
4684 string <fmt>. The query string, if any, is left intact. If a scheme and
4685 authority is found before the path, they are left intact as well. If the
4686 request doesn't have a path ("*"), this one is replaced with the format.
4687 This can be used to prepend a directory component in front of a path for
4688 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02004689
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004690 Example :
4691 # prepend the host name before the path
4692 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004693
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004694http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004695
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004696 This is used to set the queue priority class of the current request.
4697 The value must be a sample expression which converts to an integer in the
4698 range -2047..2047. Results outside this range will be truncated.
4699 The priority class determines the order in which queued requests are
4700 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004701
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004702http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004703
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004704 This is used to set the queue priority timestamp offset of the current
4705 request. The value must be a sample expression which converts to an integer
4706 in the range -524287..524287. Results outside this range will be truncated.
4707 When a request is queued, it is ordered first by the priority class, then by
4708 the current timestamp adjusted by the given offset in milliseconds. Lower
4709 values have higher priority.
4710 Note that the resulting timestamp is is only tracked with enough precision
4711 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
4712 adjusted timestamp exceeds this value, it will be misidentified as highest
4713 priority. Thus it is important to set "timeout queue" to a value, where when
4714 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004715
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004716http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004717
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004718 This rewrites the request's query string which appears after the first
4719 question mark ("?") with the result of the evaluation of format string <fmt>.
4720 The part prior to the question mark is left intact. If the request doesn't
4721 contain a question mark and the new value is not empty, then one is added at
4722 the end of the URI, followed by the new value. If a question mark was
4723 present, it will never be removed even if the value is empty. This can be
4724 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004725
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004726 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004727
4728 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004729 # replace "%3D" with "=" in the query string
4730 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004731
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004732http-request set-src <expr> [ { if | unless } <condition> ]
4733 This is used to set the source IP address to the value of specified
4734 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
4735 provides the correct IP in a HTTP header; or you want to mask source IP for
4736 privacy.
4737
4738 Arguments :
4739 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4740 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004741
Cyril Bonté78caf842010-03-10 22:41:43 +01004742 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004743 http-request set-src hdr(x-forwarded-for)
4744 http-request set-src src,ipmask(24)
4745
4746 When possible, set-src preserves the original source port as long as the
4747 address family allows it, otherwise the source port is set to 0.
4748
4749http-request set-src-port <expr> [ { if | unless } <condition> ]
4750
4751 This is used to set the source port address to the value of specified
4752 expression.
4753
4754 Arguments:
4755 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4756 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004757
Willy Tarreau20b0de52012-12-24 15:45:22 +01004758 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004759 http-request set-src-port hdr(x-port)
4760 http-request set-src-port int(4000)
4761
4762 When possible, set-src-port preserves the original source address as long as
4763 the address family supports a port, otherwise it forces the source address to
4764 IPv4 "0.0.0.0" before rewriting the port.
4765
4766http-request set-tos <tos> [ { if | unless } <condition> ]
4767
4768 This is used to set the TOS or DSCP field value of packets sent to the client
4769 to the value passed in <tos> on platforms which support this. This value
4770 represents the whole 8 bits of the IP TOS field, and can be expressed both in
4771 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
4772 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
4773 be used to adjust some routing behavior on border routers based on some
4774 information from the request.
4775
4776 See RFC 2474, 2597, 3260 and 4594 for more information.
4777
4778http-request set-uri <fmt> [ { if | unless } <condition> ]
4779
4780 This rewrites the request URI with the result of the evaluation of format
4781 string <fmt>. The scheme, authority, path and query string are all replaced
4782 at once. This can be used to rewrite hosts in front of proxies, or to
4783 perform complex modifications to the URI such as moving parts between the
4784 path and the query string.
4785 See also "http-request set-path" and "http-request set-query".
4786
4787http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4788
4789 This is used to set the contents of a variable. The variable is declared
4790 inline.
4791
4792 Arguments:
4793 <var-name> The name of the variable starts with an indication about its
4794 scope. The scopes allowed are:
4795 "proc" : the variable is shared with the whole process
4796 "sess" : the variable is shared with the whole session
4797 "txn" : the variable is shared with the transaction
4798 (request and response)
4799 "req" : the variable is shared only during request
4800 processing
4801 "res" : the variable is shared only during response
4802 processing
4803 This prefix is followed by a name. The separator is a '.'.
4804 The name may only contain characters 'a-z', 'A-Z', '0-9'
4805 and '_'.
4806
4807 <expr> Is a standard HAProxy expression formed by a sample-fetch
4808 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004809
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004810 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004811 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004812
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004813http-request send-spoe-group <engine-name> <group-name>
4814 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004815
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004816 This action is used to trigger sending of a group of SPOE messages. To do so,
4817 the SPOE engine used to send messages must be defined, as well as the SPOE
4818 group to send. Of course, the SPOE engine must refer to an existing SPOE
4819 filter. If not engine name is provided on the SPOE filter line, the SPOE
4820 agent name must be used.
4821
4822 Arguments:
4823 <engine-name> The SPOE engine name.
4824
4825 <group-name> The SPOE group name as specified in the engine
4826 configuration.
4827
4828http-request silent-drop [ { if | unless } <condition> ]
4829
4830 This stops the evaluation of the rules and makes the client-facing connection
4831 suddenly disappear using a system-dependent way that tries to prevent the
4832 client from being notified. The effect it then that the client still sees an
4833 established connection while there's none on HAProxy. The purpose is to
4834 achieve a comparable effect to "tarpit" except that it doesn't use any local
4835 resource at all on the machine running HAProxy. It can resist much higher
4836 loads than "tarpit", and slow down stronger attackers. It is important to
4837 understand the impact of using this mechanism. All stateful equipment placed
4838 between the client and HAProxy (firewalls, proxies, load balancers) will also
4839 keep the established connection for a long time and may suffer from this
4840 action.
4841 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4842 option is used to block the emission of a TCP reset. On other systems, the
4843 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4844 router, though it's still delivered to local networks. Do not use it unless
4845 you fully understand how it works.
4846
4847http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
4848
4849 This stops the evaluation of the rules and immediately blocks the request
4850 without responding for a delay specified by "timeout tarpit" or
4851 "timeout connect" if the former is not set. After that delay, if the client
4852 is still connected, an HTTP error 500 (or optionally the status code
4853 specified as an argument to "deny_status") is returned so that the client
4854 does not suspect it has been tarpitted. Logs will report the flags "PT".
4855 The goal of the tarpit rule is to slow down robots during an attack when
4856 they're limited on the number of concurrent requests. It can be very
4857 efficient against very dumb robots, and will significantly reduce the load
4858 on firewalls compared to a "deny" rule. But when facing "correctly"
4859 developed robots, it can make things worse by forcing haproxy and the front
4860 firewall to support insane number of concurrent connections.
4861 See also the "silent-drop" action.
4862
4863http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4864http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4865http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
4866
4867 This enables tracking of sticky counters from current request. These rules do
4868 not stop evaluation and do not change default action. The number of counters
4869 that may be simultaneously tracked by the same connection is set in
4870 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
4871 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
4872 "track-sc0" rule executed enables tracking of the counters of the specified
4873 table as the first set. The first "track-sc1" rule executed enables tracking
4874 of the counters of the specified table as the second set. The first
4875 "track-sc2" rule executed enables tracking of the counters of the specified
4876 table as the third set. It is a recommended practice to use the first set of
4877 counters for the per-frontend counters and the second set for the per-backend
4878 ones. But this is just a guideline, all may be used everywhere.
4879
4880 Arguments :
4881 <key> is mandatory, and is a sample expression rule as described in
4882 section 7.3. It describes what elements of the incoming request or
4883 connection will be analyzed, extracted, combined, and used to
4884 select which table entry to update the counters.
4885
4886 <table> is an optional table to be used instead of the default one, which
4887 is the stick-table declared in the current proxy. All the counters
4888 for the matches and updates for the key will then be performed in
4889 that table until the session ends.
4890
4891 Once a "track-sc*" rule is executed, the key is looked up in the table and if
4892 it is not found, an entry is allocated for it. Then a pointer to that entry
4893 is kept during all the session's life, and this entry's counters are updated
4894 as often as possible, every time the session's counters are updated, and also
4895 systematically when the session ends. Counters are only updated for events
4896 that happen after the tracking has been started. As an exception, connection
4897 counters and request counters are systematically updated so that they reflect
4898 useful information.
4899
4900 If the entry tracks concurrent connection counters, one connection is counted
4901 for as long as the entry is tracked, and the entry will not expire during
4902 that time. Tracking counters also provides a performance advantage over just
4903 checking the keys, because only one table lookup is performed for all ACL
4904 checks that make use of it.
4905
4906http-request unset-var(<var-name>) [ { if | unless } <condition> ]
4907
4908 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004909
4910 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004911 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004912
Christopher Faulet6bd406e2019-11-22 15:34:17 +01004913http-request use-service <service-name> [ { if | unless } <condition> ]
4914
4915 This directive executes the configured HTTP service to reply to the request
4916 and stops the evaluation of the rules. An HTTP service may choose to reply by
4917 sending any valid HTTP response or it may immediately close the connection
4918 without sending any response. Outside natives services, for instance the
4919 Prometheus exporter, it is possible to write your own services in Lua. No
4920 further "http-request" rules are evaluated.
4921
4922 Arguments :
4923 <service-name> is mandatory. It is the service to call
4924
4925 Example:
4926 http-request use-service prometheus-exporter if { path /metrics }
4927
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004928http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004929
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004930 This will delay the processing of the request until the SSL handshake
4931 happened. This is mostly useful to delay processing early data until we're
4932 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004933
Willy Tarreauef781042010-01-27 11:53:01 +01004934
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004935http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004936 Access control for Layer 7 responses
4937
4938 May be used in sections: defaults | frontend | listen | backend
4939 no | yes | yes | yes
4940
4941 The http-response statement defines a set of rules which apply to layer 7
4942 processing. The rules are evaluated in their declaration order when they are
4943 met in a frontend, listen or backend section. Any rule may optionally be
4944 followed by an ACL-based condition, in which case it will only be evaluated
4945 if the condition is true. Since these rules apply on responses, the backend
4946 rules are applied first, followed by the frontend's rules.
4947
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004948 The first keyword is the rule's action. The supported actions are described
4949 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004950
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004951 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004952
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004953 It is important to know that http-response rules are processed very early in
4954 the HTTP processing, before "rspdel" or "rsprep" or "rspadd" rules. That way,
4955 headers added by "add-header"/"set-header" are visible by almost all further
4956 ACL rules.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004957
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004958 Using "rspadd"/"rspdel"/"rsprep" to manipulate request headers is discouraged
4959 in newer versions (>= 1.5). But if you need to use regular expression to
4960 delete headers, you can still use "rspdel". Also please use
4961 "http-response deny" instead of "rspdeny".
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004962
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004963 Example:
4964 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004965
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004966 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06004967
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004968 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4969 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06004970
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004971 Example:
4972 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06004973
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004974 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004975
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004976 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4977 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06004978
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004979 See also : "http-request", section 3.4 about userlists and section 7 about
4980 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004981
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004982http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004983
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004984 This is used to add a new entry into an ACL. The ACL must be loaded from a
4985 file (even a dummy empty file). The file name of the ACL to be updated is
4986 passed between parentheses. It takes one argument: <key fmt>, which follows
4987 log-format rules, to collect content of the new entry. It performs a lookup
4988 in the ACL before insertion, to avoid duplicated (or more) values.
4989 This lookup is done by a linear search and can be expensive with large lists!
4990 It is the equivalent of the "add acl" command from the stats socket, but can
4991 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06004992
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004993http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004994
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004995 This appends an HTTP header field whose name is specified in <name> and whose
4996 value is defined by <fmt> which follows the log-format rules (see Custom Log
4997 Format in section 8.2.4). This may be used to send a cookie to a client for
4998 example, or to pass some internal information.
4999 This rule is not final, so it is possible to add other similar rules.
5000 Note that header addition is performed immediately, so one rule might reuse
5001 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06005002
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005003http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005004
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005005 This stops the evaluation of the rules and lets the response pass the check.
5006 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06005007
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02005008http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005009
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005010 See section 10.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06005011
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005012http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06005013
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005014 This captures sample expression <sample> from the response buffer, and
5015 converts it to a string. The resulting string is stored into the next request
5016 "capture" slot, so it will possibly appear next to some captured HTTP
5017 headers. It will then automatically appear in the logs, and it will be
5018 possible to extract it using sample fetch rules to feed it into headers or
5019 anything. Please check section 7.3 (Fetching samples) and
5020 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005021
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005022 The keyword "id" is the id of the capture slot which is used for storing the
5023 string. The capture slot must be defined in an associated frontend.
5024 This is useful to run captures in backends. The slot id can be declared by a
5025 previous directive "http-response capture" or with the "declare capture"
5026 keyword.
5027 If the slot <id> doesn't exist, then HAProxy fails parsing the configuration
5028 to prevent unexpected behavior at run time.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005029
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005030http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02005031
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005032 This is used to delete an entry from an ACL. The ACL must be loaded from a
5033 file (even a dummy empty file). The file name of the ACL to be updated is
5034 passed between parentheses. It takes one argument: <key fmt>, which follows
5035 log-format rules, to collect content of the entry to delete.
5036 It is the equivalent of the "del acl" command from the stats socket, but can
5037 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02005038
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005039http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02005040
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005041 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02005042
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005043http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02005044
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005045 This is used to delete an entry from a MAP. The MAP must be loaded from a
5046 file (even a dummy empty file). The file name of the MAP to be updated is
5047 passed between parentheses. It takes one argument: <key fmt>, which follows
5048 log-format rules, to collect content of the entry to delete.
5049 It takes one argument: "file name" It is the equivalent of the "del map"
5050 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005051
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005052http-response deny [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005053
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005054 This stops the evaluation of the rules and immediately rejects the response
5055 and emits an HTTP 502 error. No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005056
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005057http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005058
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005059 This performs an HTTP redirection based on a redirect rule.
5060 This supports a format string similarly to "http-request redirect" rules,
5061 with the exception that only the "location" type of redirect is possible on
5062 the response. See the "redirect" keyword for the rule's syntax. When a
5063 redirect rule is applied during a response, connections to the server are
5064 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02005065
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005066http-response replace-header <name> <regex-match> <replace-fmt>
5067 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02005068
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005069 This matches the regular expression in all occurrences of header field <name>
5070 according to <match-regex>, and replaces them with the <replace-fmt> argument.
5071 Format characters are allowed in replace-fmt and work like in <fmt> arguments
5072 in "add-header". The match is only case-sensitive. It is important to
5073 understand that this action only considers whole header lines, regardless of
5074 the number of values they may contain. This usage is suited to headers
5075 naturally containing commas in their value, such as Set-Cookie, Expires and
5076 so on.
William Lallemand86d0df02017-11-24 21:36:45 +01005077
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005078 Example:
5079 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02005080
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005081 # applied to:
5082 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005083
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005084 # outputs:
5085 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 +02005086
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005087 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005088
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005089http-response replace-value <name> <regex-match> <replace-fmt>
5090 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005091
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005092 This works like "replace-header" except that it matches the regex against
5093 every comma-delimited value of the header field <name> instead of the entire
5094 header. This is suited for all headers which are allowed to carry more than
5095 one value. An example could be the Accept header.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02005096
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005097 Example:
5098 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01005099
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005100 # applied to:
5101 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01005102
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005103 # outputs:
5104 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01005105
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005106http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
5107http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08005108
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005109 This action increments the GPC0 or GPC1 counter according with the sticky
5110 counter designated by <sc-id>. If an error occurs, this action silently fails
5111 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02005112
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005113http-response sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02005114
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005115 This action sets the GPT0 tag according to the sticky counter designated by
5116 <sc-id> and the value of <int>. The expected result is a boolean. If an error
5117 occurs, this action silently fails and the actions evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01005118
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005119http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02005120
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005121 This action is used to trigger sending of a group of SPOE messages. To do so,
5122 the SPOE engine used to send messages must be defined, as well as the SPOE
5123 group to send. Of course, the SPOE engine must refer to an existing SPOE
5124 filter. If not engine name is provided on the SPOE filter line, the SPOE
5125 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005126
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005127 Arguments:
5128 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005129
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005130 <group-name> The SPOE group name as specified in the engine
5131 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02005132
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005133http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005134
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005135 This does the same as "add-header" except that the header name is first
5136 removed if it existed. This is useful when passing security information to
5137 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005138
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005139http-response set-log-level <level> [ { if | unless } <condition> ]
5140
5141 This is used to change the log level of the current request when a certain
5142 condition is met. Valid levels are the 8 syslog levels (see the "log"
5143 keyword) plus the special level "silent" which disables logging for this
5144 request. This rule is not final so the last matching rule wins. This rule can
5145 be useful to disable health checks coming from another equipment.
5146
5147http-response set-map(<file-name>) <key fmt> <value fmt>
5148
5149 This is used to add a new entry into a MAP. The MAP must be loaded from a
5150 file (even a dummy empty file). The file name of the MAP to be updated is
5151 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
5152 log-format rules, used to collect MAP key, and <value fmt>, which follows
5153 log-format rules, used to collect content for the new entry. It performs a
5154 lookup in the MAP before insertion, to avoid duplicated (or more) values.
5155 This lookup is done by a linear search and can be expensive with large lists!
5156 It is the equivalent of the "set map" command from the stats socket, but can
5157 be triggered by an HTTP response.
5158
5159http-response set-mark <mark> [ { if | unless } <condition> ]
5160
5161 This is used to set the Netfilter MARK on all packets sent to the client to
5162 the value passed in <mark> on platforms which support it. This value is an
5163 unsigned 32 bit value which can be matched by netfilter and by the routing
5164 table. It can be expressed both in decimal or hexadecimal format (prefixed
5165 by "0x"). This can be useful to force certain packets to take a different
5166 route (for example a cheaper network path for bulk downloads). This works on
5167 Linux kernels 2.6.32 and above and requires admin privileges.
5168
5169http-response set-nice <nice> [ { if | unless } <condition> ]
5170
5171 This sets the "nice" factor of the current request being processed.
5172 It only has effect against the other requests being processed at the same
5173 time. The default value is 0, unless altered by the "nice" setting on the
5174 "bind" line. The accepted range is -1024..1024. The higher the value, the
5175 nicest the request will be. Lower values will make the request more important
5176 than other ones. This can be useful to improve the speed of some requests, or
5177 lower the priority of non-important requests. Using this setting without
5178 prior experimentation can cause some major slowdown.
5179
5180http-response set-status <status> [reason <str>]
5181 [ { if | unless } <condition> ]
5182
5183 This replaces the response status code with <status> which must be an integer
5184 between 100 and 999. Optionally, a custom reason text can be provided defined
5185 by <str>, or the default reason for the specified code will be used as a
5186 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08005187
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005188 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005189 # return "431 Request Header Fields Too Large"
5190 http-response set-status 431
5191 # return "503 Slow Down", custom reason
5192 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005193
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005194http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005195
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005196 This is used to set the TOS or DSCP field value of packets sent to the client
5197 to the value passed in <tos> on platforms which support this.
5198 This value represents the whole 8 bits of the IP TOS field, and can be
5199 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
5200 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
5201 always 0. This can be used to adjust some routing behavior on border routers
5202 based on some information from the request.
5203
5204 See RFC 2474, 2597, 3260 and 4594 for more information.
5205
5206http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
5207
5208 This is used to set the contents of a variable. The variable is declared
5209 inline.
5210
5211 Arguments:
5212 <var-name> The name of the variable starts with an indication about its
5213 scope. The scopes allowed are:
5214 "proc" : the variable is shared with the whole process
5215 "sess" : the variable is shared with the whole session
5216 "txn" : the variable is shared with the transaction
5217 (request and response)
5218 "req" : the variable is shared only during request
5219 processing
5220 "res" : the variable is shared only during response
5221 processing
5222 This prefix is followed by a name. The separator is a '.'.
5223 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
5224 and '_'.
5225
5226 <expr> Is a standard HAProxy expression formed by a sample-fetch
5227 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005228
5229 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005230 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005231
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005232http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005233
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005234 This stops the evaluation of the rules and makes the client-facing connection
5235 suddenly disappear using a system-dependent way that tries to prevent the
5236 client from being notified. The effect it then that the client still sees an
5237 established connection while there's none on HAProxy. The purpose is to
5238 achieve a comparable effect to "tarpit" except that it doesn't use any local
5239 resource at all on the machine running HAProxy. It can resist much higher
5240 loads than "tarpit", and slow down stronger attackers. It is important to
5241 understand the impact of using this mechanism. All stateful equipment placed
5242 between the client and HAProxy (firewalls, proxies, load balancers) will also
5243 keep the established connection for a long time and may suffer from this
5244 action.
5245 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
5246 option is used to block the emission of a TCP reset. On other systems, the
5247 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
5248 router, though it's still delivered to local networks. Do not use it unless
5249 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02005250
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005251http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
5252http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
5253http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02005254
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02005255 This enables tracking of sticky counters from current response. Please refer
5256 to "http-request track-sc" for a complete description. The only difference
5257 from "http-request track-sc" is the <key> sample expression can only make use
5258 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
5259 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
5260 supported, haproxy will fail and warn while parsing the config.
5261
5262http-response unset-var(<var-name>) [ { if | unless } <condition> ]
5263
5264 This is used to unset a variable. See "http-response set-var" for details
5265 about <var-name>.
5266
5267 Example:
5268 http-response unset-var(sess.last_redir)
5269
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02005270
Willy Tarreau30631952015-08-06 15:05:24 +02005271http-reuse { never | safe | aggressive | always }
5272 Declare how idle HTTP connections may be shared between requests
5273
5274 May be used in sections: defaults | frontend | listen | backend
5275 yes | no | yes | yes
5276
5277 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01005278 which is considered safe for reuse is moved back to the server's idle
5279 connections pool so that any other request can make use of it. This is the
5280 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02005281
5282 The argument indicates the desired connection reuse strategy :
5283
Olivier Houchard86006a52018-12-14 19:37:49 +01005284 - "never" : idle connections are never shared between sessions. This mode
5285 may be enforced to cancel a different strategy inherited from
5286 a defaults section or for troubleshooting. For example, if an
5287 old bogus application considers that multiple requests over
5288 the same connection come from the same client and it is not
5289 possible to fix the application, it may be desirable to
5290 disable connection sharing in a single backend. An example of
5291 such an application could be an old haproxy using cookie
5292 insertion in tunnel mode and not checking any request past the
5293 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02005294
Olivier Houchard86006a52018-12-14 19:37:49 +01005295 - "safe" : this is the default and the recommended strategy. The first
5296 request of a session is always sent over its own connection,
5297 and only subsequent requests may be dispatched over other
5298 existing connections. This ensures that in case the server
5299 closes the connection when the request is being sent, the
5300 browser can decide to silently retry it. Since it is exactly
5301 equivalent to regular keep-alive, there should be no side
5302 effects.
Willy Tarreau30631952015-08-06 15:05:24 +02005303
5304 - "aggressive" : this mode may be useful in webservices environments where
5305 all servers are not necessarily known and where it would be
5306 appreciable to deliver most first requests over existing
5307 connections. In this case, first requests are only delivered
5308 over existing connections that have been reused at least once,
5309 proving that the server correctly supports connection reuse.
5310 It should only be used when it's sure that the client can
5311 retry a failed request once in a while and where the benefit
Michael Prokop4438c602019-05-24 10:25:45 +02005312 of aggressive connection reuse significantly outweighs the
Willy Tarreau30631952015-08-06 15:05:24 +02005313 downsides of rare connection failures.
5314
5315 - "always" : this mode is only recommended when the path to the server is
5316 known for never breaking existing connections quickly after
5317 releasing them. It allows the first request of a session to be
5318 sent to an existing connection. This can provide a significant
5319 performance increase over the "safe" strategy when the backend
5320 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005321 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02005322 sharing. It is recommended that the "http-keep-alive" timeout
5323 remains low in this mode so that no dead connections remain
5324 usable. In most cases, this will lead to the same performance
5325 gains as "aggressive" but with more risks. It should only be
5326 used when it improves the situation over "aggressive".
5327
5328 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005329 connection properties and compatibility. Specifically :
5330 - connections made with "usesrc" followed by a client-dependent value
5331 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005332
5333 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01005334 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005335
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02005336 - connections with certain bogus authentication schemes (relying on the
5337 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02005338
Lukas Tribus79a56932019-11-06 11:50:25 +01005339 A connection pool is involved and configurable with "pool-max-conn".
Willy Tarreau30631952015-08-06 15:05:24 +02005340
5341 Note: connection reuse improves the accuracy of the "server maxconn" setting,
5342 because almost no new connection will be established while idle connections
5343 remain available. This is particularly true with the "always" strategy.
5344
5345 See also : "option http-keep-alive", "server maxconn"
5346
5347
Mark Lamourinec2247f02012-01-04 13:02:01 -05005348http-send-name-header [<header>]
5349 Add the server name to a request. Use the header string given by <header>
Mark Lamourinec2247f02012-01-04 13:02:01 -05005350 May be used in sections: defaults | frontend | listen | backend
5351 yes | no | yes | yes
Mark Lamourinec2247f02012-01-04 13:02:01 -05005352 Arguments :
Mark Lamourinec2247f02012-01-04 13:02:01 -05005353 <header> The header string to use to send the server name
5354
Willy Tarreaue0e32792019-10-07 14:58:02 +02005355 The "http-send-name-header" statement causes the header field named <header>
5356 to be set to the name of the target server at the moment the request is about
5357 to be sent on the wire. Any existing occurrences of this header are removed.
5358 Upon retries and redispatches, the header field is updated to always reflect
5359 the server being attempted to connect to. Given that this header is modified
5360 very late in the connection setup, it may have unexpected effects on already
5361 modified headers. For example using it with transport-level header such as
5362 connection, content-length, transfer-encoding and so on will likely result in
5363 invalid requests being sent to the server. Additionally it has been reported
5364 that this directive is currently being used as a way to overwrite the Host
5365 header field in outgoing requests; while this trick has been known to work
5366 as a side effect of the feature for some time, it is not officially supported
5367 and might possibly not work anymore in a future version depending on the
5368 technical difficulties this feature induces. A long-term solution instead
5369 consists in fixing the application which required this trick so that it binds
5370 to the correct host name.
Mark Lamourinec2247f02012-01-04 13:02:01 -05005371
5372 See also : "server"
5373
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005374id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02005375 Set a persistent ID to a proxy.
5376 May be used in sections : defaults | frontend | listen | backend
5377 no | yes | yes | yes
5378 Arguments : none
5379
5380 Set a persistent ID for the proxy. This ID must be unique and positive.
5381 An unused ID will automatically be assigned if unset. The first assigned
5382 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005383
5384
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005385ignore-persist { if | unless } <condition>
5386 Declare a condition to ignore persistence
5387 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005388 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005389
5390 By default, when cookie persistence is enabled, every requests containing
5391 the cookie are unconditionally persistent (assuming the target server is up
5392 and running).
5393
5394 The "ignore-persist" statement allows one to declare various ACL-based
5395 conditions which, when met, will cause a request to ignore persistence.
5396 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005397 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005398 persistence for a specific User-Agent (for example, some web crawler bots).
5399
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005400 The persistence is ignored when an "if" condition is met, or unless an
5401 "unless" condition is met.
5402
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005403 Example:
5404 acl url_static path_beg /static /images /img /css
5405 acl url_static path_end .gif .png .jpg .css .js
5406 ignore-persist if url_static
5407
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005408 See also : "force-persist", "cookie", and section 7 about ACL usage.
5409
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005410load-server-state-from-file { global | local | none }
5411 Allow seamless reload of HAProxy
5412 May be used in sections: defaults | frontend | listen | backend
5413 yes | no | yes | yes
5414
5415 This directive points HAProxy to a file where server state from previous
5416 running process has been saved. That way, when starting up, before handling
5417 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005418 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005419 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5420 loading state or load states from a file containing all backends and servers.
5421 The state file can be generated by running the command "show servers state"
5422 over the stats socket and redirect output.
5423
Davor Ocelice9ed2812017-12-25 17:49:28 +01005424 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005425 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005426 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005427
5428 Arguments:
5429 global load the content of the file pointed by the global directive
5430 named "server-state-file".
5431
5432 local load the content of the file pointed by the directive
5433 "server-state-file-name" if set. If not set, then the backend
5434 name is used as a file name.
5435
5436 none don't load any stat for this backend
5437
5438 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005439 - server's IP address is preserved across reloads by default, but the
5440 order can be changed thanks to the server's "init-addr" setting. This
5441 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005442 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005443 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005444
5445 - server's weight is applied from previous running process unless it has
5446 has changed between previous and new configuration files.
5447
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005448 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005449
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005450 global
5451 stats socket /tmp/socket
5452 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005453
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005454 defaults
5455 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005456
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005457 backend bk
5458 server s1 127.0.0.1:22 check weight 11
5459 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005460
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005461
5462 Then one can run :
5463
5464 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5465
5466 Content of the file /tmp/server_state would be like this:
5467
5468 1
5469 # <field names skipped for the doc example>
5470 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5471 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5472
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005473 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005474
5475 global
5476 stats socket /tmp/socket
5477 server-state-base /etc/haproxy/states
5478
5479 defaults
5480 load-server-state-from-file local
5481
5482 backend bk
5483 server s1 127.0.0.1:22 check weight 11
5484 server s2 127.0.0.1:22 check weight 12
5485
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005486
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005487 Then one can run :
5488
5489 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5490
5491 Content of the file /etc/haproxy/states/bk would be like this:
5492
5493 1
5494 # <field names skipped for the doc example>
5495 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5496 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5497
5498 See also: "server-state-file", "server-state-file-name", and
5499 "show servers state"
5500
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005501
Willy Tarreau2769aa02007-12-27 18:26:09 +01005502log global
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005503log <address> [len <length>] [format <format>] [sample <ranges>:<smp_size>]
5504 <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005505no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005506 Enable per-instance logging of events and traffic.
5507 May be used in sections : defaults | frontend | listen | backend
5508 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005509
5510 Prefix :
5511 no should be used when the logger list must be flushed. For example,
5512 if you don't want to inherit from the default logger list. This
5513 prefix does not allow arguments.
5514
Willy Tarreau2769aa02007-12-27 18:26:09 +01005515 Arguments :
5516 global should be used when the instance's logging parameters are the
5517 same as the global ones. This is the most common usage. "global"
5518 replaces <address>, <facility> and <level> with those of the log
5519 entries found in the "global" section. Only one "log global"
5520 statement may be used per instance, and this form takes no other
5521 parameter.
5522
5523 <address> indicates where to send the logs. It takes the same format as
5524 for the "global" section's logs, and can be one of :
5525
5526 - An IPv4 address optionally followed by a colon (':') and a UDP
5527 port. If no port is specified, 514 is used by default (the
5528 standard syslog port).
5529
David du Colombier24bb5f52011-03-17 10:40:23 +01005530 - An IPv6 address followed by a colon (':') and optionally a UDP
5531 port. If no port is specified, 514 is used by default (the
5532 standard syslog port).
5533
Willy Tarreau2769aa02007-12-27 18:26:09 +01005534 - A filesystem path to a UNIX domain socket, keeping in mind
5535 considerations for chroot (be sure the path is accessible
5536 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005537 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005538
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005539 - A file descriptor number in the form "fd@<number>", which may
5540 point to a pipe, terminal, or socket. In this case unbuffered
5541 logs are used and one writev() call per log is performed. This
5542 is a bit expensive but acceptable for most workloads. Messages
5543 sent this way will not be truncated but may be dropped, in
5544 which case the DroppedLogs counter will be incremented. The
5545 writev() call is atomic even on pipes for messages up to
5546 PIPE_BUF size, which POSIX recommends to be at least 512 and
5547 which is 4096 bytes on most modern operating systems. Any
5548 larger message may be interleaved with messages from other
5549 processes. Exceptionally for debugging purposes the file
5550 descriptor may also be directed to a file, but doing so will
5551 significantly slow haproxy down as non-blocking calls will be
5552 ignored. Also there will be no way to purge nor rotate this
5553 file without restarting the process. Note that the configured
5554 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01005555 with a TCP syslog server. See also the "short" and "raw"
5556 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005557
5558 - "stdout" / "stderr", which are respectively aliases for "fd@1"
5559 and "fd@2", see above.
5560
5561 You may want to reference some environment variables in the
5562 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005563
Willy Tarreau18324f52014-06-27 18:10:07 +02005564 <length> is an optional maximum line length. Log lines larger than this
5565 value will be truncated before being sent. The reason is that
5566 syslog servers act differently on log line length. All servers
5567 support the default value of 1024, but some servers simply drop
5568 larger lines while others do log them. If a server supports long
5569 lines, it may make sense to set this value here in order to avoid
5570 truncating long lines. Similarly, if a server drops long lines,
5571 it is preferable to truncate them before sending them. Accepted
5572 values are 80 to 65535 inclusive. The default value of 1024 is
5573 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005574 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005575
Frédéric Lécailled690dfa2019-04-25 10:52:17 +02005576 <ranges> A list of comma-separated ranges to identify the logs to sample.
5577 This is used to balance the load of the logs to send to the log
5578 server. The limits of the ranges cannot be null. They are numbered
5579 from 1. The size or period (in number of logs) of the sample must
5580 be set with <sample_size> parameter.
5581
5582 <sample_size>
5583 The size of the sample in number of logs to consider when balancing
5584 their logging loads. It is used to balance the load of the logs to
5585 send to the syslog server. This size must be greater or equal to the
5586 maximum of the high limits of the ranges.
5587 (see also <ranges> parameter).
5588
Willy Tarreauadb345d2018-11-12 07:56:13 +01005589 <format> is the log format used when generating syslog messages. It may be
5590 one of the following :
5591
5592 rfc3164 The RFC3164 syslog message format. This is the default.
5593 (https://tools.ietf.org/html/rfc3164)
5594
5595 rfc5424 The RFC5424 syslog message format.
5596 (https://tools.ietf.org/html/rfc5424)
5597
Willy Tarreaue8746a02018-11-12 08:45:00 +01005598 short A message containing only a level between angle brackets such as
5599 '<3>', followed by the text. The PID, date, time, process name
5600 and system name are omitted. This is designed to be used with a
5601 local log server. This format is compatible with what the
5602 systemd logger consumes.
5603
Willy Tarreauc1b06452018-11-12 11:57:56 +01005604 raw A message containing only the text. The level, PID, date, time,
5605 process name and system name are omitted. This is designed to
5606 be used in containers or during development, where the severity
5607 only depends on the file descriptor used (stdout/stderr).
5608
Willy Tarreau2769aa02007-12-27 18:26:09 +01005609 <facility> must be one of the 24 standard syslog facilities :
5610
Willy Tarreaue8746a02018-11-12 08:45:00 +01005611 kern user mail daemon auth syslog lpr news
5612 uucp cron auth2 ftp ntp audit alert cron2
5613 local0 local1 local2 local3 local4 local5 local6 local7
5614
Willy Tarreauc1b06452018-11-12 11:57:56 +01005615 Note that the facility is ignored for the "short" and "raw"
5616 formats, but still required as a positional field. It is
5617 recommended to use "daemon" in this case to make it clear that
5618 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005619
5620 <level> is optional and can be specified to filter outgoing messages. By
5621 default, all messages are sent. If a level is specified, only
5622 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005623 will be sent. An optional minimum level can be specified. If it
5624 is set, logs emitted with a more severe level than this one will
5625 be capped to this level. This is used to avoid sending "emerg"
5626 messages on all terminals on some default syslog configurations.
5627 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005628
5629 emerg alert crit err warning notice info debug
5630
William Lallemand0f99e342011-10-12 17:50:54 +02005631 It is important to keep in mind that it is the frontend which decides what to
5632 log from a connection, and that in case of content switching, the log entries
5633 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005634
5635 However, backend log declaration define how and where servers status changes
5636 will be logged. Level "notice" will be used to indicate a server going up,
5637 "warning" will be used for termination signals and definitive service
5638 termination, and "alert" will be used for when a server goes down.
5639
5640 Note : According to RFC3164, messages are truncated to 1024 bytes before
5641 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005642
5643 Example :
5644 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01005645 log stdout format short daemon # send log to systemd
5646 log stdout format raw daemon # send everything to stdout
5647 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005648 log 127.0.0.1:514 local0 notice # only send important events
5649 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005650 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005651
Willy Tarreau2769aa02007-12-27 18:26:09 +01005652
William Lallemand48940402012-01-30 16:47:22 +01005653log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005654 Specifies the log format string to use for traffic logs
5655 May be used in sections: defaults | frontend | listen | backend
5656 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005657
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005658 This directive specifies the log format string that will be used for all logs
5659 resulting from traffic passing through the frontend using this line. If the
5660 directive is used in a defaults section, all subsequent frontends will use
5661 the same log format. Please see section 8.2.4 which covers the log format
5662 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005663
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005664 "log-format" directive overrides previous "option tcplog", "log-format" and
5665 "option httplog" directives.
5666
Dragan Dosen7ad31542015-09-28 17:16:47 +02005667log-format-sd <string>
5668 Specifies the RFC5424 structured-data log format string
5669 May be used in sections: defaults | frontend | listen | backend
5670 yes | yes | yes | no
5671
5672 This directive specifies the RFC5424 structured-data log format string that
5673 will be used for all logs resulting from traffic passing through the frontend
5674 using this line. If the directive is used in a defaults section, all
5675 subsequent frontends will use the same log format. Please see section 8.2.4
5676 which covers the log format string in depth.
5677
5678 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5679 about the RFC5424 structured-data part.
5680
5681 Note : This log format string will be used only for loggers that have set
5682 log format to "rfc5424".
5683
5684 Example :
5685 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5686
5687
Willy Tarreau094af4e2015-01-07 15:03:42 +01005688log-tag <string>
5689 Specifies the log tag to use for all outgoing logs
5690 May be used in sections: defaults | frontend | listen | backend
5691 yes | yes | yes | yes
5692
5693 Sets the tag field in the syslog header to this string. It defaults to the
5694 log-tag set in the global section, otherwise the program name as launched
5695 from the command line, which usually is "haproxy". Sometimes it can be useful
5696 to differentiate between multiple processes running on the same host, or to
5697 differentiate customer instances running in the same process. In the backend,
5698 logs about servers up/down will use this tag. As a hint, it can be convenient
5699 to set a log-tag related to a hosted customer in a defaults section then put
5700 all the frontends and backends for that customer, then start another customer
5701 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005702
Willy Tarreauc35362a2014-04-25 13:58:37 +02005703max-keep-alive-queue <value>
5704 Set the maximum server queue size for maintaining keep-alive connections
5705 May be used in sections: defaults | frontend | listen | backend
5706 yes | no | yes | yes
5707
5708 HTTP keep-alive tries to reuse the same server connection whenever possible,
5709 but sometimes it can be counter-productive, for example if a server has a lot
5710 of connections while other ones are idle. This is especially true for static
5711 servers.
5712
5713 The purpose of this setting is to set a threshold on the number of queued
5714 connections at which haproxy stops trying to reuse the same server and prefers
5715 to find another one. The default value, -1, means there is no limit. A value
5716 of zero means that keep-alive requests will never be queued. For very close
5717 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005718 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005719 use a value of 10 or less). For remote servers suffering from a high latency,
5720 higher values might be needed to cover for the latency and/or the cost of
5721 picking a different server.
5722
5723 Note that this has no impact on responses which are maintained to the same
5724 server consecutively to a 401 response. They will still go to the same server
5725 even if they have to be queued.
5726
5727 See also : "option http-server-close", "option prefer-last-server", server
5728 "maxconn" and cookie persistence.
5729
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01005730max-session-srv-conns <nb>
5731 Set the maximum number of outgoing connections we can keep idling for a given
5732 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
5733 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02005734
Willy Tarreau2769aa02007-12-27 18:26:09 +01005735maxconn <conns>
5736 Fix the maximum number of concurrent connections on a frontend
5737 May be used in sections : defaults | frontend | listen | backend
5738 yes | yes | yes | no
5739 Arguments :
5740 <conns> is the maximum number of concurrent connections the frontend will
5741 accept to serve. Excess connections will be queued by the system
5742 in the socket's listen queue and will be served once a connection
5743 closes.
5744
5745 If the system supports it, it can be useful on big sites to raise this limit
5746 very high so that haproxy manages connection queues, instead of leaving the
5747 clients with unanswered connection attempts. This value should not exceed the
5748 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005749 of tune.bufsize (16kB by default) each, as well as some other data resulting
5750 in about 33 kB of RAM being consumed per established connection. That means
5751 that a medium system equipped with 1GB of RAM can withstand around
5752 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005753
5754 Also, when <conns> is set to large values, it is possible that the servers
5755 are not sized to accept such loads, and for this reason it is generally wise
5756 to assign them some reasonable connection limits.
5757
Willy Tarreauc8d5b952019-02-27 17:25:52 +01005758 When this value is set to zero, which is the default, the global "maxconn"
5759 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02005760
Willy Tarreau2769aa02007-12-27 18:26:09 +01005761 See also : "server", global section's "maxconn", "fullconn"
5762
5763
5764mode { tcp|http|health }
5765 Set the running mode or protocol of the instance
5766 May be used in sections : defaults | frontend | listen | backend
5767 yes | yes | yes | yes
5768 Arguments :
5769 tcp The instance will work in pure TCP mode. A full-duplex connection
5770 will be established between clients and servers, and no layer 7
5771 examination will be performed. This is the default mode. It
5772 should be used for SSL, SSH, SMTP, ...
5773
5774 http The instance will work in HTTP mode. The client request will be
5775 analyzed in depth before connecting to any server. Any request
5776 which is not RFC-compliant will be rejected. Layer 7 filtering,
5777 processing and switching will be possible. This is the mode which
5778 brings HAProxy most of its value.
5779
5780 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005781 to incoming connections and close the connection. Alternatively,
5782 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5783 instead. Nothing will be logged in either case. This mode is used
5784 to reply to external components health checks. This mode is
5785 deprecated and should not be used anymore as it is possible to do
5786 the same and even better by combining TCP or HTTP modes with the
5787 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005788
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005789 When doing content switching, it is mandatory that the frontend and the
5790 backend are in the same mode (generally HTTP), otherwise the configuration
5791 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005792
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005793 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005794 defaults http_instances
5795 mode http
5796
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005797 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005798
Willy Tarreau0ba27502007-12-24 16:55:16 +01005799
Cyril Bontéf0c60612010-02-06 14:44:47 +01005800monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005801 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005802 May be used in sections : defaults | frontend | listen | backend
5803 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005804 Arguments :
5805 if <cond> the monitor request will fail if the condition is satisfied,
5806 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005807 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005808 are met, for instance a low number of servers both in a
5809 backend and its backup.
5810
5811 unless <cond> the monitor request will succeed only if the condition is
5812 satisfied, and will fail otherwise. Such a condition may be
5813 based on a test on the presence of a minimum number of active
5814 servers in a list of backends.
5815
5816 This statement adds a condition which can force the response to a monitor
5817 request to report a failure. By default, when an external component queries
5818 the URI dedicated to monitoring, a 200 response is returned. When one of the
5819 conditions above is met, haproxy will return 503 instead of 200. This is
5820 very useful to report a site failure to an external component which may base
5821 routing advertisements between multiple sites on the availability reported by
5822 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005823 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5824 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005825
5826 Example:
5827 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005828 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005829 acl site_dead nbsrv(dynamic) lt 2
5830 acl site_dead nbsrv(static) lt 2
5831 monitor-uri /site_alive
5832 monitor fail if site_dead
5833
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005834 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005835
5836
5837monitor-net <source>
5838 Declare a source network which is limited to monitor requests
5839 May be used in sections : defaults | frontend | listen | backend
5840 yes | yes | yes | no
5841 Arguments :
5842 <source> is the source IPv4 address or network which will only be able to
5843 get monitor responses to any request. It can be either an IPv4
5844 address, a host name, or an address followed by a slash ('/')
5845 followed by a mask.
5846
5847 In TCP mode, any connection coming from a source matching <source> will cause
5848 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005849 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005850 forwarding the connection to a remote server.
5851
5852 In HTTP mode, a connection coming from a source matching <source> will be
5853 accepted, the following response will be sent without waiting for a request,
5854 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5855 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005856 running without forwarding the request to a backend server. Note that this
5857 response is sent in raw format, without any transformation. This is important
5858 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005859
Willy Tarreau82569f92012-09-27 23:48:56 +02005860 Monitor requests are processed very early, just after tcp-request connection
5861 ACLs which are the only ones able to block them. These connections are short
5862 lived and never wait for any data from the client. They cannot be logged, and
5863 it is the intended purpose. They are only used to report HAProxy's health to
5864 an upper component, nothing more. Please note that "monitor fail" rules do
5865 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005866
Willy Tarreau95cd2832010-03-04 23:36:33 +01005867 Last, please note that only one "monitor-net" statement can be specified in
5868 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005869
Willy Tarreau2769aa02007-12-27 18:26:09 +01005870 Example :
5871 # addresses .252 and .253 are just probing us.
5872 frontend www
5873 monitor-net 192.168.0.252/31
5874
5875 See also : "monitor fail", "monitor-uri"
5876
5877
5878monitor-uri <uri>
5879 Intercept a URI used by external components' monitor requests
5880 May be used in sections : defaults | frontend | listen | backend
5881 yes | yes | yes | no
5882 Arguments :
5883 <uri> is the exact URI which we want to intercept to return HAProxy's
5884 health status instead of forwarding the request.
5885
5886 When an HTTP request referencing <uri> will be received on a frontend,
5887 HAProxy will not forward it nor log it, but instead will return either
5888 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5889 conditions defined with "monitor fail". This is normally enough for any
5890 front-end HTTP probe to detect that the service is UP and running without
5891 forwarding the request to a backend server. Note that the HTTP method, the
5892 version and all headers are ignored, but the request must at least be valid
5893 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5894
Willy Tarreau721d8e02017-12-01 18:25:08 +01005895 Monitor requests are processed very early, just after the request is parsed
5896 and even before any "http-request" or "block" rulesets. The only rulesets
5897 applied before are the tcp-request ones. They cannot be logged either, and it
5898 is the intended purpose. They are only used to report HAProxy's health to an
5899 upper component, nothing more. However, it is possible to add any number of
5900 conditions using "monitor fail" and ACLs so that the result can be adjusted
5901 to whatever check can be imagined (most often the number of available servers
5902 in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005903
5904 Example :
5905 # Use /haproxy_test to report haproxy's status
5906 frontend www
5907 mode http
5908 monitor-uri /haproxy_test
5909
5910 See also : "monitor fail", "monitor-net"
5911
Willy Tarreau0ba27502007-12-24 16:55:16 +01005912
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005913option abortonclose
5914no option abortonclose
5915 Enable or disable early dropping of aborted requests pending in queues.
5916 May be used in sections : defaults | frontend | listen | backend
5917 yes | no | yes | yes
5918 Arguments : none
5919
5920 In presence of very high loads, the servers will take some time to respond.
5921 The per-instance connection queue will inflate, and the response time will
5922 increase respective to the size of the queue times the average per-session
5923 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005924 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005925 the queue, and slowing down other users, and the servers as well, because the
5926 request will eventually be served, then aborted at the first error
5927 encountered while delivering the response.
5928
5929 As there is no way to distinguish between a full STOP and a simple output
5930 close on the client side, HTTP agents should be conservative and consider
5931 that the client might only have closed its output channel while waiting for
5932 the response. However, this introduces risks of congestion when lots of users
5933 do the same, and is completely useless nowadays because probably no client at
5934 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01005935 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005936 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005937 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005938 of being the single component to break rare but valid traffic is extremely
5939 low, which adds to the temptation to be able to abort a session early while
5940 still not served and not pollute the servers.
5941
Davor Ocelice9ed2812017-12-25 17:49:28 +01005942 In HAProxy, the user can choose the desired behavior using the option
5943 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005944 compliant and aborted requests will be served. But when the option is
5945 specified, a session with an incoming channel closed will be aborted while
5946 it is still possible, either pending in the queue for a connection slot, or
5947 during the connection establishment if the server has not yet acknowledged
5948 the connection request. This considerably reduces the queue size and the load
5949 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005950 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005951
5952 If this option has been enabled in a "defaults" section, it can be disabled
5953 in a specific instance by prepending the "no" keyword before it.
5954
5955 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5956
5957
Willy Tarreau4076a152009-04-02 15:18:36 +02005958option accept-invalid-http-request
5959no option accept-invalid-http-request
5960 Enable or disable relaxing of HTTP request parsing
5961 May be used in sections : defaults | frontend | listen | backend
5962 yes | yes | yes | no
5963 Arguments : none
5964
Willy Tarreau91852eb2015-05-01 13:26:00 +02005965 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005966 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005967 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005968 forbidden characters are essentially used to build attacks exploiting server
5969 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5970 server will emit invalid header names for whatever reason (configuration,
5971 implementation) and the issue will not be immediately fixed. In such a case,
5972 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005973 even if that does not make sense, by specifying this option. Similarly, the
5974 list of characters allowed to appear in a URI is well defined by RFC3986, and
5975 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5976 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01005977 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005978 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005979 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5980 to pass through (no version specified) and multiple digits for both the major
5981 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005982
5983 This option should never be enabled by default as it hides application bugs
5984 and open security breaches. It should only be deployed after a problem has
5985 been confirmed.
5986
5987 When this option is enabled, erroneous header names will still be accepted in
5988 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005989 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5990 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005991 also helps confirming that the issue has been solved.
5992
5993 If this option has been enabled in a "defaults" section, it can be disabled
5994 in a specific instance by prepending the "no" keyword before it.
5995
5996 See also : "option accept-invalid-http-response" and "show errors" on the
5997 stats socket.
5998
5999
6000option accept-invalid-http-response
6001no option accept-invalid-http-response
6002 Enable or disable relaxing of HTTP response parsing
6003 May be used in sections : defaults | frontend | listen | backend
6004 yes | no | yes | yes
6005 Arguments : none
6006
Willy Tarreau91852eb2015-05-01 13:26:00 +02006007 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02006008 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01006009 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02006010 forbidden characters are essentially used to build attacks exploiting server
6011 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
6012 server will emit invalid header names for whatever reason (configuration,
6013 implementation) and the issue will not be immediately fixed. In such a case,
6014 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02006015 even if that does not make sense, by specifying this option. This option also
6016 relaxes the test on the HTTP version format, it allows multiple digits for
6017 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02006018
6019 This option should never be enabled by default as it hides application bugs
6020 and open security breaches. It should only be deployed after a problem has
6021 been confirmed.
6022
6023 When this option is enabled, erroneous header names will still be accepted in
6024 responses, but the complete response will be captured in order to permit
6025 later analysis using the "show errors" request on the UNIX stats socket.
6026 Doing this also helps confirming that the issue has been solved.
6027
6028 If this option has been enabled in a "defaults" section, it can be disabled
6029 in a specific instance by prepending the "no" keyword before it.
6030
6031 See also : "option accept-invalid-http-request" and "show errors" on the
6032 stats socket.
6033
6034
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006035option allbackups
6036no option allbackups
6037 Use either all backup servers at a time or only the first one
6038 May be used in sections : defaults | frontend | listen | backend
6039 yes | no | yes | yes
6040 Arguments : none
6041
6042 By default, the first operational backup server gets all traffic when normal
6043 servers are all down. Sometimes, it may be preferred to use multiple backups
6044 at once, because one will not be enough. When "option allbackups" is enabled,
6045 the load balancing will be performed among all backup servers when all normal
6046 ones are unavailable. The same load balancing algorithm will be used and the
6047 servers' weights will be respected. Thus, there will not be any priority
6048 order between the backup servers anymore.
6049
6050 This option is mostly used with static server farms dedicated to return a
6051 "sorry" page when an application is completely offline.
6052
6053 If this option has been enabled in a "defaults" section, it can be disabled
6054 in a specific instance by prepending the "no" keyword before it.
6055
6056
6057option checkcache
6058no option checkcache
Godbach7056a352013-12-11 20:01:07 +08006059 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006060 May be used in sections : defaults | frontend | listen | backend
6061 yes | no | yes | yes
6062 Arguments : none
6063
6064 Some high-level frameworks set application cookies everywhere and do not
6065 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006066 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006067 high risk of session crossing or stealing between users traversing the same
6068 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02006069 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006070
6071 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006072 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01006073 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006074 response to check if there's a risk of caching a cookie on a client-side
6075 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01006076 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01006077 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01006078 - all those with a return code other than 200, 203, 204, 206, 300, 301,
6079 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01006080 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01006081 - all those that result from a request using a method other than GET, HEAD,
6082 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01006083 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006084 - those with a 'Pragma: no-cache' header
6085 - those with a 'Cache-control: private' header
6086 - those with a 'Cache-control: no-store' header
6087 - those with a 'Cache-control: max-age=0' header
6088 - those with a 'Cache-control: s-maxage=0' header
6089 - those with a 'Cache-control: no-cache' header
6090 - those with a 'Cache-control: no-cache="set-cookie"' header
6091 - those with a 'Cache-control: no-cache="set-cookie,' header
6092 (allowing other fields after set-cookie)
6093
6094 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01006095 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006096 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006097 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006098 that admins are informed that there's something to be fixed.
6099
6100 Due to the high impact on the application, the application should be tested
6101 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006102 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01006103 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006104
6105 If this option has been enabled in a "defaults" section, it can be disabled
6106 in a specific instance by prepending the "no" keyword before it.
6107
6108
6109option clitcpka
6110no option clitcpka
6111 Enable or disable the sending of TCP keepalive packets on the client side
6112 May be used in sections : defaults | frontend | listen | backend
6113 yes | yes | yes | no
6114 Arguments : none
6115
6116 When there is a firewall or any session-aware component between a client and
6117 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006118 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006119 components decides to expire a session which has remained idle for too long.
6120
6121 Enabling socket-level TCP keep-alives makes the system regularly send packets
6122 to the other end of the connection, leaving it active. The delay between
6123 keep-alive probes is controlled by the system only and depends both on the
6124 operating system and its tuning parameters.
6125
6126 It is important to understand that keep-alive packets are neither emitted nor
6127 received at the application level. It is only the network stacks which sees
6128 them. For this reason, even if one side of the proxy already uses keep-alives
6129 to maintain its connection alive, those keep-alive packets will not be
6130 forwarded to the other side of the proxy.
6131
6132 Please note that this has nothing to do with HTTP keep-alive.
6133
6134 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
6135 client side of a connection, which should help when session expirations are
6136 noticed between HAProxy and a client.
6137
6138 If this option has been enabled in a "defaults" section, it can be disabled
6139 in a specific instance by prepending the "no" keyword before it.
6140
6141 See also : "option srvtcpka", "option tcpka"
6142
6143
Willy Tarreau0ba27502007-12-24 16:55:16 +01006144option contstats
6145 Enable continuous traffic statistics updates
6146 May be used in sections : defaults | frontend | listen | backend
6147 yes | yes | yes | no
6148 Arguments : none
6149
6150 By default, counters used for statistics calculation are incremented
6151 only when a session finishes. It works quite well when serving small
6152 objects, but with big ones (for example large images or archives) or
6153 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01006154 a hedgehog. With this option enabled counters get incremented frequently
6155 along the session, typically every 5 seconds, which is often enough to
6156 produce clean graphs. Recounting touches a hotpath directly so it is not
6157 not enabled by default, as it can cause a lot of wakeups for very large
6158 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01006159
6160
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006161option dontlog-normal
6162no option dontlog-normal
6163 Enable or disable logging of normal, successful connections
6164 May be used in sections : defaults | frontend | listen | backend
6165 yes | yes | yes | no
6166 Arguments : none
6167
6168 There are large sites dealing with several thousand connections per second
6169 and for which logging is a major pain. Some of them are even forced to turn
6170 logs off and cannot debug production issues. Setting this option ensures that
6171 normal connections, those which experience no error, no timeout, no retry nor
6172 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
6173 mode, the response status code is checked and return codes 5xx will still be
6174 logged.
6175
6176 It is strongly discouraged to use this option as most of the time, the key to
6177 complex issues is in the normal logs which will not be logged here. If you
6178 need to separate logs, see the "log-separate-errors" option instead.
6179
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006180 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006181 logging.
6182
6183
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006184option dontlognull
6185no option dontlognull
6186 Enable or disable logging of null connections
6187 May be used in sections : defaults | frontend | listen | backend
6188 yes | yes | yes | no
6189 Arguments : none
6190
6191 In certain environments, there are components which will regularly connect to
6192 various systems to ensure that they are still alive. It can be the case from
6193 another load balancer as well as from monitoring systems. By default, even a
6194 simple port probe or scan will produce a log. If those connections pollute
6195 the logs too much, it is possible to enable option "dontlognull" to indicate
6196 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02006197 which typically corresponds to those probes. Note that errors will still be
6198 returned to the client and accounted for in the stats. If this is not what is
6199 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006200
6201 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01006202 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006203 would not be logged.
6204
6205 If this option has been enabled in a "defaults" section, it can be disabled
6206 in a specific instance by prepending the "no" keyword before it.
6207
Willy Tarreau0f228a02015-05-01 15:37:53 +02006208 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
6209 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006210
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006211
Willy Tarreau87cf5142011-08-19 22:57:24 +02006212option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006213 Enable insertion of the X-Forwarded-For header to requests sent to servers
6214 May be used in sections : defaults | frontend | listen | backend
6215 yes | yes | yes | yes
6216 Arguments :
6217 <network> is an optional argument used to disable this option for sources
6218 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02006219 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01006220 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006221
6222 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
6223 their client address. This is sometimes annoying when the client's IP address
6224 is expected in server logs. To solve this problem, the well-known HTTP header
6225 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
6226 This header contains a value representing the client's IP address. Since this
6227 header is always appended at the end of the existing header list, the server
6228 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02006229 the server's manual to find how to enable use of this standard header. Note
6230 that only the last occurrence of the header must be used, since it is really
6231 possible that the client has already brought one.
6232
Willy Tarreaud72758d2010-01-12 10:42:19 +01006233 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02006234 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01006235 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01006236 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006237 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02006238 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01006239
6240 Sometimes, a same HAProxy instance may be shared between a direct client
6241 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6242 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6243 header for a known source address or network by adding the "except" keyword
6244 followed by the network address. In this case, any source IP matching the
6245 network will not cause an addition of this header. Most common uses are with
6246 private networks or 127.0.0.1.
6247
Willy Tarreau87cf5142011-08-19 22:57:24 +02006248 Alternatively, the keyword "if-none" states that the header will only be
6249 added if it is not present. This should only be used in perfectly trusted
6250 environment, as this might cause a security issue if headers reaching haproxy
6251 are under the control of the end-user.
6252
Willy Tarreauc27debf2008-01-06 08:57:02 +01006253 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02006254 least one of them uses it, the header will be added. Note that the backend's
6255 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02006256 both are defined. In the case of the "if-none" argument, if at least one of
6257 the frontend or the backend does not specify it, it wants the addition to be
6258 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006259
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02006260 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01006261 # Public HTTP address also used by stunnel on the same machine
6262 frontend www
6263 mode http
6264 option forwardfor except 127.0.0.1 # stunnel already adds the header
6265
Ross Westaf72a1d2008-08-03 10:51:45 +02006266 # Those servers want the IP Address in X-Client
6267 backend www
6268 mode http
6269 option forwardfor header X-Client
6270
Willy Tarreau87cf5142011-08-19 22:57:24 +02006271 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006272 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01006273
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006274
Christopher Fauleta99ff4d2019-07-22 16:18:24 +02006275option h1-case-adjust-bogus-client
6276no option h1-case-adjust-bogus-client
6277 Enable or disable the case adjustment of HTTP/1 headers sent to bogus clients
6278 May be used in sections : defaults | frontend | listen | backend
6279 yes | yes | yes | no
6280 Arguments : none
6281
6282 There is no standard case for header names because, as stated in RFC7230,
6283 they are case-insensitive. So applications must handle them in a case-
6284 insensitive manner. But some bogus applications violate the standards and
6285 erroneously rely on the cases most commonly used by browsers. This problem
6286 becomes critical with HTTP/2 because all header names must be exchanged in
6287 lower case, and HAProxy follows the same convention. All header names are
6288 sent in lower case to clients and servers, regardless of the HTTP version.
6289
6290 When HAProxy receives an HTTP/1 response, its header names are converted to
6291 lower case and manipulated and sent this way to the clients. If a client is
6292 known to violate the HTTP standards and to fail to process a response coming
6293 from HAProxy, it is possible to transform the lower case header names to a
6294 different format when the response is formatted and sent to the client, by
6295 enabling this option and specifying the list of headers to be reformatted
6296 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6297 must only be a temporary workaround for the time it takes the client to be
6298 fixed, because clients which require such workarounds might be vulnerable to
6299 content smuggling attacks and must absolutely be fixed.
6300
6301 Please note that this option will not affect standards-compliant clients.
6302
6303 If this option has been enabled in a "defaults" section, it can be disabled
6304 in a specific instance by prepending the "no" keyword before it.
6305
6306 See also: "option h1-case-adjust-bogus-server", "h1-case-adjust",
6307 "h1-case-adjust-file".
6308
6309
6310option h1-case-adjust-bogus-server
6311no option h1-case-adjust-bogus-server
6312 Enable or disable the case adjustment of HTTP/1 headers sent to bogus servers
6313 May be used in sections : defaults | frontend | listen | backend
6314 yes | no | yes | yes
6315 Arguments : none
6316
6317 There is no standard case for header names because, as stated in RFC7230,
6318 they are case-insensitive. So applications must handle them in a case-
6319 insensitive manner. But some bogus applications violate the standards and
6320 erroneously rely on the cases most commonly used by browsers. This problem
6321 becomes critical with HTTP/2 because all header names must be exchanged in
6322 lower case, and HAProxy follows the same convention. All header names are
6323 sent in lower case to clients and servers, regardless of the HTTP version.
6324
6325 When HAProxy receives an HTTP/1 request, its header names are converted to
6326 lower case and manipulated and sent this way to the servers. If a server is
6327 known to violate the HTTP standards and to fail to process a request coming
6328 from HAProxy, it is possible to transform the lower case header names to a
6329 different format when the request is formatted and sent to the server, by
6330 enabling this option and specifying the list of headers to be reformatted
6331 using the global directives "h1-case-adjust" or "h1-case-adjust-file". This
6332 must only be a temporary workaround for the time it takes the server to be
6333 fixed, because servers which require such workarounds might be vulnerable to
6334 content smuggling attacks and must absolutely be fixed.
6335
6336 Please note that this option will not affect standards-compliant servers.
6337
6338 If this option has been enabled in a "defaults" section, it can be disabled
6339 in a specific instance by prepending the "no" keyword before it.
6340
6341 See also: "option h1-case-adjust-bogus-client", "h1-case-adjust",
6342 "h1-case-adjust-file".
6343
6344
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006345option http-buffer-request
6346no option http-buffer-request
6347 Enable or disable waiting for whole HTTP request body before proceeding
6348 May be used in sections : defaults | frontend | listen | backend
6349 yes | yes | yes | yes
6350 Arguments : none
6351
6352 It is sometimes desirable to wait for the body of an HTTP request before
6353 taking a decision. This is what is being done by "balance url_param" for
6354 example. The first use case is to buffer requests from slow clients before
6355 connecting to the server. Another use case consists in taking the routing
6356 decision based on the request body's contents. This option placed in a
6357 frontend or backend forces the HTTP processing to wait until either the whole
6358 body is received, or the request buffer is full, or the first chunk is
6359 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01006360 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006361 the frontend and the backend, so this should definitely not be used by
6362 default.
6363
Baptiste Assmanneccdf432015-10-28 13:49:01 +01006364 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006365
6366
Willy Tarreau0f228a02015-05-01 15:37:53 +02006367option http-ignore-probes
6368no option http-ignore-probes
6369 Enable or disable logging of null connections and request timeouts
6370 May be used in sections : defaults | frontend | listen | backend
6371 yes | yes | yes | no
6372 Arguments : none
6373
6374 Recently some browsers started to implement a "pre-connect" feature
6375 consisting in speculatively connecting to some recently visited web sites
6376 just in case the user would like to visit them. This results in many
6377 connections being established to web sites, which end up in 408 Request
6378 Timeout if the timeout strikes first, or 400 Bad Request when the browser
6379 decides to close them first. These ones pollute the log and feed the error
6380 counters. There was already "option dontlognull" but it's insufficient in
6381 this case. Instead, this option does the following things :
6382 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01006383 was received over a connection before it was closed;
6384 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02006385 - prevent any error counter from being incremented
6386
6387 That way the empty connection is silently ignored. Note that it is better
6388 not to use this unless it is clear that it is needed, because it will hide
6389 real problems. The most common reason for not receiving a request and seeing
6390 a 408 is due to an MTU inconsistency between the client and an intermediary
6391 element such as a VPN, which blocks too large packets. These issues are
6392 generally seen with POST requests as well as GET with large cookies. The logs
6393 are often the only way to detect them.
6394
6395 If this option has been enabled in a "defaults" section, it can be disabled
6396 in a specific instance by prepending the "no" keyword before it.
6397
6398 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
6399
6400
Willy Tarreau16bfb022010-01-16 19:48:41 +01006401option http-keep-alive
6402no option http-keep-alive
6403 Enable or disable HTTP keep-alive from client to server
6404 May be used in sections : defaults | frontend | listen | backend
6405 yes | yes | yes | yes
6406 Arguments : none
6407
Willy Tarreau70dffda2014-01-30 03:07:23 +01006408 By default HAProxy operates in keep-alive mode with regards to persistent
6409 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02006410 leaves the connection idle on both sides between the end of a response and
6411 the start of a new request. This mode may be changed by several options such
6412 as "option http-server-close", "option httpclose" or "option http-tunnel".
6413 This option allows to set back the keep-alive mode, which can be useful when
6414 another mode was used in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01006415
6416 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
6417 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01006418 network) and the fastest session reuse on the server side at the expense
6419 of maintaining idle connections to the servers. In general, it is possible
6420 with this option to achieve approximately twice the request rate that the
6421 "http-server-close" option achieves on small objects. There are mainly two
6422 situations where this option may be useful :
6423
6424 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01006425 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01006426
6427 - when the cost of establishing the connection to the server is significant
6428 compared to the cost of retrieving the associated object from the server.
6429
6430 This last case can happen when the server is a fast static server of cache.
6431 In this case, the server will need to be properly tuned to support high enough
6432 connection counts because connections will last until the client sends another
6433 request.
6434
6435 If the client request has to go to another backend or another server due to
6436 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01006437 immediately be closed and a new one re-opened. Option "prefer-last-server" is
6438 available to try optimize server selection so that if the server currently
6439 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01006440
Willy Tarreau16bfb022010-01-16 19:48:41 +01006441 At the moment, logs will not indicate whether requests came from the same
6442 session or not. The accept date reported in the logs corresponds to the end
6443 of the previous request, and the request time corresponds to the time spent
6444 waiting for a new request. The keep-alive request time is still bound to the
6445 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6446 not set.
6447
Cyril Bonté653dcd62014-02-20 00:13:15 +01006448 This option disables and replaces any previous "option httpclose", "option
Christopher Faulet315b39c2018-09-21 16:26:19 +02006449 http-server-close" or "option http-tunnel". When backend and frontend options
6450 differ, all of these 4 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006451
Christopher Faulet315b39c2018-09-21 16:26:19 +02006452 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01006453 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01006454 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006455
6456
Willy Tarreau96e31212011-05-30 18:10:30 +02006457option http-no-delay
6458no option http-no-delay
6459 Instruct the system to favor low interactive delays over performance in HTTP
6460 May be used in sections : defaults | frontend | listen | backend
6461 yes | yes | yes | yes
6462 Arguments : none
6463
6464 In HTTP, each payload is unidirectional and has no notion of interactivity.
6465 Any agent is expected to queue data somewhat for a reasonably low delay.
6466 There are some very rare server-to-server applications that abuse the HTTP
6467 protocol and expect the payload phase to be highly interactive, with many
6468 interleaved data chunks in both directions within a single request. This is
6469 absolutely not supported by the HTTP specification and will not work across
6470 most proxies or servers. When such applications attempt to do this through
6471 haproxy, it works but they will experience high delays due to the network
6472 optimizations which favor performance by instructing the system to wait for
6473 enough data to be available in order to only send full packets. Typical
6474 delays are around 200 ms per round trip. Note that this only happens with
6475 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6476 affected.
6477
6478 When "option http-no-delay" is present in either the frontend or the backend
6479 used by a connection, all such optimizations will be disabled in order to
6480 make the exchanges as fast as possible. Of course this offers no guarantee on
6481 the functionality, as it may break at any other place. But if it works via
6482 HAProxy, it will work as fast as possible. This option should never be used
6483 by default, and should never be used at all unless such a buggy application
6484 is discovered. The impact of using this option is an increase of bandwidth
6485 usage and CPU usage, which may significantly lower performance in high
6486 latency environments.
6487
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006488 See also : "option http-buffer-request"
6489
Willy Tarreau96e31212011-05-30 18:10:30 +02006490
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006491option http-pretend-keepalive
6492no option http-pretend-keepalive
6493 Define whether haproxy will announce keepalive to the server or not
6494 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006495 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006496 Arguments : none
6497
Christopher Faulet315b39c2018-09-21 16:26:19 +02006498 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006499 adds a "Connection: close" header to the request forwarded to the server.
6500 Unfortunately, when some servers see this header, they automatically refrain
6501 from using the chunked encoding for responses of unknown length, while this
6502 is totally unrelated. The immediate effect is that this prevents haproxy from
6503 maintaining the client connection alive. A second effect is that a client or
6504 a cache could receive an incomplete response without being aware of it, and
6505 consider the response complete.
6506
6507 By setting "option http-pretend-keepalive", haproxy will make the server
6508 believe it will keep the connection alive. The server will then not fall back
6509 to the abnormal undesired above. When haproxy gets the whole response, it
6510 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006511 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006512 connection is correctly closed on the server side.
6513
6514 It is recommended not to enable this option by default, because most servers
6515 will more efficiently close the connection themselves after the last packet,
6516 and release its buffers slightly earlier. Also, the added packet on the
6517 network could slightly reduce the overall peak performance. However it is
6518 worth noting that when this option is enabled, haproxy will have slightly
6519 less work to do. So if haproxy is the bottleneck on the whole architecture,
6520 enabling this option might save a few CPU cycles.
6521
Christopher Faulet98db9762018-09-21 10:25:19 +02006522 This option may be set in backend and listen sections. Using it in a frontend
6523 section will be ignored and a warning will be reported during startup. It is
6524 a backend related option, so there is no real reason to set it on a
6525 frontend. This option may be combined with "option httpclose", which will
6526 cause keepalive to be announced to the server and close to be announced to
6527 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006528
6529 If this option has been enabled in a "defaults" section, it can be disabled
6530 in a specific instance by prepending the "no" keyword before it.
6531
Christopher Faulet315b39c2018-09-21 16:26:19 +02006532 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01006533 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006534
Willy Tarreauc27debf2008-01-06 08:57:02 +01006535
Willy Tarreaub608feb2010-01-02 22:47:18 +01006536option http-server-close
6537no option http-server-close
6538 Enable or disable HTTP connection closing on the server side
6539 May be used in sections : defaults | frontend | listen | backend
6540 yes | yes | yes | yes
6541 Arguments : none
6542
Willy Tarreau70dffda2014-01-30 03:07:23 +01006543 By default HAProxy operates in keep-alive mode with regards to persistent
6544 connections: for each connection it processes each request and response, and
6545 leaves the connection idle on both sides between the end of a response and
6546 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006547 as "option http-server-close", "option httpclose" or "option http-tunnel".
6548 Setting "option http-server-close" enables HTTP connection-close mode on the
6549 server side while keeping the ability to support HTTP keep-alive and
6550 pipelining on the client side. This provides the lowest latency on the client
6551 side (slow network) and the fastest session reuse on the server side to save
6552 server resources, similarly to "option httpclose". It also permits
6553 non-keepalive capable servers to be served in keep-alive mode to the clients
6554 if they conform to the requirements of RFC7230. Please note that some servers
6555 do not always conform to those requirements when they see "Connection: close"
6556 in the request. The effect will be that keep-alive will never be used. A
6557 workaround consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006558
6559 At the moment, logs will not indicate whether requests came from the same
6560 session or not. The accept date reported in the logs corresponds to the end
6561 of the previous request, and the request time corresponds to the time spent
6562 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01006563 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6564 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006565
6566 This option may be set both in a frontend and in a backend. It is enabled if
6567 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet315b39c2018-09-21 16:26:19 +02006568 It disables and replaces any previous "option httpclose", "option http-tunnel"
6569 or "option http-keep-alive". Please check section 4 ("Proxies") to see how
6570 this option combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006571
6572 If this option has been enabled in a "defaults" section, it can be disabled
6573 in a specific instance by prepending the "no" keyword before it.
6574
Christopher Faulet315b39c2018-09-21 16:26:19 +02006575 See also : "option httpclose", "option http-pretend-keepalive",
6576 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006577
6578
Christopher Faulet6c9bbb22019-03-26 21:37:23 +01006579option http-tunnel (deprecated)
6580no option http-tunnel (deprecated)
6581 Disable or enable HTTP connection processing after first transaction.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006582 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet4212a302018-09-21 10:42:19 +02006583 yes | yes | yes | no
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006584 Arguments : none
6585
Christopher Faulet6c9bbb22019-03-26 21:37:23 +01006586 Warning : Because it cannot work in HTTP/2, this option is deprecated and it
6587 is only supported on legacy HTTP frontends. In HTX, it is ignored and a
6588 warning is emitted during HAProxy startup.
6589
Willy Tarreau70dffda2014-01-30 03:07:23 +01006590 By default HAProxy operates in keep-alive mode with regards to persistent
6591 connections: for each connection it processes each request and response, and
6592 leaves the connection idle on both sides between the end of a response and
6593 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006594 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006595
6596 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006597 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01006598 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
6599 is normally not needed anymore unless in very specific cases such as when
6600 using an in-house protocol that looks like HTTP but is not compatible, or
6601 just to log one request per client in order to reduce log size. Note that
6602 everything which works at the HTTP level, including header parsing/addition,
6603 cookie processing or content switching will only work for the first request
6604 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006605
Christopher Faulet4212a302018-09-21 10:42:19 +02006606 This option may be set on frontend and listen sections. Using it on a backend
6607 section will be ignored and a warning will be reported during the startup. It
6608 is a frontend related option, so there is no real reason to set it on a
6609 backend.
6610
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006611 If this option has been enabled in a "defaults" section, it can be disabled
6612 in a specific instance by prepending the "no" keyword before it.
6613
Christopher Faulet315b39c2018-09-21 16:26:19 +02006614 See also : "option httpclose", "option http-server-close",
6615 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006616
6617
Willy Tarreau88d349d2010-01-25 12:15:43 +01006618option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01006619no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01006620 Make use of non-standard Proxy-Connection header instead of Connection
6621 May be used in sections : defaults | frontend | listen | backend
6622 yes | yes | yes | no
6623 Arguments : none
6624
Lukas Tribus23953682017-04-28 13:24:30 +00006625 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006626 Connection header to indicate their wish of persistent or non-persistent
6627 connections, both browsers and proxies ignore this header for proxied
6628 connections and make use of the undocumented, non-standard Proxy-Connection
6629 header instead. The issue begins when trying to put a load balancer between
6630 browsers and such proxies, because there will be a difference between what
6631 haproxy understands and what the client and the proxy agree on.
6632
6633 By setting this option in a frontend, haproxy can automatically switch to use
6634 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006635 defined here as one where the URI begins with neither a '/' nor a '*'. This
6636 is incompatible with the HTTP tunnel mode. Note that this option can only be
6637 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006638
Willy Tarreau844a7e72010-01-31 21:46:18 +01006639 Also, when this option is set, a request which requires authentication will
6640 automatically switch to use proxy authentication headers if it is itself a
6641 proxied request. That makes it possible to check or enforce authentication in
6642 front of an existing proxy.
6643
Willy Tarreau88d349d2010-01-25 12:15:43 +01006644 This option should normally never be used, except in front of a proxy.
6645
Christopher Faulet315b39c2018-09-21 16:26:19 +02006646 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006647
6648
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006649option http-use-htx
6650no option http-use-htx
6651 Switch to the new HTX internal representation for HTTP protocol elements
6652 May be used in sections : defaults | frontend | listen | backend
6653 yes | yes | yes | yes
6654 Arguments : none
6655
Christopher Faulet1d2b5862019-04-12 16:10:51 +02006656 Historically, the HTTP protocol is processed as-is. Inserting, deleting, or
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006657 modifying a header field requires to rewrite the affected part in the buffer
Christopher Faulet1d2b5862019-04-12 16:10:51 +02006658 and to move the buffer's tail accordingly. This mode is known as the legacy
6659 HTTP mode. Since this principle has deep roots in haproxy, the HTTP/2
6660 protocol is converted to HTTP/1.1 before being processed this way. It also
6661 results in the inability to establish HTTP/2 connections to servers because
6662 of the loss of HTTP/2 semantics in the HTTP/1 representation.
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006663
6664 HTX is the name of a totally new native internal representation for the HTTP
6665 protocol, that is agnostic to the version and aims at preserving semantics
6666 all along the chain. It relies on a fast parsing, tokenizing and indexing of
6667 the protocol elements so that no more memory moves are necessary and that
Christopher Faulet1d2b5862019-04-12 16:10:51 +02006668 most elements are directly accessed. It supports using either HTTP/1 or
6669 HTTP/2 on any side regardless of the other side's version. It also supports
6670 upgrades from TCP to HTTP and implicit ones from HTTP/1 to HTTP/2 (matching
6671 the HTTP/2 preface).
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006672
Christopher Faulet1d2b5862019-04-12 16:10:51 +02006673 This option indicates that HTX needs to be used. Since the version 2.0-dev3,
6674 the HTX is the default mode. To switch back on the legacy HTTP mode, the
6675 option must be explicitly disabled using the "no" prefix. For prior versions,
6676 the feature has incomplete functional coverage, so it is not enabled by
6677 default.
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006678
6679 See also : "mode http"
6680
6681
Willy Tarreaud63335a2010-02-26 12:56:52 +01006682option httpchk
6683option httpchk <uri>
6684option httpchk <method> <uri>
6685option httpchk <method> <uri> <version>
6686 Enable HTTP protocol to check on the servers health
6687 May be used in sections : defaults | frontend | listen | backend
6688 yes | no | yes | yes
6689 Arguments :
6690 <method> is the optional HTTP method used with the requests. When not set,
6691 the "OPTIONS" method is used, as it generally requires low server
6692 processing and is easy to filter out from the logs. Any method
6693 may be used, though it is not recommended to invent non-standard
6694 ones.
6695
6696 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6697 which is accessible by default on almost any server, but may be
6698 changed to any other URI. Query strings are permitted.
6699
6700 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6701 but some servers might behave incorrectly in HTTP 1.0, so turning
6702 it to HTTP/1.1 may sometimes help. Note that the Host field is
6703 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
6704 after "\r\n" following the version string.
6705
6706 By default, server health checks only consist in trying to establish a TCP
6707 connection. When "option httpchk" is specified, a complete HTTP request is
6708 sent once the TCP connection is established, and responses 2xx and 3xx are
6709 considered valid, while all other ones indicate a server failure, including
6710 the lack of any response.
6711
6712 The port and interval are specified in the server configuration.
6713
6714 This option does not necessarily require an HTTP backend, it also works with
6715 plain TCP backends. This is particularly useful to check simple scripts bound
6716 to some dedicated ports using the inetd daemon.
6717
6718 Examples :
6719 # Relay HTTPS traffic to Apache instance and check service availability
6720 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6721 backend https_relay
6722 mode tcp
6723 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6724 server apache1 192.168.1.1:443 check port 80
6725
Simon Hormanafc47ee2013-11-25 10:46:35 +09006726 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6727 "option pgsql-check", "http-check" and the "check", "port" and
6728 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006729
6730
Willy Tarreauc27debf2008-01-06 08:57:02 +01006731option httpclose
6732no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006733 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006734 May be used in sections : defaults | frontend | listen | backend
6735 yes | yes | yes | yes
6736 Arguments : none
6737
Willy Tarreau70dffda2014-01-30 03:07:23 +01006738 By default HAProxy operates in keep-alive mode with regards to persistent
6739 connections: for each connection it processes each request and response, and
6740 leaves the connection idle on both sides between the end of a response and
6741 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006742 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006743
Christopher Faulet315b39c2018-09-21 16:26:19 +02006744 If "option httpclose" is set, HAProxy will close connections with the server
6745 and the client as soon as the request and the response are received. It will
John Roesler27c754c2019-07-10 15:45:51 -05006746 also check if a "Connection: close" header is already set in each direction,
Christopher Faulet315b39c2018-09-21 16:26:19 +02006747 and will add one if missing. Any "Connection" header different from "close"
6748 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006749
Christopher Faulet315b39c2018-09-21 16:26:19 +02006750 This option may also be combined with "option http-pretend-keepalive", which
6751 will disable sending of the "Connection: close" header, but will still cause
6752 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006753
6754 This option may be set both in a frontend and in a backend. It is enabled if
6755 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01006756 It disables and replaces any previous "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006757 "option http-keep-alive" or "option http-tunnel". Please check section 4
6758 ("Proxies") to see how this option combines with others when frontend and
6759 backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006760
6761 If this option has been enabled in a "defaults" section, it can be disabled
6762 in a specific instance by prepending the "no" keyword before it.
6763
Christopher Faulet315b39c2018-09-21 16:26:19 +02006764 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006765
6766
Emeric Brun3a058f32009-06-30 18:26:00 +02006767option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006768 Enable logging of HTTP request, session state and timers
6769 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006770 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006771 Arguments :
6772 clf if the "clf" argument is added, then the output format will be
6773 the CLF format instead of HAProxy's default HTTP format. You can
6774 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006775 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006776 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006777
6778 By default, the log output format is very poor, as it only contains the
6779 source and destination addresses, and the instance name. By specifying
6780 "option httplog", each log line turns into a much richer format including,
6781 but not limited to, the HTTP request, the connection timers, the session
6782 status, the connections numbers, the captured headers and cookies, the
6783 frontend, backend and server name, and of course the source address and
6784 ports.
6785
PiBa-NLbd556bf2014-12-11 21:31:54 +01006786 Specifying only "option httplog" will automatically clear the 'clf' mode
6787 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006788
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006789 "option httplog" overrides any previous "log-format" directive.
6790
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006791 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006792
Willy Tarreau55165fe2009-05-10 12:02:55 +02006793
6794option http_proxy
6795no option http_proxy
6796 Enable or disable plain HTTP proxy mode
6797 May be used in sections : defaults | frontend | listen | backend
6798 yes | yes | yes | yes
6799 Arguments : none
6800
6801 It sometimes happens that people need a pure HTTP proxy which understands
6802 basic proxy requests without caching nor any fancy feature. In this case,
6803 it may be worth setting up an HAProxy instance with the "option http_proxy"
6804 set. In this mode, no server is declared, and the connection is forwarded to
6805 the IP address and port found in the URL after the "http://" scheme.
6806
6807 No host address resolution is performed, so this only works when pure IP
6808 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006809 it will probably be used only by experts who know they need exactly it. This
6810 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006811
6812 If this option has been enabled in a "defaults" section, it can be disabled
6813 in a specific instance by prepending the "no" keyword before it.
6814
6815 Example :
6816 # this backend understands HTTP proxy requests and forwards them directly.
6817 backend direct_forward
6818 option httpclose
6819 option http_proxy
6820
6821 See also : "option httpclose"
6822
Willy Tarreau211ad242009-10-03 21:45:07 +02006823
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006824option independent-streams
6825no option independent-streams
6826 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006827 May be used in sections : defaults | frontend | listen | backend
6828 yes | yes | yes | yes
6829 Arguments : none
6830
6831 By default, when data is sent over a socket, both the write timeout and the
6832 read timeout for that socket are refreshed, because we consider that there is
6833 activity on that socket, and we have no other means of guessing if we should
6834 receive data or not.
6835
Davor Ocelice9ed2812017-12-25 17:49:28 +01006836 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006837 exists a situation where it is desirable to disable it, and only refresh the
6838 read timeout if there are incoming data. This happens on sessions with large
6839 timeouts and low amounts of exchanged data such as telnet session. If the
6840 server suddenly disappears, the output data accumulates in the system's
6841 socket buffers, both timeouts are correctly refreshed, and there is no way
6842 to know the server does not receive them, so we don't timeout. However, when
6843 the underlying protocol always echoes sent data, it would be enough by itself
6844 to detect the issue using the read timeout. Note that this problem does not
6845 happen with more verbose protocols because data won't accumulate long in the
6846 socket buffers.
6847
6848 When this option is set on the frontend, it will disable read timeout updates
6849 on data sent to the client. There probably is little use of this case. When
6850 the option is set on the backend, it will disable read timeout updates on
6851 data sent to the server. Doing so will typically break large HTTP posts from
6852 slow lines, so use it with caution.
6853
Lukas Tribus745f15e2018-11-08 12:41:42 +01006854 Note: older versions used to call this setting "option independant-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006855 with a spelling mistake. This spelling is still supported but
6856 deprecated.
6857
Willy Tarreauce887fd2012-05-12 12:50:00 +02006858 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006859
6860
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006861option ldap-check
6862 Use LDAPv3 health checks for server testing
6863 May be used in sections : defaults | frontend | listen | backend
6864 yes | no | yes | yes
6865 Arguments : none
6866
6867 It is possible to test that the server correctly talks LDAPv3 instead of just
6868 testing that it accepts the TCP connection. When this option is set, an
6869 LDAPv3 anonymous simple bind message is sent to the server, and the response
6870 is analyzed to find an LDAPv3 bind response message.
6871
6872 The server is considered valid only when the LDAP response contains success
6873 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6874
6875 Logging of bind requests is server dependent see your documentation how to
6876 configure it.
6877
6878 Example :
6879 option ldap-check
6880
6881 See also : "option httpchk"
6882
6883
Simon Horman98637e52014-06-20 12:30:16 +09006884option external-check
6885 Use external processes for server health checks
6886 May be used in sections : defaults | frontend | listen | backend
6887 yes | no | yes | yes
6888
6889 It is possible to test the health of a server using an external command.
6890 This is achieved by running the executable set using "external-check
6891 command".
6892
6893 Requires the "external-check" global to be set.
6894
6895 See also : "external-check", "external-check command", "external-check path"
6896
6897
Willy Tarreau211ad242009-10-03 21:45:07 +02006898option log-health-checks
6899no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006900 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006901 May be used in sections : defaults | frontend | listen | backend
6902 yes | no | yes | yes
6903 Arguments : none
6904
Willy Tarreaubef1b322014-05-13 21:01:39 +02006905 By default, failed health check are logged if server is UP and successful
6906 health checks are logged if server is DOWN, so the amount of additional
6907 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006908
Willy Tarreaubef1b322014-05-13 21:01:39 +02006909 When this option is enabled, any change of the health check status or to
6910 the server's health will be logged, so that it becomes possible to know
6911 that a server was failing occasional checks before crashing, or exactly when
6912 it failed to respond a valid HTTP status, then when the port started to
6913 reject connections, then when the server stopped responding at all.
6914
Davor Ocelice9ed2812017-12-25 17:49:28 +01006915 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006916 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006917
Willy Tarreaubef1b322014-05-13 21:01:39 +02006918 See also: "option httpchk", "option ldap-check", "option mysql-check",
6919 "option pgsql-check", "option redis-check", "option smtpchk",
6920 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006921
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006922
6923option log-separate-errors
6924no option log-separate-errors
6925 Change log level for non-completely successful connections
6926 May be used in sections : defaults | frontend | listen | backend
6927 yes | yes | yes | no
6928 Arguments : none
6929
6930 Sometimes looking for errors in logs is not easy. This option makes haproxy
6931 raise the level of logs containing potentially interesting information such
6932 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6933 level changes from "info" to "err". This makes it possible to log them
6934 separately to a different file with most syslog daemons. Be careful not to
6935 remove them from the original file, otherwise you would lose ordering which
6936 provides very important information.
6937
6938 Using this option, large sites dealing with several thousand connections per
6939 second may log normal traffic to a rotating buffer and only archive smaller
6940 error logs.
6941
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006942 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006943 logging.
6944
Willy Tarreauc27debf2008-01-06 08:57:02 +01006945
6946option logasap
6947no option logasap
6948 Enable or disable early logging of HTTP requests
6949 May be used in sections : defaults | frontend | listen | backend
6950 yes | yes | yes | no
6951 Arguments : none
6952
6953 By default, HTTP requests are logged upon termination so that the total
6954 transfer time and the number of bytes appear in the logs. When large objects
6955 are being transferred, it may take a while before the request appears in the
6956 logs. Using "option logasap", the request gets logged as soon as the server
6957 sends the complete headers. The only missing information in the logs will be
6958 the total number of bytes which will indicate everything except the amount
6959 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006960 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006961 "Content-Length" response header so that the logs at least indicate how many
6962 bytes are expected to be transferred.
6963
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006964 Examples :
6965 listen http_proxy 0.0.0.0:80
6966 mode http
6967 option httplog
6968 option logasap
6969 log 192.168.2.200 local3
6970
6971 >>> Feb 6 12:14:14 localhost \
6972 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6973 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6974 "GET /image.iso HTTP/1.0"
6975
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006976 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006977 logging.
6978
6979
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006980option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006981 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006982 May be used in sections : defaults | frontend | listen | backend
6983 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006984 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006985 <username> This is the username which will be used when connecting to MySQL
6986 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006987 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006988
6989 If you specify a username, the check consists of sending two MySQL packet,
6990 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01006991 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006992 Error packet. It is a basic but useful test which does not produce error nor
6993 aborted connect on the server. However, it requires adding an authorization
6994 in the MySQL table, like this :
6995
6996 USE mysql;
6997 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6998 FLUSH PRIVILEGES;
6999
7000 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007001 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02007002 Error packet, we don't send anything in this mode. It was reported that it
7003 can generate lockout if check is too frequent and/or if there is not enough
7004 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
7005 value as if a connection is established successfully within fewer than MySQL
7006 "max_connect_errors" attempts after a previous connection was interrupted,
7007 the error count for the host is cleared to zero. If HAProxy's server get
7008 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
7009
7010 Remember that this does not check database presence nor database consistency.
7011 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007012
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02007013 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007014
7015 Most often, an incoming MySQL server needs to see the client's IP address for
7016 various purposes, including IP privilege matching and connection logging.
7017 When possible, it is often wise to masquerade the client's IP address when
7018 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007019 which requires the transparent proxy feature to be compiled in, and the MySQL
7020 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01007021
7022 See also: "option httpchk"
7023
7024
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007025option nolinger
7026no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007027 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007028 May be used in sections: defaults | frontend | listen | backend
7029 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007030 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007031
Davor Ocelice9ed2812017-12-25 17:49:28 +01007032 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007033 physically disconnected), the session timeouts triggers and the session is
7034 closed. But it will remain in FIN_WAIT1 state for some time in the system,
7035 using some resources and possibly limiting the ability to establish newer
7036 connections.
7037
7038 When this happens, it is possible to activate "option nolinger" which forces
7039 the system to immediately remove any socket's pending data on close. Thus,
7040 the session is instantly purged from the system's tables. This usually has
7041 side effects such as increased number of TCP resets due to old retransmits
7042 getting immediately rejected. Some firewalls may sometimes complain about
7043 this too.
7044
7045 For this reason, it is not recommended to use this option when not absolutely
7046 needed. You know that you need it when you have thousands of FIN_WAIT1
7047 sessions on your system (TIME_WAIT ones do not count).
7048
7049 This option may be used both on frontends and backends, depending on the side
7050 where it is required. Use it on the frontend for clients, and on the backend
7051 for servers.
7052
7053 If this option has been enabled in a "defaults" section, it can be disabled
7054 in a specific instance by prepending the "no" keyword before it.
7055
7056
Willy Tarreau55165fe2009-05-10 12:02:55 +02007057option originalto [ except <network> ] [ header <name> ]
7058 Enable insertion of the X-Original-To header to requests sent to servers
7059 May be used in sections : defaults | frontend | listen | backend
7060 yes | yes | yes | yes
7061 Arguments :
7062 <network> is an optional argument used to disable this option for sources
7063 matching <network>
7064 <name> an optional argument to specify a different "X-Original-To"
7065 header name.
7066
7067 Since HAProxy can work in transparent mode, every request from a client can
7068 be redirected to the proxy and HAProxy itself can proxy every request to a
7069 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
7070 be lost. This is annoying when you want access rules based on destination ip
7071 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
7072 added by HAProxy to all requests sent to the server. This header contains a
7073 value representing the original destination IP address. Since this must be
7074 configured to always use the last occurrence of this header only. Note that
7075 only the last occurrence of the header must be used, since it is really
7076 possible that the client has already brought one.
7077
7078 The keyword "header" may be used to supply a different header name to replace
7079 the default "X-Original-To". This can be useful where you might already
7080 have a "X-Original-To" header from a different application, and you need
7081 preserve it. Also if your backend server doesn't use the "X-Original-To"
7082 header and requires different one.
7083
7084 Sometimes, a same HAProxy instance may be shared between a direct client
7085 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
7086 used to decrypt HTTPS traffic). It is possible to disable the addition of the
7087 header for a known source address or network by adding the "except" keyword
7088 followed by the network address. In this case, any source IP matching the
7089 network will not cause an addition of this header. Most common uses are with
7090 private networks or 127.0.0.1.
7091
7092 This option may be specified either in the frontend or in the backend. If at
7093 least one of them uses it, the header will be added. Note that the backend's
7094 setting of the header subargument takes precedence over the frontend's if
7095 both are defined.
7096
Willy Tarreau55165fe2009-05-10 12:02:55 +02007097 Examples :
7098 # Original Destination address
7099 frontend www
7100 mode http
7101 option originalto except 127.0.0.1
7102
7103 # Those servers want the IP Address in X-Client-Dst
7104 backend www
7105 mode http
7106 option originalto header X-Client-Dst
7107
Christopher Faulet315b39c2018-09-21 16:26:19 +02007108 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02007109
7110
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007111option persist
7112no option persist
7113 Enable or disable forced persistence on down servers
7114 May be used in sections: defaults | frontend | listen | backend
7115 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007116 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007117
7118 When an HTTP request reaches a backend with a cookie which references a dead
7119 server, by default it is redispatched to another server. It is possible to
7120 force the request to be sent to the dead server first using "option persist"
7121 if absolutely needed. A common use case is when servers are under extreme
7122 load and spend their time flapping. In this case, the users would still be
7123 directed to the server they opened the session on, in the hope they would be
7124 correctly served. It is recommended to use "option redispatch" in conjunction
7125 with this option so that in the event it would not be possible to connect to
7126 the server at all (server definitely dead), the client would finally be
7127 redirected to another valid server.
7128
7129 If this option has been enabled in a "defaults" section, it can be disabled
7130 in a specific instance by prepending the "no" keyword before it.
7131
Willy Tarreau4de91492010-01-22 19:10:05 +01007132 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007133
7134
Willy Tarreau0c122822013-12-15 18:49:01 +01007135option pgsql-check [ user <username> ]
7136 Use PostgreSQL health checks for server testing
7137 May be used in sections : defaults | frontend | listen | backend
7138 yes | no | yes | yes
7139 Arguments :
7140 <username> This is the username which will be used when connecting to
7141 PostgreSQL server.
7142
7143 The check sends a PostgreSQL StartupMessage and waits for either
7144 Authentication request or ErrorResponse message. It is a basic but useful
7145 test which does not produce error nor aborted connect on the server.
7146 This check is identical with the "mysql-check".
7147
7148 See also: "option httpchk"
7149
7150
Willy Tarreau9420b122013-12-15 18:58:25 +01007151option prefer-last-server
7152no option prefer-last-server
7153 Allow multiple load balanced requests to remain on the same server
7154 May be used in sections: defaults | frontend | listen | backend
7155 yes | no | yes | yes
7156 Arguments : none
7157
7158 When the load balancing algorithm in use is not deterministic, and a previous
7159 request was sent to a server to which haproxy still holds a connection, it is
7160 sometimes desirable that subsequent requests on a same session go to the same
7161 server as much as possible. Note that this is different from persistence, as
7162 we only indicate a preference which haproxy tries to apply without any form
7163 of warranty. The real use is for keep-alive connections sent to servers. When
7164 this option is used, haproxy will try to reuse the same connection that is
7165 attached to the server instead of rebalancing to another server, causing a
7166 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01007167 not make much sense to use this in combination with hashing algorithms. Note,
7168 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02007169 to a proxy which sends a 407 (authentication required), when the load
7170 balancing algorithm is not deterministic. This is mandatory for use with the
7171 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01007172 troubleshooting some faulty applications. Option prefer-last-server might be
7173 desirable in these environments as well, to avoid redistributing the traffic
7174 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01007175
7176 If this option has been enabled in a "defaults" section, it can be disabled
7177 in a specific instance by prepending the "no" keyword before it.
7178
7179 See also: "option http-keep-alive"
7180
7181
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007182option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07007183option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007184no option redispatch
7185 Enable or disable session redistribution in case of connection failure
7186 May be used in sections: defaults | frontend | listen | backend
7187 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07007188 Arguments :
7189 <interval> The optional integer value that controls how often redispatches
7190 occur when retrying connections. Positive value P indicates a
7191 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01007192 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07007193 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007194 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07007195 positive value of 1 would indicate a redispatch on every retry,
7196 and a positive value of 3 would indicate a redispatch on every
7197 third retry. You can disable redispatches with a value of 0.
7198
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007199
7200 In HTTP mode, if a server designated by a cookie is down, clients may
7201 definitely stick to it because they cannot flush the cookie, so they will not
7202 be able to access the service anymore.
7203
Willy Tarreau59884a62019-01-02 14:48:31 +01007204 Specifying "option redispatch" will allow the proxy to break cookie or
7205 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007206
Joseph Lynch726ab712015-05-11 23:25:34 -07007207 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007208 connection failures. Of course, it requires having "retries" set to a nonzero
7209 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007210
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007211 This form is the preferred form, which replaces both the "redispatch" and
7212 "redisp" keywords.
7213
7214 If this option has been enabled in a "defaults" section, it can be disabled
7215 in a specific instance by prepending the "no" keyword before it.
7216
Willy Tarreau4de91492010-01-22 19:10:05 +01007217 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007218
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007219
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007220option redis-check
7221 Use redis health checks for server testing
7222 May be used in sections : defaults | frontend | listen | backend
7223 yes | no | yes | yes
7224 Arguments : none
7225
7226 It is possible to test that the server correctly talks REDIS protocol instead
7227 of just testing that it accepts the TCP connection. When this option is set,
7228 a PING redis command is sent to the server, and the response is analyzed to
7229 find the "+PONG" response message.
7230
7231 Example :
7232 option redis-check
7233
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03007234 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02007235
7236
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007237option smtpchk
7238option smtpchk <hello> <domain>
7239 Use SMTP health checks for server testing
7240 May be used in sections : defaults | frontend | listen | backend
7241 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01007242 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007243 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02007244 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007245 values will be turned into the default command ("HELO").
7246
7247 <domain> is the domain name to present to the server. It may only be
7248 specified (and is mandatory) if the hello command has been
7249 specified. By default, "localhost" is used.
7250
7251 When "option smtpchk" is set, the health checks will consist in TCP
7252 connections followed by an SMTP command. By default, this command is
7253 "HELO localhost". The server's return code is analyzed and only return codes
7254 starting with a "2" will be considered as valid. All other responses,
7255 including a lack of response will constitute an error and will indicate a
7256 dead server.
7257
7258 This test is meant to be used with SMTP servers or relays. Depending on the
7259 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01007260 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007261 25 is often easier than adjusting the configuration.
7262
7263 Most often, an incoming SMTP server needs to see the client's IP address for
7264 various purposes, including spam filtering, anti-spoofing and logging. When
7265 possible, it is often wise to masquerade the client's IP address when
7266 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02007267 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007268
7269 Example :
7270 option smtpchk HELO mydomain.org
7271
7272 See also : "option httpchk", "source"
7273
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007274
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02007275option socket-stats
7276no option socket-stats
7277
7278 Enable or disable collecting & providing separate statistics for each socket.
7279 May be used in sections : defaults | frontend | listen | backend
7280 yes | yes | yes | no
7281
7282 Arguments : none
7283
7284
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007285option splice-auto
7286no option splice-auto
7287 Enable or disable automatic kernel acceleration on sockets in both directions
7288 May be used in sections : defaults | frontend | listen | backend
7289 yes | yes | yes | yes
7290 Arguments : none
7291
7292 When this option is enabled either on a frontend or on a backend, haproxy
7293 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007294 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007295 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007296 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007297 are not much aggressive in order to limit excessive use of splicing. This
7298 option requires splicing to be enabled at compile time, and may be globally
7299 disabled with the global option "nosplice". Since splice uses pipes, using it
7300 requires that there are enough spare pipes.
7301
7302 Important note: kernel-based TCP splicing is a Linux-specific feature which
7303 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
7304 transfer data between sockets without copying these data to user-space, thus
7305 providing noticeable performance gains and CPU cycles savings. Since many
7306 early implementations are buggy, corrupt data and/or are inefficient, this
7307 feature is not enabled by default, and it should be used with extreme care.
7308 While it is not possible to detect the correctness of an implementation,
7309 2.6.29 is the first version offering a properly working implementation. In
7310 case of doubt, splicing may be globally disabled using the global "nosplice"
7311 keyword.
7312
7313 Example :
7314 option splice-auto
7315
7316 If this option has been enabled in a "defaults" section, it can be disabled
7317 in a specific instance by prepending the "no" keyword before it.
7318
7319 See also : "option splice-request", "option splice-response", and global
7320 options "nosplice" and "maxpipes"
7321
7322
7323option splice-request
7324no option splice-request
7325 Enable or disable automatic kernel acceleration on sockets for requests
7326 May be used in sections : defaults | frontend | listen | backend
7327 yes | yes | yes | yes
7328 Arguments : none
7329
7330 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007331 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007332 the client to the server. It might still use the recv/send scheme if there
7333 are no spare pipes left. This option requires splicing to be enabled at
7334 compile time, and may be globally disabled with the global option "nosplice".
7335 Since splice uses pipes, using it requires that there are enough spare pipes.
7336
7337 Important note: see "option splice-auto" for usage limitations.
7338
7339 Example :
7340 option splice-request
7341
7342 If this option has been enabled in a "defaults" section, it can be disabled
7343 in a specific instance by prepending the "no" keyword before it.
7344
7345 See also : "option splice-auto", "option splice-response", and global options
7346 "nosplice" and "maxpipes"
7347
7348
7349option splice-response
7350no option splice-response
7351 Enable or disable automatic kernel acceleration on sockets for responses
7352 May be used in sections : defaults | frontend | listen | backend
7353 yes | yes | yes | yes
7354 Arguments : none
7355
7356 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04007357 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01007358 the server to the client. It might still use the recv/send scheme if there
7359 are no spare pipes left. This option requires splicing to be enabled at
7360 compile time, and may be globally disabled with the global option "nosplice".
7361 Since splice uses pipes, using it requires that there are enough spare pipes.
7362
7363 Important note: see "option splice-auto" for usage limitations.
7364
7365 Example :
7366 option splice-response
7367
7368 If this option has been enabled in a "defaults" section, it can be disabled
7369 in a specific instance by prepending the "no" keyword before it.
7370
7371 See also : "option splice-auto", "option splice-request", and global options
7372 "nosplice" and "maxpipes"
7373
7374
Christopher Fauletba7bc162016-11-07 21:07:38 +01007375option spop-check
7376 Use SPOP health checks for server testing
7377 May be used in sections : defaults | frontend | listen | backend
7378 no | no | no | yes
7379 Arguments : none
7380
7381 It is possible to test that the server correctly talks SPOP protocol instead
7382 of just testing that it accepts the TCP connection. When this option is set,
7383 a HELLO handshake is performed between HAProxy and the server, and the
7384 response is analyzed to check no error is reported.
7385
7386 Example :
7387 option spop-check
7388
7389 See also : "option httpchk"
7390
7391
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007392option srvtcpka
7393no option srvtcpka
7394 Enable or disable the sending of TCP keepalive packets on the server side
7395 May be used in sections : defaults | frontend | listen | backend
7396 yes | no | yes | yes
7397 Arguments : none
7398
7399 When there is a firewall or any session-aware component between a client and
7400 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007401 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007402 components decides to expire a session which has remained idle for too long.
7403
7404 Enabling socket-level TCP keep-alives makes the system regularly send packets
7405 to the other end of the connection, leaving it active. The delay between
7406 keep-alive probes is controlled by the system only and depends both on the
7407 operating system and its tuning parameters.
7408
7409 It is important to understand that keep-alive packets are neither emitted nor
7410 received at the application level. It is only the network stacks which sees
7411 them. For this reason, even if one side of the proxy already uses keep-alives
7412 to maintain its connection alive, those keep-alive packets will not be
7413 forwarded to the other side of the proxy.
7414
7415 Please note that this has nothing to do with HTTP keep-alive.
7416
7417 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
7418 server side of a connection, which should help when session expirations are
7419 noticed between HAProxy and a server.
7420
7421 If this option has been enabled in a "defaults" section, it can be disabled
7422 in a specific instance by prepending the "no" keyword before it.
7423
7424 See also : "option clitcpka", "option tcpka"
7425
7426
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007427option ssl-hello-chk
7428 Use SSLv3 client hello health checks for server testing
7429 May be used in sections : defaults | frontend | listen | backend
7430 yes | no | yes | yes
7431 Arguments : none
7432
7433 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
7434 possible to test that the server correctly talks SSL instead of just testing
7435 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
7436 SSLv3 client hello messages are sent once the connection is established to
7437 the server, and the response is analyzed to find an SSL server hello message.
7438 The server is considered valid only when the response contains this server
7439 hello message.
7440
7441 All servers tested till there correctly reply to SSLv3 client hello messages,
7442 and most servers tested do not even log the requests containing only hello
7443 messages, which is appreciable.
7444
Willy Tarreau763a95b2012-10-04 23:15:39 +02007445 Note that this check works even when SSL support was not built into haproxy
7446 because it forges the SSL message. When SSL support is available, it is best
7447 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007448
Willy Tarreau763a95b2012-10-04 23:15:39 +02007449 See also: "option httpchk", "check-ssl"
7450
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007451
Willy Tarreaued179852013-12-16 01:07:00 +01007452option tcp-check
7453 Perform health checks using tcp-check send/expect sequences
7454 May be used in sections: defaults | frontend | listen | backend
7455 yes | no | yes | yes
7456
7457 This health check method is intended to be combined with "tcp-check" command
7458 lists in order to support send/expect types of health check sequences.
7459
7460 TCP checks currently support 4 modes of operations :
7461 - no "tcp-check" directive : the health check only consists in a connection
7462 attempt, which remains the default mode.
7463
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007464 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01007465 used to send a string along with a connection opening. With some
7466 protocols, it helps sending a "QUIT" message for example that prevents
7467 the server from logging a connection error for each health check. The
7468 check result will still be based on the ability to open the connection
7469 only.
7470
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007471 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01007472 The connection is opened and haproxy waits for the server to present some
7473 contents which must validate some rules. The check result will be based
7474 on the matching between the contents and the rules. This is suited for
7475 POP, IMAP, SMTP, FTP, SSH, TELNET.
7476
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007477 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007478 used to test a hello-type protocol. HAProxy sends a message, the server
7479 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007480 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007481 suited for protocols which require a binding or a request/response model.
7482 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7483 already all have their dedicated checks with a deeper understanding of
7484 the respective protocols.
7485 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007486 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007487
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007488 A fifth mode can be used to insert comments in different steps of the
7489 script.
7490
7491 For each tcp-check rule you create, you can add a "comment" directive,
7492 followed by a string. This string will be reported in the log and stderr
7493 in debug mode. It is useful to make user-friendly error reporting.
7494 The "comment" is of course optional.
7495
7496
Willy Tarreaued179852013-12-16 01:07:00 +01007497 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007498 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007499 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007500 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007501
Davor Ocelice9ed2812017-12-25 17:49:28 +01007502 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007503 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007504 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007505
7506 # look for the redis master server after ensuring it speaks well
7507 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007508 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007509 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007510 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007511 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007512 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007513 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007514 tcp-check send info\ replication\r\n
7515 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007516 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007517 tcp-check send QUIT\r\n
7518 tcp-check expect string +OK
7519
Davor Ocelice9ed2812017-12-25 17:49:28 +01007520 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007521 (send many headers before analyzing)
7522 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007523 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007524 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7525 tcp-check send Host:\ www.mydomain.com\r\n
7526 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7527 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007528 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007529
7530
7531 See also : "tcp-check expect", "tcp-check send"
7532
7533
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007534option tcp-smart-accept
7535no option tcp-smart-accept
7536 Enable or disable the saving of one ACK packet during the accept sequence
7537 May be used in sections : defaults | frontend | listen | backend
7538 yes | yes | yes | no
7539 Arguments : none
7540
7541 When an HTTP connection request comes in, the system acknowledges it on
7542 behalf of HAProxy, then the client immediately sends its request, and the
7543 system acknowledges it too while it is notifying HAProxy about the new
7544 connection. HAProxy then reads the request and responds. This means that we
7545 have one TCP ACK sent by the system for nothing, because the request could
7546 very well be acknowledged by HAProxy when it sends its response.
7547
7548 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7549 sending this useless ACK on platforms which support it (currently at least
7550 Linux). It must not cause any problem, because the system will send it anyway
7551 after 40 ms if the response takes more time than expected to come.
7552
7553 During complex network debugging sessions, it may be desirable to disable
7554 this optimization because delayed ACKs can make troubleshooting more complex
7555 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007556 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007557
7558 It is also possible to force it for non-HTTP proxies by simply specifying
7559 "option tcp-smart-accept". For instance, it can make sense with some services
7560 such as SMTP where the server speaks first.
7561
7562 It is recommended to avoid forcing this option in a defaults section. In case
7563 of doubt, consider setting it back to automatic values by prepending the
7564 "default" keyword before it, or disabling it using the "no" keyword.
7565
Willy Tarreaud88edf22009-06-14 15:48:17 +02007566 See also : "option tcp-smart-connect"
7567
7568
7569option tcp-smart-connect
7570no option tcp-smart-connect
7571 Enable or disable the saving of one ACK packet during the connect sequence
7572 May be used in sections : defaults | frontend | listen | backend
7573 yes | no | yes | yes
7574 Arguments : none
7575
7576 On certain systems (at least Linux), HAProxy can ask the kernel not to
7577 immediately send an empty ACK upon a connection request, but to directly
7578 send the buffer request instead. This saves one packet on the network and
7579 thus boosts performance. It can also be useful for some servers, because they
7580 immediately get the request along with the incoming connection.
7581
7582 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7583 It is not enabled by default because it makes network troubleshooting more
7584 complex.
7585
7586 It only makes sense to enable it with protocols where the client speaks first
7587 such as HTTP. In other situations, if there is no data to send in place of
7588 the ACK, a normal ACK is sent.
7589
7590 If this option has been enabled in a "defaults" section, it can be disabled
7591 in a specific instance by prepending the "no" keyword before it.
7592
7593 See also : "option tcp-smart-accept"
7594
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007595
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007596option tcpka
7597 Enable or disable the sending of TCP keepalive packets on both sides
7598 May be used in sections : defaults | frontend | listen | backend
7599 yes | yes | yes | yes
7600 Arguments : none
7601
7602 When there is a firewall or any session-aware component between a client and
7603 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007604 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007605 components decides to expire a session which has remained idle for too long.
7606
7607 Enabling socket-level TCP keep-alives makes the system regularly send packets
7608 to the other end of the connection, leaving it active. The delay between
7609 keep-alive probes is controlled by the system only and depends both on the
7610 operating system and its tuning parameters.
7611
7612 It is important to understand that keep-alive packets are neither emitted nor
7613 received at the application level. It is only the network stacks which sees
7614 them. For this reason, even if one side of the proxy already uses keep-alives
7615 to maintain its connection alive, those keep-alive packets will not be
7616 forwarded to the other side of the proxy.
7617
7618 Please note that this has nothing to do with HTTP keep-alive.
7619
7620 Using option "tcpka" enables the emission of TCP keep-alive probes on both
7621 the client and server sides of a connection. Note that this is meaningful
7622 only in "defaults" or "listen" sections. If this option is used in a
7623 frontend, only the client side will get keep-alives, and if this option is
7624 used in a backend, only the server side will get keep-alives. For this
7625 reason, it is strongly recommended to explicitly use "option clitcpka" and
7626 "option srvtcpka" when the configuration is split between frontends and
7627 backends.
7628
7629 See also : "option clitcpka", "option srvtcpka"
7630
Willy Tarreau844e3c52008-01-11 16:28:18 +01007631
7632option tcplog
7633 Enable advanced logging of TCP connections with session state and timers
7634 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007635 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007636 Arguments : none
7637
7638 By default, the log output format is very poor, as it only contains the
7639 source and destination addresses, and the instance name. By specifying
7640 "option tcplog", each log line turns into a much richer format including, but
7641 not limited to, the connection timers, the session status, the connections
7642 numbers, the frontend, backend and server name, and of course the source
7643 address and ports. This option is useful for pure TCP proxies in order to
7644 find which of the client or server disconnects or times out. For normal HTTP
7645 proxies, it's better to use "option httplog" which is even more complete.
7646
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007647 "option tcplog" overrides any previous "log-format" directive.
7648
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007649 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007650
7651
Willy Tarreau844e3c52008-01-11 16:28:18 +01007652option transparent
7653no option transparent
7654 Enable client-side transparent proxying
7655 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007656 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007657 Arguments : none
7658
7659 This option was introduced in order to provide layer 7 persistence to layer 3
7660 load balancers. The idea is to use the OS's ability to redirect an incoming
7661 connection for a remote address to a local process (here HAProxy), and let
7662 this process know what address was initially requested. When this option is
7663 used, sessions without cookies will be forwarded to the original destination
7664 IP address of the incoming request (which should match that of another
7665 equipment), while requests with cookies will still be forwarded to the
7666 appropriate server.
7667
7668 Note that contrary to a common belief, this option does NOT make HAProxy
7669 present the client's IP to the server when establishing the connection.
7670
Willy Tarreaua1146052011-03-01 09:51:54 +01007671 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007672 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007673
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007674
Simon Horman98637e52014-06-20 12:30:16 +09007675external-check command <command>
7676 Executable to run when performing an external-check
7677 May be used in sections : defaults | frontend | listen | backend
7678 yes | no | yes | yes
7679
7680 Arguments :
7681 <command> is the external command to run
7682
Simon Horman98637e52014-06-20 12:30:16 +09007683 The arguments passed to the to the command are:
7684
Cyril Bonté777be862014-12-02 21:21:35 +01007685 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007686
Cyril Bonté777be862014-12-02 21:21:35 +01007687 The <proxy_address> and <proxy_port> are derived from the first listener
7688 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7689 listener the proxy_address will be the path of the socket and the
7690 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7691 possible to determine a listener, and both <proxy_address> and <proxy_port>
7692 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007693
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007694 Some values are also provided through environment variables.
7695
7696 Environment variables :
7697 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7698 applicable, for example in a "backend" section).
7699
7700 HAPROXY_PROXY_ID The backend id.
7701
7702 HAPROXY_PROXY_NAME The backend name.
7703
7704 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7705 applicable, for example in a "backend" section or
7706 for a UNIX socket).
7707
7708 HAPROXY_SERVER_ADDR The server address.
7709
7710 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7711
7712 HAPROXY_SERVER_ID The server id.
7713
7714 HAPROXY_SERVER_MAXCONN The server max connections.
7715
7716 HAPROXY_SERVER_NAME The server name.
7717
7718 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7719 socket).
7720
7721 PATH The PATH environment variable used when executing
7722 the command may be set using "external-check path".
7723
William Lallemand4d03e432019-06-14 15:35:37 +02007724 See also "2.3. Environment variables" for other variables.
7725
Simon Horman98637e52014-06-20 12:30:16 +09007726 If the command executed and exits with a zero status then the check is
7727 considered to have passed, otherwise the check is considered to have
7728 failed.
7729
7730 Example :
7731 external-check command /bin/true
7732
7733 See also : "external-check", "option external-check", "external-check path"
7734
7735
7736external-check path <path>
7737 The value of the PATH environment variable used when running an external-check
7738 May be used in sections : defaults | frontend | listen | backend
7739 yes | no | yes | yes
7740
7741 Arguments :
7742 <path> is the path used when executing external command to run
7743
7744 The default path is "".
7745
7746 Example :
7747 external-check path "/usr/bin:/bin"
7748
7749 See also : "external-check", "option external-check",
7750 "external-check command"
7751
7752
Emeric Brun647caf12009-06-30 17:57:00 +02007753persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007754persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007755 Enable RDP cookie-based persistence
7756 May be used in sections : defaults | frontend | listen | backend
7757 yes | no | yes | yes
7758 Arguments :
7759 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007760 default cookie name "msts" will be used. There currently is no
7761 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007762
7763 This statement enables persistence based on an RDP cookie. The RDP cookie
7764 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007765 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007766 and if an RDP cookie is found, it is decoded. If it matches a known server
7767 which is still UP (or if "option persist" is set), then the connection is
7768 forwarded to this server.
7769
7770 Note that this only makes sense in a TCP backend, but for this to work, the
7771 frontend must have waited long enough to ensure that an RDP cookie is present
7772 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007773 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007774 a single "listen" section.
7775
Willy Tarreau61e28f22010-05-16 22:31:05 +02007776 Also, it is important to understand that the terminal server will emit this
7777 RDP cookie only if it is configured for "token redirection mode", which means
7778 that the "IP address redirection" option is disabled.
7779
Emeric Brun647caf12009-06-30 17:57:00 +02007780 Example :
7781 listen tse-farm
7782 bind :3389
7783 # wait up to 5s for an RDP cookie in the request
7784 tcp-request inspect-delay 5s
7785 tcp-request content accept if RDP_COOKIE
7786 # apply RDP cookie persistence
7787 persist rdp-cookie
7788 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007789 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007790 balance rdp-cookie
7791 server srv1 1.1.1.1:3389
7792 server srv2 1.1.1.2:3389
7793
Simon Hormanab814e02011-06-24 14:50:20 +09007794 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7795 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007796
7797
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007798rate-limit sessions <rate>
7799 Set a limit on the number of new sessions accepted per second on a frontend
7800 May be used in sections : defaults | frontend | listen | backend
7801 yes | yes | yes | no
7802 Arguments :
7803 <rate> The <rate> parameter is an integer designating the maximum number
7804 of new sessions per second to accept on the frontend.
7805
7806 When the frontend reaches the specified number of new sessions per second, it
7807 stops accepting new connections until the rate drops below the limit again.
7808 During this time, the pending sessions will be kept in the socket's backlog
7809 (in system buffers) and haproxy will not even be aware that sessions are
7810 pending. When applying very low limit on a highly loaded service, it may make
7811 sense to increase the socket's backlog using the "backlog" keyword.
7812
7813 This feature is particularly efficient at blocking connection-based attacks
7814 or service abuse on fragile servers. Since the session rate is measured every
7815 millisecond, it is extremely accurate. Also, the limit applies immediately,
7816 no delay is needed at all to detect the threshold.
7817
7818 Example : limit the connection rate on SMTP to 10 per second max
7819 listen smtp
7820 mode tcp
7821 bind :25
7822 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007823 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007824
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007825 Note : when the maximum rate is reached, the frontend's status is not changed
7826 but its sockets appear as "WAITING" in the statistics if the
7827 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007828
7829 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7830
7831
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007832redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7833redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7834redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007835 Return an HTTP redirection if/unless a condition is matched
7836 May be used in sections : defaults | frontend | listen | backend
7837 no | yes | yes | yes
7838
7839 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007840 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007841
Willy Tarreau0140f252008-11-19 21:07:09 +01007842 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007843 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007844 the HTTP "Location" header. When used in an "http-request" rule,
7845 <loc> value follows the log-format rules and can include some
7846 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007847
7848 <pfx> With "redirect prefix", the "Location" header is built from the
7849 concatenation of <pfx> and the complete URI path, including the
7850 query string, unless the "drop-query" option is specified (see
7851 below). As a special case, if <pfx> equals exactly "/", then
7852 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007853 redirect to the same URL (for instance, to insert a cookie). When
7854 used in an "http-request" rule, <pfx> value follows the log-format
7855 rules and can include some dynamic values (see Custom Log Format
7856 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007857
7858 <sch> With "redirect scheme", then the "Location" header is built by
7859 concatenating <sch> with "://" then the first occurrence of the
7860 "Host" header, and then the URI path, including the query string
7861 unless the "drop-query" option is specified (see below). If no
7862 path is found or if the path is "*", then "/" is used instead. If
7863 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007864 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007865 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007866 HTTPS. When used in an "http-request" rule, <sch> value follows
7867 the log-format rules and can include some dynamic values (see
7868 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007869
7870 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007871 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7872 with 302 used by default if no code is specified. 301 means
7873 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007874 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007875 cache the redirection. 303 is equivalent to 302 except that the
7876 browser will fetch the location with a GET method. 307 is just
7877 like 302 but makes it clear that the same method must be reused.
7878 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007879
7880 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007881 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007882
7883 - "drop-query"
7884 When this keyword is used in a prefix-based redirection, then the
7885 location will be set without any possible query-string, which is useful
7886 for directing users to a non-secure page for instance. It has no effect
7887 with a location-type redirect.
7888
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007889 - "append-slash"
7890 This keyword may be used in conjunction with "drop-query" to redirect
7891 users who use a URL not ending with a '/' to the same one with the '/'.
7892 It can be useful to ensure that search engines will only see one URL.
7893 For this, a return code 301 is preferred.
7894
Willy Tarreau0140f252008-11-19 21:07:09 +01007895 - "set-cookie NAME[=value]"
7896 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7897 to the response. This is sometimes used to indicate that a user has
7898 been seen, for instance to protect against some types of DoS. No other
7899 cookie option is added, so the cookie will be a session cookie. Note
7900 that for a browser, a sole cookie name without an equal sign is
7901 different from a cookie with an equal sign.
7902
7903 - "clear-cookie NAME[=]"
7904 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7905 with the "Max-Age" attribute set to zero. This will tell the browser to
7906 delete this cookie. It is useful for instance on logout pages. It is
7907 important to note that clearing the cookie "NAME" will not remove a
7908 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7909 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007910
7911 Example: move the login URL only to HTTPS.
7912 acl clear dst_port 80
7913 acl secure dst_port 8080
7914 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007915 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007916 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007917 acl cookie_set hdr_sub(cookie) SEEN=1
7918
7919 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007920 redirect prefix https://mysite.com if login_page !secure
7921 redirect prefix http://mysite.com drop-query if login_page !uid_given
7922 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007923 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007924
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007925 Example: send redirects for request for articles without a '/'.
7926 acl missing_slash path_reg ^/article/[^/]*$
7927 redirect code 301 prefix / drop-query append-slash if missing_slash
7928
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007929 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007930 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007931
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007932 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007933 http-request redirect code 301 location \
7934 http://www.%[hdr(host)]%[capture.req.uri] \
7935 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007936
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007937 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007938
7939
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007940redisp (deprecated)
7941redispatch (deprecated)
7942 Enable or disable session redistribution in case of connection failure
7943 May be used in sections: defaults | frontend | listen | backend
7944 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007945 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007946
7947 In HTTP mode, if a server designated by a cookie is down, clients may
7948 definitely stick to it because they cannot flush the cookie, so they will not
7949 be able to access the service anymore.
7950
7951 Specifying "redispatch" will allow the proxy to break their persistence and
7952 redistribute them to a working server.
7953
7954 It also allows to retry last connection to another server in case of multiple
7955 connection failures. Of course, it requires having "retries" set to a nonzero
7956 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007957
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007958 This form is deprecated, do not use it in any new configuration, use the new
7959 "option redispatch" instead.
7960
7961 See also : "option redispatch"
7962
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007963
Willy Tarreau96d51952019-05-22 20:34:35 +02007964reqadd <string> [{if | unless} <cond>] (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01007965 Add a header at the end of the HTTP request
7966 May be used in sections : defaults | frontend | listen | backend
7967 no | yes | yes | yes
7968 Arguments :
7969 <string> is the complete line to be added. Any space or known delimiter
7970 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007971 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007972
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007973 <cond> is an optional matching condition built from ACLs. It makes it
7974 possible to ignore this rule when other conditions are not met.
7975
Willy Tarreau303c0352008-01-17 19:01:39 +01007976 A new line consisting in <string> followed by a line feed will be added after
7977 the last header of an HTTP request.
7978
7979 Header transformations only apply to traffic which passes through HAProxy,
7980 and not to traffic generated by HAProxy, such as health-checks or error
7981 responses.
7982
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007983 Example : add "X-Proto: SSL" to requests coming via port 81
7984 acl is-ssl dst_port 81
7985 reqadd X-Proto:\ SSL if is-ssl
7986
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007987 See also: "rspadd", "http-request", section 6 about HTTP header manipulation,
7988 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007989
7990
Willy Tarreau96d51952019-05-22 20:34:35 +02007991reqallow <search> [{if | unless} <cond>] (deprecated)
7992reqiallow <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01007993 Definitely allow an HTTP request if a line matches a regular expression
7994 May be used in sections : defaults | frontend | listen | backend
7995 no | yes | yes | yes
7996 Arguments :
7997 <search> is the regular expression applied to HTTP headers and to the
7998 request line. This is an extended regular expression. Parenthesis
7999 grouping is supported and no preliminary backslash is required.
8000 Any space or known delimiter must be escaped using a backslash
8001 ('\'). The pattern applies to a full line at a time. The
8002 "reqallow" keyword strictly matches case while "reqiallow"
8003 ignores case.
8004
Willy Tarreau5321c422010-01-28 20:35:13 +01008005 <cond> is an optional matching condition built from ACLs. It makes it
8006 possible to ignore this rule when other conditions are not met.
8007
Willy Tarreau303c0352008-01-17 19:01:39 +01008008 A request containing any line which matches extended regular expression
8009 <search> will mark the request as allowed, even if any later test would
8010 result in a deny. The test applies both to the request line and to request
8011 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01008012 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01008013
8014 It is easier, faster and more powerful to use ACLs to write access policies.
8015 Reqdeny, reqallow and reqpass should be avoided in new designs.
8016
8017 Example :
8018 # allow www.* but refuse *.local
8019 reqiallow ^Host:\ www\.
8020 reqideny ^Host:\ .*\.local
8021
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008022 See also: "reqdeny", "block", "http-request", section 6 about HTTP header
8023 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008024
8025
Willy Tarreau96d51952019-05-22 20:34:35 +02008026reqdel <search> [{if | unless} <cond>] (deprecated)
8027reqidel <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008028 Delete all headers matching a regular expression in an HTTP request
8029 May be used in sections : defaults | frontend | listen | backend
8030 no | yes | yes | yes
8031 Arguments :
8032 <search> is the regular expression applied to HTTP headers and to the
8033 request line. This is an extended regular expression. Parenthesis
8034 grouping is supported and no preliminary backslash is required.
8035 Any space or known delimiter must be escaped using a backslash
8036 ('\'). The pattern applies to a full line at a time. The "reqdel"
8037 keyword strictly matches case while "reqidel" ignores case.
8038
Willy Tarreau5321c422010-01-28 20:35:13 +01008039 <cond> is an optional matching condition built from ACLs. It makes it
8040 possible to ignore this rule when other conditions are not met.
8041
Willy Tarreau303c0352008-01-17 19:01:39 +01008042 Any header line matching extended regular expression <search> in the request
8043 will be completely deleted. Most common use of this is to remove unwanted
8044 and/or dangerous headers or cookies from a request before passing it to the
8045 next servers.
8046
8047 Header transformations only apply to traffic which passes through HAProxy,
8048 and not to traffic generated by HAProxy, such as health-checks or error
8049 responses. Keep in mind that header names are not case-sensitive.
8050
8051 Example :
8052 # remove X-Forwarded-For header and SERVER cookie
8053 reqidel ^X-Forwarded-For:.*
8054 reqidel ^Cookie:.*SERVER=
8055
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008056 See also: "reqadd", "reqrep", "rspdel", "http-request", section 6 about
8057 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008058
8059
Willy Tarreau96d51952019-05-22 20:34:35 +02008060reqdeny <search> [{if | unless} <cond>] (deprecated)
8061reqideny <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008062 Deny an HTTP request if a line matches a regular expression
8063 May be used in sections : defaults | frontend | listen | backend
8064 no | yes | yes | yes
8065 Arguments :
8066 <search> is the regular expression applied to HTTP headers and to the
8067 request line. This is an extended regular expression. Parenthesis
8068 grouping is supported and no preliminary backslash is required.
8069 Any space or known delimiter must be escaped using a backslash
8070 ('\'). The pattern applies to a full line at a time. The
8071 "reqdeny" keyword strictly matches case while "reqideny" ignores
8072 case.
8073
Willy Tarreau5321c422010-01-28 20:35:13 +01008074 <cond> is an optional matching condition built from ACLs. It makes it
8075 possible to ignore this rule when other conditions are not met.
8076
Willy Tarreau303c0352008-01-17 19:01:39 +01008077 A request containing any line which matches extended regular expression
8078 <search> will mark the request as denied, even if any later test would
8079 result in an allow. The test applies both to the request line and to request
8080 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01008081 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01008082
Willy Tarreauced27012008-01-17 20:35:34 +01008083 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008084 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01008085 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01008086
Willy Tarreau303c0352008-01-17 19:01:39 +01008087 It is easier, faster and more powerful to use ACLs to write access policies.
8088 Reqdeny, reqallow and reqpass should be avoided in new designs.
8089
8090 Example :
8091 # refuse *.local, then allow www.*
8092 reqideny ^Host:\ .*\.local
8093 reqiallow ^Host:\ www\.
8094
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008095 See also: "reqallow", "rspdeny", "block", "http-request", section 6 about
8096 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008097
8098
Willy Tarreau96d51952019-05-22 20:34:35 +02008099reqpass <search> [{if | unless} <cond>] (deprecated)
8100reqipass <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008101 Ignore any HTTP request line matching a regular expression in next rules
8102 May be used in sections : defaults | frontend | listen | backend
8103 no | yes | yes | yes
8104 Arguments :
8105 <search> is the regular expression applied to HTTP headers and to the
8106 request line. This is an extended regular expression. Parenthesis
8107 grouping is supported and no preliminary backslash is required.
8108 Any space or known delimiter must be escaped using a backslash
8109 ('\'). The pattern applies to a full line at a time. The
8110 "reqpass" keyword strictly matches case while "reqipass" ignores
8111 case.
8112
Willy Tarreau5321c422010-01-28 20:35:13 +01008113 <cond> is an optional matching condition built from ACLs. It makes it
8114 possible to ignore this rule when other conditions are not met.
8115
Willy Tarreau303c0352008-01-17 19:01:39 +01008116 A request containing any line which matches extended regular expression
8117 <search> will skip next rules, without assigning any deny or allow verdict.
8118 The test applies both to the request line and to request headers. Keep in
8119 mind that URLs in request line are case-sensitive while header names are not.
8120
8121 It is easier, faster and more powerful to use ACLs to write access policies.
8122 Reqdeny, reqallow and reqpass should be avoided in new designs.
8123
8124 Example :
8125 # refuse *.local, then allow www.*, but ignore "www.private.local"
8126 reqipass ^Host:\ www.private\.local
8127 reqideny ^Host:\ .*\.local
8128 reqiallow ^Host:\ www\.
8129
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008130 See also: "reqallow", "reqdeny", "block", "http-request", section 6 about
8131 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008132
8133
Willy Tarreau96d51952019-05-22 20:34:35 +02008134reqrep <search> <string> [{if | unless} <cond>] (deprecated)
8135reqirep <search> <string> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008136 Replace a regular expression with a string in an HTTP request line
8137 May be used in sections : defaults | frontend | listen | backend
8138 no | yes | yes | yes
8139 Arguments :
8140 <search> is the regular expression applied to HTTP headers and to the
8141 request line. This is an extended regular expression. Parenthesis
8142 grouping is supported and no preliminary backslash is required.
8143 Any space or known delimiter must be escaped using a backslash
8144 ('\'). The pattern applies to a full line at a time. The "reqrep"
8145 keyword strictly matches case while "reqirep" ignores case.
8146
8147 <string> is the complete line to be added. Any space or known delimiter
8148 must be escaped using a backslash ('\'). References to matched
8149 pattern groups are possible using the common \N form, with N
8150 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008151 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01008152
Willy Tarreau5321c422010-01-28 20:35:13 +01008153 <cond> is an optional matching condition built from ACLs. It makes it
8154 possible to ignore this rule when other conditions are not met.
8155
Willy Tarreau303c0352008-01-17 19:01:39 +01008156 Any line matching extended regular expression <search> in the request (both
8157 the request line and header lines) will be completely replaced with <string>.
8158 Most common use of this is to rewrite URLs or domain names in "Host" headers.
8159
8160 Header transformations only apply to traffic which passes through HAProxy,
8161 and not to traffic generated by HAProxy, such as health-checks or error
8162 responses. Note that for increased readability, it is suggested to add enough
8163 spaces between the request and the response. Keep in mind that URLs in
8164 request line are case-sensitive while header names are not.
8165
8166 Example :
8167 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04008168 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01008169 # replace "www.mydomain.com" with "www" in the host name.
8170 reqirep ^Host:\ www.mydomain.com Host:\ www
8171
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008172 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", "http-request",
8173 section 6 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008174
8175
Willy Tarreau96d51952019-05-22 20:34:35 +02008176reqtarpit <search> [{if | unless} <cond>] (deprecated)
8177reqitarpit <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008178 Tarpit an HTTP request containing a line matching a regular expression
8179 May be used in sections : defaults | frontend | listen | backend
8180 no | yes | yes | yes
8181 Arguments :
8182 <search> is the regular expression applied to HTTP headers and to the
8183 request line. This is an extended regular expression. Parenthesis
8184 grouping is supported and no preliminary backslash is required.
8185 Any space or known delimiter must be escaped using a backslash
8186 ('\'). The pattern applies to a full line at a time. The
8187 "reqtarpit" keyword strictly matches case while "reqitarpit"
8188 ignores case.
8189
Willy Tarreau5321c422010-01-28 20:35:13 +01008190 <cond> is an optional matching condition built from ACLs. It makes it
8191 possible to ignore this rule when other conditions are not met.
8192
Willy Tarreau303c0352008-01-17 19:01:39 +01008193 A request containing any line which matches extended regular expression
8194 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01008195 be kept open for a pre-defined time, then will return an HTTP error 500 so
8196 that the attacker does not suspect it has been tarpitted. The status 500 will
8197 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01008198 delay is defined by "timeout tarpit", or "timeout connect" if the former is
8199 not set.
8200
8201 The goal of the tarpit is to slow down robots attacking servers with
8202 identifiable requests. Many robots limit their outgoing number of connections
8203 and stay connected waiting for a reply which can take several minutes to
8204 come. Depending on the environment and attack, it may be particularly
8205 efficient at reducing the load on the network and firewalls.
8206
Willy Tarreau5321c422010-01-28 20:35:13 +01008207 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01008208 # ignore user-agents reporting any flavor of "Mozilla" or "MSIE", but
Willy Tarreau303c0352008-01-17 19:01:39 +01008209 # block all others.
8210 reqipass ^User-Agent:\.*(Mozilla|MSIE)
8211 reqitarpit ^User-Agent:
8212
Willy Tarreau5321c422010-01-28 20:35:13 +01008213 # block bad guys
8214 acl badguys src 10.1.0.3 172.16.13.20/28
8215 reqitarpit . if badguys
8216
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008217 See also: "reqallow", "reqdeny", "reqpass", "http-request", section 6
8218 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008219
8220
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02008221retries <value>
8222 Set the number of retries to perform on a server after a connection failure
8223 May be used in sections: defaults | frontend | listen | backend
8224 yes | no | yes | yes
8225 Arguments :
8226 <value> is the number of times a connection attempt should be retried on
8227 a server when a connection either is refused or times out. The
8228 default value is 3.
8229
8230 It is important to understand that this value applies to the number of
8231 connection attempts, not full requests. When a connection has effectively
8232 been established to a server, there will be no more retry.
8233
8234 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07008235 a turn-around timer of min("timeout connect", one second) is applied before
8236 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02008237
8238 When "option redispatch" is set, the last retry may be performed on another
8239 server even if a cookie references a different server.
8240
8241 See also : "option redispatch"
8242
8243
Olivier Houcharda254a372019-04-05 15:30:12 +02008244retry-on [list of keywords]
8245 Specify when to attempt to automatically retry a failed request
8246 May be used in sections: defaults | frontend | listen | backend
8247 yes | no | yes | yes
8248 Arguments :
8249 <keywords> is a list of keywords or HTTP status codes, each representing a
8250 type of failure event on which an attempt to retry the request
8251 is desired. Please read the notes at the bottom before changing
8252 this setting. The following keywords are supported :
8253
8254 none never retry
8255
8256 conn-failure retry when the connection or the SSL handshake failed
8257 and the request could not be sent. This is the default.
8258
8259 empty-response retry when the server connection was closed after part
8260 of the request was sent, and nothing was received from
8261 the server. This type of failure may be caused by the
8262 request timeout on the server side, poor network
8263 condition, or a server crash or restart while
8264 processing the request.
8265
Olivier Houcharde3249a92019-05-03 23:01:47 +02008266 junk-response retry when the server returned something not looking
8267 like a complete HTTP response. This includes partial
8268 responses headers as well as non-HTTP contents. It
8269 usually is a bad idea to retry on such events, which
8270 may be caused a configuration issue (wrong server port)
8271 or by the request being harmful to the server (buffer
8272 overflow attack for example).
8273
Olivier Houcharda254a372019-04-05 15:30:12 +02008274 response-timeout the server timeout stroke while waiting for the server
8275 to respond to the request. This may be caused by poor
8276 network condition, the reuse of an idle connection
8277 which has expired on the path, or by the request being
8278 extremely expensive to process. It generally is a bad
8279 idea to retry on such events on servers dealing with
8280 heavy database processing (full scans, etc) as it may
8281 amplify denial of service attacks.
8282
Olivier Houchard865d8392019-05-03 22:46:27 +02008283 0rtt-rejected retry requests which were sent over early data and were
8284 rejected by the server. These requests are generally
8285 considered to be safe to retry.
8286
Olivier Houcharda254a372019-04-05 15:30:12 +02008287 <status> any HTTP status code among "404" (Not Found), "408"
8288 (Request Timeout), "425" (Too Early), "500" (Server
8289 Error), "501" (Not Implemented), "502" (Bad Gateway),
8290 "503" (Service Unavailable), "504" (Gateway Timeout).
8291
Olivier Houchardddf0e032019-05-10 18:05:40 +02008292 all-retryable-errors
8293 retry request for any error that are considered
8294 retryable. This currently activates "conn-failure",
8295 "empty-response", "junk-response", "response-timeout",
8296 "0rtt-rejected", "500", "502", "503", and "504".
8297
Olivier Houcharda254a372019-04-05 15:30:12 +02008298 Using this directive replaces any previous settings with the new ones; it is
8299 not cumulative.
8300
8301 Please note that using anything other than "none" and "conn-failure" requires
8302 to allocate a buffer and copy the whole request into it, so it has memory and
8303 performance impacts. Requests not fitting in a single buffer will never be
8304 retried (see the global tune.bufsize setting).
8305
8306 You have to make sure the application has a replay protection mechanism built
8307 in such as a unique transaction IDs passed in requests, or that replaying the
8308 same request has no consequence, or it is very dangerous to use any retry-on
8309 value beside "conn-failure" and "none". Static file servers and caches are
8310 generally considered safe against any type of retry. Using a status code can
8311 be useful to quickly leave a server showing an abnormal behavior (out of
8312 memory, file system issues, etc), but in this case it may be a good idea to
8313 immediately redispatch the connection to another server (please see "option
8314 redispatch" for this). Last, it is important to understand that most causes
8315 of failures are the requests themselves and that retrying a request causing a
8316 server to misbehave will often make the situation even worse for this server,
8317 or for the whole service in case of redispatch.
8318
8319 Unless you know exactly how the application deals with replayed requests, you
8320 should not use this directive.
8321
8322 The default is "conn-failure".
8323
8324 See also: "retries", "option redispatch", "tune.bufsize"
8325
Willy Tarreau96d51952019-05-22 20:34:35 +02008326rspadd <string> [{if | unless} <cond>] (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008327 Add a header at the end of the HTTP response
8328 May be used in sections : defaults | frontend | listen | backend
8329 no | yes | yes | yes
8330 Arguments :
8331 <string> is the complete line to be added. Any space or known delimiter
8332 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008333 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01008334
Willy Tarreaufdb563c2010-01-31 15:43:27 +01008335 <cond> is an optional matching condition built from ACLs. It makes it
8336 possible to ignore this rule when other conditions are not met.
8337
Willy Tarreau303c0352008-01-17 19:01:39 +01008338 A new line consisting in <string> followed by a line feed will be added after
8339 the last header of an HTTP response.
8340
8341 Header transformations only apply to traffic which passes through HAProxy,
8342 and not to traffic generated by HAProxy, such as health-checks or error
8343 responses.
8344
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008345 See also: "rspdel" "reqadd", "http-response", section 6 about HTTP header
8346 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008347
8348
Willy Tarreau96d51952019-05-22 20:34:35 +02008349rspdel <search> [{if | unless} <cond>] (deprecated)
8350rspidel <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008351 Delete all headers matching a regular expression in an HTTP response
8352 May be used in sections : defaults | frontend | listen | backend
8353 no | yes | yes | yes
8354 Arguments :
8355 <search> is the regular expression applied to HTTP headers and to the
8356 response line. This is an extended regular expression, so
8357 parenthesis grouping is supported and no preliminary backslash
8358 is required. Any space or known delimiter must be escaped using
8359 a backslash ('\'). The pattern applies to a full line at a time.
8360 The "rspdel" keyword strictly matches case while "rspidel"
8361 ignores case.
8362
Willy Tarreaufdb563c2010-01-31 15:43:27 +01008363 <cond> is an optional matching condition built from ACLs. It makes it
8364 possible to ignore this rule when other conditions are not met.
8365
Willy Tarreau303c0352008-01-17 19:01:39 +01008366 Any header line matching extended regular expression <search> in the response
8367 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008368 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01008369 client.
8370
8371 Header transformations only apply to traffic which passes through HAProxy,
8372 and not to traffic generated by HAProxy, such as health-checks or error
8373 responses. Keep in mind that header names are not case-sensitive.
8374
8375 Example :
8376 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02008377 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01008378
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008379 See also: "rspadd", "rsprep", "reqdel", "http-response", section 6 about
8380 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008381
8382
Willy Tarreau96d51952019-05-22 20:34:35 +02008383rspdeny <search> [{if | unless} <cond>] (deprecated)
8384rspideny <search> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008385 Block an HTTP response if a line matches a regular expression
8386 May be used in sections : defaults | frontend | listen | backend
8387 no | yes | yes | yes
8388 Arguments :
8389 <search> is the regular expression applied to HTTP headers and to the
8390 response line. This is an extended regular expression, so
8391 parenthesis grouping is supported and no preliminary backslash
8392 is required. Any space or known delimiter must be escaped using
8393 a backslash ('\'). The pattern applies to a full line at a time.
8394 The "rspdeny" keyword strictly matches case while "rspideny"
8395 ignores case.
8396
Willy Tarreaufdb563c2010-01-31 15:43:27 +01008397 <cond> is an optional matching condition built from ACLs. It makes it
8398 possible to ignore this rule when other conditions are not met.
8399
Willy Tarreau303c0352008-01-17 19:01:39 +01008400 A response containing any line which matches extended regular expression
8401 <search> will mark the request as denied. The test applies both to the
8402 response line and to response headers. Keep in mind that header names are not
8403 case-sensitive.
8404
8405 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01008406 block the response before it reaches the client. If a response is denied, it
8407 will be replaced with an HTTP 502 error so that the client never retrieves
8408 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01008409
8410 It is easier, faster and more powerful to use ACLs to write access policies.
8411 Rspdeny should be avoided in new designs.
8412
8413 Example :
8414 # Ensure that no content type matching ms-word will leak
8415 rspideny ^Content-type:\.*/ms-word
8416
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008417 See also: "reqdeny", "acl", "block", "http-response", section 6 about
8418 HTTP header manipulation and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008419
8420
Willy Tarreau96d51952019-05-22 20:34:35 +02008421rsprep <search> <string> [{if | unless} <cond>] (deprecated)
8422rspirep <search> <string> [{if | unless} <cond>] (ignore case) (deprecated)
Willy Tarreau303c0352008-01-17 19:01:39 +01008423 Replace a regular expression with a string in an HTTP response line
8424 May be used in sections : defaults | frontend | listen | backend
8425 no | yes | yes | yes
8426 Arguments :
8427 <search> is the regular expression applied to HTTP headers and to the
8428 response line. This is an extended regular expression, so
8429 parenthesis grouping is supported and no preliminary backslash
8430 is required. Any space or known delimiter must be escaped using
8431 a backslash ('\'). The pattern applies to a full line at a time.
8432 The "rsprep" keyword strictly matches case while "rspirep"
8433 ignores case.
8434
8435 <string> is the complete line to be added. Any space or known delimiter
8436 must be escaped using a backslash ('\'). References to matched
8437 pattern groups are possible using the common \N form, with N
8438 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008439 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01008440
Willy Tarreaufdb563c2010-01-31 15:43:27 +01008441 <cond> is an optional matching condition built from ACLs. It makes it
8442 possible to ignore this rule when other conditions are not met.
8443
Willy Tarreau303c0352008-01-17 19:01:39 +01008444 Any line matching extended regular expression <search> in the response (both
8445 the response line and header lines) will be completely replaced with
8446 <string>. Most common use of this is to rewrite Location headers.
8447
8448 Header transformations only apply to traffic which passes through HAProxy,
8449 and not to traffic generated by HAProxy, such as health-checks or error
8450 responses. Note that for increased readability, it is suggested to add enough
8451 spaces between the request and the response. Keep in mind that header names
8452 are not case-sensitive.
8453
8454 Example :
8455 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
8456 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
8457
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08008458 See also: "rspadd", "rspdel", "reqrep", "http-response", section 6 about
8459 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01008460
8461
David du Colombier486df472011-03-17 10:40:26 +01008462server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008463 Declare a server in a backend
8464 May be used in sections : defaults | frontend | listen | backend
8465 no | no | yes | yes
8466 Arguments :
8467 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01008468 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05008469 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008470
David du Colombier486df472011-03-17 10:40:26 +01008471 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
8472 resolvable hostname is supported, but this name will be resolved
8473 during start-up. Address "0.0.0.0" or "*" has a special meaning.
8474 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02008475 address as the one from the client connection. This is useful in
8476 transparent proxy architectures where the client's connection is
8477 intercepted and haproxy must forward to the original destination
8478 address. This is more or less what the "transparent" keyword does
8479 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01008480 to report statistics. Optionally, an address family prefix may be
8481 used before the address to force the family regardless of the
8482 address format, which can be useful to specify a path to a unix
8483 socket with no slash ('/'). Currently supported prefixes are :
8484 - 'ipv4@' -> address is always IPv4
8485 - 'ipv6@' -> address is always IPv6
8486 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008487 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02008488 - 'sockpair@' -> address is the FD of a connected unix
8489 socket or of a socketpair. During a connection, the
8490 backend creates a pair of connected sockets, and passes
8491 one of them over the FD. The bind part will use the
8492 received socket as the client FD. Should be used
8493 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02008494 You may want to reference some environment variables in the
8495 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01008496 variables. The "init-addr" setting can be used to modify the way
8497 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008498
Willy Tarreaub6205fd2012-09-24 12:27:33 +02008499 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008500 be sent to this port. If unset, the same port the client
8501 connected to will be used. The port may also be prefixed by a "+"
8502 or a "-". In this case, the server's port will be determined by
8503 adding this value to the client's port.
8504
8505 <param*> is a list of parameters for this server. The "server" keywords
8506 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008507 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008508
8509 Examples :
8510 server first 10.1.1.1:1080 cookie first check inter 1000
8511 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01008512 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02008513 server backup "${SRV_BACKUP}:1080" backup
8514 server www1_dc1 "${LAN_DC1}.101:80"
8515 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008516
Willy Tarreau55dcaf62015-09-27 15:03:15 +02008517 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
8518 sun_path length is used for the address length. Some other programs
8519 such as socat use the string length only by default. Pass the option
8520 ",unix-tightsocklen=0" to any abstract socket definition in socat to
8521 make it compatible with HAProxy's.
8522
Mark Lamourinec2247f02012-01-04 13:02:01 -05008523 See also: "default-server", "http-send-name-header" and section 5 about
8524 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008525
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02008526server-state-file-name [<file>]
8527 Set the server state file to read, load and apply to servers available in
8528 this backend. It only applies when the directive "load-server-state-from-file"
8529 is set to "local". When <file> is not provided or if this directive is not
8530 set, then backend name is used. If <file> starts with a slash '/', then it is
8531 considered as an absolute path. Otherwise, <file> is concatenated to the
8532 global directive "server-state-file-base".
8533
8534 Example: the minimal configuration below would make HAProxy look for the
8535 state server file '/etc/haproxy/states/bk':
8536
8537 global
8538 server-state-file-base /etc/haproxy/states
8539
8540 backend bk
8541 load-server-state-from-file
8542
8543 See also: "server-state-file-base", "load-server-state-from-file", and
8544 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008545
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02008546server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
8547 Set a template to initialize servers with shared parameters.
8548 The names of these servers are built from <prefix> and <num | range> parameters.
8549 May be used in sections : defaults | frontend | listen | backend
8550 no | no | yes | yes
8551
8552 Arguments:
8553 <prefix> A prefix for the server names to be built.
8554
8555 <num | range>
8556 If <num> is provided, this template initializes <num> servers
8557 with 1 up to <num> as server name suffixes. A range of numbers
8558 <num_low>-<num_high> may also be used to use <num_low> up to
8559 <num_high> as server name suffixes.
8560
8561 <fqdn> A FQDN for all the servers this template initializes.
8562
8563 <port> Same meaning as "server" <port> argument (see "server" keyword).
8564
8565 <params*>
8566 Remaining server parameters among all those supported by "server"
8567 keyword.
8568
8569 Examples:
8570 # Initializes 3 servers with srv1, srv2 and srv3 as names,
8571 # google.com as FQDN, and health-check enabled.
8572 server-template srv 1-3 google.com:80 check
8573
8574 # or
8575 server-template srv 3 google.com:80 check
8576
8577 # would be equivalent to:
8578 server srv1 google.com:80 check
8579 server srv2 google.com:80 check
8580 server srv3 google.com:80 check
8581
8582
8583
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008584source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02008585source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008586source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008587 Set the source address for outgoing connections
8588 May be used in sections : defaults | frontend | listen | backend
8589 yes | no | yes | yes
8590 Arguments :
8591 <addr> is the IPv4 address HAProxy will bind to before connecting to a
8592 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01008593
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008594 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01008595 the most appropriate address to reach its destination. Optionally
8596 an address family prefix may be used before the address to force
8597 the family regardless of the address format, which can be useful
8598 to specify a path to a unix socket with no slash ('/'). Currently
8599 supported prefixes are :
8600 - 'ipv4@' -> address is always IPv4
8601 - 'ipv6@' -> address is always IPv6
8602 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008603 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02008604 You may want to reference some environment variables in the
8605 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008606
8607 <port> is an optional port. It is normally not needed but may be useful
8608 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02008609 the system will select a free port. Note that port ranges are not
8610 supported in the backend. If you want to force port ranges, you
8611 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008612
8613 <addr2> is the IP address to present to the server when connections are
8614 forwarded in full transparent proxy mode. This is currently only
8615 supported on some patched Linux kernels. When this address is
8616 specified, clients connecting to the server will be presented
8617 with this address, while health checks will still use the address
8618 <addr>.
8619
8620 <port2> is the optional port to present to the server when connections
8621 are forwarded in full transparent proxy mode (see <addr2> above).
8622 The default value of zero means the system will select a free
8623 port.
8624
Willy Tarreaubce70882009-09-07 11:51:47 +02008625 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
8626 This is the name of a comma-separated header list which can
8627 contain multiple IP addresses. By default, the last occurrence is
8628 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01008629 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02008630 by previous proxy, typically Stunnel. In order to use another
8631 occurrence from the last one, please see the <occ> parameter
8632 below. When the header (or occurrence) is not found, no binding
8633 is performed so that the proxy's default IP address is used. Also
8634 keep in mind that the header name is case insensitive, as for any
8635 HTTP header.
8636
8637 <occ> is the occurrence number of a value to be used in a multi-value
8638 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008639 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02008640 address. Positive values indicate a position from the first
8641 occurrence, 1 being the first one. Negative values indicate
8642 positions relative to the last one, -1 being the last one. This
8643 is helpful for situations where an X-Forwarded-For header is set
8644 at the entry point of an infrastructure and must be used several
8645 proxy layers away. When this value is not specified, -1 is
8646 assumed. Passing a zero here disables the feature.
8647
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008648 <name> is an optional interface name to which to bind to for outgoing
8649 traffic. On systems supporting this features (currently, only
8650 Linux), this allows one to bind all traffic to the server to
8651 this interface even if it is not the one the system would select
8652 based on routing tables. This should be used with extreme care.
8653 Note that using this option requires root privileges.
8654
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008655 The "source" keyword is useful in complex environments where a specific
8656 address only is allowed to connect to the servers. It may be needed when a
8657 private address must be used through a public gateway for instance, and it is
8658 known that the system cannot determine the adequate source address by itself.
8659
8660 An extension which is available on certain patched Linux kernels may be used
8661 through the "usesrc" optional keyword. It makes it possible to connect to the
8662 servers with an IP address which does not belong to the system itself. This
8663 is called "full transparent proxy mode". For this to work, the destination
8664 servers have to route their traffic back to this address through the machine
8665 running HAProxy, and IP forwarding must generally be enabled on this machine.
8666
8667 In this "full transparent proxy" mode, it is possible to force a specific IP
8668 address to be presented to the servers. This is not much used in fact. A more
8669 common use is to tell HAProxy to present the client's IP address. For this,
8670 there are two methods :
8671
8672 - present the client's IP and port addresses. This is the most transparent
8673 mode, but it can cause problems when IP connection tracking is enabled on
8674 the machine, because a same connection may be seen twice with different
8675 states. However, this solution presents the huge advantage of not
8676 limiting the system to the 64k outgoing address+port couples, because all
8677 of the client ranges may be used.
8678
8679 - present only the client's IP address and select a spare port. This
8680 solution is still quite elegant but slightly less transparent (downstream
8681 firewalls logs will not match upstream's). It also presents the downside
8682 of limiting the number of concurrent connections to the usual 64k ports.
8683 However, since the upstream and downstream ports are different, local IP
8684 connection tracking on the machine will not be upset by the reuse of the
8685 same session.
8686
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008687 This option sets the default source for all servers in the backend. It may
8688 also be specified in a "defaults" section. Finer source address specification
8689 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008690 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008691
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008692 In order to work, "usesrc" requires root privileges.
8693
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008694 Examples :
8695 backend private
8696 # Connect to the servers using our 192.168.1.200 source address
8697 source 192.168.1.200
8698
8699 backend transparent_ssl1
8700 # Connect to the SSL farm from the client's source address
8701 source 192.168.1.200 usesrc clientip
8702
8703 backend transparent_ssl2
8704 # Connect to the SSL farm from the client's source address and port
8705 # not recommended if IP conntrack is present on the local machine.
8706 source 192.168.1.200 usesrc client
8707
8708 backend transparent_ssl3
8709 # Connect to the SSL farm from the client's source address. It
8710 # is more conntrack-friendly.
8711 source 192.168.1.200 usesrc clientip
8712
8713 backend transparent_smtp
8714 # Connect to the SMTP farm from the client's source address/port
8715 # with Tproxy version 4.
8716 source 0.0.0.0 usesrc clientip
8717
Willy Tarreaubce70882009-09-07 11:51:47 +02008718 backend transparent_http
8719 # Connect to the servers using the client's IP as seen by previous
8720 # proxy.
8721 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8722
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008723 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008724 the Linux kernel on www.balabit.com, the "bind" keyword.
8725
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008726
Willy Tarreau844e3c52008-01-11 16:28:18 +01008727srvtimeout <timeout> (deprecated)
8728 Set the maximum inactivity time on the server side.
8729 May be used in sections : defaults | frontend | listen | backend
8730 yes | no | yes | yes
8731 Arguments :
8732 <timeout> is the timeout value specified in milliseconds by default, but
8733 can be in any other unit if the number is suffixed by the unit,
8734 as explained at the top of this document.
8735
8736 The inactivity timeout applies when the server is expected to acknowledge or
8737 send data. In HTTP mode, this timeout is particularly important to consider
8738 during the first phase of the server's response, when it has to send the
8739 headers, as it directly represents the server's processing time for the
8740 request. To find out what value to put there, it's often good to start with
8741 what would be considered as unacceptable response times, then check the logs
8742 to observe the response time distribution, and adjust the value accordingly.
8743
8744 The value is specified in milliseconds by default, but can be in any other
8745 unit if the number is suffixed by the unit, as specified at the top of this
8746 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
8747 recommended that the client timeout remains equal to the server timeout in
8748 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008749 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01008750 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +01008751 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01008752
8753 This parameter is specific to backends, but can be specified once for all in
8754 "defaults" sections. This is in fact one of the easiest solutions not to
8755 forget about it. An unspecified timeout results in an infinite timeout, which
8756 is not recommended. Such a usage is accepted and works but reports a warning
8757 during startup because it may results in accumulation of expired sessions in
8758 the system if the system's timeouts are not configured either.
8759
8760 This parameter is provided for compatibility but is currently deprecated.
8761 Please use "timeout server" instead.
8762
Willy Tarreauce887fd2012-05-12 12:50:00 +02008763 See also : "timeout server", "timeout tunnel", "timeout client" and
8764 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01008765
8766
Cyril Bonté66c327d2010-10-12 00:14:37 +02008767stats admin { if | unless } <cond>
8768 Enable statistics admin level if/unless a condition is matched
8769 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008770 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008771
8772 This statement enables the statistics admin level if/unless a condition is
8773 matched.
8774
8775 The admin level allows to enable/disable servers from the web interface. By
8776 default, statistics page is read-only for security reasons.
8777
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008778 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8779 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008780 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008781
Cyril Bonté23b39d92011-02-10 22:54:44 +01008782 Currently, the POST request is limited to the buffer size minus the reserved
8783 buffer space, which means that if the list of servers is too long, the
8784 request won't be processed. It is recommended to alter few servers at a
8785 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008786
8787 Example :
8788 # statistics admin level only for localhost
8789 backend stats_localhost
8790 stats enable
8791 stats admin if LOCALHOST
8792
8793 Example :
8794 # statistics admin level always enabled because of the authentication
8795 backend stats_auth
8796 stats enable
8797 stats auth admin:AdMiN123
8798 stats admin if TRUE
8799
8800 Example :
8801 # statistics admin level depends on the authenticated user
8802 userlist stats-auth
8803 group admin users admin
8804 user admin insecure-password AdMiN123
8805 group readonly users haproxy
8806 user haproxy insecure-password haproxy
8807
8808 backend stats_auth
8809 stats enable
8810 acl AUTH http_auth(stats-auth)
8811 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8812 stats http-request auth unless AUTH
8813 stats admin if AUTH_ADMIN
8814
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008815 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8816 "bind-process", section 3.4 about userlists and section 7 about
8817 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008818
8819
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008820stats auth <user>:<passwd>
8821 Enable statistics with authentication and grant access to an account
8822 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008823 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008824 Arguments :
8825 <user> is a user name to grant access to
8826
8827 <passwd> is the cleartext password associated to this user
8828
8829 This statement enables statistics with default settings, and restricts access
8830 to declared users only. It may be repeated as many times as necessary to
8831 allow as many users as desired. When a user tries to access the statistics
8832 without a valid account, a "401 Forbidden" response will be returned so that
8833 the browser asks the user to provide a valid user and password. The real
8834 which will be returned to the browser is configurable using "stats realm".
8835
8836 Since the authentication method is HTTP Basic Authentication, the passwords
8837 circulate in cleartext on the network. Thus, it was decided that the
8838 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008839 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008840
8841 It is also possible to reduce the scope of the proxies which appear in the
8842 report using "stats scope".
8843
8844 Though this statement alone is enough to enable statistics reporting, it is
8845 recommended to set all other settings in order to avoid relying on default
8846 unobvious parameters.
8847
8848 Example :
8849 # public access (limited to this backend only)
8850 backend public_www
8851 server srv1 192.168.0.1:80
8852 stats enable
8853 stats hide-version
8854 stats scope .
8855 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008856 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008857 stats auth admin1:AdMiN123
8858 stats auth admin2:AdMiN321
8859
8860 # internal monitoring access (unlimited)
8861 backend private_monitoring
8862 stats enable
8863 stats uri /admin?stats
8864 stats refresh 5s
8865
8866 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8867
8868
8869stats enable
8870 Enable statistics reporting with default settings
8871 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008872 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008873 Arguments : none
8874
8875 This statement enables statistics reporting with default settings defined
8876 at build time. Unless stated otherwise, these settings are used :
8877 - stats uri : /haproxy?stats
8878 - stats realm : "HAProxy Statistics"
8879 - stats auth : no authentication
8880 - stats scope : no restriction
8881
8882 Though this statement alone is enough to enable statistics reporting, it is
8883 recommended to set all other settings in order to avoid relying on default
8884 unobvious parameters.
8885
8886 Example :
8887 # public access (limited to this backend only)
8888 backend public_www
8889 server srv1 192.168.0.1:80
8890 stats enable
8891 stats hide-version
8892 stats scope .
8893 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008894 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008895 stats auth admin1:AdMiN123
8896 stats auth admin2:AdMiN321
8897
8898 # internal monitoring access (unlimited)
8899 backend private_monitoring
8900 stats enable
8901 stats uri /admin?stats
8902 stats refresh 5s
8903
8904 See also : "stats auth", "stats realm", "stats uri"
8905
8906
Willy Tarreaud63335a2010-02-26 12:56:52 +01008907stats hide-version
8908 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008909 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008910 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008911 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008912
Willy Tarreaud63335a2010-02-26 12:56:52 +01008913 By default, the stats page reports some useful status information along with
8914 the statistics. Among them is HAProxy's version. However, it is generally
8915 considered dangerous to report precise version to anyone, as it can help them
8916 target known weaknesses with specific attacks. The "stats hide-version"
8917 statement removes the version from the statistics report. This is recommended
8918 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008919
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008920 Though this statement alone is enough to enable statistics reporting, it is
8921 recommended to set all other settings in order to avoid relying on default
8922 unobvious parameters.
8923
Willy Tarreaud63335a2010-02-26 12:56:52 +01008924 Example :
8925 # public access (limited to this backend only)
8926 backend public_www
8927 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008928 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008929 stats hide-version
8930 stats scope .
8931 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008932 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008933 stats auth admin1:AdMiN123
8934 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008935
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008936 # internal monitoring access (unlimited)
8937 backend private_monitoring
8938 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008939 stats uri /admin?stats
8940 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008941
Willy Tarreaud63335a2010-02-26 12:56:52 +01008942 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008943
Willy Tarreau983e01e2010-01-11 18:42:06 +01008944
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008945stats http-request { allow | deny | auth [realm <realm>] }
8946 [ { if | unless } <condition> ]
8947 Access control for statistics
8948
8949 May be used in sections: defaults | frontend | listen | backend
8950 no | no | yes | yes
8951
8952 As "http-request", these set of options allow to fine control access to
8953 statistics. Each option may be followed by if/unless and acl.
8954 First option with matched condition (or option without condition) is final.
8955 For "deny" a 403 error will be returned, for "allow" normal processing is
8956 performed, for "auth" a 401/407 error code is returned so the client
8957 should be asked to enter a username and password.
8958
8959 There is no fixed limit to the number of http-request statements per
8960 instance.
8961
8962 See also : "http-request", section 3.4 about userlists and section 7
8963 about ACL usage.
8964
8965
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008966stats realm <realm>
8967 Enable statistics and set authentication realm
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 :
8971 <realm> is the name of the HTTP Basic Authentication realm reported to
8972 the browser. The browser uses it to display it in the pop-up
8973 inviting the user to enter a valid username and password.
8974
8975 The realm is read as a single word, so any spaces in it should be escaped
8976 using a backslash ('\').
8977
8978 This statement is useful only in conjunction with "stats auth" since it is
8979 only related to authentication.
8980
8981 Though this statement alone is enough to enable statistics reporting, it is
8982 recommended to set all other settings in order to avoid relying on default
8983 unobvious parameters.
8984
8985 Example :
8986 # public access (limited to this backend only)
8987 backend public_www
8988 server srv1 192.168.0.1:80
8989 stats enable
8990 stats hide-version
8991 stats scope .
8992 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008993 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008994 stats auth admin1:AdMiN123
8995 stats auth admin2:AdMiN321
8996
8997 # internal monitoring access (unlimited)
8998 backend private_monitoring
8999 stats enable
9000 stats uri /admin?stats
9001 stats refresh 5s
9002
9003 See also : "stats auth", "stats enable", "stats uri"
9004
9005
9006stats refresh <delay>
9007 Enable statistics with automatic refresh
9008 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009009 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009010 Arguments :
9011 <delay> is the suggested refresh delay, specified in seconds, which will
9012 be returned to the browser consulting the report page. While the
9013 browser is free to apply any delay, it will generally respect it
9014 and refresh the page this every seconds. The refresh interval may
9015 be specified in any other non-default time unit, by suffixing the
9016 unit after the value, as explained at the top of this document.
9017
9018 This statement is useful on monitoring displays with a permanent page
9019 reporting the load balancer's activity. When set, the HTML report page will
9020 include a link "refresh"/"stop refresh" so that the user can select whether
9021 he wants automatic refresh of the page or not.
9022
9023 Though this statement alone is enough to enable statistics reporting, it is
9024 recommended to set all other settings in order to avoid relying on default
9025 unobvious parameters.
9026
9027 Example :
9028 # public access (limited to this backend only)
9029 backend public_www
9030 server srv1 192.168.0.1:80
9031 stats enable
9032 stats hide-version
9033 stats scope .
9034 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009035 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009036 stats auth admin1:AdMiN123
9037 stats auth admin2:AdMiN321
9038
9039 # internal monitoring access (unlimited)
9040 backend private_monitoring
9041 stats enable
9042 stats uri /admin?stats
9043 stats refresh 5s
9044
9045 See also : "stats auth", "stats enable", "stats realm", "stats uri"
9046
9047
9048stats scope { <name> | "." }
9049 Enable statistics and limit access scope
9050 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009051 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009052 Arguments :
9053 <name> is the name of a listen, frontend or backend section to be
9054 reported. The special name "." (a single dot) designates the
9055 section in which the statement appears.
9056
9057 When this statement is specified, only the sections enumerated with this
9058 statement will appear in the report. All other ones will be hidden. This
9059 statement may appear as many times as needed if multiple sections need to be
9060 reported. Please note that the name checking is performed as simple string
9061 comparisons, and that it is never checked that a give section name really
9062 exists.
9063
9064 Though this statement alone is enough to enable statistics reporting, it is
9065 recommended to set all other settings in order to avoid relying on default
9066 unobvious parameters.
9067
9068 Example :
9069 # public access (limited to this backend only)
9070 backend public_www
9071 server srv1 192.168.0.1:80
9072 stats enable
9073 stats hide-version
9074 stats scope .
9075 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009076 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009077 stats auth admin1:AdMiN123
9078 stats auth admin2:AdMiN321
9079
9080 # internal monitoring access (unlimited)
9081 backend private_monitoring
9082 stats enable
9083 stats uri /admin?stats
9084 stats refresh 5s
9085
9086 See also : "stats auth", "stats enable", "stats realm", "stats uri"
9087
Willy Tarreaud63335a2010-02-26 12:56:52 +01009088
Willy Tarreauc9705a12010-07-27 20:05:50 +02009089stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01009090 Enable reporting of a description on the statistics page.
9091 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009092 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01009093
Willy Tarreauc9705a12010-07-27 20:05:50 +02009094 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01009095 description from global section is automatically used instead.
9096
9097 This statement is useful for users that offer shared services to their
9098 customers, where node or description should be different for each customer.
9099
9100 Though this statement alone is enough to enable statistics reporting, it is
9101 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01009102 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01009103
9104 Example :
9105 # internal monitoring access (unlimited)
9106 backend private_monitoring
9107 stats enable
9108 stats show-desc Master node for Europe, Asia, Africa
9109 stats uri /admin?stats
9110 stats refresh 5s
9111
9112 See also: "show-node", "stats enable", "stats uri" and "description" in
9113 global section.
9114
9115
9116stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02009117 Enable reporting additional information on the statistics page
9118 May be used in sections : defaults | frontend | listen | backend
9119 yes | yes | yes | yes
9120 Arguments : none
9121
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009122 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01009123 - cap: capabilities (proxy)
9124 - mode: one of tcp, http or health (proxy)
9125 - id: SNMP ID (proxy, socket, server)
9126 - IP (socket, server)
9127 - cookie (backend, server)
9128
9129 Though this statement alone is enough to enable statistics reporting, it is
9130 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01009131 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01009132
9133 See also: "stats enable", "stats uri".
9134
9135
9136stats show-node [ <name> ]
9137 Enable reporting of a host name on the statistics page.
9138 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009139 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01009140 Arguments:
9141 <name> is an optional name to be reported. If unspecified, the
9142 node name from global section is automatically used instead.
9143
9144 This statement is useful for users that offer shared services to their
9145 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01009146 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01009147
9148 Though this statement alone is enough to enable statistics reporting, it is
9149 recommended to set all other settings in order to avoid relying on default
9150 unobvious parameters.
9151
9152 Example:
9153 # internal monitoring access (unlimited)
9154 backend private_monitoring
9155 stats enable
9156 stats show-node Europe-1
9157 stats uri /admin?stats
9158 stats refresh 5s
9159
9160 See also: "show-desc", "stats enable", "stats uri", and "node" in global
9161 section.
9162
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009163
9164stats uri <prefix>
9165 Enable statistics and define the URI prefix to access them
9166 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02009167 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009168 Arguments :
9169 <prefix> is the prefix of any URI which will be redirected to stats. This
9170 prefix may contain a question mark ('?') to indicate part of a
9171 query string.
9172
9173 The statistics URI is intercepted on the relayed traffic, so it appears as a
9174 page within the normal application. It is strongly advised to ensure that the
9175 selected URI will never appear in the application, otherwise it will never be
9176 possible to reach it in the application.
9177
9178 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009179 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009180 It is generally a good idea to include a question mark in the URI so that
9181 intermediate proxies refrain from caching the results. Also, since any string
9182 beginning with the prefix will be accepted as a stats request, the question
9183 mark helps ensuring that no valid URI will begin with the same words.
9184
9185 It is sometimes very convenient to use "/" as the URI prefix, and put that
9186 statement in a "listen" instance of its own. That makes it easy to dedicate
9187 an address or a port to statistics only.
9188
9189 Though this statement alone is enough to enable statistics reporting, it is
9190 recommended to set all other settings in order to avoid relying on default
9191 unobvious parameters.
9192
9193 Example :
9194 # public access (limited to this backend only)
9195 backend public_www
9196 server srv1 192.168.0.1:80
9197 stats enable
9198 stats hide-version
9199 stats scope .
9200 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01009201 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009202 stats auth admin1:AdMiN123
9203 stats auth admin2:AdMiN321
9204
9205 # internal monitoring access (unlimited)
9206 backend private_monitoring
9207 stats enable
9208 stats uri /admin?stats
9209 stats refresh 5s
9210
9211 See also : "stats auth", "stats enable", "stats realm"
9212
9213
Willy Tarreaud63335a2010-02-26 12:56:52 +01009214stick match <pattern> [table <table>] [{if | unless} <cond>]
9215 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01009216 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01009217 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009218
9219 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009220 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009221 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01009222 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009223 stickiness table. This rule is mandatory.
9224
9225 <table> is an optional stickiness table name. If unspecified, the same
9226 backend's table is used. A stickiness table is declared using
9227 the "stick-table" statement.
9228
9229 <cond> is an optional matching condition. It makes it possible to match
9230 on a certain criterion only when other conditions are met (or
9231 not met). For instance, it could be used to match on a source IP
9232 address except when a request passes through a known proxy, in
9233 which case we'd match on a header containing that IP address.
9234
9235 Some protocols or applications require complex stickiness rules and cannot
9236 always simply rely on cookies nor hashing. The "stick match" statement
9237 describes a rule to extract the stickiness criterion from an incoming request
9238 or connection. See section 7 for a complete list of possible patterns and
9239 transformation rules.
9240
9241 The table has to be declared using the "stick-table" statement. It must be of
9242 a type compatible with the pattern. By default it is the one which is present
9243 in the same backend. It is possible to share a table with other backends by
9244 referencing it using the "table" keyword. If another table is referenced,
9245 the server's ID inside the backends are used. By default, all server IDs
9246 start at 1 in each backend, so the server ordering is enough. But in case of
9247 doubt, it is highly recommended to force server IDs using their "id" setting.
9248
9249 It is possible to restrict the conditions where a "stick match" statement
9250 will apply, using "if" or "unless" followed by a condition. See section 7 for
9251 ACL based conditions.
9252
9253 There is no limit on the number of "stick match" statements. The first that
9254 applies and matches will cause the request to be directed to the same server
9255 as was used for the request which created the entry. That way, multiple
9256 matches can be used as fallbacks.
9257
9258 The stick rules are checked after the persistence cookies, so they will not
9259 affect stickiness if a cookie has already been used to select a server. That
9260 way, it becomes very easy to insert cookies and match on IP addresses in
9261 order to maintain stickiness between HTTP and HTTPS.
9262
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009263 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9264 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009265 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009266
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009267 Example :
9268 # forward SMTP users to the same server they just used for POP in the
9269 # last 30 minutes
9270 backend pop
9271 mode tcp
9272 balance roundrobin
9273 stick store-request src
9274 stick-table type ip size 200k expire 30m
9275 server s1 192.168.1.1:110
9276 server s2 192.168.1.1:110
9277
9278 backend smtp
9279 mode tcp
9280 balance roundrobin
9281 stick match src table pop
9282 server s1 192.168.1.1:25
9283 server s2 192.168.1.1:25
9284
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009285 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02009286 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009287
9288
9289stick on <pattern> [table <table>] [{if | unless} <condition>]
9290 Define a request pattern to associate a user to a server
9291 May be used in sections : defaults | frontend | listen | backend
9292 no | no | yes | yes
9293
9294 Note : This form is exactly equivalent to "stick match" followed by
9295 "stick store-request", all with the same arguments. Please refer
9296 to both keywords for details. It is only provided as a convenience
9297 for writing more maintainable configurations.
9298
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009299 Note : Consider not using this feature in multi-process mode (nbproc > 1)
9300 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01009301 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009302
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009303 Examples :
9304 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01009305 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009306
9307 # ...is strictly equivalent to this one :
9308 stick match src table pop if !localhost
9309 stick store-request src table pop if !localhost
9310
9311
9312 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
9313 # well as HTTP without cookie. Share the same table between both accesses.
9314 backend http
9315 mode http
9316 balance roundrobin
9317 stick on src table https
9318 cookie SRV insert indirect nocache
9319 server s1 192.168.1.1:80 cookie s1
9320 server s2 192.168.1.1:80 cookie s2
9321
9322 backend https
9323 mode tcp
9324 balance roundrobin
9325 stick-table type ip size 200k expire 30m
9326 stick on src
9327 server s1 192.168.1.1:443
9328 server s2 192.168.1.1:443
9329
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009330 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009331
9332
9333stick store-request <pattern> [table <table>] [{if | unless} <condition>]
9334 Define a request pattern used to create an entry in a stickiness table
9335 May be used in sections : defaults | frontend | listen | backend
9336 no | no | yes | yes
9337
9338 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009339 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009340 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01009341 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009342 server is selected.
9343
9344 <table> is an optional stickiness table name. If unspecified, the same
9345 backend's table is used. A stickiness table is declared using
9346 the "stick-table" statement.
9347
9348 <cond> is an optional storage condition. It makes it possible to store
9349 certain criteria only when some conditions are met (or not met).
9350 For instance, it could be used to store the source IP address
9351 except when the request passes through a known proxy, in which
9352 case we'd store a converted form of a header containing that IP
9353 address.
9354
9355 Some protocols or applications require complex stickiness rules and cannot
9356 always simply rely on cookies nor hashing. The "stick store-request" statement
9357 describes a rule to decide what to extract from the request and when to do
9358 it, in order to store it into a stickiness table for further requests to
9359 match it using the "stick match" statement. Obviously the extracted part must
9360 make sense and have a chance to be matched in a further request. Storing a
9361 client's IP address for instance often makes sense. Storing an ID found in a
9362 URL parameter also makes sense. Storing a source port will almost never make
9363 any sense because it will be randomly matched. See section 7 for a complete
9364 list of possible patterns and transformation rules.
9365
9366 The table has to be declared using the "stick-table" statement. It must be of
9367 a type compatible with the pattern. By default it is the one which is present
9368 in the same backend. It is possible to share a table with other backends by
9369 referencing it using the "table" keyword. If another table is referenced,
9370 the server's ID inside the backends are used. By default, all server IDs
9371 start at 1 in each backend, so the server ordering is enough. But in case of
9372 doubt, it is highly recommended to force server IDs using their "id" setting.
9373
9374 It is possible to restrict the conditions where a "stick store-request"
9375 statement will apply, using "if" or "unless" followed by a condition. This
9376 condition will be evaluated while parsing the request, so any criteria can be
9377 used. See section 7 for ACL based conditions.
9378
9379 There is no limit on the number of "stick store-request" statements, but
9380 there is a limit of 8 simultaneous stores per request or response. This
9381 makes it possible to store up to 8 criteria, all extracted from either the
9382 request or the response, regardless of the number of rules. Only the 8 first
9383 ones which match will be kept. Using this, it is possible to feed multiple
9384 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009385 another protocol or access method. Using multiple store-request rules with
9386 the same table is possible and may be used to find the best criterion to rely
9387 on, by arranging the rules by decreasing preference order. Only the first
9388 extracted criterion for a given table will be stored. All subsequent store-
9389 request rules referencing the same table will be skipped and their ACLs will
9390 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009391
9392 The "store-request" rules are evaluated once the server connection has been
9393 established, so that the table will contain the real server that processed
9394 the request.
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 Example :
9401 # forward SMTP users to the same server they just used for POP in the
9402 # last 30 minutes
9403 backend pop
9404 mode tcp
9405 balance roundrobin
9406 stick store-request src
9407 stick-table type ip size 200k expire 30m
9408 server s1 192.168.1.1:110
9409 server s2 192.168.1.1:110
9410
9411 backend smtp
9412 mode tcp
9413 balance roundrobin
9414 stick match src table pop
9415 server s1 192.168.1.1:25
9416 server s2 192.168.1.1:25
9417
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009418 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02009419 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009420
9421
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009422stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02009423 size <size> [expire <expire>] [nopurge] [peers <peersect>]
9424 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08009425 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009426 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009427 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009428
9429 Arguments :
9430 ip a table declared with "type ip" will only store IPv4 addresses.
9431 This form is very compact (about 50 bytes per entry) and allows
9432 very fast entry lookup and stores with almost no overhead. This
9433 is mainly used to store client source IP addresses.
9434
David du Colombier9a6d3c92011-03-17 10:40:24 +01009435 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
9436 This form is very compact (about 60 bytes per entry) and allows
9437 very fast entry lookup and stores with almost no overhead. This
9438 is mainly used to store client source IP addresses.
9439
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009440 integer a table declared with "type integer" will store 32bit integers
9441 which can represent a client identifier found in a request for
9442 instance.
9443
9444 string a table declared with "type string" will store substrings of up
9445 to <len> characters. If the string provided by the pattern
9446 extractor is larger than <len>, it will be truncated before
9447 being stored. During matching, at most <len> characters will be
9448 compared between the string in the table and the extracted
9449 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009450 to 32 characters.
9451
9452 binary a table declared with "type binary" will store binary blocks
9453 of <len> bytes. If the block provided by the pattern
9454 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02009455 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009456 is shorter than <len>, it will be padded by 0. When not
9457 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009458
9459 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02009460 "string" type table (See type "string" above). Or the number
9461 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009462 changing this parameter as memory usage will proportionally
9463 increase.
9464
9465 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01009466 value directly impacts memory usage. Count approximately
9467 50 bytes per entry, plus the size of a string if any. The size
9468 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009469
9470 [nopurge] indicates that we refuse to purge older entries when the table
9471 is full. When not specified and the table is full when haproxy
9472 wants to store an entry in it, it will flush a few of the oldest
9473 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01009474 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009475 be desirable to refuse new entries instead of purging the older
9476 ones. That may be the case when the amount of data to store is
9477 far above the hardware limits and we prefer not to offer access
9478 to new clients than to reject the ones already connected. When
9479 using this parameter, be sure to properly set the "expire"
9480 parameter (see below).
9481
Emeric Brunf099e792010-09-27 12:05:28 +02009482 <peersect> is the name of the peers section to use for replication. Entries
9483 which associate keys to server IDs are kept synchronized with
9484 the remote peers declared in this section. All entries are also
9485 automatically learned from the local peer (old process) during a
9486 soft restart.
9487
Willy Tarreau1abc6732015-05-01 19:21:02 +02009488 NOTE : each peers section may be referenced only by tables
9489 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01009490
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009491 <expire> defines the maximum duration of an entry in the table since it
9492 was last created, refreshed or matched. The expiration delay is
9493 defined using the standard time format, similarly as the various
9494 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009495 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02009496 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009497 be removed once full. Be sure not to use the "nopurge" parameter
9498 if not expiration delay is specified.
9499
Willy Tarreau08d5f982010-06-06 13:34:54 +02009500 <data_type> is used to store additional information in the stick-table. This
9501 may be used by ACLs in order to control various criteria related
9502 to the activity of the client matching the stick-table. For each
9503 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02009504 that the additional data can fit. Several data types may be
9505 stored with an entry. Multiple data types may be specified after
9506 the "store" keyword, as a comma-separated list. Alternatively,
9507 it is possible to repeat the "store" keyword followed by one or
9508 several data types. Except for the "server_id" type which is
9509 automatically detected and enabled, all data types must be
9510 explicitly declared to be stored. If an ACL references a data
9511 type which is not stored, the ACL will simply not match. Some
9512 data types require an argument which must be passed just after
9513 the type between parenthesis. See below for the supported data
9514 types and their arguments.
9515
9516 The data types that can be stored with an entry are the following :
9517 - server_id : this is an integer which holds the numeric ID of the server a
9518 request was assigned to. It is used by the "stick match", "stick store",
9519 and "stick on" rules. It is automatically enabled when referenced.
9520
9521 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
9522 integer which may be used for anything. Most of the time it will be used
9523 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01009524 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02009525
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009526 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
9527 over a period. It is a positive 32-bit integer integer which may be used
9528 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01009529 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009530 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01009531 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009532
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009533 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
9534 integer which may be used for anything. Most of the time it will be used
9535 to put a special tag on some entries, for instance to note that a
9536 specific behavior was detected and must be known for future matches.
9537
9538 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
9539 over a period. It is a positive 32-bit integer integer which may be used
9540 for anything. Just like <gpc1>, it counts events, but instead of keeping
9541 a cumulative number, it maintains the rate at which the counter is
9542 incremented. Most of the time it will be used to measure the frequency of
9543 occurrence of certain events (e.g. requests to a specific URL).
9544
Willy Tarreauc9705a12010-07-27 20:05:50 +02009545 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
9546 the absolute number of connections received from clients which matched
9547 this entry. It does not mean the connections were accepted, just that
9548 they were received.
9549
9550 - conn_cur : Current Connections. It is a positive 32-bit integer which
9551 stores the concurrent connection counts for the entry. It is incremented
9552 once an incoming connection matches the entry, and decremented once the
9553 connection leaves. That way it is possible to know at any time the exact
9554 number of concurrent connections for an entry.
9555
9556 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9557 integer parameter <period> which indicates in milliseconds the length
9558 of the period over which the average is measured. It reports the average
9559 incoming connection rate over that period, in connections per period. The
9560 result is an integer which can be matched using ACLs.
9561
9562 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
9563 the absolute number of sessions received from clients which matched this
9564 entry. A session is a connection that was accepted by the layer 4 rules.
9565
9566 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9567 integer parameter <period> which indicates in milliseconds the length
9568 of the period over which the average is measured. It reports the average
9569 incoming session rate over that period, in sessions per period. The
9570 result is an integer which can be matched using ACLs.
9571
9572 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
9573 counts the absolute number of HTTP requests received from clients which
9574 matched this entry. It does not matter whether they are valid requests or
9575 not. Note that this is different from sessions when keep-alive is used on
9576 the client side.
9577
9578 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9579 integer parameter <period> which indicates in milliseconds the length
9580 of the period over which the average is measured. It reports the average
9581 HTTP request rate over that period, in requests per period. The result is
9582 an integer which can be matched using ACLs. It does not matter whether
9583 they are valid requests or not. Note that this is different from sessions
9584 when keep-alive is used on the client side.
9585
9586 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
9587 counts the absolute number of HTTP requests errors induced by clients
9588 which matched this entry. Errors are counted on invalid and truncated
9589 requests, as well as on denied or tarpitted requests, and on failed
9590 authentications. If the server responds with 4xx, then the request is
9591 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01009592 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02009593
9594 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9595 integer parameter <period> which indicates in milliseconds the length
9596 of the period over which the average is measured. It reports the average
9597 HTTP request error rate over that period, in requests per period (see
9598 http_err_cnt above for what is accounted as an error). The result is an
9599 integer which can be matched using ACLs.
9600
9601 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009602 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02009603 which matched this entry. Headers are included in the count. This may be
9604 used to limit abuse of upload features on photo or video servers.
9605
9606 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9607 integer parameter <period> which indicates in milliseconds the length
9608 of the period over which the average is measured. It reports the average
9609 incoming bytes rate over that period, in bytes per period. It may be used
9610 to detect users which upload too much and too fast. Warning: with large
9611 uploads, it is possible that the amount of uploaded data will be counted
9612 once upon termination, thus causing spikes in the average transfer speed
9613 instead of having a smooth one. This may partially be smoothed with
9614 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
9615 recommended for better fairness.
9616
9617 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009618 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02009619 matched this entry. Headers are included in the count. This may be used
9620 to limit abuse of bots sucking the whole site.
9621
9622 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
9623 an integer parameter <period> which indicates in milliseconds the length
9624 of the period over which the average is measured. It reports the average
9625 outgoing bytes rate over that period, in bytes per period. It may be used
9626 to detect users which download too much and too fast. Warning: with large
9627 transfers, it is possible that the amount of transferred data will be
9628 counted once upon termination, thus causing spikes in the average
9629 transfer speed instead of having a smooth one. This may partially be
9630 smoothed with "option contstats" though this is not perfect yet. Use of
9631 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02009632
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009633 There is only one stick-table per proxy. At the moment of writing this doc,
9634 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009635 to be required, simply create a dummy backend with a stick-table in it and
9636 reference it.
9637
9638 It is important to understand that stickiness based on learning information
9639 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01009640 lost upon restart unless peers are properly configured to transfer such
9641 information upon restart (recommended). In general it can be good as a
9642 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009643
Willy Tarreauc9705a12010-07-27 20:05:50 +02009644 Last, memory requirements may be important when storing many data types.
9645 Indeed, storing all indicators above at once in each entry requires 116 bytes
9646 per entry, or 116 MB for a 1-million entries table. This is definitely not
9647 something that can be ignored.
9648
9649 Example:
9650 # Keep track of counters of up to 1 million IP addresses over 5 minutes
9651 # and store a general purpose counter and the average connection rate
9652 # computed over a sliding window of 30 seconds.
9653 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
9654
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009655 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01009656 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009657
9658
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009659stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01009660 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009661 May be used in sections : defaults | frontend | listen | backend
9662 no | no | yes | yes
9663
9664 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009665 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009666 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009667 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009668 server is selected.
9669
9670 <table> is an optional stickiness table name. If unspecified, the same
9671 backend's table is used. A stickiness table is declared using
9672 the "stick-table" statement.
9673
9674 <cond> is an optional storage condition. It makes it possible to store
9675 certain criteria only when some conditions are met (or not met).
9676 For instance, it could be used to store the SSL session ID only
9677 when the response is a SSL server hello.
9678
9679 Some protocols or applications require complex stickiness rules and cannot
9680 always simply rely on cookies nor hashing. The "stick store-response"
9681 statement describes a rule to decide what to extract from the response and
9682 when to do it, in order to store it into a stickiness table for further
9683 requests to match it using the "stick match" statement. Obviously the
9684 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009685 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009686 See section 7 for a complete list of possible patterns and transformation
9687 rules.
9688
9689 The table has to be declared using the "stick-table" statement. It must be of
9690 a type compatible with the pattern. By default it is the one which is present
9691 in the same backend. It is possible to share a table with other backends by
9692 referencing it using the "table" keyword. If another table is referenced,
9693 the server's ID inside the backends are used. By default, all server IDs
9694 start at 1 in each backend, so the server ordering is enough. But in case of
9695 doubt, it is highly recommended to force server IDs using their "id" setting.
9696
9697 It is possible to restrict the conditions where a "stick store-response"
9698 statement will apply, using "if" or "unless" followed by a condition. This
9699 condition will be evaluated while parsing the response, so any criteria can
9700 be used. See section 7 for ACL based conditions.
9701
9702 There is no limit on the number of "stick store-response" statements, but
9703 there is a limit of 8 simultaneous stores per request or response. This
9704 makes it possible to store up to 8 criteria, all extracted from either the
9705 request or the response, regardless of the number of rules. Only the 8 first
9706 ones which match will be kept. Using this, it is possible to feed multiple
9707 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009708 another protocol or access method. Using multiple store-response rules with
9709 the same table is possible and may be used to find the best criterion to rely
9710 on, by arranging the rules by decreasing preference order. Only the first
9711 extracted criterion for a given table will be stored. All subsequent store-
9712 response rules referencing the same table will be skipped and their ACLs will
9713 not be evaluated. However, even if a store-request rule references a table, a
9714 store-response rule may also use the same table. This means that each table
9715 may learn exactly one element from the request and one element from the
9716 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009717
9718 The table will contain the real server that processed the request.
9719
9720 Example :
9721 # Learn SSL session ID from both request and response and create affinity.
9722 backend https
9723 mode tcp
9724 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009725 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009726 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009727
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009728 acl clienthello req_ssl_hello_type 1
9729 acl serverhello rep_ssl_hello_type 2
9730
9731 # use tcp content accepts to detects ssl client and server hello.
9732 tcp-request inspect-delay 5s
9733 tcp-request content accept if clienthello
9734
9735 # no timeout on response inspect delay by default.
9736 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009737
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009738 # SSL session ID (SSLID) may be present on a client or server hello.
9739 # Its length is coded on 1 byte at offset 43 and its value starts
9740 # at offset 44.
9741
9742 # Match and learn on request if client hello.
9743 stick on payload_lv(43,1) if clienthello
9744
9745 # Learn on response if server hello.
9746 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009747
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009748 server s1 192.168.1.1:443
9749 server s2 192.168.1.1:443
9750
9751 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9752 extraction.
9753
9754
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009755tcp-check connect [params*]
9756 Opens a new connection
9757 May be used in sections: defaults | frontend | listen | backend
9758 no | no | yes | yes
9759
9760 When an application lies on more than a single TCP port or when HAProxy
9761 load-balance many services in a single backend, it makes sense to probe all
9762 the services individually before considering a server as operational.
9763
9764 When there are no TCP port configured on the server line neither server port
9765 directive, then the 'tcp-check connect port <port>' must be the first step
9766 of the sequence.
9767
9768 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9769 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9770 do.
9771
9772 Parameters :
9773 They are optional and can be used to describe how HAProxy should open and
9774 use the TCP connection.
9775
9776 port if not set, check port or server port is used.
9777 It tells HAProxy where to open the connection to.
9778 <port> must be a valid TCP port source integer, from 1 to 65535.
9779
9780 send-proxy send a PROXY protocol string
9781
9782 ssl opens a ciphered connection
9783
9784 Examples:
9785 # check HTTP and HTTPs services on a server.
9786 # first open port 80 thanks to server line port directive, then
9787 # tcp-check opens port 443, ciphered and run a request on it:
9788 option tcp-check
9789 tcp-check connect
9790 tcp-check send GET\ /\ HTTP/1.0\r\n
9791 tcp-check send Host:\ haproxy.1wt.eu\r\n
9792 tcp-check send \r\n
9793 tcp-check expect rstring (2..|3..)
9794 tcp-check connect port 443 ssl
9795 tcp-check send GET\ /\ HTTP/1.0\r\n
9796 tcp-check send Host:\ haproxy.1wt.eu\r\n
9797 tcp-check send \r\n
9798 tcp-check expect rstring (2..|3..)
9799 server www 10.0.0.1 check port 80
9800
9801 # check both POP and IMAP from a single server:
9802 option tcp-check
9803 tcp-check connect port 110
9804 tcp-check expect string +OK\ POP3\ ready
9805 tcp-check connect port 143
9806 tcp-check expect string *\ OK\ IMAP4\ ready
9807 server mail 10.0.0.1 check
9808
9809 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9810
9811
9812tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009813 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009814 May be used in sections: defaults | frontend | listen | backend
9815 no | no | yes | yes
9816
9817 Arguments :
9818 <match> is a keyword indicating how to look for a specific pattern in the
9819 response. The keyword may be one of "string", "rstring" or
9820 binary.
9821 The keyword may be preceded by an exclamation mark ("!") to negate
9822 the match. Spaces are allowed between the exclamation mark and the
9823 keyword. See below for more details on the supported keywords.
9824
9825 <pattern> is the pattern to look for. It may be a string or a regular
9826 expression. If the pattern contains spaces, they must be escaped
9827 with the usual backslash ('\').
9828 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009829 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009830 two digits will represent a byte. The hexadecimal digits may be
9831 used upper or lower case.
9832
9833
9834 The available matches are intentionally similar to their http-check cousins :
9835
9836 string <string> : test the exact string matches in the response buffer.
9837 A health check response will be considered valid if the
9838 response's buffer contains this exact string. If the
9839 "string" keyword is prefixed with "!", then the response
9840 will be considered invalid if the body contains this
9841 string. This can be used to look for a mandatory pattern
9842 in a protocol response, or to detect a failure when a
9843 specific error appears in a protocol banner.
9844
9845 rstring <regex> : test a regular expression on the response buffer.
9846 A health check response will be considered valid if the
9847 response's buffer matches this expression. If the
9848 "rstring" keyword is prefixed with "!", then the response
9849 will be considered invalid if the body matches the
9850 expression.
9851
9852 binary <hexstring> : test the exact string in its hexadecimal form matches
9853 in the response buffer. A health check response will
9854 be considered valid if the response's buffer contains
9855 this exact hexadecimal string.
9856 Purpose is to match data on binary protocols.
9857
9858 It is important to note that the responses will be limited to a certain size
9859 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9860 Thus, too large responses may not contain the mandatory pattern when using
9861 "string", "rstring" or binary. If a large response is absolutely required, it
9862 is possible to change the default max size by setting the global variable.
9863 However, it is worth keeping in mind that parsing very large responses can
9864 waste some CPU cycles, especially when regular expressions are used, and that
9865 it is always better to focus the checks on smaller resources. Also, in its
9866 current state, the check will not find any string nor regex past a null
9867 character in the response. Similarly it is not possible to request matching
9868 the null character.
9869
9870 Examples :
9871 # perform a POP check
9872 option tcp-check
9873 tcp-check expect string +OK\ POP3\ ready
9874
9875 # perform an IMAP check
9876 option tcp-check
9877 tcp-check expect string *\ OK\ IMAP4\ ready
9878
9879 # look for the redis master server
9880 option tcp-check
9881 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009882 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009883 tcp-check send info\ replication\r\n
9884 tcp-check expect string role:master
9885 tcp-check send QUIT\r\n
9886 tcp-check expect string +OK
9887
9888
9889 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9890 "tcp-check send-binary", "http-check expect", tune.chksize
9891
9892
9893tcp-check send <data>
9894 Specify a string to be sent as a question during a generic health check
9895 May be used in sections: defaults | frontend | listen | backend
9896 no | no | yes | yes
9897
9898 <data> : the data to be sent as a question during a generic health check
9899 session. For now, <data> must be a string.
9900
9901 Examples :
9902 # look for the redis master server
9903 option tcp-check
9904 tcp-check send info\ replication\r\n
9905 tcp-check expect string role:master
9906
9907 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9908 "tcp-check send-binary", tune.chksize
9909
9910
Davor Ocelice9ed2812017-12-25 17:49:28 +01009911tcp-check send-binary <hexstring>
9912 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009913 tcp health check
9914 May be used in sections: defaults | frontend | listen | backend
9915 no | no | yes | yes
9916
9917 <data> : the data to be sent as a question during a generic health check
9918 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009919 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009920 response buffer. A health check response will be considered
9921 valid if the response's buffer contains this exact
9922 hexadecimal string.
9923 Purpose is to send binary data to ask on binary protocols.
9924
9925 Examples :
9926 # redis check in binary
9927 option tcp-check
9928 tcp-check send-binary 50494e470d0a # PING\r\n
9929 tcp-check expect binary 2b504F4e47 # +PONG
9930
9931
9932 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9933 "tcp-check send", tune.chksize
9934
9935
Willy Tarreaue9656522010-08-17 15:40:09 +02009936tcp-request connection <action> [{if | unless} <condition>]
9937 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009938 May be used in sections : defaults | frontend | listen | backend
9939 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009940 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009941 <action> defines the action to perform if the condition applies. See
9942 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009943
Willy Tarreaue9656522010-08-17 15:40:09 +02009944 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009945
9946 Immediately after acceptance of a new incoming connection, it is possible to
9947 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009948 or dropped or have its counters tracked. Those conditions cannot make use of
9949 any data contents because the connection has not been read from yet, and the
9950 buffers are not yet allocated. This is used to selectively and very quickly
9951 accept or drop connections from various sources with a very low overhead. If
9952 some contents need to be inspected in order to take the decision, the
9953 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009954
Willy Tarreaue9656522010-08-17 15:40:09 +02009955 The "tcp-request connection" rules are evaluated in their exact declaration
9956 order. If no rule matches or if there is no rule, the default action is to
9957 accept the incoming connection. There is no specific limit to the number of
9958 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009959
Willy Tarreaua9083d02015-05-08 15:27:59 +02009960 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009961 - accept :
9962 accepts the connection if the condition is true (when used with "if")
9963 or false (when used with "unless"). The first such rule executed ends
9964 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009965
Willy Tarreaue9656522010-08-17 15:40:09 +02009966 - reject :
9967 rejects the connection if the condition is true (when used with "if")
9968 or false (when used with "unless"). The first such rule executed ends
9969 the rules evaluation. Rejected connections do not even become a
9970 session, which is why they are accounted separately for in the stats,
9971 as "denied connections". They are not considered for the session
9972 rate-limit and are not logged either. The reason is that these rules
9973 should only be used to filter extremely high connection rates such as
9974 the ones encountered during a massive DDoS attack. Under these extreme
9975 conditions, the simple action of logging each event would make the
9976 system collapse and would considerably lower the filtering capacity. If
9977 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009978 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009979
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009980 - expect-proxy layer4 :
9981 configures the client-facing connection to receive a PROXY protocol
9982 header before any byte is read from the socket. This is equivalent to
9983 having the "accept-proxy" keyword on the "bind" line, except that using
9984 the TCP rule allows the PROXY protocol to be accepted only for certain
9985 IP address ranges using an ACL. This is convenient when multiple layers
9986 of load balancers are passed through by traffic coming from public
9987 hosts.
9988
Bertrand Jacquin90759682016-06-06 15:35:39 +01009989 - expect-netscaler-cip layer4 :
9990 configures the client-facing connection to receive a NetScaler Client
9991 IP insertion protocol header before any byte is read from the socket.
9992 This is equivalent to having the "accept-netscaler-cip" keyword on the
9993 "bind" line, except that using the TCP rule allows the PROXY protocol
9994 to be accepted only for certain IP address ranges using an ACL. This
9995 is convenient when multiple layers of load balancers are passed
9996 through by traffic coming from public hosts.
9997
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009998 - capture <sample> len <length> :
9999 This only applies to "tcp-request content" rules. It captures sample
10000 expression <sample> from the request buffer, and converts it to a
10001 string of at most <len> characters. The resulting string is stored into
10002 the next request "capture" slot, so it will possibly appear next to
10003 some captured HTTP headers. It will then automatically appear in the
10004 logs, and it will be possible to extract it using sample fetch rules to
10005 feed it into headers or anything. The length should be limited given
10006 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +020010007 session life. Please check section 7.3 (Fetching samples) and "capture
10008 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +020010009
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010010 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +020010011 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020010012 rules do not stop evaluation and do not change default action. The
10013 number of counters that may be simultaneously tracked by the same
10014 connection is set in MAX_SESS_STKCTR at build time (reported in
John Roesler27c754c2019-07-10 15:45:51 -050010015 haproxy -vv) which defaults to 3, so the track-sc number is between 0
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020010016 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
10017 tracking of the counters of the specified table as the first set. The
10018 first "track-sc1" rule executed enables tracking of the counters of the
10019 specified table as the second set. The first "track-sc2" rule executed
10020 enables tracking of the counters of the specified table as the third
10021 set. It is a recommended practice to use the first set of counters for
10022 the per-frontend counters and the second set for the per-backend ones.
10023 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010024
Willy Tarreaue9656522010-08-17 15:40:09 +020010025 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +020010026 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +020010027 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +010010028 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010029 and used to select which table entry to update the counters.
10030 Note that "tcp-request connection" cannot use content-based
10031 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010032
Willy Tarreaue9656522010-08-17 15:40:09 +020010033 <table> is an optional table to be used instead of the default one,
10034 which is the stick-table declared in the current proxy. All
10035 the counters for the matches and updates for the key will
10036 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010037
Willy Tarreaue9656522010-08-17 15:40:09 +020010038 Once a "track-sc*" rule is executed, the key is looked up in the table
10039 and if it is not found, an entry is allocated for it. Then a pointer to
10040 that entry is kept during all the session's life, and this entry's
10041 counters are updated as often as possible, every time the session's
10042 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010043 Counters are only updated for events that happen after the tracking has
10044 been started. For example, connection counters will not be updated when
10045 tracking layer 7 information, since the connection event happens before
10046 layer7 information is extracted.
10047
Willy Tarreaue9656522010-08-17 15:40:09 +020010048 If the entry tracks concurrent connection counters, one connection is
10049 counted for as long as the entry is tracked, and the entry will not
10050 expire during that time. Tracking counters also provides a performance
10051 advantage over just checking the keys, because only one table lookup is
10052 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010053
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020010054 - sc-inc-gpc0(<sc-id>):
10055 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
10056 counter designated by <sc-id>. If an error occurs, this action silently
10057 fails and the actions evaluation continues.
10058
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010059 - sc-inc-gpc1(<sc-id>):
10060 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
10061 counter designated by <sc-id>. If an error occurs, this action silently
10062 fails and the actions evaluation continues.
10063
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010064 - sc-set-gpt0(<sc-id>) <int>:
10065 This action sets the GPT0 tag according to the sticky counter designated
10066 by <sc-id> and the value of <int>. The expected result is a boolean. If
10067 an error occurs, this action silently fails and the actions evaluation
10068 continues.
10069
William Lallemand2e785f22016-05-25 01:48:42 +020010070 - set-src <expr> :
10071 Is used to set the source IP address to the value of specified
10072 expression. Useful if you want to mask source IP for privacy.
10073 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020010074 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +020010075
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020010076 Arguments:
10077 <expr> Is a standard HAProxy expression formed by a sample-fetch
10078 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +020010079
10080 Example:
William Lallemand2e785f22016-05-25 01:48:42 +020010081 tcp-request connection set-src src,ipmask(24)
10082
Willy Tarreau0c630532016-10-21 17:52:58 +020010083 When possible, set-src preserves the original source port as long as the
10084 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +020010085
William Lallemand44be6402016-05-25 01:51:35 +020010086 - set-src-port <expr> :
10087 Is used to set the source port address to the value of specified
10088 expression.
10089
Cyril Bonté6c81d5f2018-10-17 00:14:51 +020010090 Arguments:
10091 <expr> Is a standard HAProxy expression formed by a sample-fetch
10092 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +020010093
10094 Example:
William Lallemand44be6402016-05-25 01:51:35 +020010095 tcp-request connection set-src-port int(4000)
10096
Willy Tarreau0c630532016-10-21 17:52:58 +020010097 When possible, set-src-port preserves the original source address as long
10098 as the address family supports a port, otherwise it forces the source
10099 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +020010100
William Lallemand13e9b0c2016-05-25 02:34:07 +020010101 - set-dst <expr> :
10102 Is used to set the destination IP address to the value of specified
10103 expression. Useful if you want to mask IP for privacy in log.
10104 If you want to provide an IP from a HTTP header use "http-request
10105 set-dst". If you want to connect to the new address/port, use
10106 '0.0.0.0:0' as a server address in the backend.
10107
10108 <expr> Is a standard HAProxy expression formed by a sample-fetch
10109 followed by some converters.
10110
10111 Example:
10112
10113 tcp-request connection set-dst dst,ipmask(24)
10114 tcp-request connection set-dst ipv4(10.0.0.1)
10115
Willy Tarreau0c630532016-10-21 17:52:58 +020010116 When possible, set-dst preserves the original destination port as long as
10117 the address family allows it, otherwise the destination port is set to 0.
10118
William Lallemand13e9b0c2016-05-25 02:34:07 +020010119 - set-dst-port <expr> :
10120 Is used to set the destination port address to the value of specified
10121 expression. If you want to connect to the new address/port, use
10122 '0.0.0.0:0' as a server address in the backend.
10123
10124
10125 <expr> Is a standard HAProxy expression formed by a sample-fetch
10126 followed by some converters.
10127
10128 Example:
10129
10130 tcp-request connection set-dst-port int(4000)
10131
Willy Tarreau0c630532016-10-21 17:52:58 +020010132 When possible, set-dst-port preserves the original destination address as
10133 long as the address family supports a port, otherwise it forces the
10134 destination address to IPv4 "0.0.0.0" before rewriting the port.
10135
Willy Tarreau2d392c22015-08-24 01:43:45 +020010136 - "silent-drop" :
10137 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010010138 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020010139 to prevent the client from being notified. The effect it then that the
10140 client still sees an established connection while there's none on
10141 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
10142 except that it doesn't use any local resource at all on the machine
10143 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010010144 slow down stronger attackers. It is important to understand the impact
10145 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010146 client and HAProxy (firewalls, proxies, load balancers) will also keep
10147 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010148 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010149 TCP_REPAIR socket option is used to block the emission of a TCP
10150 reset. On other systems, the socket's TTL is reduced to 1 so that the
10151 TCP reset doesn't pass the first router, though it's still delivered to
10152 local networks. Do not use it unless you fully understand how it works.
10153
Willy Tarreaue9656522010-08-17 15:40:09 +020010154 Note that the "if/unless" condition is optional. If no condition is set on
10155 the action, it is simply performed unconditionally. That can be useful for
10156 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010157
Willy Tarreaue9656522010-08-17 15:40:09 +020010158 Example: accept all connections from white-listed hosts, reject too fast
10159 connection without counting them, and track accepted connections.
10160 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010161
Willy Tarreaue9656522010-08-17 15:40:09 +020010162 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010163 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010164 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010165
Willy Tarreaue9656522010-08-17 15:40:09 +020010166 Example: accept all connections from white-listed hosts, count all other
10167 connections and reject too fast ones. This results in abusive ones
10168 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010169
Willy Tarreaue9656522010-08-17 15:40:09 +020010170 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010171 tcp-request connection track-sc0 src
10172 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010173
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010174 Example: enable the PROXY protocol for traffic coming from all known proxies.
10175
10176 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10177
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010178 See section 7 about ACL usage.
10179
Willy Tarreau4f614292016-10-21 17:49:36 +020010180 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010181
10182
Willy Tarreaue9656522010-08-17 15:40:09 +020010183tcp-request content <action> [{if | unless} <condition>]
10184 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +020010185 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020010186 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +020010187 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020010188 <action> defines the action to perform if the condition applies. See
10189 below.
Willy Tarreau62644772008-07-16 18:36:06 +020010190
Willy Tarreaue9656522010-08-17 15:40:09 +020010191 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +020010192
Davor Ocelice9ed2812017-12-25 17:49:28 +010010193 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +020010194 called "TCP content inspection". During this stage, ACL-based rules are
10195 evaluated every time the request contents are updated, until either an
10196 "accept" or a "reject" rule matches, or the TCP request inspection delay
10197 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +020010198
Willy Tarreaue9656522010-08-17 15:40:09 +020010199 The first difference between these rules and "tcp-request connection" rules
10200 is that "tcp-request content" rules can make use of contents to take a
10201 decision. Most often, these decisions will consider a protocol recognition or
10202 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +010010203 both frontends and backends. In case of HTTP keep-alive with the client, all
10204 tcp-request content rules are evaluated again, so haproxy keeps a record of
10205 what sticky counters were assigned by a "tcp-request connection" versus a
10206 "tcp-request content" rule, and flushes all the content-related ones after
10207 processing an HTTP request, so that they may be evaluated again by the rules
10208 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010209 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +010010210 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +020010211
Willy Tarreaue9656522010-08-17 15:40:09 +020010212 Content-based rules are evaluated in their exact declaration order. If no
10213 rule matches or if there is no rule, the default action is to accept the
10214 contents. There is no specific limit to the number of rules which may be
10215 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +020010216
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010217 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +020010218 - accept : the request is accepted
Baptiste Assmann333939c2019-01-21 08:34:50 +010010219 - do-resolve: perform a DNS resolution
Willy Tarreau18bf01e2014-06-13 16:18:52 +020010220 - reject : the request is rejected and the connection is closed
10221 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -040010222 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010223 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020010224 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010225 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +020010226 - sc-set-gpt0(<sc-id>) <int>
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020010227 - set-dst <expr>
10228 - set-dst-port <expr>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010229 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010230 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +020010231 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +010010232 - send-spoe-group <engine-name> <group-name>
Christopher Faulet6bd406e2019-11-22 15:34:17 +010010233 - use-service <service-name>
Willy Tarreau62644772008-07-16 18:36:06 +020010234
Willy Tarreaue9656522010-08-17 15:40:09 +020010235 They have the same meaning as their counter-parts in "tcp-request connection"
10236 so please refer to that section for a complete description.
Baptiste Assmann333939c2019-01-21 08:34:50 +010010237 For "do-resolve" action, please check the "http-request do-resolve"
10238 configuration section.
Willy Tarreau62644772008-07-16 18:36:06 +020010239
Willy Tarreauf3338342014-01-28 21:40:28 +010010240 While there is nothing mandatory about it, it is recommended to use the
10241 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
10242 content" rules in the frontend, and track-sc2 for "tcp-request content"
10243 rules in the backend, because that makes the configuration more readable
10244 and easier to troubleshoot, but this is just a guideline and all counters
10245 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +020010246
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010247 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +020010248 the action, it is simply performed unconditionally. That can be useful for
10249 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +020010250
Willy Tarreaue9656522010-08-17 15:40:09 +020010251 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +020010252 rules, since HTTP-specific ACL matches are able to preliminarily parse the
10253 contents of a buffer before extracting the required data. If the buffered
10254 contents do not parse as a valid HTTP message, then the ACL does not match.
10255 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +010010256 processing, so there is no risk of parsing something differently. In an HTTP
10257 backend connected to from an HTTP frontend, it is guaranteed that HTTP
10258 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +020010259
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010260 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020010261 are present when the rule is processed. The rule processing engine is able to
10262 wait until the inspect delay expires when the data to be tracked is not yet
10263 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010264
Baptiste Assmanne1afd4f2019-04-18 16:21:13 +020010265 The "set-dst" and "set-dst-port" are used to set respectively the destination
10266 IP and port. More information on how to use it at "http-request set-dst".
10267
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010268 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +020010269 declared inline. For "tcp-request session" rules, only session-level
10270 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010271
Daniel Schneller0b547052016-03-21 20:46:57 +010010272 <var-name> The name of the variable starts with an indication about
10273 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010010274 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010010275 "sess" : the variable is shared with the whole session
10276 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010277 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010010278 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010279 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010010280 "res" : the variable is shared only during response
10281 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010282 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010010283 The name may only contain characters 'a-z', 'A-Z', '0-9',
10284 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010285
10286 <expr> Is a standard HAProxy expression formed by a sample-fetch
10287 followed by some converters.
10288
Christopher Faulet85d79c92016-11-09 16:54:56 +010010289 The "unset-var" is used to unset a variable. See above for details about
10290 <var-name>.
10291
Patrick Hemmer268a7072018-05-11 12:52:31 -040010292 The "set-priority-class" is used to set the queue priority class of the
10293 current request. The value must be a sample expression which converts to an
10294 integer in the range -2047..2047. Results outside this range will be
10295 truncated. The priority class determines the order in which queued requests
10296 are processed. Lower values have higher priority.
10297
10298 The "set-priority-offset" is used to set the queue priority timestamp offset
10299 of the current request. The value must be a sample expression which converts
10300 to an integer in the range -524287..524287. Results outside this range will be
10301 truncated. When a request is queued, it is ordered first by the priority
10302 class, then by the current timestamp adjusted by the given offset in
10303 milliseconds. Lower values have higher priority.
10304 Note that the resulting timestamp is is only tracked with enough precision for
10305 524,287ms (8m44s287ms). If the request is queued long enough to where the
10306 adjusted timestamp exceeds this value, it will be misidentified as highest
10307 priority. Thus it is important to set "timeout queue" to a value, where when
10308 combined with the offset, does not exceed this limit.
10309
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010310 The "send-spoe-group" is used to trigger sending of a group of SPOE
10311 messages. To do so, the SPOE engine used to send messages must be defined, as
10312 well as the SPOE group to send. Of course, the SPOE engine must refer to an
10313 existing SPOE filter. If not engine name is provided on the SPOE filter line,
10314 the SPOE agent name must be used.
10315
10316 <engine-name> The SPOE engine name.
10317
10318 <group-name> The SPOE group name as specified in the engine configuration.
10319
Christopher Faulet6bd406e2019-11-22 15:34:17 +010010320 The "use-service" is used to executes a TCP service which will reply to the
10321 request and stop the evaluation of the rules. This service may choose to
10322 reply by sending any valid response or it may immediately close the
10323 connection without sending anything. Outside natives services, it is possible
10324 to write your own services in Lua. No further "tcp-request" rules are
10325 evaluated.
10326
10327 Example:
10328 tcp-request content use-service lua.deny { src -f /etc/haproxy/blacklist.lst }
10329
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010330 Example:
10331
10332 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +010010333 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010334
Willy Tarreau62644772008-07-16 18:36:06 +020010335 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +020010336 # Accept HTTP requests containing a Host header saying "example.com"
10337 # and reject everything else.
10338 acl is_host_com hdr(Host) -i example.com
10339 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +020010340 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +020010341 tcp-request content reject
10342
10343 Example:
Willy Tarreau62644772008-07-16 18:36:06 +020010344 # reject SMTP connection if client speaks first
10345 tcp-request inspect-delay 30s
10346 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010347 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +020010348
10349 # Forward HTTPS connection only if client speaks
10350 tcp-request inspect-delay 30s
10351 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +020010352 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +020010353 tcp-request content reject
10354
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010355 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010356 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010357 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020010358 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010359 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
10360 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010361
10362 Example:
10363 # track request counts per "base" (concatenation of Host+URL)
10364 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +020010365 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +010010366
Willy Tarreaue9656522010-08-17 15:40:09 +020010367 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010368 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +020010369
10370 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +010010371 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +020010372 # protecting all our sites
10373 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010374 tcp-request connection track-sc0 src
10375 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +020010376 ...
10377 use_backend http_dynamic if { path_end .php }
10378
10379 backend http_dynamic
10380 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010381 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +020010382 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010383 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010384 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020010385 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +020010386 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +020010387
Willy Tarreauc57f0e22009-05-10 13:12:33 +020010388 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +020010389
Jarno Huuskonen95b012b2017-04-06 13:59:14 +030010390 See also : "tcp-request connection", "tcp-request session",
10391 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +020010392
10393
10394tcp-request inspect-delay <timeout>
10395 Set the maximum allowed time to wait for data during content inspection
10396 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +020010397 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +020010398 Arguments :
10399 <timeout> is the timeout value specified in milliseconds by default, but
10400 can be in any other unit if the number is suffixed by the unit,
10401 as explained at the top of this document.
10402
10403 People using haproxy primarily as a TCP relay are often worried about the
10404 risk of passing any type of protocol to a server without any analysis. In
10405 order to be able to analyze the request contents, we must first withhold
10406 the data then analyze them. This statement simply enables withholding of
10407 data for at most the specified amount of time.
10408
Willy Tarreaufb356202010-08-03 14:02:05 +020010409 TCP content inspection applies very early when a connection reaches a
10410 frontend, then very early when the connection is forwarded to a backend. This
10411 means that a connection may experience a first delay in the frontend and a
10412 second delay in the backend if both have tcp-request rules.
10413
Willy Tarreau62644772008-07-16 18:36:06 +020010414 Note that when performing content inspection, haproxy will evaluate the whole
10415 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010416 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +020010417 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +010010418 contents are definitive. If no delay is set, haproxy will not wait at all
10419 and will immediately apply a verdict based on the available information.
10420 Obviously this is unlikely to be very useful and might even be racy, so such
10421 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +020010422
10423 As soon as a rule matches, the request is released and continues as usual. If
10424 the timeout is reached and no rule matches, the default policy will be to let
10425 it pass through unaffected.
10426
10427 For most protocols, it is enough to set it to a few seconds, as most clients
10428 send the full request immediately upon connection. Add 3 or more seconds to
10429 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +010010430 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +010010431 before the server (e.g. SMTP), or to wait for a client to talk before passing
10432 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +020010433 least the inspection delay, otherwise it will expire first. If the client
10434 closes the connection or if the buffer is full, the delay immediately expires
10435 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +020010436
Willy Tarreau55165fe2009-05-10 12:02:55 +020010437 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +020010438 "timeout client".
10439
10440
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010441tcp-response content <action> [{if | unless} <condition>]
10442 Perform an action on a session response depending on a layer 4-7 condition
10443 May be used in sections : defaults | frontend | listen | backend
10444 no | no | yes | yes
10445 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +020010446 <action> defines the action to perform if the condition applies. See
10447 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010448
10449 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
10450
Davor Ocelice9ed2812017-12-25 17:49:28 +010010451 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010452 called "TCP content inspection". During this stage, ACL-based rules are
10453 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010454 "accept", "close" or a "reject" rule matches, or a TCP response inspection
10455 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010456
10457 Most often, these decisions will consider a protocol recognition or validity.
10458
10459 Content-based rules are evaluated in their exact declaration order. If no
10460 rule matches or if there is no rule, the default action is to accept the
10461 contents. There is no specific limit to the number of rules which may be
10462 inserted.
10463
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010464 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010465 - accept :
10466 accepts the response if the condition is true (when used with "if")
10467 or false (when used with "unless"). The first such rule executed ends
10468 the rules evaluation.
10469
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010470 - close :
10471 immediately closes the connection with the server if the condition is
10472 true (when used with "if"), or false (when used with "unless"). The
10473 first such rule executed ends the rules evaluation. The main purpose of
10474 this action is to force a connection to be finished between a client
10475 and a server after an exchange when the application protocol expects
10476 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010477 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +020010478 protocols.
10479
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010480 - reject :
10481 rejects the response if the condition is true (when used with "if")
10482 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010483 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010484
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010485 - set-var(<var-name>) <expr>
10486 Sets a variable.
10487
Christopher Faulet85d79c92016-11-09 16:54:56 +010010488 - unset-var(<var-name>)
10489 Unsets a variable.
10490
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +020010491 - sc-inc-gpc0(<sc-id>):
10492 This action increments the GPC0 counter according to the sticky
10493 counter designated by <sc-id>. If an error occurs, this action fails
10494 silently and the actions evaluation continues.
10495
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010496 - sc-inc-gpc1(<sc-id>):
10497 This action increments the GPC1 counter according to the sticky
10498 counter designated by <sc-id>. If an error occurs, this action fails
10499 silently and the actions evaluation continues.
10500
Thierry FOURNIER236657b2015-08-19 08:25:14 +020010501 - sc-set-gpt0(<sc-id>) <int> :
10502 This action sets the GPT0 tag according to the sticky counter designated
10503 by <sc-id> and the value of <int>. The expected result is a boolean. If
10504 an error occurs, this action silently fails and the actions evaluation
10505 continues.
10506
Willy Tarreau2d392c22015-08-24 01:43:45 +020010507 - "silent-drop" :
10508 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +010010509 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +020010510 to prevent the client from being notified. The effect it then that the
10511 client still sees an established connection while there's none on
10512 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
10513 except that it doesn't use any local resource at all on the machine
10514 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +010010515 slow down stronger attackers. It is important to understand the impact
10516 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010517 client and HAProxy (firewalls, proxies, load balancers) will also keep
10518 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010519 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +020010520 TCP_REPAIR socket option is used to block the emission of a TCP
10521 reset. On other systems, the socket's TTL is reduced to 1 so that the
10522 TCP reset doesn't pass the first router, though it's still delivered to
10523 local networks. Do not use it unless you fully understand how it works.
10524
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010525 - send-spoe-group <engine-name> <group-name>
10526 Send a group of SPOE messages.
10527
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010528 Note that the "if/unless" condition is optional. If no condition is set on
10529 the action, it is simply performed unconditionally. That can be useful for
10530 for changing the default action to a reject.
10531
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010532 It is perfectly possible to match layer 7 contents with "tcp-response
10533 content" rules, but then it is important to ensure that a full response has
10534 been buffered, otherwise no contents will match. In order to achieve this,
10535 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010536 period.
10537
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010538 The "set-var" is used to set the content of a variable. The variable is
10539 declared inline.
10540
Daniel Schneller0b547052016-03-21 20:46:57 +010010541 <var-name> The name of the variable starts with an indication about
10542 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010010543 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010010544 "sess" : the variable is shared with the whole session
10545 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010546 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010010547 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010548 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010010549 "res" : the variable is shared only during response
10550 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010551 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010010552 The name may only contain characters 'a-z', 'A-Z', '0-9',
10553 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010554
10555 <expr> Is a standard HAProxy expression formed by a sample-fetch
10556 followed by some converters.
10557
10558 Example:
10559
10560 tcp-request content set-var(sess.my_var) src
10561
Christopher Faulet85d79c92016-11-09 16:54:56 +010010562 The "unset-var" is used to unset a variable. See above for details about
10563 <var-name>.
10564
10565 Example:
10566
10567 tcp-request content unset-var(sess.my_var)
10568
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010569 The "send-spoe-group" is used to trigger sending of a group of SPOE
10570 messages. To do so, the SPOE engine used to send messages must be defined, as
10571 well as the SPOE group to send. Of course, the SPOE engine must refer to an
10572 existing SPOE filter. If not engine name is provided on the SPOE filter line,
10573 the SPOE agent name must be used.
10574
10575 <engine-name> The SPOE engine name.
10576
10577 <group-name> The SPOE group name as specified in the engine configuration.
10578
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010579 See section 7 about ACL usage.
10580
10581 See also : "tcp-request content", "tcp-response inspect-delay"
10582
10583
Willy Tarreau4f614292016-10-21 17:49:36 +020010584tcp-request session <action> [{if | unless} <condition>]
10585 Perform an action on a validated session depending on a layer 5 condition
10586 May be used in sections : defaults | frontend | listen | backend
10587 no | yes | yes | no
10588 Arguments :
10589 <action> defines the action to perform if the condition applies. See
10590 below.
10591
10592 <condition> is a standard layer5-only ACL-based condition (see section 7).
10593
Davor Ocelice9ed2812017-12-25 17:49:28 +010010594 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020010595 it is possible to evaluate some conditions to decide whether this session
10596 must be accepted or dropped or have its counters tracked. Those conditions
10597 cannot make use of any data contents because no buffers are allocated yet and
10598 the processing cannot wait at this stage. The main use case it to copy some
10599 early information into variables (since variables are accessible in the
10600 session), or to keep track of some information collected after the handshake,
10601 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010010602 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020010603 extracted information can thus be copied to a variable or tracked using
10604 "track-sc" rules. Of course it is also possible to decide to accept/reject as
10605 with other rulesets. Most operations performed here could also be performed
10606 in "tcp-request content" rules, except that in HTTP these rules are evaluated
10607 for each new request, and that might not always be acceptable. For example a
10608 rule might increment a counter on each evaluation. It would also be possible
10609 that a country is resolved by geolocation from the source IP address,
10610 assigned to a session-wide variable, then the source address rewritten from
10611 an HTTP header for all requests. If some contents need to be inspected in
10612 order to take the decision, the "tcp-request content" statements must be used
10613 instead.
10614
10615 The "tcp-request session" rules are evaluated in their exact declaration
10616 order. If no rule matches or if there is no rule, the default action is to
10617 accept the incoming session. There is no specific limit to the number of
10618 rules which may be inserted.
10619
10620 Several types of actions are supported :
10621 - accept : the request is accepted
10622 - reject : the request is rejected and the connection is closed
10623 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
10624 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010625 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010626 - sc-set-gpt0(<sc-id>) <int>
10627 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010628 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010629 - silent-drop
10630
10631 These actions have the same meaning as their respective counter-parts in
10632 "tcp-request connection" and "tcp-request content", so please refer to these
10633 sections for a complete description.
10634
10635 Note that the "if/unless" condition is optional. If no condition is set on
10636 the action, it is simply performed unconditionally. That can be useful for
10637 "track-sc*" actions as well as for changing the default action to a reject.
10638
10639 Example: track the original source address by default, or the one advertised
10640 in the PROXY protocol header for connection coming from the local
10641 proxies. The first connection-level rule enables receipt of the
10642 PROXY protocol for these ones, the second rule tracks whatever
10643 address we decide to keep after optional decoding.
10644
10645 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10646 tcp-request session track-sc0 src
10647
10648 Example: accept all sessions from white-listed hosts, reject too fast
10649 sessions without counting them, and track accepted sessions.
10650 This results in session rate being capped from abusive sources.
10651
10652 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10653 tcp-request session reject if { src_sess_rate gt 10 }
10654 tcp-request session track-sc0 src
10655
10656 Example: accept all sessions from white-listed hosts, count all other
10657 sessions and reject too fast ones. This results in abusive ones
10658 being blocked as long as they don't slow down.
10659
10660 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10661 tcp-request session track-sc0 src
10662 tcp-request session reject if { sc0_sess_rate gt 10 }
10663
10664 See section 7 about ACL usage.
10665
10666 See also : "tcp-request connection", "tcp-request content", "stick-table"
10667
10668
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010669tcp-response inspect-delay <timeout>
10670 Set the maximum allowed time to wait for a response during content inspection
10671 May be used in sections : defaults | frontend | listen | backend
10672 no | no | yes | yes
10673 Arguments :
10674 <timeout> is the timeout value specified in milliseconds by default, but
10675 can be in any other unit if the number is suffixed by the unit,
10676 as explained at the top of this document.
10677
10678 See also : "tcp-response content", "tcp-request inspect-delay".
10679
10680
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010681timeout check <timeout>
10682 Set additional check timeout, but only after a connection has been already
10683 established.
10684
10685 May be used in sections: defaults | frontend | listen | backend
10686 yes | no | yes | yes
10687 Arguments:
10688 <timeout> is the timeout value specified in milliseconds by default, but
10689 can be in any other unit if the number is suffixed by the unit,
10690 as explained at the top of this document.
10691
10692 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
10693 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010694 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010695 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010010696 (Please also note that there is no valid reason to have such long connect
10697 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
10698 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010699
10700 If "timeout check" is not set haproxy uses "inter" for complete check
10701 timeout (connect + read) exactly like all <1.3.15 version.
10702
10703 In most cases check request is much simpler and faster to handle than normal
10704 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010010705 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010706
10707 This parameter is specific to backends, but can be specified once for all in
10708 "defaults" sections. This is in fact one of the easiest solutions not to
10709 forget about it.
10710
Willy Tarreau41a340d2008-01-22 12:25:31 +010010711 See also: "timeout connect", "timeout queue", "timeout server",
10712 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010713
10714
Willy Tarreau0ba27502007-12-24 16:55:16 +010010715timeout client <timeout>
10716timeout clitimeout <timeout> (deprecated)
10717 Set the maximum inactivity time on the client side.
10718 May be used in sections : defaults | frontend | listen | backend
10719 yes | yes | yes | no
10720 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010721 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010722 can be in any other unit if the number is suffixed by the unit,
10723 as explained at the top of this document.
10724
10725 The inactivity timeout applies when the client is expected to acknowledge or
10726 send data. In HTTP mode, this timeout is particularly important to consider
10727 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010728 response while it is reading data sent by the server. That said, for the
10729 first phase, it is preferable to set the "timeout http-request" to better
10730 protect HAProxy from Slowloris like attacks. The value is specified in
10731 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010732 suffixed by the unit, as specified at the top of this document. In TCP mode
10733 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10734 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010735 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010736 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010737 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10738 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010739 which overrides "timeout client" and "timeout server" for tunnels, as well as
10740 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010741
10742 This parameter is specific to frontends, but can be specified once for all in
10743 "defaults" sections. This is in fact one of the easiest solutions not to
10744 forget about it. An unspecified timeout results in an infinite timeout, which
10745 is not recommended. Such a usage is accepted and works but reports a warning
John Roesler27c754c2019-07-10 15:45:51 -050010746 during startup because it may result in accumulation of expired sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010747 the system if the system's timeouts are not configured either.
10748
Willy Tarreau95c4e142017-11-26 12:18:55 +010010749 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010750
Willy Tarreau0ba27502007-12-24 16:55:16 +010010751 This parameter replaces the old, deprecated "clitimeout". It is recommended
10752 to use it to write new configurations. The form "timeout clitimeout" is
10753 provided only by backwards compatibility but its use is strongly discouraged.
10754
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010755 See also : "clitimeout", "timeout server", "timeout tunnel",
10756 "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010757
10758
Willy Tarreau05cdd962014-05-10 14:30:07 +020010759timeout client-fin <timeout>
10760 Set the inactivity timeout on the client side for half-closed connections.
10761 May be used in sections : defaults | frontend | listen | backend
10762 yes | yes | yes | no
10763 Arguments :
10764 <timeout> is the timeout value specified in milliseconds by default, but
10765 can be in any other unit if the number is suffixed by the unit,
10766 as explained at the top of this document.
10767
10768 The inactivity timeout applies when the client is expected to acknowledge or
10769 send data while one direction is already shut down. This timeout is different
10770 from "timeout client" in that it only applies to connections which are closed
10771 in one direction. This is particularly useful to avoid keeping connections in
10772 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10773 problem is particularly common long connections such as RDP or WebSocket.
10774 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010775 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10776 frame was sent, often indicating an expectation that the connection quickly
10777 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010778
10779 This parameter is specific to frontends, but can be specified once for all in
10780 "defaults" sections. By default it is not set, so half-closed connections
10781 will use the other timeouts (timeout.client or timeout.tunnel).
10782
10783 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10784
10785
Willy Tarreau0ba27502007-12-24 16:55:16 +010010786timeout connect <timeout>
10787timeout contimeout <timeout> (deprecated)
10788 Set the maximum time to wait for a connection attempt to a server to succeed.
10789 May be used in sections : defaults | frontend | listen | backend
10790 yes | no | yes | yes
10791 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010792 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010793 can be in any other unit if the number is suffixed by the unit,
10794 as explained at the top of this document.
10795
10796 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010797 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010798 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010799 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010800 connect timeout also presets both queue and tarpit timeouts to the same value
10801 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010802
10803 This parameter is specific to backends, but can be specified once for all in
10804 "defaults" sections. This is in fact one of the easiest solutions not to
10805 forget about it. An unspecified timeout results in an infinite timeout, which
10806 is not recommended. Such a usage is accepted and works but reports a warning
John Roesler27c754c2019-07-10 15:45:51 -050010807 during startup because it may result in accumulation of failed sessions in
Willy Tarreau0ba27502007-12-24 16:55:16 +010010808 the system if the system's timeouts are not configured either.
10809
10810 This parameter replaces the old, deprecated "contimeout". It is recommended
10811 to use it to write new configurations. The form "timeout contimeout" is
10812 provided only by backwards compatibility but its use is strongly discouraged.
10813
Willy Tarreau41a340d2008-01-22 12:25:31 +010010814 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
10815 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010816
10817
Willy Tarreaub16a5742010-01-10 14:46:16 +010010818timeout http-keep-alive <timeout>
10819 Set the maximum allowed time to wait for a new HTTP request to appear
10820 May be used in sections : defaults | frontend | listen | backend
10821 yes | yes | yes | yes
10822 Arguments :
10823 <timeout> is the timeout value specified in milliseconds by default, but
10824 can be in any other unit if the number is suffixed by the unit,
10825 as explained at the top of this document.
10826
10827 By default, the time to wait for a new request in case of keep-alive is set
10828 by "timeout http-request". However this is not always convenient because some
10829 people want very short keep-alive timeouts in order to release connections
10830 faster, and others prefer to have larger ones but still have short timeouts
10831 once the request has started to present itself.
10832
10833 The "http-keep-alive" timeout covers these needs. It will define how long to
10834 wait for a new HTTP request to start coming after a response was sent. Once
10835 the first byte of request has been seen, the "http-request" timeout is used
10836 to wait for the complete request to come. Note that empty lines prior to a
10837 new request do not refresh the timeout and are not counted as a new request.
10838
10839 There is also another difference between the two timeouts : when a connection
10840 expires during timeout http-keep-alive, no error is returned, the connection
10841 just closes. If the connection expires in "http-request" while waiting for a
10842 connection to complete, a HTTP 408 error is returned.
10843
10844 In general it is optimal to set this value to a few tens to hundreds of
10845 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010846 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010847 1 millisecond) it will probably only accept pipelined requests but not the
10848 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010849 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010850
10851 If this parameter is not set, the "http-request" timeout applies, and if both
10852 are not set, "timeout client" still applies at the lower level. It should be
10853 set in the frontend to take effect, unless the frontend is in TCP mode, in
10854 which case the HTTP backend's timeout will be used.
10855
Willy Tarreau95c4e142017-11-26 12:18:55 +010010856 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10857 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010858 (where we only have one connection per client and a connection setup).
10859
Willy Tarreaub16a5742010-01-10 14:46:16 +010010860 See also : "timeout http-request", "timeout client".
10861
10862
Willy Tarreau036fae02008-01-06 13:24:40 +010010863timeout http-request <timeout>
10864 Set the maximum allowed time to wait for a complete HTTP request
10865 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010866 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010867 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010868 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010869 can be in any other unit if the number is suffixed by the unit,
10870 as explained at the top of this document.
10871
10872 In order to offer DoS protection, it may be required to lower the maximum
10873 accepted time to receive a complete HTTP request without affecting the client
10874 timeout. This helps protecting against established connections on which
10875 nothing is sent. The client timeout cannot offer a good protection against
10876 this abuse because it is an inactivity timeout, which means that if the
10877 attacker sends one character every now and then, the timeout will not
10878 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010879 types, the request will be aborted if it does not complete in time. When the
10880 timeout expires, an HTTP 408 response is sent to the client to inform it
10881 about the problem, and the connection is closed. The logs will report
10882 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010883 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010884 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10885 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010886
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010887 By default, this timeout only applies to the header part of the request,
10888 and not to any data. As soon as the empty line is received, this timeout is
10889 not used anymore. When combined with "option http-buffer-request", this
10890 timeout also applies to the body of the request..
10891 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010892 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010893
10894 Generally it is enough to set it to a few seconds, as most clients send the
10895 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010896 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010897 generally work on local networks as long as there are no packet losses. This
10898 will prevent people from sending bare HTTP requests using telnet.
10899
10900 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010901 chunk of the incoming request. It should be set in the frontend to take
10902 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10903 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010904
Willy Tarreau0f228a02015-05-01 15:37:53 +020010905 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010906 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010907
Willy Tarreau844e3c52008-01-11 16:28:18 +010010908
10909timeout queue <timeout>
10910 Set the maximum time to wait in the queue for a connection slot to be free
10911 May be used in sections : defaults | frontend | listen | backend
10912 yes | no | yes | yes
10913 Arguments :
10914 <timeout> is the timeout value specified in milliseconds by default, but
10915 can be in any other unit if the number is suffixed by the unit,
10916 as explained at the top of this document.
10917
10918 When a server's maxconn is reached, connections are left pending in a queue
10919 which may be server-specific or global to the backend. In order not to wait
10920 indefinitely, a timeout is applied to requests pending in the queue. If the
10921 timeout is reached, it is considered that the request will almost never be
10922 served, so it is dropped and a 503 error is returned to the client.
10923
10924 The "timeout queue" statement allows to fix the maximum time for a request to
10925 be left pending in a queue. If unspecified, the same value as the backend's
10926 connection timeout ("timeout connect") is used, for backwards compatibility
10927 with older versions with no "timeout queue" parameter.
10928
10929 See also : "timeout connect", "contimeout".
10930
10931
10932timeout server <timeout>
10933timeout srvtimeout <timeout> (deprecated)
10934 Set the maximum inactivity time on the server side.
10935 May be used in sections : defaults | frontend | listen | backend
10936 yes | no | yes | yes
10937 Arguments :
10938 <timeout> is the timeout value specified in milliseconds by default, but
10939 can be in any other unit if the number is suffixed by the unit,
10940 as explained at the top of this document.
10941
10942 The inactivity timeout applies when the server is expected to acknowledge or
10943 send data. In HTTP mode, this timeout is particularly important to consider
10944 during the first phase of the server's response, when it has to send the
10945 headers, as it directly represents the server's processing time for the
10946 request. To find out what value to put there, it's often good to start with
10947 what would be considered as unacceptable response times, then check the logs
10948 to observe the response time distribution, and adjust the value accordingly.
10949
10950 The value is specified in milliseconds by default, but can be in any other
10951 unit if the number is suffixed by the unit, as specified at the top of this
10952 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10953 recommended that the client timeout remains equal to the server timeout in
10954 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010955 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010956 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010957 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10958 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010959 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10960 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010961
10962 This parameter is specific to backends, but can be specified once for all in
10963 "defaults" sections. This is in fact one of the easiest solutions not to
10964 forget about it. An unspecified timeout results in an infinite timeout, which
10965 is not recommended. Such a usage is accepted and works but reports a warning
John Roesler27c754c2019-07-10 15:45:51 -050010966 during startup because it may result in accumulation of expired sessions in
Willy Tarreau844e3c52008-01-11 16:28:18 +010010967 the system if the system's timeouts are not configured either.
10968
10969 This parameter replaces the old, deprecated "srvtimeout". It is recommended
10970 to use it to write new configurations. The form "timeout srvtimeout" is
10971 provided only by backwards compatibility but its use is strongly discouraged.
10972
Willy Tarreauce887fd2012-05-12 12:50:00 +020010973 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010974
Willy Tarreau05cdd962014-05-10 14:30:07 +020010975
10976timeout server-fin <timeout>
10977 Set the inactivity timeout on the server side for half-closed connections.
10978 May be used in sections : defaults | frontend | listen | backend
10979 yes | no | yes | yes
10980 Arguments :
10981 <timeout> is the timeout value specified in milliseconds by default, but
10982 can be in any other unit if the number is suffixed by the unit,
10983 as explained at the top of this document.
10984
10985 The inactivity timeout applies when the server is expected to acknowledge or
10986 send data while one direction is already shut down. This timeout is different
10987 from "timeout server" in that it only applies to connections which are closed
10988 in one direction. This is particularly useful to avoid keeping connections in
10989 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10990 This problem is particularly common long connections such as RDP or WebSocket.
10991 Note that this timeout can override "timeout tunnel" when a connection shuts
10992 down in one direction. This setting was provided for completeness, but in most
10993 situations, it should not be needed.
10994
10995 This parameter is specific to backends, but can be specified once for all in
10996 "defaults" sections. By default it is not set, so half-closed connections
10997 will use the other timeouts (timeout.server or timeout.tunnel).
10998
10999 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
11000
Willy Tarreau844e3c52008-01-11 16:28:18 +010011001
11002timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010011003 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010011004 May be used in sections : defaults | frontend | listen | backend
11005 yes | yes | yes | yes
11006 Arguments :
11007 <timeout> is the tarpit duration specified in milliseconds by default, but
11008 can be in any other unit if the number is suffixed by the unit,
11009 as explained at the top of this document.
11010
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030011011 When a connection is tarpitted using "http-request tarpit" or
11012 "reqtarpit", it is maintained open with no activity for a certain
11013 amount of time, then closed. "timeout tarpit" defines how long it will
11014 be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011015
11016 The value is specified in milliseconds by default, but can be in any other
11017 unit if the number is suffixed by the unit, as specified at the top of this
11018 document. If unspecified, the same value as the backend's connection timeout
11019 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010011020 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011021
11022 See also : "timeout connect", "contimeout".
11023
11024
Willy Tarreauce887fd2012-05-12 12:50:00 +020011025timeout tunnel <timeout>
11026 Set the maximum inactivity time on the client and server side for tunnels.
11027 May be used in sections : defaults | frontend | listen | backend
11028 yes | no | yes | yes
11029 Arguments :
11030 <timeout> is the timeout value specified in milliseconds by default, but
11031 can be in any other unit if the number is suffixed by the unit,
11032 as explained at the top of this document.
11033
Jamie Gloudonaaa21002012-08-25 00:18:33 -040011034 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020011035 between a client and a server, and the connection remains inactive in both
11036 directions. This timeout supersedes both the client and server timeouts once
11037 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010011038 analyzer remains attached to either connection (e.g. tcp content rules are
11039 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020011040 when switching to the WebSocket protocol, or forwarding a CONNECT request
11041 to a proxy), or after the first response when no keepalive/close option is
11042 specified.
11043
Willy Tarreau05cdd962014-05-10 14:30:07 +020011044 Since this timeout is usually used in conjunction with long-lived connections,
11045 it usually is a good idea to also set "timeout client-fin" to handle the
11046 situation where a client suddenly disappears from the net and does not
11047 acknowledge a close, or sends a shutdown and does not acknowledge pending
11048 data anymore. This can happen in lossy networks where firewalls are present,
11049 and is detected by the presence of large amounts of sessions in a FIN_WAIT
11050 state.
11051
Willy Tarreauce887fd2012-05-12 12:50:00 +020011052 The value is specified in milliseconds by default, but can be in any other
11053 unit if the number is suffixed by the unit, as specified at the top of this
11054 document. Whatever the expected normal idle time, it is a good practice to
11055 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010011056 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020011057
11058 This parameter is specific to backends, but can be specified once for all in
11059 "defaults" sections. This is in fact one of the easiest solutions not to
11060 forget about it.
11061
11062 Example :
11063 defaults http
11064 option http-server-close
11065 timeout connect 5s
11066 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020011067 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020011068 timeout server 30s
11069 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
11070
Willy Tarreau05cdd962014-05-10 14:30:07 +020011071 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020011072
11073
Willy Tarreau844e3c52008-01-11 16:28:18 +010011074transparent (deprecated)
11075 Enable client-side transparent proxying
11076 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010011077 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010011078 Arguments : none
11079
11080 This keyword was introduced in order to provide layer 7 persistence to layer
11081 3 load balancers. The idea is to use the OS's ability to redirect an incoming
11082 connection for a remote address to a local process (here HAProxy), and let
11083 this process know what address was initially requested. When this option is
11084 used, sessions without cookies will be forwarded to the original destination
11085 IP address of the incoming request (which should match that of another
11086 equipment), while requests with cookies will still be forwarded to the
11087 appropriate server.
11088
11089 The "transparent" keyword is deprecated, use "option transparent" instead.
11090
11091 Note that contrary to a common belief, this option does NOT make HAProxy
11092 present the client's IP to the server when establishing the connection.
11093
Willy Tarreau844e3c52008-01-11 16:28:18 +010011094 See also: "option transparent"
11095
William Lallemanda73203e2012-03-12 12:48:57 +010011096unique-id-format <string>
11097 Generate a unique ID for each request.
11098 May be used in sections : defaults | frontend | listen | backend
11099 yes | yes | yes | no
11100 Arguments :
11101 <string> is a log-format string.
11102
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011103 This keyword creates a ID for each request using the custom log format. A
11104 unique ID is useful to trace a request passing through many components of
11105 a complex infrastructure. The newly created ID may also be logged using the
11106 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010011107
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011108 The format should be composed from elements that are guaranteed to be
11109 unique when combined together. For instance, if multiple haproxy instances
11110 are involved, it might be important to include the node name. It is often
11111 needed to log the incoming connection's source and destination addresses
11112 and ports. Note that since multiple requests may be performed over the same
11113 connection, including a request counter may help differentiate them.
11114 Similarly, a timestamp may protect against a rollover of the counter.
11115 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010011116
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011117 It is recommended to use hexadecimal notation for many fields since it
11118 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010011119
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011120 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010011121
Julien Vehentf21be322014-03-07 08:27:34 -050011122 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010011123
11124 will generate:
11125
11126 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
11127
11128 See also: "unique-id-header"
11129
11130unique-id-header <name>
11131 Add a unique ID header in the HTTP request.
11132 May be used in sections : defaults | frontend | listen | backend
11133 yes | yes | yes | no
11134 Arguments :
11135 <name> is the name of the header.
11136
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011137 Add a unique-id header in the HTTP request sent to the server, using the
11138 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010011139
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011140 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010011141
Julien Vehentf21be322014-03-07 08:27:34 -050011142 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010011143 unique-id-header X-Unique-ID
11144
11145 will generate:
11146
11147 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
11148
11149 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010011150
Willy Tarreauf51658d2014-04-23 01:21:56 +020011151use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020011152 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011153 May be used in sections : defaults | frontend | listen | backend
11154 no | yes | yes | no
11155 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010011156 <backend> is the name of a valid backend or "listen" section, or a
11157 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011158
Willy Tarreauf51658d2014-04-23 01:21:56 +020011159 <condition> is a condition composed of ACLs, as described in section 7. If
11160 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011161
11162 When doing content-switching, connections arrive on a frontend and are then
11163 dispatched to various backends depending on a number of conditions. The
11164 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020011165 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011166 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020011167 source address validation) or combined with a "tcp-request" rule to wait for
11168 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010011169
11170 There may be as many "use_backend" rules as desired. All of these rules are
11171 evaluated in their declaration order, and the first one which matches will
11172 assign the backend.
11173
11174 In the first form, the backend will be used if the condition is met. In the
11175 second form, the backend will be used if the condition is not met. If no
11176 condition is valid, the backend defined with "default_backend" will be used.
11177 If no default backend is defined, either the servers in the same section are
11178 used (in case of a "listen" section) or, in case of a frontend, no server is
11179 used and a 503 service unavailable response is returned.
11180
Willy Tarreau51aecc72009-07-12 09:47:04 +020011181 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011182 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020011183 and backend processing will immediately follow, or the backend will wait for
11184 a complete HTTP request to get in. This feature is useful when a frontend
11185 must decode several protocols on a unique port, one of them being HTTP.
11186
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010011187 When <backend> is a simple name, it is resolved at configuration time, and an
11188 error is reported if the specified backend does not exist. If <backend> is
11189 a log-format string instead, no check may be done at configuration time, so
11190 the backend name is resolved dynamically at run time. If the resulting
11191 backend name does not correspond to any valid backend, no other rule is
11192 evaluated, and the default_backend directive is applied instead. Note that
11193 when using dynamic backend names, it is highly recommended to use a prefix
11194 that no other backend uses in order to ensure that an unauthorized backend
11195 cannot be forced from the request.
11196
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011197 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010011198 used to detect the association between frontends and backends to compute the
11199 backend's "fullconn" setting. This cannot be done for dynamic names.
11200
11201 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
11202 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010011203
Willy Tarreau036fae02008-01-06 13:24:40 +010011204
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011205use-server <server> if <condition>
11206use-server <server> unless <condition>
11207 Only use a specific server if/unless an ACL-based condition is matched.
11208 May be used in sections : defaults | frontend | listen | backend
11209 no | no | yes | yes
11210 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020011211 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011212
11213 <condition> is a condition composed of ACLs, as described in section 7.
11214
11215 By default, connections which arrive to a backend are load-balanced across
11216 the available servers according to the configured algorithm, unless a
11217 persistence mechanism such as a cookie is used and found in the request.
11218
11219 Sometimes it is desirable to forward a particular request to a specific
11220 server without having to declare a dedicated backend for this server. This
11221 can be achieved using the "use-server" rules. These rules are evaluated after
11222 the "redirect" rules and before evaluating cookies, and they have precedence
11223 on them. There may be as many "use-server" rules as desired. All of these
11224 rules are evaluated in their declaration order, and the first one which
11225 matches will assign the server.
11226
11227 If a rule designates a server which is down, and "option persist" is not used
11228 and no force-persist rule was validated, it is ignored and evaluation goes on
11229 with the next rules until one matches.
11230
11231 In the first form, the server will be used if the condition is met. In the
11232 second form, the server will be used if the condition is not met. If no
11233 condition is valid, the processing continues and the server will be assigned
11234 according to other persistence mechanisms.
11235
11236 Note that even if a rule is matched, cookie processing is still performed but
11237 does not assign the server. This allows prefixed cookies to have their prefix
11238 stripped.
11239
11240 The "use-server" statement works both in HTTP and TCP mode. This makes it
11241 suitable for use with content-based inspection. For instance, a server could
11242 be selected in a farm according to the TLS SNI field. And if these servers
11243 have their weight set to zero, they will not be used for other traffic.
11244
11245 Example :
11246 # intercept incoming TLS requests based on the SNI field
11247 use-server www if { req_ssl_sni -i www.example.com }
11248 server www 192.168.0.1:443 weight 0
11249 use-server mail if { req_ssl_sni -i mail.example.com }
11250 server mail 192.168.0.1:587 weight 0
11251 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000011252 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011253 # all the rest is forwarded to this server
11254 server default 192.168.0.2:443 check
11255
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011256 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011257
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011258
Davor Ocelice9ed2812017-12-25 17:49:28 +0100112595. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011260--------------------------
11261
11262The "bind", "server" and "default-server" keywords support a number of settings
11263depending on some build options and on the system HAProxy was built on. These
11264settings generally each consist in one word sometimes followed by a value,
11265written on the same line as the "bind" or "server" line. All these options are
11266described in this section.
11267
11268
112695.1. Bind options
11270-----------------
11271
11272The "bind" keyword supports a certain number of settings which are all passed
11273as arguments on the same line. The order in which those arguments appear makes
11274no importance, provided that they appear after the bind address. All of these
11275parameters are optional. Some of them consist in a single words (booleans),
11276while other ones expect a value after them. In this case, the value must be
11277provided immediately after the setting name.
11278
11279The currently supported settings are the following ones.
11280
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011281accept-netscaler-cip <magic number>
11282 Enforces the use of the NetScaler Client IP insertion protocol over any
11283 connection accepted by any of the TCP sockets declared on the same line. The
11284 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
11285 the incoming connection to be used everywhere an address is used, with the
11286 only exception of "tcp-request connection" rules which will only see the
11287 real connection address. Logs will reflect the addresses indicated in the
11288 protocol, unless it is violated, in which case the real address will still
11289 be used. This keyword combined with support from external components can be
11290 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010011291 mechanism which is not always reliable and not even always usable. See also
11292 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
11293 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010011294
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011295accept-proxy
11296 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020011297 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
11298 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011299 3/4 addresses of the incoming connection to be used everywhere an address is
11300 used, with the only exception of "tcp-request connection" rules which will
11301 only see the real connection address. Logs will reflect the addresses
11302 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010011303 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011304 components can be used as an efficient and reliable alternative to the
11305 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020011306 usable. See also "tcp-request connection expect-proxy" for a finer-grained
11307 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011308
Olivier Houchardc2aae742017-09-22 18:26:28 +020011309allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010011310 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010011311 due to security considerations. Because it is vulnerable to replay attacks,
John Roesler27c754c2019-07-10 15:45:51 -050011312 you should only allow if for requests that are safe to replay, i.e. requests
Olivier Houchard69752962019-01-08 15:35:32 +010011313 that are idempotent. You can use the "wait-for-handshake" action for any
11314 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020011315
Willy Tarreauab861d32013-04-02 02:30:41 +020011316alpn <protocols>
11317 This enables the TLS ALPN extension and advertises the specified protocol
11318 list as supported on top of ALPN. The protocol list consists in a comma-
11319 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roesler27c754c2019-07-10 15:45:51 -050011320 quotes). This requires that the SSL library is built with support for TLS
Willy Tarreauab861d32013-04-02 02:30:41 +020011321 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010011322 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
11323 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
11324 now obsolete NPN extension. At the time of writing this, most browsers still
11325 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
11326 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
11327 are expected to be supported, both versions can be advertised, in order of
11328 preference, like below :
11329
11330 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020011331
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011332backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010011333 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011334 backlog is used instead, which generally defaults to the maxconn value.
11335
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010011336curves <curves>
11337 This setting is only available when support for OpenSSL was built in. It sets
11338 the string describing the list of elliptic curves algorithms ("curve suite")
11339 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
11340 string is a colon-delimited list of curve name.
11341 Example: "X25519:P-256" (without quote)
11342 When "curves" is set, "ecdhe" parameter is ignored.
11343
Emeric Brun7fb34422012-09-28 15:26:15 +020011344ecdhe <named curve>
11345 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010011346 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
11347 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020011348
Emeric Brunfd33a262012-10-11 16:28:27 +020011349ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020011350 This setting is only available when support for OpenSSL was built in. It
11351 designates a PEM file from which to load CA certificates used to verify
11352 client's certificate.
11353
Emeric Brunb6dc9342012-09-28 17:55:37 +020011354ca-ignore-err [all|<errorID>,...]
11355 This setting is only available when support for OpenSSL was built in.
11356 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
11357 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
11358 error is ignored.
11359
Christopher Faulet31af49d2015-06-09 17:29:50 +020011360ca-sign-file <cafile>
11361 This setting is only available when support for OpenSSL was built in. It
11362 designates a PEM file containing both the CA certificate and the CA private
11363 key used to create and sign server's certificates. This is a mandatory
11364 setting when the dynamic generation of certificates is enabled. See
11365 'generate-certificates' for details.
11366
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000011367ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020011368 This setting is only available when support for OpenSSL was built in. It is
11369 the CA private key passphrase. This setting is optional and used only when
11370 the dynamic generation of certificates is enabled. See
11371 'generate-certificates' for details.
11372
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011373ciphers <ciphers>
11374 This setting is only available when support for OpenSSL was built in. It sets
11375 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000011376 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011377 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011378 information and recommendations see e.g.
11379 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
11380 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
11381 cipher configuration, please check the "ciphersuites" keyword.
11382
11383ciphersuites <ciphersuites>
11384 This setting is only available when support for OpenSSL was built in and
11385 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
11386 the list of cipher algorithms ("cipher suite") that are negotiated during the
11387 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011388 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
11389 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011390
Emeric Brunfd33a262012-10-11 16:28:27 +020011391crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020011392 This setting is only available when support for OpenSSL was built in. It
11393 designates a PEM file from which to load certificate revocation list used
11394 to verify client's certificate.
11395
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011396crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000011397 This setting is only available when support for OpenSSL was built in. It
11398 designates a PEM file containing both the required certificates and any
11399 associated private keys. This file can be built by concatenating multiple
11400 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
11401 requires an intermediate certificate, this can also be concatenated into this
11402 file.
11403
11404 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
11405 are loaded.
11406
11407 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010011408 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010011409 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
11410 specified multiple times in order to load certificates from multiple files or
11411 directories. The certificates will be presented to clients who provide a
11412 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010011413 subjects. Wildcards are supported, where a wildcard character '*' is used
11414 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010011415 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000011416
11417 If no SNI is provided by the client or if the SSL library does not support
11418 TLS extensions, or if the client provides an SNI hostname which does not
11419 match any certificate, then the first loaded certificate will be presented.
11420 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010011421 recommended to load the default one first as a file or to ensure that it will
11422 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000011423
Emeric Brune032bfa2012-09-28 13:01:45 +020011424 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011425
Davor Ocelice9ed2812017-12-25 17:49:28 +010011426 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000011427 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010011428 choose a web server that the CA believes requires an intermediate CA (for
11429 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000011430 others, e.g. nginx, result in a wrong bundle that will not work for some
11431 clients).
11432
Emeric Brun4147b2e2014-06-16 18:36:30 +020011433 For each PEM file, haproxy checks for the presence of file at the same path
11434 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
11435 Status Request extension (also known as "OCSP stapling") is automatically
11436 enabled. The content of this file is optional. If not empty, it must contain
11437 a valid OCSP Response in DER format. In order to be valid an OCSP Response
11438 must comply with the following rules: it has to indicate a good status,
11439 it has to be a single response for the certificate of the PEM file, and it
11440 has to be valid at the moment of addition. If these rules are not respected
11441 the OCSP Response is ignored and a warning is emitted. In order to identify
11442 which certificate an OCSP Response applies to, the issuer's certificate is
11443 necessary. If the issuer's certificate is not found in the PEM file, it will
11444 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
11445 if it exists otherwise it will fail with an error.
11446
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010011447 For each PEM file, haproxy also checks for the presence of file at the same
11448 path suffixed by ".sctl". If such file is found, support for Certificate
11449 Transparency (RFC6962) TLS extension is enabled. The file must contain a
11450 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
11451 to check basic syntax, but no signatures are verified.
11452
yanbzhu6c25e9e2016-01-05 12:52:02 -050011453 There are cases where it is desirable to support multiple key types, e.g. RSA
11454 and ECDSA in the cipher suites offered to the clients. This allows clients
11455 that support EC certificates to be able to use EC ciphers, while
11456 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050011457
11458 In order to provide this functionality, multiple PEM files, each with a
11459 different key type, are required. To associate these PEM files into a
11460 "cert bundle" that is recognized by haproxy, they must be named in the
11461 following way: All PEM files that are to be bundled must have the same base
11462 name, with a suffix indicating the key type. Currently, three suffixes are
11463 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
11464 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
11465 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
11466 suffix matters. To load this bundle into haproxy, specify the base name only:
11467
11468 Example : bind :8443 ssl crt example.pem
11469
yanbzhu6c25e9e2016-01-05 12:52:02 -050011470 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050011471 a cert bundle.
11472
Davor Ocelice9ed2812017-12-25 17:49:28 +010011473 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050011474 support multiple key types. PEM files are combined based on Common Name
11475 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
11476 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
11477 entries in the certificates in that bundle, haproxy will not be able to
11478 provide multi-cert support.
11479
11480 Assuming bundle in the example above contained the following:
11481
11482 Filename | CN | SAN
11483 -------------------+-----------------+-------------------
11484 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050011485 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050011486 example.pem.ecdsa | www.example.com | ecdsa.example.com
11487 -------------------+-----------------+-------------------
11488
11489 Users connecting with an SNI of "www.example.com" will be able
11490 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
11491 "rsa.example.com" will only be able to use RSA cipher suites, and users
11492 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020011493 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
11494 no need to bundle certificates. ECDSA certificate will be preferred if client
11495 support it.
yanbzhud19630c2015-12-14 15:10:25 -050011496
11497 If a directory name is given as the <cert> argument, haproxy will
11498 automatically search and load bundled files in that directory.
11499
11500 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
11501 bundling. Each certificate can have its own .ocsp and .issuer file. At this
11502 time, sctl is not supported in multi-certificate bundling.
11503
Emeric Brunb6dc9342012-09-28 17:55:37 +020011504crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000011505 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010011506 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011507 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000011508 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020011509
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011510crt-list <file>
11511 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011512 designates a list of PEM file with an optional ssl configuration and a SNI
11513 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011514
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011515 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
11516
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011517 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
11518 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020011519 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011520 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011521
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020011522 Wildcards are supported in the SNI filter. Negative filter are also supported,
11523 only useful in combination with a wildcard filter to exclude a particular SNI.
11524 The certificates will be presented to clients who provide a valid TLS Server
11525 Name Indication field matching one of the SNI filters. If no SNI filter is
11526 specified, the CN and alt subjects are used. This directive may be specified
11527 multiple times. See the "crt" option for more information. The default
11528 certificate is still needed to meet OpenSSL expectations. If it is not used,
11529 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010011530
yanbzhu6c25e9e2016-01-05 12:52:02 -050011531 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020011532 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020011533 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
11534 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
11535 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050011536
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011537 crt-list file example:
11538 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011539 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011540 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011541 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011542
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011543defer-accept
11544 Is an optional keyword which is supported only on certain Linux kernels. It
11545 states that a connection will only be accepted once some data arrive on it,
11546 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010011547 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011548 performance by ensuring that most of the request is already available when
11549 the connection is accepted. On the other hand, it will not be able to detect
11550 connections which don't talk. It is important to note that this option is
11551 broken in all kernels up to 2.6.31, as the connection is never accepted until
11552 the client talks. This can cause issues with front firewalls which would see
11553 an established connection while the proxy will only see it in SYN_RECV. This
11554 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
11555
William Lallemandf6975e92017-05-26 17:42:10 +020011556expose-fd listeners
11557 This option is only usable with the stats socket. It gives your stats socket
11558 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020011559 During a reload with the master-worker mode, the process is automatically
11560 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011561 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020011562
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011563force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011564 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011565 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011566 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011567 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011568
11569force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011570 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011571 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011572 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011573
11574force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011575 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011576 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011577 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011578
11579force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011580 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011581 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011582 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011583
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011584force-tlsv13
11585 This option enforces use of TLSv1.3 only on SSL connections instantiated from
11586 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011587 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011588
Christopher Faulet31af49d2015-06-09 17:29:50 +020011589generate-certificates
11590 This setting is only available when support for OpenSSL was built in. It
11591 enables the dynamic SSL certificates generation. A CA certificate and its
11592 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
11593 a transparent forward proxy, SSL requests generate errors because of a common
11594 name mismatch on the certificate presented to the client. With this option
11595 enabled, HAProxy will try to forge a certificate using the SNI hostname
11596 indicated by the client. This is done only if no certificate matches the SNI
11597 hostname (see 'crt-list'). If an error occurs, the default certificate is
11598 used, else the 'strict-sni' option is set.
11599 It can also be used when HAProxy is configured as a reverse proxy to ease the
11600 deployment of an architecture with many backends.
11601
11602 Creating a SSL certificate is an expensive operation, so a LRU cache is used
11603 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010011604 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020011605 certificate is used many times.
11606
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011607gid <gid>
11608 Sets the group of the UNIX sockets to the designated system gid. It can also
11609 be set by default in the global section's "unix-bind" statement. Note that
11610 some platforms simply ignore this. This setting is equivalent to the "group"
11611 setting except that the group ID is used instead of its name. This setting is
11612 ignored by non UNIX sockets.
11613
11614group <group>
11615 Sets the group of the UNIX sockets to the designated system group. It can
11616 also be set by default in the global section's "unix-bind" statement. Note
11617 that some platforms simply ignore this. This setting is equivalent to the
11618 "gid" setting except that the group name is used instead of its gid. This
11619 setting is ignored by non UNIX sockets.
11620
11621id <id>
11622 Fixes the socket ID. By default, socket IDs are automatically assigned, but
11623 sometimes it is more convenient to fix them to ease monitoring. This value
11624 must be strictly positive and unique within the listener/frontend. This
11625 option can only be used when defining only a single socket.
11626
11627interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010011628 Restricts the socket to a specific interface. When specified, only packets
11629 received from that particular interface are processed by the socket. This is
11630 currently only supported on Linux. The interface must be a primary system
11631 interface, not an aliased interface. It is also possible to bind multiple
11632 frontends to the same address if they are bound to different interfaces. Note
11633 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010011634 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
11635 uses the same interface as inbound traffic, and its associated routing table,
11636 even if there are explicit routes through different interfaces configured.
11637 This can prove useful to address asymmetric routing issues when the same
11638 client IP addresses need to be able to reach frontends hosted on different
11639 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011640
Willy Tarreauabb175f2012-09-24 12:43:26 +020011641level <level>
11642 This setting is used with the stats sockets only to restrict the nature of
11643 the commands that can be issued on the socket. It is ignored by other
11644 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010011645 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020011646 read, and no change is allowed. It would make sense on systems where it
11647 is not easy to restrict access to the socket.
11648 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011649 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020011650 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010011651 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020011652 all counters).
11653
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020011654severity-output <format>
11655 This setting is used with the stats sockets only to configure severity
11656 level output prepended to informational feedback messages. Severity
11657 level of messages can range between 0 and 7, conforming to syslog
11658 rfc5424. Valid and successful socket commands requesting data
11659 (i.e. "show map", "get acl foo" etc.) will never have a severity level
11660 prepended. It is ignored by other sockets. <format> can be one of :
11661 - "none" (default) no severity level is prepended to feedback messages.
11662 - "number" severity level is prepended as a number.
11663 - "string" severity level is prepended as a string following the
11664 rfc5424 convention.
11665
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011666maxconn <maxconn>
11667 Limits the sockets to this number of concurrent connections. Extraneous
11668 connections will remain in the system's backlog until a connection is
11669 released. If unspecified, the limit will be the same as the frontend's
11670 maxconn. Note that in case of port ranges or multiple addresses, the same
11671 value will be applied to each socket. This setting enables different
11672 limitations on expensive sockets, for instance SSL entries which may easily
11673 eat all memory.
11674
11675mode <mode>
11676 Sets the octal mode used to define access permissions on the UNIX socket. It
11677 can also be set by default in the global section's "unix-bind" statement.
11678 Note that some platforms simply ignore this. This setting is ignored by non
11679 UNIX sockets.
11680
11681mss <maxseg>
11682 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
11683 connections. This can be used to force a lower MSS for certain specific
11684 ports, for instance for connections passing through a VPN. Note that this
11685 relies on a kernel feature which is theoretically supported under Linux but
11686 was buggy in all versions prior to 2.6.28. It may or may not work on other
11687 operating systems. It may also not change the advertised value but change the
11688 effective size of outgoing segments. The commonly advertised value for TCPv4
11689 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
11690 positive, it will be used as the advertised MSS. If it is negative, it will
11691 indicate by how much to reduce the incoming connection's advertised MSS for
11692 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
11693
11694name <name>
11695 Sets an optional name for these sockets, which will be reported on the stats
11696 page.
11697
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011698namespace <name>
11699 On Linux, it is possible to specify which network namespace a socket will
11700 belong to. This directive makes it possible to explicitly bind a listener to
11701 a namespace different from the default one. Please refer to your operating
11702 system's documentation to find more details about network namespaces.
11703
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011704nice <nice>
11705 Sets the 'niceness' of connections initiated from the socket. Value must be
11706 in the range -1024..1024 inclusive, and defaults to zero. Positive values
11707 means that such connections are more friendly to others and easily offer
11708 their place in the scheduler. On the opposite, negative values mean that
11709 connections want to run with a higher priority than others. The difference
11710 only happens under high loads when the system is close to saturation.
11711 Negative values are appropriate for low-latency or administration services,
11712 and high values are generally recommended for CPU intensive tasks such as SSL
11713 processing or bulk transfers which are less sensible to latency. For example,
11714 it may make sense to use a positive value for an SMTP socket and a negative
11715 one for an RDP socket.
11716
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011717no-ca-names
11718 This setting is only available when support for OpenSSL was built in. It
11719 prevents from send CA names in server hello message when ca-file is used.
11720
Emeric Brun9b3009b2012-10-05 11:55:06 +020011721no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011722 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011723 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011724 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011725 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011726 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11727 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011728
Emeric Brun90ad8722012-10-02 14:00:59 +020011729no-tls-tickets
11730 This setting is only available when support for OpenSSL was built in. It
11731 disables the stateless session resumption (RFC 5077 TLS Ticket
11732 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011733 session resumption is more expensive in CPU usage. This option is also
11734 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020011735
Emeric Brun9b3009b2012-10-05 11:55:06 +020011736no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011737 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011738 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011739 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011740 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011741 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11742 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011743
Emeric Brun9b3009b2012-10-05 11:55:06 +020011744no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011745 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011746 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011747 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011748 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011749 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11750 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011751
Emeric Brun9b3009b2012-10-05 11:55:06 +020011752no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011753 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011754 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011755 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011756 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011757 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11758 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011759
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011760no-tlsv13
11761 This setting is only available when support for OpenSSL was built in. It
11762 disables support for TLSv1.3 on any sockets instantiated from the listener
11763 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11764 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011765 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11766 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011767
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011768npn <protocols>
11769 This enables the NPN TLS extension and advertises the specified protocol list
11770 as supported on top of NPN. The protocol list consists in a comma-delimited
11771 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roesler27c754c2019-07-10 15:45:51 -050011772 This requires that the SSL library is built with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011773 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011774 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11775 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11776 version of OpenSSL, NPN might still be used as most clients still support it
11777 at the time of writing this. It is possible to enable both NPN and ALPN
11778 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011779
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011780prefer-client-ciphers
11781 Use the client's preference when selecting the cipher suite, by default
11782 the server's preference is enforced. This option is also available on
11783 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011784 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11785 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11786 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011787
Christopher Fauletc644fa92017-11-23 22:44:11 +010011788process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010011789 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010011790 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011791 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011792 between the two is applied. If in the end the listener is not allowed to run
11793 on any remaining process, a warning is emitted, and the listener will either
11794 run on the first process of the listener if a single process was specified,
11795 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011796 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010011797 connections for this listener, for the the process set. If multiple processes
11798 and threads are configured, a warning is emitted, as it either results from a
11799 configuration error or a misunderstanding of these models. For the unlikely
11800 case where several ranges are needed, this directive may be repeated.
11801 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010011802
11803 all | odd | even | number[-[number]]
11804
11805 Ranges can be partially defined. The higher bound can be omitted. In such
11806 case, it is replaced by the corresponding maximum value. The main purpose of
11807 this directive is to be used with the stats sockets and have one different
11808 socket per process. The second purpose is to have multiple bind lines sharing
11809 the same IP:port but not the same process in a listener, so that the system
11810 can distribute the incoming connections into multiple queues and allow a
11811 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11812 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011813
Christopher Fauleta717b992018-04-10 14:43:00 +020011814proto <name>
11815 Forces the multiplexer's protocol to use for the incoming connections. It
11816 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11817 be usable on the frontend side. The list of available protocols is reported
11818 in haproxy -vv.
11819 Idea behind this optipon is to bypass the selection of the best multiplexer's
11820 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011821 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011822 h2" on the bind line.
11823
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011824ssl
11825 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011826 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011827 certificate is necessary (see "crt" above). All contents in the buffers will
11828 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011829 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11830 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011831
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011832ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11833 This option enforces use of <version> or lower on SSL connections instantiated
11834 from this listener. This option is also available on global statement
11835 "ssl-default-bind-options". See also "ssl-min-ver".
11836
11837ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11838 This option enforces use of <version> or upper on SSL connections instantiated
11839 from this listener. This option is also available on global statement
11840 "ssl-default-bind-options". See also "ssl-max-ver".
11841
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011842strict-sni
11843 This setting is only available when support for OpenSSL was built in. The
11844 SSL/TLS negotiation is allow only if the client provided an SNI which match
11845 a certificate. The default certificate is not used.
11846 See the "crt" option for more information.
11847
Willy Tarreau2af207a2015-02-04 00:45:58 +010011848tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011849 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011850 listening socket. This option is available on Linux since version 2.6.37. It
11851 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011852 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011853 useful on long-lived connections experiencing long idle periods such as
11854 remote terminals or database connection pools, where the client and server
11855 timeouts must remain high to allow a long period of idle, but where it is
11856 important to detect that the client has disappeared in order to release all
11857 resources associated with its connection (and the server's session). The
11858 argument is a delay expressed in milliseconds by default. This only works
11859 for regular TCP connections, and is ignored for other protocols.
11860
Willy Tarreau1c862c52012-10-05 16:21:00 +020011861tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011862 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011863 enables TCP Fast Open on the listening socket, which means that clients which
11864 support this feature will be able to send a request and receive a response
11865 during the 3-way handshake starting from second connection, thus saving one
11866 round-trip after the first connection. This only makes sense with protocols
11867 that use high connection rates and where each round trip matters. This can
11868 possibly cause issues with many firewalls which do not accept data on SYN
11869 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011870 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11871 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11872 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011873
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011874tls-ticket-keys <keyfile>
11875 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010011876 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
11877 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
11878 The first key determines the key length used for next keys: you can't mix
11879 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
11880 build option (default 3) and at least as many keys need to be present in
11881 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
11882 penultimate one for encryption. This enables easy key rotation by just
11883 appending new key to the file and reloading the process. Keys must be
11884 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
11885 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011886 storage such as hard drives (hint: use tmpfs and don't swap those files).
11887 Lifetime hint can be changed using tune.ssl.timeout.
11888
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011889transparent
11890 Is an optional keyword which is supported only on certain Linux kernels. It
11891 indicates that the addresses will be bound even if they do not belong to the
11892 local machine, and that packets targeting any of these addresses will be
11893 intercepted just as if the addresses were locally configured. This normally
11894 requires that IP forwarding is enabled. Caution! do not use this with the
11895 default address '*', as it would redirect any traffic for the specified port.
11896 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11897 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11898 kernel version. Some distribution kernels include backports of the feature,
11899 so check for support with your vendor.
11900
Willy Tarreau77e3af92012-11-24 15:07:23 +010011901v4v6
11902 Is an optional keyword which is supported only on most recent systems
11903 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11904 and IPv6 when it uses the default address. Doing so is sometimes necessary
11905 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011906 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011907
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011908v6only
11909 Is an optional keyword which is supported only on most recent systems
11910 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11911 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011912 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11913 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011914
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011915uid <uid>
11916 Sets the owner of the UNIX sockets to the designated system uid. It can also
11917 be set by default in the global section's "unix-bind" statement. Note that
11918 some platforms simply ignore this. This setting is equivalent to the "user"
11919 setting except that the user numeric ID is used instead of its name. This
11920 setting is ignored by non UNIX sockets.
11921
11922user <user>
11923 Sets the owner of the UNIX sockets to the designated system user. It can also
11924 be set by default in the global section's "unix-bind" statement. Note that
11925 some platforms simply ignore this. This setting is equivalent to the "uid"
11926 setting except that the user name is used instead of its uid. This setting is
11927 ignored by non UNIX sockets.
11928
Emeric Brun1a073b42012-09-28 17:07:34 +020011929verify [none|optional|required]
11930 This setting is only available when support for OpenSSL was built in. If set
11931 to 'none', client certificate is not requested. This is the default. In other
11932 cases, a client certificate is requested. If the client does not provide a
11933 certificate after the request and if 'verify' is set to 'required', then the
11934 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011935 certificate provided by the client is always verified using CAs from
11936 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11937 is aborted, regardless of the 'verify' option, unless the error code exactly
11938 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011939
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200119405.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011941------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011942
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011943The "server" and "default-server" keywords support a certain number of settings
11944which are all passed as arguments on the server line. The order in which those
11945arguments appear does not count, and they are all optional. Some of those
11946settings are single words (booleans) while others expect one or several values
11947after them. In this case, the values must immediately follow the setting name.
11948Except default-server, all those settings must be specified after the server's
11949address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011950
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011951 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011952 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011953
Frédéric Lécailled2376272017-03-21 18:52:12 +010011954Note that all these settings are supported both by "server" and "default-server"
11955keywords, except "id" which is only supported by "server".
11956
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011957The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011958
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011959addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011960 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011961 to send health-checks or to probe the agent-check. On some servers, it may be
11962 desirable to dedicate an IP address to specific component able to perform
11963 complex tests which are more suitable to health-checks than the application.
11964 This parameter is ignored if the "check" parameter is not set. See also the
11965 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011966
Simon Hormand60d6912013-11-25 10:46:36 +090011967agent-check
11968 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011969 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010011970 to the port set by the "agent-port" parameter and reading an ASCII string
11971 terminated by the first '\r' or '\n' met. The string is made of a series of
11972 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011973
Willy Tarreau81f5d942013-12-09 20:51:51 +010011974 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011975 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011976 weight of a server as configured when haproxy starts. Note that a zero
11977 weight is reported on the stats page as "DRAIN" since it has the same
11978 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011979
Davor Ocelice9ed2812017-12-25 17:49:28 +010011980 - The string "maxconn:" followed by an integer (no space between). Values
11981 in this format will set the maxconn of a server. The maximum number of
11982 connections advertised needs to be multiplied by the number of load
11983 balancers and different backends that use this health check to get the
11984 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011985
Willy Tarreau81f5d942013-12-09 20:51:51 +010011986 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011987 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011988
Willy Tarreau81f5d942013-12-09 20:51:51 +010011989 - The word "drain". This will turn the server's administrative state to the
11990 DRAIN mode, thus it will not accept any new connections other than those
11991 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011992
Willy Tarreau81f5d942013-12-09 20:51:51 +010011993 - The word "maint". This will turn the server's administrative state to the
11994 MAINT mode, thus it will not accept any new connections at all, and health
11995 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011996
Willy Tarreau81f5d942013-12-09 20:51:51 +010011997 - The words "down", "failed", or "stopped", optionally followed by a
11998 description string after a sharp ('#'). All of these mark the server's
11999 operating state as DOWN, but since the word itself is reported on the stats
12000 page, the difference allows an administrator to know if the situation was
12001 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010012002 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010012003 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090012004
Willy Tarreau81f5d942013-12-09 20:51:51 +010012005 - The word "up" sets back the server's operating state as UP if health checks
12006 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090012007
Willy Tarreau81f5d942013-12-09 20:51:51 +010012008 Parameters which are not advertised by the agent are not changed. For
12009 example, an agent might be designed to monitor CPU usage and only report a
12010 relative weight and never interact with the operating status. Similarly, an
12011 agent could be designed as an end-user interface with 3 radio buttons
12012 allowing an administrator to change only the administrative state. However,
12013 it is important to consider that only the agent may revert its own actions,
12014 so if a server is set to DRAIN mode or to DOWN state using the agent, the
12015 agent must implement the other equivalent actions to bring the service into
12016 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090012017
Simon Horman2f1f9552013-11-25 10:46:37 +090012018 Failure to connect to the agent is not considered an error as connectivity
12019 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010012020 parameter. Warning though, it is not a good idea to stop an agent after it
12021 reports "down", since only an agent reporting "up" will be able to turn the
12022 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010012023 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090012024
Willy Tarreau81f5d942013-12-09 20:51:51 +010012025 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010012026 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090012027
James Brown55f9ff12015-10-21 18:19:05 -070012028agent-send <string>
12029 If this option is specified, haproxy will send the given string (verbatim)
12030 to the agent server upon connection. You could, for example, encode
12031 the backend name into this string, which would enable your agent to send
12032 different responses based on the backend. Make sure to include a '\n' if
12033 you want to terminate your request with a newline.
12034
Simon Hormand60d6912013-11-25 10:46:36 +090012035agent-inter <delay>
12036 The "agent-inter" parameter sets the interval between two agent checks
12037 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
12038
12039 Just as with every other time-based parameter, it may be entered in any
12040 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
12041 parameter also serves as a timeout for agent checks "timeout check" is
12042 not set. In order to reduce "resonance" effects when multiple servers are
12043 hosted on the same hardware, the agent and health checks of all servers
12044 are started with a small time offset between them. It is also possible to
12045 add some random noise in the agent and health checks interval using the
12046 global "spread-checks" keyword. This makes sense for instance when a lot
12047 of backends use the same servers.
12048
12049 See also the "agent-check" and "agent-port" parameters.
12050
Misiek768d8602017-01-09 09:52:43 +010012051agent-addr <addr>
12052 The "agent-addr" parameter sets address for agent check.
12053
12054 You can offload agent-check to another target, so you can make single place
12055 managing status and weights of servers defined in haproxy in case you can't
12056 make self-aware and self-managing services. You can specify both IP or
12057 hostname, it will be resolved.
12058
Simon Hormand60d6912013-11-25 10:46:36 +090012059agent-port <port>
12060 The "agent-port" parameter sets the TCP port used for agent checks.
12061
12062 See also the "agent-check" and "agent-inter" parameters.
12063
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020012064allow-0rtt
12065 Allow sending early data to the server when using TLS 1.3.
Olivier Houchard22c9b442019-05-06 19:01:04 +020012066 Note that early data will be sent only if the client used early data, or
12067 if the backend uses "retry-on" with the "0rtt-rejected" keyword.
Olivier Houchard8cb2d2e2019-05-06 18:58:48 +020012068
Olivier Houchardc7566002018-11-20 23:33:50 +010012069alpn <protocols>
12070 This enables the TLS ALPN extension and advertises the specified protocol
12071 list as supported on top of ALPN. The protocol list consists in a comma-
12072 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
John Roesler27c754c2019-07-10 15:45:51 -050012073 quotes). This requires that the SSL library is built with support for TLS
Olivier Houchardc7566002018-11-20 23:33:50 +010012074 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
12075 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
12076 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
12077 now obsolete NPN extension.
12078 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
12079 be advertised, in order of preference, like below :
12080
12081 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
12082
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012083backup
12084 When "backup" is present on a server line, the server is only used in load
12085 balancing when all other non-backup servers are unavailable. Requests coming
12086 with a persistence cookie referencing the server will always be served
12087 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010012088 the "allbackups" option is set in the backend. See also the "no-backup" and
12089 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012090
Emeric Brunef42d922012-10-11 16:11:36 +020012091ca-file <cafile>
12092 This setting is only available when support for OpenSSL was built in. It
12093 designates a PEM file from which to load CA certificates used to verify
12094 server's certificate.
12095
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012096check
12097 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010012098 always considered available. If "check" is set, the server is available when
12099 accepting periodic TCP connections, to ensure that it is really able to serve
12100 requests. The default address and port to send the tests to are those of the
12101 server, and the default source is the same as the one defined in the
12102 backend. It is possible to change the address using the "addr" parameter, the
12103 port using the "port" parameter, the source address using the "source"
12104 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090012105 parameters. The request method is define in the backend using the "httpchk",
12106 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010012107 refer to those options and parameters for more information. See also
12108 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012109
Willy Tarreau6c16adc2012-10-05 00:04:16 +020012110check-send-proxy
12111 This option forces emission of a PROXY protocol line with outgoing health
12112 checks, regardless of whether the server uses send-proxy or not for the
12113 normal traffic. By default, the PROXY protocol is enabled for health checks
12114 if it is already enabled for normal traffic and if no "port" nor "addr"
12115 directive is present. However, if such a directive is present, the
12116 "check-send-proxy" option needs to be used to force the use of the
12117 protocol. See also the "send-proxy" option for more information.
12118
Olivier Houchard92150142018-12-21 19:47:01 +010012119check-alpn <protocols>
12120 Defines which protocols to advertise with ALPN. The protocol list consists in
12121 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
12122 (without quotes). If it is not set, the server ALPN is used.
12123
Jérôme Magninae9bb762018-12-09 16:08:26 +010012124check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020012125 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010012126 over SSL. It is only possible to use a string to set <sni>. If you want to
12127 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020012128
Willy Tarreau763a95b2012-10-04 23:15:39 +020012129check-ssl
12130 This option forces encryption of all health checks over SSL, regardless of
12131 whether the server uses SSL or not for the normal traffic. This is generally
12132 used when an explicit "port" or "addr" directive is specified and SSL health
12133 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012134 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020012135 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
12136 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012137 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010012138 See the "ssl" option for more information and "no-check-ssl" to disable
12139 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012140
Alexander Liu2a54bb72019-05-22 19:44:48 +080012141check-via-socks4
John Roesler27c754c2019-07-10 15:45:51 -050012142 This option enables outgoing health checks using upstream socks4 proxy. By
Alexander Liu2a54bb72019-05-22 19:44:48 +080012143 default, the health checks won't go through socks tunnel even it was enabled
12144 for normal traffic.
12145
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012146ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012147 This setting is only available when support for OpenSSL was built in. This
12148 option sets the string describing the list of cipher algorithms that is
12149 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012150 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
12151 information and recommendations see e.g.
12152 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
12153 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
12154 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012155
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012156ciphersuites <ciphersuites>
12157 This setting is only available when support for OpenSSL was built in and
12158 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
12159 describing the list of cipher algorithms that is negotiated during the TLS
12160 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000012161 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
12162 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
12163 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020012164
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012165cookie <value>
12166 The "cookie" parameter sets the cookie value assigned to the server to
12167 <value>. This value will be checked in incoming requests, and the first
12168 operational server possessing the same value will be selected. In return, in
12169 cookie insertion or rewrite modes, this value will be assigned to the cookie
12170 sent to the client. There is nothing wrong in having several servers sharing
12171 the same cookie value, and it is in fact somewhat common between normal and
12172 backup servers. See also the "cookie" keyword in backend section.
12173
Emeric Brunef42d922012-10-11 16:11:36 +020012174crl-file <crlfile>
12175 This setting is only available when support for OpenSSL was built in. It
12176 designates a PEM file from which to load certificate revocation list used
12177 to verify server's certificate.
12178
Emeric Bruna7aa3092012-10-26 12:58:00 +020012179crt <cert>
12180 This setting is only available when support for OpenSSL was built in.
12181 It designates a PEM file from which to load both a certificate and the
12182 associated private key. This file can be built by concatenating both PEM
12183 files into one. This certificate will be sent if the server send a client
12184 certificate request.
12185
Willy Tarreau96839092010-03-29 10:02:24 +020012186disabled
12187 The "disabled" keyword starts the server in the "disabled" state. That means
12188 that it is marked down in maintenance mode, and no connection other than the
12189 ones allowed by persist mode will reach it. It is very well suited to setup
12190 new servers, because normal traffic will never reach them, while it is still
12191 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012192 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020012193
Frédéric Lécailled2376272017-03-21 18:52:12 +010012194enabled
12195 This option may be used as 'server' setting to reset any 'disabled'
12196 setting which would have been inherited from 'default-server' directive as
12197 default value.
12198 It may also be used as 'default-server' setting to reset any previous
12199 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020012200
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012201error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010012202 If health observing is enabled, the "error-limit" parameter specifies the
12203 number of consecutive errors that triggers event selected by the "on-error"
12204 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012205
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012206 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012207
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012208fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012209 The "fall" parameter states that a server will be considered as dead after
12210 <count> consecutive unsuccessful health checks. This value defaults to 3 if
12211 unspecified. See also the "check", "inter" and "rise" parameters.
12212
Emeric Brun8694b9a2012-10-05 14:39:07 +020012213force-sslv3
12214 This option enforces use of SSLv3 only when SSL is used to communicate with
12215 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012216 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012217 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012218
12219force-tlsv10
12220 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012221 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012222 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012223
12224force-tlsv11
12225 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012226 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012227 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012228
12229force-tlsv12
12230 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012231 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012232 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020012233
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012234force-tlsv13
12235 This option enforces use of TLSv1.3 only when SSL is used to communicate with
12236 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012237 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012238
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012239id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020012240 Set a persistent ID for the server. This ID must be positive and unique for
12241 the proxy. An unused ID will automatically be assigned if unset. The first
12242 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012243
Willy Tarreau6a031d12016-11-07 19:42:35 +010012244init-addr {last | libc | none | <ip>},[...]*
12245 Indicate in what order the server's address should be resolved upon startup
12246 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012247 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010012248 method which succeeds is used. If the end of the list is reached without
12249 finding a working method, an error is thrown. Method "last" suggests to pick
12250 the address which appears in the state file (see "server-state-file"). Method
12251 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
12252 depending on the operating system and build options). Method "none"
12253 specifically indicates that the server should start without any valid IP
12254 address in a down state. It can be useful to ignore some DNS issues upon
12255 startup, waiting for the situation to get fixed later. Finally, an IP address
12256 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012257 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010012258 server used to catch old sessions and present them with a decent error
12259 message for example. When the "first" load balancing algorithm is used, this
12260 IP address could point to a fake server used to trigger the creation of new
12261 instances on the fly. This option defaults to "last,libc" indicating that the
12262 previous address found in the state file (if any) is used first, otherwise
12263 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012264 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010012265
12266 Example:
12267 defaults
12268 # never fail on address resolution
12269 default-server init-addr last,libc,none
12270
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012271inter <delay>
12272fastinter <delay>
12273downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012274 The "inter" parameter sets the interval between two consecutive health checks
12275 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
12276 It is also possible to use "fastinter" and "downinter" to optimize delays
12277 between checks depending on the server state :
12278
Pieter Baauw44fc9df2015-09-17 21:30:46 +020012279 Server state | Interval used
12280 ----------------------------------------+----------------------------------
12281 UP 100% (non-transitional) | "inter"
12282 ----------------------------------------+----------------------------------
12283 Transitionally UP (going down "fall"), | "fastinter" if set,
12284 Transitionally DOWN (going up "rise"), | "inter" otherwise.
12285 or yet unchecked. |
12286 ----------------------------------------+----------------------------------
12287 DOWN 100% (non-transitional) | "downinter" if set,
12288 | "inter" otherwise.
12289 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010012290
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012291 Just as with every other time-based parameter, they can be entered in any
12292 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
12293 serves as a timeout for health checks sent to servers if "timeout check" is
12294 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090012295 hosted on the same hardware, the agent and health checks of all servers
12296 are started with a small time offset between them. It is also possible to
12297 add some random noise in the agent and health checks interval using the
12298 global "spread-checks" keyword. This makes sense for instance when a lot
12299 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012300
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012301maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012302 The "maxconn" parameter specifies the maximal number of concurrent
12303 connections that will be sent to this server. If the number of incoming
Tim Duesterhus50cfb312019-11-27 22:35:27 +010012304 concurrent connections goes higher than this value, they will be queued,
12305 waiting for a slot to be released. This parameter is very important as it can
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012306 save fragile servers from going down under extreme loads. If a "minconn"
12307 parameter is specified, the limit becomes dynamic. The default value is "0"
12308 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
12309 the backend's "fullconn" keyword.
12310
Tim Duesterhus50cfb312019-11-27 22:35:27 +010012311 In HTTP mode this parameter limits the number of concurrent requests instead
12312 of the number of connections. Multiple requests might be multiplexed over a
12313 single TCP connection to the server. As an example if you specify a maxconn
12314 of 50 you might see between 1 and 50 actual server connections, but no more
12315 than 50 concurrent requests.
12316
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012317maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012318 The "maxqueue" parameter specifies the maximal number of connections which
12319 will wait in the queue for this server. If this limit is reached, next
12320 requests will be redispatched to other servers instead of indefinitely
12321 waiting to be served. This will break persistence but may allow people to
12322 quickly re-log in when the server they try to connect to is dying. The
12323 default value is "0" which means the queue is unlimited. See also the
12324 "maxconn" and "minconn" parameters.
12325
Willy Tarreau9c538e02019-01-23 10:21:49 +010012326max-reuse <count>
12327 The "max-reuse" argument indicates the HTTP connection processors that they
12328 should not reuse a server connection more than this number of times to send
12329 new requests. Permitted values are -1 (the default), which disables this
12330 limit, or any positive value. Value zero will effectively disable keep-alive.
12331 This is only used to work around certain server bugs which cause them to leak
12332 resources over time. The argument is not necessarily respected by the lower
12333 layers as there might be technical limitations making it impossible to
12334 enforce. At least HTTP/2 connections to servers will respect it.
12335
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012336minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012337 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
12338 limit following the backend's load. The server will always accept at least
12339 <minconn> connections, never more than <maxconn>, and the limit will be on
12340 the ramp between both values when the backend has less than <fullconn>
12341 concurrent connections. This makes it possible to limit the load on the
12342 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012343 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012344 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012345
Willy Tarreaud72f0f32015-10-13 14:50:22 +020012346namespace <name>
12347 On Linux, it is possible to specify which network namespace a socket will
12348 belong to. This directive makes it possible to explicitly bind a server to
12349 a namespace different from the default one. Please refer to your operating
12350 system's documentation to find more details about network namespaces.
12351
Frédéric Lécailled2376272017-03-21 18:52:12 +010012352no-agent-check
12353 This option may be used as "server" setting to reset any "agent-check"
12354 setting which would have been inherited from "default-server" directive as
12355 default value.
12356 It may also be used as "default-server" setting to reset any previous
12357 "default-server" "agent-check" setting.
12358
12359no-backup
12360 This option may be used as "server" setting to reset any "backup"
12361 setting which would have been inherited from "default-server" directive as
12362 default value.
12363 It may also be used as "default-server" setting to reset any previous
12364 "default-server" "backup" setting.
12365
12366no-check
12367 This option may be used as "server" setting to reset any "check"
12368 setting which would have been inherited from "default-server" directive as
12369 default value.
12370 It may also be used as "default-server" setting to reset any previous
12371 "default-server" "check" setting.
12372
12373no-check-ssl
12374 This option may be used as "server" setting to reset any "check-ssl"
12375 setting which would have been inherited from "default-server" directive as
12376 default value.
12377 It may also be used as "default-server" setting to reset any previous
12378 "default-server" "check-ssl" setting.
12379
Frédéric Lécailled2376272017-03-21 18:52:12 +010012380no-send-proxy
12381 This option may be used as "server" setting to reset any "send-proxy"
12382 setting which would have been inherited from "default-server" directive as
12383 default value.
12384 It may also be used as "default-server" setting to reset any previous
12385 "default-server" "send-proxy" setting.
12386
12387no-send-proxy-v2
12388 This option may be used as "server" setting to reset any "send-proxy-v2"
12389 setting which would have been inherited from "default-server" directive as
12390 default value.
12391 It may also be used as "default-server" setting to reset any previous
12392 "default-server" "send-proxy-v2" setting.
12393
12394no-send-proxy-v2-ssl
12395 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
12396 setting which would have been inherited from "default-server" directive as
12397 default value.
12398 It may also be used as "default-server" setting to reset any previous
12399 "default-server" "send-proxy-v2-ssl" setting.
12400
12401no-send-proxy-v2-ssl-cn
12402 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
12403 setting which would have been inherited from "default-server" directive as
12404 default value.
12405 It may also be used as "default-server" setting to reset any previous
12406 "default-server" "send-proxy-v2-ssl-cn" setting.
12407
12408no-ssl
12409 This option may be used as "server" setting to reset any "ssl"
12410 setting which would have been inherited from "default-server" directive as
12411 default value.
12412 It may also be used as "default-server" setting to reset any previous
12413 "default-server" "ssl" setting.
12414
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010012415no-ssl-reuse
12416 This option disables SSL session reuse when SSL is used to communicate with
12417 the server. It will force the server to perform a full handshake for every
12418 new connection. It's probably only useful for benchmarking, troubleshooting,
12419 and for paranoid users.
12420
Emeric Brun9b3009b2012-10-05 11:55:06 +020012421no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012422 This option disables support for SSLv3 when SSL is used to communicate with
12423 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012424 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012425
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012426 Supported in default-server: No
12427
Emeric Brunf9c5c472012-10-11 15:28:34 +020012428no-tls-tickets
12429 This setting is only available when support for OpenSSL was built in. It
12430 disables the stateless session resumption (RFC 5077 TLS Ticket
12431 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012432 session resumption is more expensive in CPU usage for servers. This option
12433 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010012434 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020012435
Emeric Brun9b3009b2012-10-05 11:55:06 +020012436no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020012437 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020012438 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12439 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012440 often makes sense to disable it when communicating with local servers. This
12441 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012442 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012443
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012444 Supported in default-server: No
12445
Emeric Brun9b3009b2012-10-05 11:55:06 +020012446no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020012447 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020012448 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12449 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012450 often makes sense to disable it when communicating with local servers. This
12451 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012452 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012453
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012454 Supported in default-server: No
12455
Emeric Brun9b3009b2012-10-05 11:55:06 +020012456no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020012457 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012458 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12459 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010012460 often makes sense to disable it when communicating with local servers. This
12461 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012462 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020012463
12464 Supported in default-server: No
12465
12466no-tlsv13
12467 This option disables support for TLSv1.3 when SSL is used to communicate with
12468 the server. Note that SSLv2 is disabled in the code and cannot be enabled
12469 using any configuration option. TLSv1 is more expensive than SSLv3 so it
12470 often makes sense to disable it when communicating with local servers. This
12471 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012472 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012473
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020012474 Supported in default-server: No
12475
Frédéric Lécailled2376272017-03-21 18:52:12 +010012476no-verifyhost
12477 This option may be used as "server" setting to reset any "verifyhost"
12478 setting which would have been inherited from "default-server" directive as
12479 default value.
12480 It may also be used as "default-server" setting to reset any previous
12481 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012482
Frédéric Lécailleaeeb1c92019-07-04 14:19:06 +020012483no-tfo
12484 This option may be used as "server" setting to reset any "tfo"
12485 setting which would have been inherited from "default-server" directive as
12486 default value.
12487 It may also be used as "default-server" setting to reset any previous
12488 "default-server" "tfo" setting.
12489
Simon Hormanfa461682011-06-25 09:39:49 +090012490non-stick
12491 Never add connections allocated to this sever to a stick-table.
12492 This may be used in conjunction with backup to ensure that
12493 stick-table persistence is disabled for backup servers.
12494
Olivier Houchardc7566002018-11-20 23:33:50 +010012495npn <protocols>
12496 This enables the NPN TLS extension and advertises the specified protocol list
12497 as supported on top of NPN. The protocol list consists in a comma-delimited
12498 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
John Roesler27c754c2019-07-10 15:45:51 -050012499 This requires that the SSL library is built with support for TLS extensions
Olivier Houchardc7566002018-11-20 23:33:50 +010012500 enabled (check with haproxy -vv). Note that the NPN extension has been
12501 replaced with the ALPN extension (see the "alpn" keyword), though this one is
12502 only available starting with OpenSSL 1.0.2.
12503
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012504observe <mode>
12505 This option enables health adjusting based on observing communication with
12506 the server. By default this functionality is disabled and enabling it also
12507 requires to enable health checks. There are two supported modes: "layer4" and
12508 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
12509 significant. In layer7, which is only allowed for http proxies, responses
12510 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010012511 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012512
12513 See also the "check", "on-error" and "error-limit".
12514
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012515on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010012516 Select what should happen when enough consecutive errors are detected.
12517 Currently, four modes are available:
12518 - fastinter: force fastinter
12519 - fail-check: simulate a failed check, also forces fastinter (default)
12520 - sudden-death: simulate a pre-fatal failed health check, one more failed
12521 check will mark a server down, forces fastinter
12522 - mark-down: mark the server immediately down and force fastinter
12523
12524 See also the "check", "observe" and "error-limit".
12525
Simon Hormane0d1bfb2011-06-21 14:34:58 +090012526on-marked-down <action>
12527 Modify what occurs when a server is marked down.
12528 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012529 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
12530 all connections to the server are immediately terminated when the server
12531 goes down. It might be used if the health check detects more complex cases
12532 than a simple connection status, and long timeouts would cause the service
12533 to remain unresponsive for too long a time. For instance, a health check
12534 might detect that a database is stuck and that there's no chance to reuse
12535 existing connections anymore. Connections killed this way are logged with
12536 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090012537
12538 Actions are disabled by default
12539
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012540on-marked-up <action>
12541 Modify what occurs when a server is marked up.
12542 Currently one action is available:
12543 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
12544 done only if the server is not in backup state and if it is not disabled
12545 (it must have an effective weight > 0). This can be used sometimes to force
12546 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010012547 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
12548 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070012549 with extreme care. Sessions killed because a server comes up are logged
12550 with an 'U' termination code (for "Up").
12551
12552 Actions are disabled by default
12553
Olivier Houchard006e3102018-12-10 18:30:32 +010012554pool-max-conn <max>
12555 Set the maximum number of idling connections for a server. -1 means unlimited
12556 connections, 0 means no idle connections. The default is -1. When idle
12557 connections are enabled, orphaned idle connections which do not belong to any
12558 client session anymore are moved to a dedicated pool so that they remain
12559 usable by future clients. This only applies to connections that can be shared
12560 according to the same principles as those applying to "http-reuse".
12561
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010012562pool-purge-delay <delay>
12563 Sets the delay to start purging idle connections. Each <delay> interval, half
Olivier Houcharda56eebf2019-03-19 16:44:02 +010012564 of the idle connections are closed. 0 means we don't keep any idle connection.
Willy Tarreaufb553652019-06-04 14:06:31 +020012565 The default is 5s.
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010012566
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012567port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012568 Using the "port" parameter, it becomes possible to use a different port to
12569 send health-checks. On some servers, it may be desirable to dedicate a port
12570 to a specific component able to perform complex tests which are more suitable
12571 to health-checks than the application. It is common to run a simple script in
12572 inetd for instance. This parameter is ignored if the "check" parameter is not
12573 set. See also the "addr" parameter.
12574
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020012575proto <name>
12576
12577 Forces the multiplexer's protocol to use for the outgoing connections to this
12578 server. It must be compatible with the mode of the backend (TCP or HTTP). It
12579 must also be usable on the backend side. The list of available protocols is
12580 reported in haproxy -vv.
12581 Idea behind this optipon is to bypass the selection of the best multiplexer's
12582 protocol for all connections established to this server.
12583
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012584redir <prefix>
12585 The "redir" parameter enables the redirection mode for all GET and HEAD
12586 requests addressing this server. This means that instead of having HAProxy
12587 forward the request to the server, it will send an "HTTP 302" response with
12588 the "Location" header composed of this prefix immediately followed by the
12589 requested URI beginning at the leading '/' of the path component. That means
12590 that no trailing slash should be used after <prefix>. All invalid requests
12591 will be rejected, and all non-GET or HEAD requests will be normally served by
12592 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012593 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012594 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012595 users to a remote location in case of local disaster. Main use consists in
12596 increasing bandwidth for static servers by having the clients directly
12597 connect to them. Note: never use a relative location here, it would cause a
12598 loop between the client and HAProxy!
12599
12600 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
12601
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012602rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012603 The "rise" parameter states that a server will be considered as operational
12604 after <count> consecutive successful health checks. This value defaults to 2
12605 if unspecified. See also the "check", "inter" and "fall" parameters.
12606
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020012607resolve-opts <option>,<option>,...
12608 Comma separated list of options to apply to DNS resolution linked to this
12609 server.
12610
12611 Available options:
12612
12613 * allow-dup-ip
12614 By default, HAProxy prevents IP address duplication in a backend when DNS
12615 resolution at runtime is in operation.
12616 That said, for some cases, it makes sense that two servers (in the same
12617 backend, being resolved by the same FQDN) have the same IP address.
12618 For such case, simply enable this option.
12619 This is the opposite of prevent-dup-ip.
12620
12621 * prevent-dup-ip
12622 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
12623 an IP address already set to a server in the same backend and sharing the
12624 same fqdn.
12625 This is the opposite of allow-dup-ip.
12626
12627 Example:
12628 backend b_myapp
12629 default-server init-addr none resolvers dns
12630 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
12631 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
12632
12633 With the option allow-dup-ip set:
12634 * if the nameserver returns a single IP address, then both servers will use
12635 it
12636 * If the nameserver returns 2 IP addresses, then each server will pick up a
12637 different address
12638
12639 Default value: not set
12640
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012641resolve-prefer <family>
12642 When DNS resolution is enabled for a server and multiple IP addresses from
12643 different families are returned, HAProxy will prefer using an IP address
12644 from the family mentioned in the "resolve-prefer" parameter.
12645 Available families: "ipv4" and "ipv6"
12646
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020012647 Default value: ipv6
12648
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012649 Example:
12650
12651 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012652
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012653resolve-net <network>[,<network[,...]]
John Roesler27c754c2019-07-10 15:45:51 -050012654 This option prioritizes the choice of an ip address matching a network. This is
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012655 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010012656 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010012657 different datacenters. The latency between datacenter is not negligible, so
12658 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012659 configured network, another address is selected.
12660
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012661 Example:
12662
12663 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012664
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012665resolvers <id>
12666 Points to an existing "resolvers" section to resolve current server's
12667 hostname.
12668
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012669 Example:
12670
12671 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012672
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012673 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012674
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012675send-proxy
12676 The "send-proxy" parameter enforces use of the PROXY protocol over any
12677 connection established to this server. The PROXY protocol informs the other
12678 end about the layer 3/4 addresses of the incoming connection, so that it can
12679 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012680 upper layer protocol. For connections accepted by an "accept-proxy" or
12681 "accept-netscaler-cip" listener, the advertised address will be used. Only
12682 TCPv4 and TCPv6 address families are supported. Other families such as
12683 Unix sockets, will report an UNKNOWN family. Servers using this option can
12684 fully be chained to another instance of haproxy listening with an
12685 "accept-proxy" setting. This setting must not be used if the server isn't
12686 aware of the protocol. When health checks are sent to the server, the PROXY
12687 protocol is automatically used when this option is set, unless there is an
12688 explicit "port" or "addr" directive, in which case an explicit
12689 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012690 See also the "no-send-proxy" option of this section and "accept-proxy" and
12691 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012692
David Safb76832014-05-08 23:42:08 -040012693send-proxy-v2
12694 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
12695 over any connection established to this server. The PROXY protocol informs
12696 the other end about the layer 3/4 addresses of the incoming connection, so
12697 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020012698 whatever the upper layer protocol. It also send ALPN information if an alpn
12699 have been negotiated. This setting must not be used if the server isn't aware
12700 of this version of the protocol. See also the "no-send-proxy-v2" option of
12701 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012702
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012703proxy-v2-options <option>[,<option>]*
12704 The "proxy-v2-options" parameter add option to send in PROXY protocol version
12705 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010012706 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
12707 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010012708 certificate, "cert-key": key algorithm of the used certificate), "authority":
12709 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010012710 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012711
David Safb76832014-05-08 23:42:08 -040012712send-proxy-v2-ssl
12713 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12714 2 over any connection established to this server. The PROXY protocol informs
12715 the other end about the layer 3/4 addresses of the incoming connection, so
12716 that it can know the client's address or the public address it accessed to,
12717 whatever the upper layer protocol. In addition, the SSL information extension
12718 of the PROXY protocol is added to the PROXY protocol header. This setting
12719 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012720 See also the "no-send-proxy-v2-ssl" option of this section and the
12721 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012722
12723send-proxy-v2-ssl-cn
12724 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12725 2 over any connection established to this server. The PROXY protocol informs
12726 the other end about the layer 3/4 addresses of the incoming connection, so
12727 that it can know the client's address or the public address it accessed to,
12728 whatever the upper layer protocol. In addition, the SSL information extension
12729 of the PROXY protocol, along along with the Common Name from the subject of
12730 the client certificate (if any), is added to the PROXY protocol header. This
12731 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012732 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
12733 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012734
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012735slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012736 The "slowstart" parameter for a server accepts a value in milliseconds which
12737 indicates after how long a server which has just come back up will run at
12738 full speed. Just as with every other time-based parameter, it can be entered
12739 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
12740 linearly from 0 to 100% during this time. The limitation applies to two
12741 parameters :
12742
12743 - maxconn: the number of connections accepted by the server will grow from 1
12744 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
12745
12746 - weight: when the backend uses a dynamic weighted algorithm, the weight
12747 grows linearly from 1 to 100%. In this case, the weight is updated at every
12748 health-check. For this reason, it is important that the "inter" parameter
12749 is smaller than the "slowstart", in order to maximize the number of steps.
12750
12751 The slowstart never applies when haproxy starts, otherwise it would cause
12752 trouble to running servers. It only applies when a server has been previously
12753 seen as failed.
12754
Willy Tarreau732eac42015-07-09 11:40:25 +020012755sni <expression>
12756 The "sni" parameter evaluates the sample fetch expression, converts it to a
12757 string and uses the result as the host name sent in the SNI TLS extension to
12758 the server. A typical use case is to send the SNI received from the client in
12759 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012760 expression, though alternatives such as req.hdr(host) can also make sense. If
12761 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012762 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010012763 "verify" directive for more details. If you want to set a SNI for health
12764 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012765
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012766source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012767source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012768source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012769 The "source" parameter sets the source address which will be used when
12770 connecting to the server. It follows the exact same parameters and principle
12771 as the backend "source" keyword, except that it only applies to the server
12772 referencing it. Please consult the "source" keyword for details.
12773
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012774 Additionally, the "source" statement on a server line allows one to specify a
12775 source port range by indicating the lower and higher bounds delimited by a
12776 dash ('-'). Some operating systems might require a valid IP address when a
12777 source port range is specified. It is permitted to have the same IP/range for
12778 several servers. Doing so makes it possible to bypass the maximum of 64k
12779 total concurrent connections. The limit will then reach 64k connections per
12780 server.
12781
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012782 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12783 specifying the source address without port(s).
12784
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012785ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012786 This option enables SSL ciphering on outgoing connections to the server. It
12787 is critical to verify server certificates using "verify" when using SSL to
12788 connect to servers, otherwise the communication is prone to trivial man in
12789 the-middle attacks rendering SSL useless. When this option is used, health
12790 checks are automatically sent in SSL too unless there is a "port" or an
12791 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012792 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12793 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012794
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012795ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12796 This option enforces use of <version> or lower when SSL is used to communicate
12797 with the server. This option is also available on global statement
12798 "ssl-default-server-options". See also "ssl-min-ver".
12799
12800ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12801 This option enforces use of <version> or upper when SSL is used to communicate
12802 with the server. This option is also available on global statement
12803 "ssl-default-server-options". See also "ssl-max-ver".
12804
Frédéric Lécailled2376272017-03-21 18:52:12 +010012805ssl-reuse
12806 This option may be used as "server" setting to reset any "no-ssl-reuse"
12807 setting which would have been inherited from "default-server" directive as
12808 default value.
12809 It may also be used as "default-server" setting to reset any previous
12810 "default-server" "no-ssl-reuse" setting.
12811
12812stick
12813 This option may be used as "server" setting to reset any "non-stick"
12814 setting which would have been inherited from "default-server" directive as
12815 default value.
12816 It may also be used as "default-server" setting to reset any previous
12817 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012818
Alexander Liu2a54bb72019-05-22 19:44:48 +080012819socks4 <addr>:<port>
John Roesler27c754c2019-07-10 15:45:51 -050012820 This option enables upstream socks4 tunnel for outgoing connections to the
Alexander Liu2a54bb72019-05-22 19:44:48 +080012821 server. Using this option won't force the health check to go via socks4 by
12822 default. You will have to use the keyword "check-via-socks4" to enable it.
12823
Willy Tarreau163d4622015-10-13 16:16:41 +020012824tcp-ut <delay>
12825 Sets the TCP User Timeout for all outgoing connections to this server. This
12826 option is available on Linux since version 2.6.37. It allows haproxy to
12827 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012828 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012829 long-lived connections experiencing long idle periods such as remote
12830 terminals or database connection pools, where the client and server timeouts
12831 must remain high to allow a long period of idle, but where it is important to
12832 detect that the server has disappeared in order to release all resources
12833 associated with its connection (and the client's session). One typical use
12834 case is also to force dead server connections to die when health checks are
12835 too slow or during a soft reload since health checks are then disabled. The
12836 argument is a delay expressed in milliseconds by default. This only works for
12837 regular TCP connections, and is ignored for other protocols.
12838
Willy Tarreau034c88c2017-01-23 23:36:45 +010012839tfo
12840 This option enables using TCP fast open when connecting to servers, on
12841 systems that support it (currently only the Linux kernel >= 4.11).
12842 See the "tfo" bind option for more information about TCP fast open.
12843 Please note that when using tfo, you should also use the "conn-failure",
12844 "empty-response" and "response-timeout" keywords for "retry-on", or haproxy
Frédéric Lécailleaeeb1c92019-07-04 14:19:06 +020012845 won't be able to retry the connection on failure. See also "no-tfo".
Willy Tarreau034c88c2017-01-23 23:36:45 +010012846
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012847track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012848 This option enables ability to set the current state of the server by tracking
12849 another one. It is possible to track a server which itself tracks another
12850 server, provided that at the end of the chain, a server has health checks
12851 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012852 used, it has to be enabled on both proxies.
12853
Frédéric Lécailled2376272017-03-21 18:52:12 +010012854tls-tickets
12855 This option may be used as "server" setting to reset any "no-tls-tickets"
12856 setting which would have been inherited from "default-server" directive as
12857 default value.
12858 It may also be used as "default-server" setting to reset any previous
12859 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012860
Emeric Brunef42d922012-10-11 16:11:36 +020012861verify [none|required]
12862 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012863 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012864 certificate provided by the server is verified using CAs from 'ca-file' and
12865 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012866 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012867 the name passed using the "sni" directive, or if not provided, the static
12868 host name passed using the "verifyhost" directive. When no name is found, the
12869 certificate's names are ignored. For this reason, without SNI it's important
12870 to use "verifyhost". On verification failure the handshake is aborted. It is
12871 critically important to verify server certificates when using SSL to connect
12872 to servers, otherwise the communication is prone to trivial man-in-the-middle
12873 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12874 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012875
Evan Broderbe554312013-06-27 00:05:25 -070012876verifyhost <hostname>
12877 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012878 only takes effect if 'verify required' is also specified. This directive sets
12879 a default static hostname to check the server's certificate against when no
12880 SNI was used to connect to the server. If SNI is not used, this is the only
12881 way to enable hostname verification. This static hostname, when set, will
12882 also be used for health checks (which cannot provide an SNI value). If none
12883 of the hostnames in the certificate match the specified hostname, the
12884 handshake is aborted. The hostnames in the server-provided certificate may
12885 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012886
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012887weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012888 The "weight" parameter is used to adjust the server's weight relative to
12889 other servers. All servers will receive a load proportional to their weight
12890 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012891 load. The default weight is 1, and the maximal value is 256. A value of 0
12892 means the server will not participate in load-balancing but will still accept
12893 persistent connections. If this parameter is used to distribute the load
12894 according to server's capacity, it is recommended to start with values which
12895 can both grow and shrink, for instance between 10 and 100 to leave enough
12896 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012897
12898
Cyril Bonté46175dd2015-07-02 22:45:32 +0200128995.3. Server IP address resolution using DNS
12900-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012901
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012902HAProxy allows using a host name on the server line to retrieve its IP address
12903using name servers. By default, HAProxy resolves the name when parsing the
12904configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012905This is not sufficient in some cases, such as in Amazon where a server's IP
12906can change after a reboot or an ELB Virtual IP can change based on current
12907workload.
12908This chapter describes how HAProxy can be configured to process server's name
12909resolution at run time.
12910Whether run time server name resolution has been enable or not, HAProxy will
12911carry on doing the first resolution when parsing the configuration.
12912
12913
Cyril Bonté46175dd2015-07-02 22:45:32 +0200129145.3.1. Global overview
12915----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012916
12917As we've seen in introduction, name resolution in HAProxy occurs at two
12918different steps of the process life:
12919
12920 1. when starting up, HAProxy parses the server line definition and matches a
12921 host name. It uses libc functions to get the host name resolved. This
12922 resolution relies on /etc/resolv.conf file.
12923
Christopher Faulet67957bd2017-09-27 11:00:59 +020012924 2. at run time, HAProxy performs periodically name resolutions for servers
12925 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012926
12927A few other events can trigger a name resolution at run time:
12928 - when a server's health check ends up in a connection timeout: this may be
12929 because the server has a new IP address. So we need to trigger a name
12930 resolution to know this new IP.
12931
Christopher Faulet67957bd2017-09-27 11:00:59 +020012932When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012933HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012934SRV label is specified, then the corresponding SRV records will be retrieved
12935from the DNS server, and the provided hostnames will be used. The SRV label
12936will be checked periodically, and if any server are added or removed, haproxy
12937will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012938
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012939A few things important to notice:
John Roesler27c754c2019-07-10 15:45:51 -050012940 - all the name servers are queried in the meantime. HAProxy will process the
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012941 first valid response.
12942
12943 - a resolution is considered as invalid (NX, timeout, refused), when all the
12944 servers return an error.
12945
12946
Cyril Bonté46175dd2015-07-02 22:45:32 +0200129475.3.2. The resolvers section
12948----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012949
12950This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012951HAProxy. There can be as many as resolvers section as needed. Each section can
12952contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012953
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012954When multiple name servers are configured in a resolvers section, then HAProxy
12955uses the first valid response. In case of invalid responses, only the last one
12956is treated. Purpose is to give the chance to a slow server to deliver a valid
12957answer after a fast faulty or outdated server.
12958
12959When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012960used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012961
Christopher Faulet67957bd2017-09-27 11:00:59 +020012962 1. HAProxy retries the same DNS query with a new query type. The A queries are
12963 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12964 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012965
Christopher Faulet67957bd2017-09-27 11:00:59 +020012966 2. When the fallback on the query type was done (or not applicable), HAProxy
12967 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012968
Christopher Faulet67957bd2017-09-27 11:00:59 +020012969 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12970 response is received after that, it stops the DNS resolution and reports
12971 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012972
Christopher Faulet67957bd2017-09-27 11:00:59 +020012973For example, with 2 name servers configured in a resolvers section, the
12974following scenarios are possible:
12975
12976 - First response is valid and is applied directly, second response is
12977 ignored
12978
12979 - First response is invalid and second one is valid, then second response is
12980 applied
12981
12982 - First response is a NX domain and second one a truncated response, then
12983 HAProxy retries the query with a new type
12984
12985 - First response is a NX domain and second one is a timeout, then HAProxy
12986 retries the query with a new type
12987
12988 - Query timed out for both name servers, then HAProxy retries it with the
12989 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012990
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012991As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12992a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012993<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012994
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012995
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012996resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012997 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012998
12999A resolvers section accept the following parameters:
13000
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020013001accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010013002 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020013003 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020013004 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
13005 by RFC 6891)
13006
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020013007 Note: the maximum allowed value is 8192.
13008
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013009nameserver <id> <ip>:<port>
13010 DNS server description:
13011 <id> : label of the server, should be unique
13012 <ip> : IP address of the server
13013 <port> : port where the DNS service actually runs
13014
Ben Draut44e609b2018-05-29 15:40:08 -060013015parse-resolv-conf
13016 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
13017 list. Ordered as if each nameserver in /etc/resolv.conf was individually
13018 placed in the resolvers section in place of this directive.
13019
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013020hold <status> <period>
13021 Defines <period> during which the last name resolution should be kept based
13022 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010013023 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020013024 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013025 <period> : interval between two successive name resolution when the last
13026 answer was in <status>. It follows the HAProxy time format.
13027 <period> is in milliseconds by default.
13028
Baptiste Assmann686408b2017-08-18 10:15:42 +020013029 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013030
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013031resolve_retries <nb>
13032 Defines the number <nb> of queries to send to resolve a server name before
13033 giving up.
13034 Default value: 3
13035
Baptiste Assmann62b75b42015-09-09 01:11:36 +020013036 A retry occurs on name server timeout or when the full sequence of DNS query
13037 type failover is over and we need to start up from the default ANY query
13038 type.
13039
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013040timeout <event> <time>
13041 Defines timeouts related to name resolution
13042 <event> : the event on which the <time> timeout period applies to.
13043 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013044 - resolve : default time to trigger name resolutions when no
13045 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020013046 Default value: 1s
13047 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013048 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020013049 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013050 <time> : time related to the event. It follows the HAProxy time format.
13051 <time> is expressed in milliseconds.
13052
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020013053 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013054
13055 resolvers mydns
13056 nameserver dns1 10.0.0.1:53
13057 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060013058 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013059 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020013060 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013061 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010013062 hold other 30s
13063 hold refused 30s
13064 hold nx 30s
13065 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013066 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020013067 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020013068
13069
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200130706. HTTP header manipulation
13071---------------------------
13072
13073In HTTP mode, it is possible to rewrite, add or delete some of the request and
13074response headers based on regular expressions. It is also possible to block a
13075request or a response if a particular header matches a regular expression,
13076which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010013077against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013078
Willy Tarreau70dffda2014-01-30 03:07:23 +010013079If HAProxy encounters an "Informational Response" (status code 1xx), it is able
13080to process all rsp* rules which can allow, deny, rewrite or delete a header,
13081but it will refuse to add a header to any such messages as this is not
13082HTTP-compliant. The reason for still processing headers in such responses is to
13083stop and/or fix any possible information leak which may happen, for instance
13084because another downstream equipment would unconditionally add a header, or if
13085a server name appears there. When such messages are seen, normal processing
13086still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020013087
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013088This section covers common usage of the following keywords, described in detail
13089in section 4.2 :
13090
13091 - reqadd <string>
13092 - reqallow <search>
13093 - reqiallow <search>
13094 - reqdel <search>
13095 - reqidel <search>
13096 - reqdeny <search>
13097 - reqideny <search>
13098 - reqpass <search>
13099 - reqipass <search>
13100 - reqrep <search> <replace>
13101 - reqirep <search> <replace>
13102 - reqtarpit <search>
13103 - reqitarpit <search>
13104 - rspadd <string>
13105 - rspdel <search>
13106 - rspidel <search>
13107 - rspdeny <search>
13108 - rspideny <search>
13109 - rsprep <search> <replace>
13110 - rspirep <search> <replace>
13111
13112With all these keywords, the same conventions are used. The <search> parameter
13113is a POSIX extended regular expression (regex) which supports grouping through
13114parenthesis (without the backslash). Spaces and other delimiters must be
13115prefixed with a backslash ('\') to avoid confusion with a field delimiter.
13116Other characters may be prefixed with a backslash to change their meaning :
13117
13118 \t for a tab
13119 \r for a carriage return (CR)
13120 \n for a new line (LF)
13121 \ to mark a space and differentiate it from a delimiter
13122 \# to mark a sharp and differentiate it from a comment
13123 \\ to use a backslash in a regex
13124 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
13125 \xXX to write the ASCII hex code XX as in the C language
13126
13127The <replace> parameter contains the string to be used to replace the largest
13128portion of text matching the regex. It can make use of the special characters
13129above, and can reference a substring which is delimited by parenthesis in the
13130regex, by writing a backslash ('\') immediately followed by one digit from 0 to
131319 indicating the group position (0 designating the entire line). This practice
13132is very common to users of the "sed" program.
13133
13134The <string> parameter represents the string which will systematically be added
13135after the last header line. It can also use special character sequences above.
13136
13137Notes related to these keywords :
13138---------------------------------
13139 - these keywords are not always convenient to allow/deny based on header
13140 contents. It is strongly recommended to use ACLs with the "block" keyword
13141 instead, resulting in far more flexible and manageable rules.
13142
13143 - lines are always considered as a whole. It is not possible to reference
13144 a header name only or a value only. This is important because of the way
13145 headers are written (notably the number of spaces after the colon).
13146
13147 - the first line is always considered as a header, which makes it possible to
13148 rewrite or filter HTTP requests URIs or response codes, but in turn makes
13149 it harder to distinguish between headers and request line. The regex prefix
13150 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
13151 ^[^ \t:]*: matches any header name followed by a colon.
13152
13153 - for performances reasons, the number of characters added to a request or to
13154 a response is limited at build time to values between 1 and 4 kB. This
13155 should normally be far more than enough for most usages. If it is too short
13156 on occasional usages, it is possible to gain some space by removing some
13157 useless headers before adding new ones.
13158
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010013159 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013160 without the 'i' letter except that they ignore case when matching patterns.
13161
13162 - when a request passes through a frontend then a backend, all req* rules
13163 from the frontend will be evaluated, then all req* rules from the backend
13164 will be evaluated. The reverse path is applied to responses.
13165
13166 - req* statements are applied after "block" statements, so that "block" is
13167 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010013168 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013169
13170
Willy Tarreau74ca5042013-06-11 23:12:07 +0200131717. Using ACLs and fetching samples
13172----------------------------------
13173
Davor Ocelice9ed2812017-12-25 17:49:28 +010013174HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020013175client or server information, from tables, environmental information etc...
13176The action of extracting such data is called fetching a sample. Once retrieved,
13177these samples may be used for various purposes such as a key to a stick-table,
13178but most common usages consist in matching them against predefined constant
13179data called patterns.
13180
13181
131827.1. ACL basics
13183---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013184
13185The use of Access Control Lists (ACL) provides a flexible solution to perform
13186content switching and generally to take decisions based on content extracted
13187from the request, the response or any environmental status. The principle is
13188simple :
13189
Willy Tarreau74ca5042013-06-11 23:12:07 +020013190 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013191 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020013192 - apply one or multiple pattern matching methods on this sample
13193 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013194
Willy Tarreau74ca5042013-06-11 23:12:07 +020013195The actions generally consist in blocking a request, selecting a backend, or
13196adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013197
13198In order to define a test, the "acl" keyword is used. The syntax is :
13199
Willy Tarreau74ca5042013-06-11 23:12:07 +020013200 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013201
13202This creates a new ACL <aclname> or completes an existing one with new tests.
13203Those tests apply to the portion of request/response specified in <criterion>
13204and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013205an operator which may be specified before the set of values. Optionally some
13206conversion operators may be applied to the sample, and they will be specified
13207as a comma-delimited list of keywords just after the first keyword. The values
13208are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013209
13210ACL names must be formed from upper and lower case letters, digits, '-' (dash),
13211'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
13212which means that "my_acl" and "My_Acl" are two different ACLs.
13213
13214There is no enforced limit to the number of ACLs. The unused ones do not affect
13215performance, they just consume a small amount of memory.
13216
Willy Tarreau74ca5042013-06-11 23:12:07 +020013217The criterion generally is the name of a sample fetch method, or one of its ACL
13218specific declinations. The default test method is implied by the output type of
13219this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013220methods of a same sample fetch method. The sample fetch methods are the only
13221ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013222
13223Sample fetch methods return data which can be of the following types :
13224 - boolean
13225 - integer (signed or unsigned)
13226 - IPv4 or IPv6 address
13227 - string
13228 - data block
13229
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013230Converters transform any of these data into any of these. For example, some
13231converters might convert a string to a lower-case string while other ones
13232would turn a string to an IPv4 address, or apply a netmask to an IP address.
13233The resulting sample is of the type of the last converter applied to the list,
13234which defaults to the type of the sample fetch method.
13235
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013236Each sample or converter returns data of a specific type, specified with its
13237keyword in this documentation. When an ACL is declared using a standard sample
13238fetch method, certain types automatically involved a default matching method
13239which are summarized in the table below :
13240
13241 +---------------------+-----------------+
13242 | Sample or converter | Default |
13243 | output type | matching method |
13244 +---------------------+-----------------+
13245 | boolean | bool |
13246 +---------------------+-----------------+
13247 | integer | int |
13248 +---------------------+-----------------+
13249 | ip | ip |
13250 +---------------------+-----------------+
13251 | string | str |
13252 +---------------------+-----------------+
13253 | binary | none, use "-m" |
13254 +---------------------+-----------------+
13255
13256Note that in order to match a binary samples, it is mandatory to specify a
13257matching method, see below.
13258
Willy Tarreau74ca5042013-06-11 23:12:07 +020013259The ACL engine can match these types against patterns of the following types :
13260 - boolean
13261 - integer or integer range
13262 - IP address / network
13263 - string (exact, substring, suffix, prefix, subdir, domain)
13264 - regular expression
13265 - hex block
13266
Willy Tarreauc57f0e22009-05-10 13:12:33 +020013267The following ACL flags are currently supported :
13268
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013269 -i : ignore case during matching of all subsequent patterns.
13270 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013271 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010013272 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010013273 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010013274 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020013275 -- : force end of flags. Useful when a string looks like one of the flags.
13276
Willy Tarreau74ca5042013-06-11 23:12:07 +020013277The "-f" flag is followed by the name of a file from which all lines will be
13278read as individual values. It is even possible to pass multiple "-f" arguments
13279if the patterns are to be loaded from multiple files. Empty lines as well as
13280lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
13281will be stripped. If it is absolutely necessary to insert a valid pattern
13282beginning with a sharp, just prefix it with a space so that it is not taken for
13283a comment. Depending on the data type and match method, haproxy may load the
13284lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
13285exact string matching. In this case, duplicates will automatically be removed.
13286
Thierry FOURNIER9860c412014-01-29 14:23:29 +010013287The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
13288parsed as two column file. The first column contains the patterns used by the
13289ACL, and the second column contain the samples. The sample can be used later by
13290a map. This can be useful in some rare cases where an ACL would just be used to
13291check for the existence of a pattern in a map before a mapping is applied.
13292
Thierry FOURNIER3534d882014-01-20 17:01:44 +010013293The "-u" flag forces the unique id of the ACL. This unique id is used with the
13294socket interface to identify ACL and dynamically change its values. Note that a
13295file is always identified by its name even if an id is set.
13296
Willy Tarreau74ca5042013-06-11 23:12:07 +020013297Also, note that the "-i" flag applies to subsequent entries and not to entries
13298loaded from files preceding it. For instance :
13299
13300 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
13301
13302In this example, each line of "exact-ua.lst" will be exactly matched against
13303the "user-agent" header of the request. Then each line of "generic-ua" will be
13304case-insensitively matched. Then the word "test" will be insensitively matched
13305as well.
13306
13307The "-m" flag is used to select a specific pattern matching method on the input
13308sample. All ACL-specific criteria imply a pattern matching method and generally
13309do not need this flag. However, this flag is useful with generic sample fetch
13310methods to describe how they're going to be matched against the patterns. This
13311is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010013312obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020013313followed by a pattern matching method name, this method is used instead of the
13314default one for the criterion. This makes it possible to match contents in ways
13315that were not initially planned, or with sample fetch methods which return a
13316string. The matching method also affects the way the patterns are parsed.
13317
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010013318The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
13319By default, if the parser cannot parse ip address it considers that the parsed
13320string is maybe a domain name and try dns resolution. The flag "-n" disable this
13321resolution. It is useful for detecting malformed ip lists. Note that if the DNS
13322server is not reachable, the haproxy configuration parsing may last many minutes
John Roesler27c754c2019-07-10 15:45:51 -050013323waiting for the timeout. During this time no error messages are displayed. The
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010013324flag "-n" disable this behavior. Note also that during the runtime, this
13325function is disabled for the dynamic acl modifications.
13326
Willy Tarreau74ca5042013-06-11 23:12:07 +020013327There are some restrictions however. Not all methods can be used with all
13328sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
13329be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020013330
13331 - "found" : only check if the requested sample could be found in the stream,
13332 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020013333 to pass any pattern to avoid confusion. This matching method is
13334 particularly useful to detect presence of certain contents such
13335 as headers, cookies, etc... even if they are empty and without
13336 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013337
13338 - "bool" : check the value as a boolean. It can only be applied to fetches
13339 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013340 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013341
13342 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020013343 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013344
13345 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020013346 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013347
Davor Ocelice9ed2812017-12-25 17:49:28 +010013348 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020013349 binary sequence. This may be used with binary or string samples.
13350
13351 - "len" : match the sample's length as an integer. This may be used with
13352 binary or string samples.
13353
Willy Tarreau74ca5042013-06-11 23:12:07 +020013354 - "str" : exact match : match the contents against a string. This may be
13355 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013356
Willy Tarreau74ca5042013-06-11 23:12:07 +020013357 - "sub" : substring match : check that the contents contain at least one of
13358 the provided string patterns. This may be used with binary or
13359 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013360
Willy Tarreau74ca5042013-06-11 23:12:07 +020013361 - "reg" : regex match : match the contents against a list of regular
13362 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013363
Willy Tarreau74ca5042013-06-11 23:12:07 +020013364 - "beg" : prefix match : check that the contents begin like the provided
13365 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013366
Willy Tarreau74ca5042013-06-11 23:12:07 +020013367 - "end" : suffix match : check that the contents end like the provided
13368 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013369
Willy Tarreau74ca5042013-06-11 23:12:07 +020013370 - "dir" : subdir match : check that a slash-delimited portion of the
13371 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013372 This may be used with binary or string samples.
13373
Willy Tarreau74ca5042013-06-11 23:12:07 +020013374 - "dom" : domain match : check that a dot-delimited portion of the contents
13375 exactly match one of the provided string patterns. This may be
13376 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020013377
13378For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
13379request, it is possible to do :
13380
13381 acl jsess_present cook(JSESSIONID) -m found
13382
13383In order to apply a regular expression on the 500 first bytes of data in the
13384buffer, one would use the following acl :
13385
13386 acl script_tag payload(0,500) -m reg -i <script>
13387
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013388On systems where the regex library is much slower when using "-i", it is
13389possible to convert the sample to lowercase before matching, like this :
13390
13391 acl script_tag payload(0,500),lower -m reg <script>
13392
Willy Tarreau74ca5042013-06-11 23:12:07 +020013393All ACL-specific criteria imply a default matching method. Most often, these
13394criteria are composed by concatenating the name of the original sample fetch
13395method and the matching method. For example, "hdr_beg" applies the "beg" match
13396to samples retrieved using the "hdr" fetch method. Since all ACL-specific
13397criteria rely on a sample fetch method, it is always possible instead to use
13398the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013399
Willy Tarreau74ca5042013-06-11 23:12:07 +020013400If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030013401the matching method is simply applied to the underlying sample fetch method.
13402For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013403
Willy Tarreau74ca5042013-06-11 23:12:07 +020013404 acl short_form hdr_beg(host) www.
13405 acl alternate1 hdr_beg(host) -m beg www.
13406 acl alternate2 hdr_dom(host) -m beg www.
13407 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013408
Willy Tarreau2b5285d2010-05-09 23:45:24 +020013409
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013410The table below summarizes the compatibility matrix between sample or converter
13411types and the pattern types to fetch against. It indicates for each compatible
13412combination the name of the matching method to be used, surrounded with angle
13413brackets ">" and "<" when the method is the default one and will work by
13414default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010013415
Willy Tarreau74ca5042013-06-11 23:12:07 +020013416 +-------------------------------------------------+
13417 | Input sample type |
13418 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013419 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013420 +----------------------+---------+---------+---------+---------+---------+
13421 | none (presence only) | found | found | found | found | found |
13422 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013423 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013424 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013425 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013426 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013427 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013428 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013429 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013430 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020013431 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013432 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013433 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013434 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013435 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013436 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013437 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013438 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013439 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013440 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013441 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013442 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010013443 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020013444 +----------------------+---------+---------+---------+---------+---------+
13445 | hex block | | | | bin | bin |
13446 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020013447
13448
Willy Tarreau74ca5042013-06-11 23:12:07 +0200134497.1.1. Matching booleans
13450------------------------
13451
13452In order to match a boolean, no value is needed and all values are ignored.
13453Boolean matching is used by default for all fetch methods of type "boolean".
13454When boolean matching is used, the fetched value is returned as-is, which means
13455that a boolean "true" will always match and a boolean "false" will never match.
13456
13457Boolean matching may also be enforced using "-m bool" on fetch methods which
13458return an integer value. Then, integer value 0 is converted to the boolean
13459"false" and all other values are converted to "true".
13460
Willy Tarreau6a06a402007-07-15 20:15:28 +020013461
Willy Tarreau74ca5042013-06-11 23:12:07 +0200134627.1.2. Matching integers
13463------------------------
13464
13465Integer matching applies by default to integer fetch methods. It can also be
13466enforced on boolean fetches using "-m int". In this case, "false" is converted
13467to the integer 0, and "true" is converted to the integer 1.
13468
13469Integer matching also supports integer ranges and operators. Note that integer
13470matching only applies to positive values. A range is a value expressed with a
13471lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013472
13473For instance, "1024:65535" is a valid range to represent a range of
13474unprivileged ports, and "1024:" would also work. "0:1023" is a valid
13475representation of privileged ports, and ":1023" would also work.
13476
Willy Tarreau62644772008-07-16 18:36:06 +020013477As a special case, some ACL functions support decimal numbers which are in fact
13478two integers separated by a dot. This is used with some version checks for
13479instance. All integer properties apply to those decimal numbers, including
13480ranges and operators.
13481
Willy Tarreau6a06a402007-07-15 20:15:28 +020013482For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010013483operators with ranges does not make much sense and is strongly discouraged.
13484Similarly, it does not make much sense to perform order comparisons with a set
13485of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013486
Willy Tarreau0ba27502007-12-24 16:55:16 +010013487Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020013488
13489 eq : true if the tested value equals at least one value
13490 ge : true if the tested value is greater than or equal to at least one value
13491 gt : true if the tested value is greater than at least one value
13492 le : true if the tested value is less than or equal to at least one value
13493 lt : true if the tested value is less than at least one value
13494
Willy Tarreau0ba27502007-12-24 16:55:16 +010013495For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020013496
13497 acl negative-length hdr_val(content-length) lt 0
13498
Willy Tarreau62644772008-07-16 18:36:06 +020013499This one matches SSL versions between 3.0 and 3.1 (inclusive) :
13500
13501 acl sslv3 req_ssl_ver 3:3.1
13502
Willy Tarreau6a06a402007-07-15 20:15:28 +020013503
Willy Tarreau74ca5042013-06-11 23:12:07 +0200135047.1.3. Matching strings
13505-----------------------
13506
13507String matching applies to string or binary fetch methods, and exists in 6
13508different forms :
13509
13510 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013511 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020013512
13513 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010013514 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020013515
13516 - prefix match (-m beg) : the patterns are compared with the beginning of
13517 the extracted string, and the ACL matches if any of them matches.
13518
13519 - suffix match (-m end) : the patterns are compared with the end of the
13520 extracted string, and the ACL matches if any of them matches.
13521
Baptiste Assmann33db6002016-03-06 23:32:10 +010013522 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020013523 string, delimited with slashes ("/"), and the ACL matches if any of them
13524 matches.
13525
13526 - domain match (-m dom) : the patterns are looked up inside the extracted
13527 string, delimited with dots ("."), and the ACL matches if any of them
13528 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013529
13530String matching applies to verbatim strings as they are passed, with the
13531exception of the backslash ("\") which makes it possible to escape some
13532characters such as the space. If the "-i" flag is passed before the first
13533string, then the matching will be performed ignoring the case. In order
13534to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010013535before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020013536
Mathias Weiersmuellerb2fe2232019-12-02 09:43:40 +010013537Do not use string matches for binary fetches which might contain null bytes
13538(0x00), as the comparison stops at the occurrence of the first null byte.
13539Instead, convert the binary fetch to a hex string with the hex converter first.
13540
13541Example:
13542 # matches if the string <tag> is present in the binary sample
13543 acl tag_found req.payload(0,0),hex -m sub 3C7461673E
13544
Willy Tarreau6a06a402007-07-15 20:15:28 +020013545
Willy Tarreau74ca5042013-06-11 23:12:07 +0200135467.1.4. Matching regular expressions (regexes)
13547---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013548
13549Just like with string matching, regex matching applies to verbatim strings as
13550they are passed, with the exception of the backslash ("\") which makes it
13551possible to escape some characters such as the space. If the "-i" flag is
13552passed before the first regex, then the matching will be performed ignoring
13553the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010013554the "--" flag before the first string. Same principle applies of course to
13555match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020013556
13557
Willy Tarreau74ca5042013-06-11 23:12:07 +0200135587.1.5. Matching arbitrary data blocks
13559-------------------------------------
13560
13561It is possible to match some extracted samples against a binary block which may
13562not safely be represented as a string. For this, the patterns must be passed as
13563a series of hexadecimal digits in an even number, when the match method is set
13564to binary. Each sequence of two digits will represent a byte. The hexadecimal
13565digits may be used upper or lower case.
13566
13567Example :
13568 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
13569 acl hello payload(0,6) -m bin 48656c6c6f0a
13570
13571
135727.1.6. Matching IPv4 and IPv6 addresses
13573---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020013574
13575IPv4 addresses values can be specified either as plain addresses or with a
13576netmask appended, in which case the IPv4 address matches whenever it is
13577within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010013578host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010013579difficult to read and debug configurations. If hostnames are used, you should
13580at least ensure that they are present in /etc/hosts so that the configuration
13581does not depend on any random DNS match at the moment the configuration is
13582parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013583
Daniel Schnellereba56342016-04-13 00:26:52 +020013584The dotted IPv4 address notation is supported in both regular as well as the
13585abbreviated form with all-0-octets omitted:
13586
13587 +------------------+------------------+------------------+
13588 | Example 1 | Example 2 | Example 3 |
13589 +------------------+------------------+------------------+
13590 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
13591 | 192.168.1 | 10.12 | 127.1 |
13592 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
13593 | 192.168.1/22 | 10.12/8 | 127.1/8 |
13594 +------------------+------------------+------------------+
13595
13596Notice that this is different from RFC 4632 CIDR address notation in which
13597192.168.42/24 would be equivalent to 192.168.42.0/24.
13598
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013599IPv6 may be entered in their usual form, with or without a netmask appended.
13600Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
13601trouble with randomly resolved IP addresses, host names are never allowed in
13602IPv6 patterns.
13603
13604HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
13605following situations :
13606 - tested address is IPv4, pattern address is IPv4, the match applies
13607 in IPv4 using the supplied mask if any.
13608 - tested address is IPv6, pattern address is IPv6, the match applies
13609 in IPv6 using the supplied mask if any.
13610 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
13611 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
13612 ::IPV4 or ::ffff:IPV4, otherwise it fails.
13613 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
13614 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
13615 applied in IPv6 using the supplied IPv6 mask.
13616
Willy Tarreau74ca5042013-06-11 23:12:07 +020013617
136187.2. Using ACLs to form conditions
13619----------------------------------
13620
13621Some actions are only performed upon a valid condition. A condition is a
13622combination of ACLs with operators. 3 operators are supported :
13623
13624 - AND (implicit)
13625 - OR (explicit with the "or" keyword or the "||" operator)
13626 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020013627
Willy Tarreau74ca5042013-06-11 23:12:07 +020013628A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013629
Willy Tarreau74ca5042013-06-11 23:12:07 +020013630 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020013631
Willy Tarreau74ca5042013-06-11 23:12:07 +020013632Such conditions are generally used after an "if" or "unless" statement,
13633indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020013634
Willy Tarreau74ca5042013-06-11 23:12:07 +020013635For instance, to block HTTP requests to the "*" URL with methods other than
13636"OPTIONS", as well as POST requests without content-length, and GET or HEAD
13637requests with a content-length greater than 0, and finally every request which
13638is not either GET/HEAD/POST/OPTIONS !
13639
13640 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013641 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
13642 http-request deny if METH_GET HTTP_CONTENT
13643 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020013644
13645To select a different backend for requests to static contents on the "www" site
13646and to every request on the "img", "video", "download" and "ftp" hosts :
13647
13648 acl url_static path_beg /static /images /img /css
13649 acl url_static path_end .gif .png .jpg .css .js
13650 acl host_www hdr_beg(host) -i www
13651 acl host_static hdr_beg(host) -i img. video. download. ftp.
13652
Davor Ocelice9ed2812017-12-25 17:49:28 +010013653 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020013654 # of host "www". Use backend "www" for the rest.
13655 use_backend static if host_static or host_www url_static
13656 use_backend www if host_www
13657
13658It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
13659expressions that are built on the fly without needing to be declared. They must
13660be enclosed between braces, with a space before and after each brace (because
13661the braces must be seen as independent words). Example :
13662
13663 The following rule :
13664
13665 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013666 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020013667
13668 Can also be written that way :
13669
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013670 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020013671
13672It is generally not recommended to use this construct because it's a lot easier
13673to leave errors in the configuration when written that way. However, for very
13674simple rules matching only one source IP address for instance, it can make more
13675sense to use them than to declare ACLs with random names. Another example of
13676good use is the following :
13677
13678 With named ACLs :
13679
13680 acl site_dead nbsrv(dynamic) lt 2
13681 acl site_dead nbsrv(static) lt 2
13682 monitor fail if site_dead
13683
13684 With anonymous ACLs :
13685
13686 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
13687
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013688See section 4.2 for detailed help on the "http-request deny" and "use_backend"
13689keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013690
13691
136927.3. Fetching samples
13693---------------------
13694
13695Historically, sample fetch methods were only used to retrieve data to match
13696against patterns using ACLs. With the arrival of stick-tables, a new class of
13697sample fetch methods was created, most often sharing the same syntax as their
13698ACL counterpart. These sample fetch methods are also known as "fetches". As
13699of now, ACLs and fetches have converged. All ACL fetch methods have been made
13700available as fetch methods, and ACLs may use any sample fetch method as well.
13701
13702This section details all available sample fetch methods and their output type.
13703Some sample fetch methods have deprecated aliases that are used to maintain
13704compatibility with existing configurations. They are then explicitly marked as
13705deprecated and should not be used in new setups.
13706
13707The ACL derivatives are also indicated when available, with their respective
13708matching methods. These ones all have a well defined default pattern matching
13709method, so it is never necessary (though allowed) to pass the "-m" option to
13710indicate how the sample will be matched using ACLs.
13711
13712As indicated in the sample type versus matching compatibility matrix above,
13713when using a generic sample fetch method in an ACL, the "-m" option is
13714mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13715the same keyword exists as an ACL keyword and as a standard fetch method, the
13716ACL engine will automatically pick the ACL-only one by default.
13717
13718Some of these keywords support one or multiple mandatory arguments, and one or
13719multiple optional arguments. These arguments are strongly typed and are checked
13720when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013721incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13722are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013723is optional, it will be indicated below between square brackets ('[ ]'). When
13724all arguments are optional, the parenthesis may be omitted.
13725
13726Thus, the syntax of a standard sample fetch method is one of the following :
13727 - name
13728 - name(arg1)
13729 - name(arg1,arg2)
13730
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013731
137327.3.1. Converters
13733-----------------
13734
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013735Sample fetch methods may be combined with transformations to be applied on top
13736of the fetched sample (also called "converters"). These combinations form what
13737is called "sample expressions" and the result is a "sample". Initially this
13738was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013739has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013740unique-id-format, add-header, ...).
13741
13742These transformations are enumerated as a series of specific keywords after the
13743sample fetch method. These keywords may equally be appended immediately after
13744the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013745support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013746
Willy Tarreau97707872015-01-27 15:12:13 +010013747A certain category of converters are bitwise and arithmetic operators which
13748support performing basic operations on integers. Some bitwise operations are
13749supported (and, or, xor, cpl) and some arithmetic operations are supported
13750(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13751bool) which make it possible to report a match without having to write an ACL.
13752
Willy Tarreau74ca5042013-06-11 23:12:07 +020013753The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013754
Ben Shillitof25e8e52016-12-02 14:25:37 +00001375551d.single(<prop>[,<prop>*])
13756 Returns values for the properties requested as a string, where values are
13757 separated by the delimiter specified with "51degrees-property-separator".
13758 The device is identified using the User-Agent header passed to the
13759 converter. The function can be passed up to five property names, and if a
13760 property name can't be found, the value "NoData" is returned.
13761
13762 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013763 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13764 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013765 # User-Agent passed to the converter.
13766 frontend http-in
13767 bind *:8081
13768 default_backend servers
13769 http-request set-header X-51D-DeviceTypeMobileTablet \
13770 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13771
Willy Tarreau97707872015-01-27 15:12:13 +010013772add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013773 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013774 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013775 name. The name of the variable starts with an indication about its scope. The
13776 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013777 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013778 "sess" : the variable is shared with the whole session
13779 "txn" : the variable is shared with the transaction (request and response)
13780 "req" : the variable is shared only during request processing
13781 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013782 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013783 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013784
Nenad Merdanovicc31499d2019-03-23 11:00:32 +010013785aes_gcm_dec(<bits>,<nonce>,<key>,<aead_tag>)
13786 Decrypts the raw byte input using the AES128-GCM, AES192-GCM or
13787 AES256-GCM algorithm, depending on the <bits> parameter. All other parameters
13788 need to be base64 encoded and the returned result is in raw byte format.
13789 If the <aead_tag> validation fails, the converter doesn't return any data.
13790 The <nonce>, <key> and <aead_tag> can either be strings or variables. This
13791 converter requires at least OpenSSL 1.0.1.
13792
13793 Example:
13794 http-response set-header X-Decrypted-Text %[var(txn.enc),\
13795 aes_gcm_dec(128,txn.nonce,Zm9vb2Zvb29mb29wZm9vbw==,txn.aead_tag)]
13796
Willy Tarreau97707872015-01-27 15:12:13 +010013797and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013798 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013799 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013800 numeric value or a variable name. The name of the variable starts with an
13801 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013802 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013803 "sess" : the variable is shared with the whole session
13804 "txn" : the variable is shared with the transaction (request and response)
13805 "req" : the variable is shared only during request processing
13806 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013807 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013808 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013809
Holger Just1bfc24b2017-05-06 00:56:53 +020013810b64dec
13811 Converts (decodes) a base64 encoded input string to its binary
13812 representation. It performs the inverse operation of base64().
13813
Emeric Brun53d1a982014-04-30 18:21:37 +020013814base64
13815 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013816 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013817 an SSL ID can be copied in a header).
13818
Willy Tarreau97707872015-01-27 15:12:13 +010013819bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013820 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013821 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013822 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013823 presence of a flag).
13824
Emeric Brun54c4ac82014-11-03 15:32:43 +010013825bytes(<offset>[,<length>])
13826 Extracts some bytes from an input binary sample. The result is a binary
13827 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013828 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013829
Willy Tarreau280f42b2018-02-19 15:34:12 +010013830concat([<start>],[<var>],[<end>])
13831 Concatenates up to 3 fields after the current sample which is then turned to
13832 a string. The first one, <start>, is a constant string, that will be appended
13833 immediately after the existing sample. It may be omitted if not used. The
13834 second one, <var>, is a variable name. The variable will be looked up, its
13835 contents converted to a string, and it will be appended immediately after the
13836 <first> part. If the variable is not found, nothing is appended. It may be
13837 omitted as well. The third field, <end> is a constant string that will be
13838 appended after the variable. It may also be omitted. Together, these elements
13839 allow to concatenate variables with delimiters to an existing set of
13840 variables. This can be used to build new variables made of a succession of
John Roesler27c754c2019-07-10 15:45:51 -050013841 other variables, such as colon-delimited values. Note that due to the config
Willy Tarreau280f42b2018-02-19 15:34:12 +010013842 parser, it is not possible to use a comma nor a closing parenthesis as
John Roesler27c754c2019-07-10 15:45:51 -050013843 delimiters.
Willy Tarreau280f42b2018-02-19 15:34:12 +010013844
13845 Example:
13846 tcp-request session set-var(sess.src) src
13847 tcp-request session set-var(sess.dn) ssl_c_s_dn
13848 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13849 http-request set-header x-hap-sig %[var(txn.sig)]
13850
Willy Tarreau97707872015-01-27 15:12:13 +010013851cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013852 Takes the input value of type signed integer, applies a ones-complement
13853 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013854
Willy Tarreau80599772015-01-20 19:35:24 +010013855crc32([<avalanche>])
13856 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13857 hash function. Optionally, it is possible to apply a full avalanche hash
13858 function to the output if the optional <avalanche> argument equals 1. This
13859 converter uses the same functions as used by the various hash-based load
13860 balancing algorithms, so it will provide exactly the same results. It is
13861 provided for compatibility with other software which want a CRC32 to be
13862 computed on some input keys, so it follows the most common implementation as
13863 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13864 but may provide a better or at least less predictable distribution. It must
13865 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013866 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13867
13868crc32c([<avalanche>])
13869 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13870 hash function. Optionally, it is possible to apply a full avalanche hash
13871 function to the output if the optional <avalanche> argument equals 1. This
13872 converter uses the same functions as described in RFC4960, Appendix B [8].
13873 It is provided for compatibility with other software which want a CRC32C to be
13874 computed on some input keys. It is slower than the other algorithms and it must
13875 not be used for security purposes as a 32-bit hash is trivial to break. See
13876 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013877
David Carlier29b3ca32015-09-25 14:09:21 +010013878da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013879 Asks the DeviceAtlas converter to identify the User Agent string passed on
13880 input, and to emit a string made of the concatenation of the properties
13881 enumerated in argument, delimited by the separator defined by the global
13882 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013883 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013884 configuration language.
13885
13886 Example:
13887 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013888 bind *:8881
13889 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013890 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 +020013891
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013892debug
13893 This converter is used as debug tool. It dumps on screen the content and the
13894 type of the input sample. The sample is returned as is on its output. This
13895 converter only exists when haproxy was built with debugging enabled.
13896
Willy Tarreau97707872015-01-27 15:12:13 +010013897div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013898 Divides the input value of type signed integer by <value>, and returns the
13899 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013900 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013901 variable name. The name of the variable starts with an indication about its
13902 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013903 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013904 "sess" : the variable is shared with the whole session
13905 "txn" : the variable is shared with the transaction (request and response)
13906 "req" : the variable is shared only during request processing
13907 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013908 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013909 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013910
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013911djb2([<avalanche>])
13912 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13913 hash function. Optionally, it is possible to apply a full avalanche hash
13914 function to the output if the optional <avalanche> argument equals 1. This
13915 converter uses the same functions as used by the various hash-based load
13916 balancing algorithms, so it will provide exactly the same results. It is
13917 mostly intended for debugging, but can be used as a stick-table entry to
13918 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013919 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13920 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013921
Willy Tarreau97707872015-01-27 15:12:13 +010013922even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013923 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013924 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13925
Marcin Deranek9631a282018-04-16 14:30:46 +020013926field(<index>,<delimiters>[,<count>])
13927 Extracts the substring at the given index counting from the beginning
13928 (positive index) or from the end (negative index) considering given delimiters
13929 from an input string. Indexes start at 1 or -1 and delimiters are a string
13930 formatted list of chars. Optionally you can specify <count> of fields to
13931 extract (default: 1). Value of 0 indicates extraction of all remaining
13932 fields.
13933
13934 Example :
13935 str(f1_f2_f3__f5),field(5,_) # f5
13936 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13937 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13938 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13939 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013940
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013941hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013942 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013943 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013944 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 +020013945 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013946
Dragan Dosen3f957b22017-10-24 09:27:34 +020013947hex2i
13948 Converts a hex string containing two hex digits per input byte to an
John Roesler27c754c2019-07-10 15:45:51 -050013949 integer. If the input value cannot be converted, then zero is returned.
Dragan Dosen3f957b22017-10-24 09:27:34 +020013950
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013951http_date([<offset>])
13952 Converts an integer supposed to contain a date since epoch to a string
13953 representing this date in a format suitable for use in HTTP header fields. If
13954 an offset value is specified, then it is a number of seconds that is added to
13955 the date before the conversion is operated. This is particularly useful to
13956 emit Date header fields, Expires values in responses when combined with a
13957 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013958
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013959in_table(<table>)
13960 Uses the string representation of the input sample to perform a look up in
13961 the specified table. If the key is not found in the table, a boolean false
13962 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013963 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013964 or not a source IP address or an Authorization header was already seen).
13965
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013966ipmask(<mask4>, [<mask6>])
13967 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013968 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013969 table entries and as such use the same server. The mask4 can be passed in
13970 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13971 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13972 If no mask6 is given IPv6 addresses will fail to convert for backwards
13973 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013974
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013975json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013976 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013977 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013978 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013979 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13980 of errors:
13981 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13982 bytes, ...)
13983 - invalid range (the decoded value is within a UTF-8 prohibited range),
13984 - code overlong (the value is encoded with more bytes than necessary).
13985
13986 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13987 character is greater than 0xffff because the JSON string escape specification
13988 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13989 in 4 variants designated by a combination of two suffix letters : "p" for
13990 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13991 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013992 - "ascii" : never fails;
13993 - "utf8" : fails on any detected errors;
13994 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013995 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013996 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013997 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13998 characters corresponding to the other errors.
13999
14000 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010014001 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014002
14003 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014004 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020014005 capture request header user-agent len 150
14006 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020014007
14008 Input request from client 127.0.0.1:
14009 GET / HTTP/1.0
14010 User-Agent: Very "Ugly" UA 1/2
14011
14012 Output log:
14013 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
14014
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014015language(<value>[,<default>])
14016 Returns the value with the highest q-factor from a list as extracted from the
14017 "accept-language" header using "req.fhdr". Values with no q-factor have a
14018 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
14019 belong to the list of semi-colon delimited <values> will be considered. The
14020 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
14021 given list and a default value is provided, it is returned. Note that language
14022 names may have a variant after a dash ('-'). If this variant is present in the
14023 list, it will be matched, but if it is not, only the base language is checked.
14024 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010014025 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014026 ordering of the values in the request counts, as the first value among
14027 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020014028
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014029 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020014030
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014031 # this configuration switches to the backend matching a
14032 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020014033
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014034 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
14035 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
14036 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
14037 use_backend spanish if es
14038 use_backend french if fr
14039 use_backend english if en
14040 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020014041
Willy Tarreau60a2ee72017-12-15 07:13:48 +010014042length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010014043 Get the length of the string. This can only be placed after a string
14044 sample fetch function or after a transformation keyword returning a string
14045 type. The result is of type integer.
14046
Willy Tarreauffcb2e42014-07-10 16:29:08 +020014047lower
14048 Convert a string sample to lower case. This can only be placed after a string
14049 sample fetch function or after a transformation keyword returning a string
14050 type. The result is of type string.
14051
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014052ltime(<format>[,<offset>])
14053 Converts an integer supposed to contain a date since epoch to a string
14054 representing this date in local time using a format defined by the <format>
14055 string using strftime(3). The purpose is to allow any date format to be used
14056 in logs. An optional <offset> in seconds may be applied to the input date
14057 (positive or negative). See the strftime() man page for the format supported
14058 by your operating system. See also the utime converter.
14059
14060 Example :
14061
14062 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010014063 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014064 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
14065
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014066map(<map_file>[,<default_value>])
14067map_<match_type>(<map_file>[,<default_value>])
14068map_<match_type>_<output_type>(<map_file>[,<default_value>])
14069 Search the input value from <map_file> using the <match_type> matching method,
14070 and return the associated value converted to the type <output_type>. If the
14071 input value cannot be found in the <map_file>, the converter returns the
14072 <default_value>. If the <default_value> is not set, the converter fails and
14073 acts as if no input value could be fetched. If the <match_type> is not set, it
14074 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
14075 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
14076 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014077
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014078 It is important to avoid overlapping between the keys : IP addresses and
14079 strings are stored in trees, so the first of the finest match will be used.
14080 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014081
Tim Düsterhus4896c442016-11-29 02:15:19 +010014082 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014083 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014084
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014085 input type | match method | output type str | output type int | output type ip
14086 -----------+--------------+-----------------+-----------------+---------------
14087 str | str | map_str | map_str_int | map_str_ip
14088 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020014089 str | beg | map_beg | map_beg_int | map_end_ip
14090 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014091 str | sub | map_sub | map_sub_int | map_sub_ip
14092 -----------+--------------+-----------------+-----------------+---------------
14093 str | dir | map_dir | map_dir_int | map_dir_ip
14094 -----------+--------------+-----------------+-----------------+---------------
14095 str | dom | map_dom | map_dom_int | map_dom_ip
14096 -----------+--------------+-----------------+-----------------+---------------
14097 str | end | map_end | map_end_int | map_end_ip
14098 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080014099 str | reg | map_reg | map_reg_int | map_reg_ip
14100 -----------+--------------+-----------------+-----------------+---------------
14101 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014102 -----------+--------------+-----------------+-----------------+---------------
14103 int | int | map_int | map_int_int | map_int_ip
14104 -----------+--------------+-----------------+-----------------+---------------
14105 ip | ip | map_ip | map_ip_int | map_ip_ip
14106 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014107
Thierry Fournier8feaa662016-02-10 22:55:20 +010014108 The special map called "map_regm" expect matching zone in the regular
14109 expression and modify the output replacing back reference (like "\1") by
14110 the corresponding match text.
14111
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014112 The file contains one key + value per line. Lines which start with '#' are
14113 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
14114 is then the first "word" (series of non-space/tabs characters), and the value
14115 is what follows this series of space/tab till the end of the line excluding
14116 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014117
Thierry FOURNIER060762e2014-04-23 13:29:15 +020014118 Example :
14119
14120 # this is a comment and is ignored
14121 2.22.246.0/23 United Kingdom \n
14122 <-><-----------><--><------------><---->
14123 | | | | `- trailing spaces ignored
14124 | | | `---------- value
14125 | | `-------------------- middle spaces ignored
14126 | `---------------------------- key
14127 `------------------------------------ leading spaces ignored
14128
Willy Tarreau97707872015-01-27 15:12:13 +010014129mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014130 Divides the input value of type signed integer by <value>, and returns the
14131 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014132 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014133 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014134 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014135 "sess" : the variable is shared with the whole session
14136 "txn" : the variable is shared with the transaction (request and response)
14137 "req" : the variable is shared only during request processing
14138 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014139 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014140 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014141
14142mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014143 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020014144 the product as an signed integer. In case of overflow, the largest possible
14145 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014146 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014147 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014148 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014149 "sess" : the variable is shared with the whole session
14150 "txn" : the variable is shared with the transaction (request and response)
14151 "req" : the variable is shared only during request processing
14152 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014153 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014154 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014155
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010014156nbsrv
14157 Takes an input value of type string, interprets it as a backend name and
14158 returns the number of usable servers in that backend. Can be used in places
14159 where we want to look up a backend from a dynamic name, like a result of a
14160 map lookup.
14161
Willy Tarreau97707872015-01-27 15:12:13 +010014162neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014163 Takes the input value of type signed integer, computes the opposite value,
14164 and returns the remainder as an signed integer. 0 is identity. This operator
14165 is provided for reversed subtracts : in order to subtract the input from a
14166 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010014167
14168not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014169 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010014170 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010014171 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010014172 absence of a flag).
14173
14174odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014175 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010014176 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
14177
14178or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014179 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014180 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010014181 numeric value or a variable name. The name of the variable starts with an
14182 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014183 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014184 "sess" : the variable is shared with the whole session
14185 "txn" : the variable is shared with the transaction (request and response)
14186 "req" : the variable is shared only during request processing
14187 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010014188 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014189 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014190
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014191protobuf(<field_number>,[<field_type>])
14192 This extracts the protocol buffers message field in raw mode of an input binary
14193 sample representation of a protocol buffer message with <field_number> as field
14194 number (dotted notation) if <field_type> is not present, or as an integer sample
14195 if this field is present (see also "ungrpc" below).
14196 The list of the authorized types is the following one: "int32", "int64", "uint32",
14197 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
14198 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
14199 "float" for the wire type 5. Note that "string" is considered as a length-delimited
14200 type, so it does not require any <field_type> argument to be extracted.
14201 More information may be found here about the protocol buffers message field types:
14202 https://developers.google.com/protocol-buffers/docs/encoding
14203
Willy Tarreauc4dc3502015-01-23 20:39:28 +010014204regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010014205 Applies a regex-based substitution to the input string. It does the same
14206 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
14207 default it will replace in the input string the first occurrence of the
14208 largest part matching the regular expression <regex> with the substitution
14209 string <subst>. It is possible to replace all occurrences instead by adding
14210 the flag "g" in the third argument <flags>. It is also possible to make the
14211 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
14212 string, it is made up from the concatenation of all desired flags. Thus if
14213 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
14214 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010014215 configuration parser, some characters such as closing parenthesis, closing
14216 square brackets or comma are not possible to use in the arguments. The first
14217 use of this converter is to replace certain characters or sequence of
14218 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010014219
14220 Example :
14221
14222 # de-duplicate "/" in header "x-path".
14223 # input: x-path: /////a///b/c/xzxyz/
14224 # output: x-path: /a/b/c/xzxyz/
14225 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
14226
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020014227capture-req(<id>)
14228 Capture the string entry in the request slot <id> and returns the entry as
14229 is. If the slot doesn't exist, the capture fails silently.
14230
14231 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020014232 "http-response capture", "capture.req.hdr" and
14233 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020014234
14235capture-res(<id>)
14236 Capture the string entry in the response slot <id> and returns the entry as
14237 is. If the slot doesn't exist, the capture fails silently.
14238
14239 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020014240 "http-response capture", "capture.req.hdr" and
14241 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020014242
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014243sdbm([<avalanche>])
14244 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
14245 hash function. Optionally, it is possible to apply a full avalanche hash
14246 function to the output if the optional <avalanche> argument equals 1. This
14247 converter uses the same functions as used by the various hash-based load
14248 balancing algorithms, so it will provide exactly the same results. It is
14249 mostly intended for debugging, but can be used as a stick-table entry to
14250 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010014251 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
14252 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014253
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014254set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010014255 Sets a variable with the input content and returns the content on the output
14256 as-is. The variable keeps the value and the associated input type. The name of
14257 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014258 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014259 "sess" : the variable is shared with the whole session
14260 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014261 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014262 "req" : the variable is shared only during request processing,
14263 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014264 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014265 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014266
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020014267sha1
14268 Converts a binary input sample to a SHA1 digest. The result is a binary
14269 sample with length of 20 bytes.
14270
Tim Duesterhusca097c12018-04-27 21:18:45 +020014271strcmp(<var>)
14272 Compares the contents of <var> with the input value of type string. Returns
14273 the result as a signed integer compatible with strcmp(3): 0 if both strings
14274 are identical. A value less than 0 if the left string is lexicographically
14275 smaller than the right string or if the left string is shorter. A value greater
14276 than 0 otherwise (right string greater than left string or the right string is
14277 shorter).
14278
14279 Example :
14280
14281 http-request set-var(txn.host) hdr(host)
14282 # Check whether the client is attempting domain fronting.
14283 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
14284
14285
Willy Tarreau97707872015-01-27 15:12:13 +010014286sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014287 Subtracts <value> from the input value of type signed integer, and returns
14288 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014289 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010014290 or a variable name. The name of the variable starts with an indication about
14291 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014292 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014293 "sess" : the variable is shared with the whole session
14294 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014295 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014296 "req" : the variable is shared only during request processing,
14297 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014298 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014299 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014300
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014301table_bytes_in_rate(<table>)
14302 Uses the string representation of the input sample to perform a look up in
14303 the specified table. If the key is not found in the table, integer value zero
14304 is returned. Otherwise the converter returns the average client-to-server
14305 bytes rate associated with the input sample in the designated table, measured
14306 in amount of bytes over the period configured in the table. See also the
14307 sc_bytes_in_rate sample fetch keyword.
14308
14309
14310table_bytes_out_rate(<table>)
14311 Uses the string representation of the input sample to perform a look up in
14312 the specified table. If the key is not found in the table, integer value zero
14313 is returned. Otherwise the converter returns the average server-to-client
14314 bytes rate associated with the input sample in the designated table, measured
14315 in amount of bytes over the period configured in the table. See also the
14316 sc_bytes_out_rate sample fetch keyword.
14317
14318table_conn_cnt(<table>)
14319 Uses the string representation of the input sample to perform a look up in
14320 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014321 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014322 connections associated with the input sample in the designated table. See
14323 also the sc_conn_cnt sample fetch keyword.
14324
14325table_conn_cur(<table>)
14326 Uses the string representation of the input sample to perform a look up in
14327 the specified table. If the key is not found in the table, integer value zero
14328 is returned. Otherwise the converter returns the current amount of concurrent
14329 tracked connections associated with the input sample in the designated table.
14330 See also the sc_conn_cur sample fetch keyword.
14331
14332table_conn_rate(<table>)
14333 Uses the string representation of the input sample to perform a look up in
14334 the specified table. If the key is not found in the table, integer value zero
14335 is returned. Otherwise the converter returns the average incoming connection
14336 rate associated with the input sample in the designated table. See also the
14337 sc_conn_rate sample fetch keyword.
14338
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014339table_gpt0(<table>)
14340 Uses the string representation of the input sample to perform a look up in
14341 the specified table. If the key is not found in the table, boolean value zero
14342 is returned. Otherwise the converter returns the current value of the first
14343 general purpose tag associated with the input sample in the designated table.
14344 See also the sc_get_gpt0 sample fetch keyword.
14345
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014346table_gpc0(<table>)
14347 Uses the string representation of the input sample to perform a look up in
14348 the specified table. If the key is not found in the table, integer value zero
14349 is returned. Otherwise the converter returns the current value of the first
14350 general purpose counter associated with the input sample in the designated
14351 table. See also the sc_get_gpc0 sample fetch keyword.
14352
14353table_gpc0_rate(<table>)
14354 Uses the string representation of the input sample to perform a look up in
14355 the specified table. If the key is not found in the table, integer value zero
14356 is returned. Otherwise the converter returns the frequency which the gpc0
14357 counter was incremented over the configured period in the table, associated
14358 with the input sample in the designated table. See also the sc_get_gpc0_rate
14359 sample fetch keyword.
14360
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014361table_gpc1(<table>)
14362 Uses the string representation of the input sample to perform a look up in
14363 the specified table. If the key is not found in the table, integer value zero
14364 is returned. Otherwise the converter returns the current value of the second
14365 general purpose counter associated with the input sample in the designated
14366 table. See also the sc_get_gpc1 sample fetch keyword.
14367
14368table_gpc1_rate(<table>)
14369 Uses the string representation of the input sample to perform a look up in
14370 the specified table. If the key is not found in the table, integer value zero
14371 is returned. Otherwise the converter returns the frequency which the gpc1
14372 counter was incremented over the configured period in the table, associated
14373 with the input sample in the designated table. See also the sc_get_gpc1_rate
14374 sample fetch keyword.
14375
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014376table_http_err_cnt(<table>)
14377 Uses the string representation of the input sample to perform a look up in
14378 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014379 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014380 errors associated with the input sample in the designated table. See also the
14381 sc_http_err_cnt sample fetch keyword.
14382
14383table_http_err_rate(<table>)
14384 Uses the string representation of the input sample to perform a look up in
14385 the specified table. If the key is not found in the table, integer value zero
14386 is returned. Otherwise the average rate of HTTP errors associated with the
14387 input sample in the designated table, measured in amount of errors over the
14388 period configured in the table. See also the sc_http_err_rate sample fetch
14389 keyword.
14390
14391table_http_req_cnt(<table>)
14392 Uses the string representation of the input sample to perform a look up in
14393 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014394 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014395 requests associated with the input sample in the designated table. See also
14396 the sc_http_req_cnt sample fetch keyword.
14397
14398table_http_req_rate(<table>)
14399 Uses the string representation of the input sample to perform a look up in
14400 the specified table. If the key is not found in the table, integer value zero
14401 is returned. Otherwise the average rate of HTTP requests associated with the
14402 input sample in the designated table, measured in amount of requests over the
14403 period configured in the table. See also the sc_http_req_rate sample fetch
14404 keyword.
14405
14406table_kbytes_in(<table>)
14407 Uses the string representation of the input sample to perform a look up in
14408 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014409 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014410 to-server data associated with the input sample in the designated table,
14411 measured in kilobytes. The test is currently performed on 32-bit integers,
14412 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
14413 keyword.
14414
14415table_kbytes_out(<table>)
14416 Uses the string representation of the input sample to perform a look up in
14417 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010014418 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014419 to-client data associated with the input sample in the designated table,
14420 measured in kilobytes. The test is currently performed on 32-bit integers,
14421 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
14422 keyword.
14423
14424table_server_id(<table>)
14425 Uses the string representation of the input sample to perform a look up in
14426 the specified table. If the key is not found in the table, integer value zero
14427 is returned. Otherwise the converter returns the server ID associated with
14428 the input sample in the designated table. A server ID is associated to a
14429 sample by a "stick" rule when a connection to a server succeeds. A server ID
14430 zero means that no server is associated with this key.
14431
14432table_sess_cnt(<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
Davor Ocelice9ed2812017-12-25 17:49:28 +010014435 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020014436 sessions associated with the input sample in the designated table. Note that
14437 a session here refers to an incoming connection being accepted by the
14438 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
14439 keyword.
14440
14441table_sess_rate(<table>)
14442 Uses the string representation of the input sample to perform a look up in
14443 the specified table. If the key is not found in the table, integer value zero
14444 is returned. Otherwise the converter returns the average incoming session
14445 rate associated with the input sample in the designated table. Note that a
14446 session here refers to an incoming connection being accepted by the
14447 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
14448 keyword.
14449
14450table_trackers(<table>)
14451 Uses the string representation of the input sample to perform a look up in
14452 the specified table. If the key is not found in the table, integer value zero
14453 is returned. Otherwise the converter returns the current amount of concurrent
14454 connections tracking the same key as the input sample in the designated
14455 table. It differs from table_conn_cur in that it does not rely on any stored
14456 information but on the table's reference count (the "use" value which is
14457 returned by "show table" on the CLI). This may sometimes be more suited for
14458 layer7 tracking. It can be used to tell a server how many concurrent
14459 connections there are from a given address for example. See also the
14460 sc_trackers sample fetch keyword.
14461
Willy Tarreauffcb2e42014-07-10 16:29:08 +020014462upper
14463 Convert a string sample to upper case. This can only be placed after a string
14464 sample fetch function or after a transformation keyword returning a string
14465 type. The result is of type string.
14466
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020014467url_dec
14468 Takes an url-encoded string provided as input and returns the decoded
14469 version as output. The input and the output are of type string.
14470
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014471ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014472 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014473 sample representation of a gRPC message with <field_number> as field number
14474 (dotted notation) if <field_type> is not present, or as an integer sample if this
14475 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014476 The list of the authorized types is the following one: "int32", "int64", "uint32",
14477 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
14478 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
14479 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014480 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014481 More information may be found here about the protocol buffers message field types:
14482 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014483
14484 Example:
14485 // with such a protocol buffer .proto file content adapted from
14486 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
14487
14488 message Point {
14489 int32 latitude = 1;
14490 int32 longitude = 2;
14491 }
14492
14493 message PPoint {
14494 Point point = 59;
14495 }
14496
14497 message Rectangle {
14498 // One corner of the rectangle.
14499 PPoint lo = 48;
14500 // The other corner of the rectangle.
14501 PPoint hi = 49;
14502 }
14503
14504 let's say a body request is made of a "Rectangle" object value (two PPoint
14505 protocol buffers messages), the four protocol buffers fields could be
14506 extracted with these "ungrpc" directives:
14507
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014508 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
14509 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
John Roesler27c754c2019-07-10 15:45:51 -050014510 req.body,ungrpc(49.59.1,int32) # "latitude" of "hi" second PPoint
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014511 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
14512
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014513 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014514
Frédéric Lécaille93d33162019-03-06 09:35:59 +010014515 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010014516
John Roesler27c754c2019-07-10 15:45:51 -050014517 As a gRPC message is always made of a gRPC header followed by protocol buffers
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010014518 messages, in the previous example the "latitude" of "lo" first PPoint
14519 could be extracted with these equivalent directives:
14520
14521 req.body,ungrpc(48.59),protobuf(1,int32)
14522 req.body,ungrpc(48),protobuf(59.1,int32)
14523 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
14524
14525 Note that the first convert must be "ungrpc", the remaining ones must be
14526 "protobuf" and only the last one may have or not a second argument to
14527 interpret the previous binary sample.
14528
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010014529
Christopher Faulet85d79c92016-11-09 16:54:56 +010014530unset-var(<var name>)
14531 Unsets a variable if the input content is defined. The name of the variable
14532 starts with an indication about its scope. The scopes allowed are:
14533 "proc" : the variable is shared with the whole process
14534 "sess" : the variable is shared with the whole session
14535 "txn" : the variable is shared with the transaction (request and
14536 response),
14537 "req" : the variable is shared only during request processing,
14538 "res" : the variable is shared only during response processing.
14539 This prefix is followed by a name. The separator is a '.'. The name may only
14540 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
14541
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014542utime(<format>[,<offset>])
14543 Converts an integer supposed to contain a date since epoch to a string
14544 representing this date in UTC time using a format defined by the <format>
14545 string using strftime(3). The purpose is to allow any date format to be used
14546 in logs. An optional <offset> in seconds may be applied to the input date
14547 (positive or negative). See the strftime() man page for the format supported
14548 by your operating system. See also the ltime converter.
14549
14550 Example :
14551
14552 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010014553 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020014554 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
14555
Marcin Deranek9631a282018-04-16 14:30:46 +020014556word(<index>,<delimiters>[,<count>])
14557 Extracts the nth word counting from the beginning (positive index) or from
14558 the end (negative index) considering given delimiters from an input string.
14559 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
14560 Optionally you can specify <count> of words to extract (default: 1).
14561 Value of 0 indicates extraction of all remaining words.
14562
14563 Example :
14564 str(f1_f2_f3__f5),word(4,_) # f5
14565 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
14566 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
14567 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
14568 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010014569
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014570wt6([<avalanche>])
14571 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
14572 hash function. Optionally, it is possible to apply a full avalanche hash
14573 function to the output if the optional <avalanche> argument equals 1. This
14574 converter uses the same functions as used by the various hash-based load
14575 balancing algorithms, so it will provide exactly the same results. It is
14576 mostly intended for debugging, but can be used as a stick-table entry to
14577 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010014578 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
14579 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020014580
Willy Tarreau97707872015-01-27 15:12:13 +010014581xor(<value>)
14582 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014583 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014584 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010014585 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014586 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014587 "sess" : the variable is shared with the whole session
14588 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014589 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014590 "req" : the variable is shared only during request processing,
14591 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014592 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014593 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014594
Thierry FOURNIER01e09742016-12-26 11:46:11 +010014595xxh32([<seed>])
14596 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
14597 variant of the XXHash hash function. This hash supports a seed which defaults
14598 to zero but a different value maybe passed as the <seed> argument. This hash
14599 is known to be very good and very fast so it can be used to hash URLs and/or
14600 URL parameters for use as stick-table keys to collect statistics with a low
14601 collision rate, though care must be taken as the algorithm is not considered
14602 as cryptographically secure.
14603
14604xxh64([<seed>])
14605 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
14606 variant of the XXHash hash function. This hash supports a seed which defaults
14607 to zero but a different value maybe passed as the <seed> argument. This hash
14608 is known to be very good and very fast so it can be used to hash URLs and/or
14609 URL parameters for use as stick-table keys to collect statistics with a low
14610 collision rate, though care must be taken as the algorithm is not considered
14611 as cryptographically secure.
14612
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014613
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200146147.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020014615--------------------------------------------
14616
14617A first set of sample fetch methods applies to internal information which does
14618not even relate to any client information. These ones are sometimes used with
14619"monitor-fail" directives to report an internal status to external watchers.
14620The sample fetch methods described in this section are usable anywhere.
14621
14622always_false : boolean
14623 Always returns the boolean "false" value. It may be used with ACLs as a
14624 temporary replacement for another one when adjusting configurations.
14625
14626always_true : boolean
14627 Always returns the boolean "true" value. It may be used with ACLs as a
14628 temporary replacement for another one when adjusting configurations.
14629
14630avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014631 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020014632 divided by the number of active servers. The current backend is used if no
14633 backend is specified. This is very similar to "queue" except that the size of
14634 the farm is considered, in order to give a more accurate measurement of the
14635 time it may take for a new connection to be processed. The main usage is with
14636 ACL to return a sorry page to new users when it becomes certain they will get
14637 a degraded service, or to pass to the backend servers in a header so that
14638 they decide to work in degraded mode or to disable some functions to speed up
14639 the processing a bit. Note that in the event there would not be any active
14640 server anymore, twice the number of queued connections would be considered as
14641 the measured value. This is a fair estimate, as we expect one server to get
14642 back soon anyway, but we still prefer to send new traffic to another backend
14643 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
14644 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010014645
Willy Tarreau74ca5042013-06-11 23:12:07 +020014646be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020014647 Applies to the number of currently established connections on the backend,
14648 possibly including the connection being evaluated. If no backend name is
14649 specified, the current one is used. But it is also possible to check another
14650 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014651 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
14652
14653be_conn_free([<backend>]) : integer
14654 Returns an integer value corresponding to the number of available connections
14655 across available servers in the backend. Queue slots are not included. Backup
14656 servers are also not included, unless all other servers are down. If no
14657 backend name is specified, the current one is used. But it is also possible
14658 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014659 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
14660 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014661
14662 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
14663 (meaning unlimited), then this fetch clearly does not make sense, in which
14664 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014665
Willy Tarreau74ca5042013-06-11 23:12:07 +020014666be_sess_rate([<backend>]) : integer
14667 Returns an integer value corresponding to the sessions creation rate on the
14668 backend, in number of new sessions per second. This is used with ACLs to
14669 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010014670 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020014671 online dictionary). It can also be useful to add this element to logs using a
14672 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014673
14674 Example :
14675 # Redirect to an error page if the dictionary is requested too often
14676 backend dynamic
14677 mode http
14678 acl being_scanned be_sess_rate gt 100
14679 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010014680
Davor Ocelice9ed2812017-12-25 17:49:28 +010014681bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014682 Returns a binary chain. The input is the hexadecimal representation
14683 of the string.
14684
14685bool(<bool>) : bool
14686 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
14687 'false' and '0' are the same. 'true' and '1' are the same.
14688
Willy Tarreau74ca5042013-06-11 23:12:07 +020014689connslots([<backend>]) : integer
14690 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014691 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020014692 connections on all servers and the maximum queue size. This is probably only
14693 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050014694
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014695 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020014696 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014697 usage; see "use_backend" keyword) can be redirected to a different backend.
14698
Willy Tarreau55165fe2009-05-10 12:02:55 +020014699 'connslots' = number of available server connection slots, + number of
14700 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014701
Willy Tarreaua36af912009-10-10 12:02:45 +020014702 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020014703 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020014704 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020014705 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010014706 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020014707 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020014708 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014709
Willy Tarreau55165fe2009-05-10 12:02:55 +020014710 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
14711 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020014712 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020014713 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014714
Willy Tarreau70fe9442018-11-22 16:07:39 +010014715cpu_calls : integer
14716 Returns the number of calls to the task processing the stream or current
14717 request since it was allocated. This number is reset for each new request on
14718 the same connections in case of HTTP keep-alive. This value should usually be
14719 low and stable (around 2 calls for a typically simple request) but may become
14720 high if some processing (compression, caching or analysis) is performed. This
14721 is purely for performance monitoring purposes.
14722
14723cpu_ns_avg : integer
14724 Returns the average number of nanoseconds spent in each call to the task
14725 processing the stream or current request. This number is reset for each new
14726 request on the same connections in case of HTTP keep-alive. This value
14727 indicates the overall cost of processing the request or the connection for
14728 each call. There is no good nor bad value but the time spent in a call
14729 automatically causes latency for other processing (see lat_ns_avg below),
14730 and may affect other connection's apparent response time. Certain operations
14731 like compression, complex regex matching or heavy Lua operations may directly
14732 affect this value, and having it in the logs will make it easier to spot the
14733 faulty processing that needs to be fixed to recover decent performance.
14734 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
14735
14736cpu_ns_tot : integer
14737 Returns the total number of nanoseconds spent in each call to the task
14738 processing the stream or current request. This number is reset for each new
14739 request on the same connections in case of HTTP keep-alive. This value
14740 indicates the overall cost of processing the request or the connection for
14741 each call. There is no good nor bad value but the time spent in a call
14742 automatically causes latency for other processing (see lat_ns_avg below),
14743 induces CPU costs on the machine, and may affect other connection's apparent
14744 response time. Certain operations like compression, complex regex matching or
14745 heavy Lua operations may directly affect this value, and having it in the
14746 logs will make it easier to spot the faulty processing that needs to be fixed
14747 to recover decent performance. The value may be artificially high due to a
14748 high cpu_calls count, for example when processing many HTTP chunks, and for
14749 this reason it is often preferred to log cpu_ns_avg instead.
14750
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014751date([<offset>]) : integer
14752 Returns the current date as the epoch (number of seconds since 01/01/1970).
14753 If an offset value is specified, then it is a number of seconds that is added
14754 to the current date before returning the value. This is particularly useful
14755 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020014756 It is useful combined with the http_date converter.
14757
14758 Example :
14759
14760 # set an expires header to now+1 hour in every response
14761 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014762
Etienne Carrierea792a0a2018-01-17 13:43:24 +010014763date_us : integer
14764 Return the microseconds part of the date (the "second" part is returned by
14765 date sample). This sample is coherent with the date sample as it is comes
14766 from the same timeval structure.
14767
Willy Tarreaud716f9b2017-10-13 11:03:15 +020014768distcc_body(<token>[,<occ>]) : binary
14769 Parses a distcc message and returns the body associated to occurrence #<occ>
14770 of the token <token>. Occurrences start at 1, and when unspecified, any may
14771 match though in practice only the first one is checked for now. This can be
14772 used to extract file names or arguments in files built using distcc through
14773 haproxy. Please refer to distcc's protocol documentation for the complete
14774 list of supported tokens.
14775
14776distcc_param(<token>[,<occ>]) : integer
14777 Parses a distcc message and returns the parameter associated to occurrence
14778 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
14779 any may match though in practice only the first one is checked for now. This
14780 can be used to extract certain information such as the protocol version, the
14781 file size or the argument in files built using distcc through haproxy.
14782 Another use case consists in waiting for the start of the preprocessed file
14783 contents before connecting to the server to avoid keeping idle connections.
14784 Please refer to distcc's protocol documentation for the complete list of
14785 supported tokens.
14786
14787 Example :
14788 # wait up to 20s for the pre-processed file to be uploaded
14789 tcp-request inspect-delay 20s
14790 tcp-request content accept if { distcc_param(DOTI) -m found }
14791 # send large files to the big farm
14792 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14793
Willy Tarreau595ec542013-06-12 21:34:28 +020014794env(<name>) : string
14795 Returns a string containing the value of environment variable <name>. As a
14796 reminder, environment variables are per-process and are sampled when the
14797 process starts. This can be useful to pass some information to a next hop
14798 server, or with ACLs to take specific action when the process is started a
14799 certain way.
14800
14801 Examples :
14802 # Pass the Via header to next hop with the local hostname in it
14803 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14804
14805 # reject cookie-less requests when the STOP environment variable is set
14806 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14807
Willy Tarreau74ca5042013-06-11 23:12:07 +020014808fe_conn([<frontend>]) : integer
14809 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014810 possibly including the connection being evaluated. If no frontend name is
14811 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014812 frontend. It can be used to return a sorry page before hard-blocking, or to
14813 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014814 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014815 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14816 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014817
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014818fe_req_rate([<frontend>]) : integer
14819 Returns an integer value corresponding to the number of HTTP requests per
14820 second sent to a frontend. This number can differ from "fe_sess_rate" in
14821 situations where client-side keep-alive is enabled.
14822
Willy Tarreau74ca5042013-06-11 23:12:07 +020014823fe_sess_rate([<frontend>]) : integer
14824 Returns an integer value corresponding to the sessions creation rate on the
14825 frontend, in number of new sessions per second. This is used with ACLs to
14826 limit the incoming session rate to an acceptable range in order to prevent
14827 abuse of service at the earliest moment, for example when combined with other
14828 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14829 down below the limit. It can also be useful to add this element to logs using
14830 a log-format directive. See also the "rate-limit sessions" directive for use
14831 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014832
14833 Example :
14834 # This frontend limits incoming mails to 10/s with a max of 100
14835 # concurrent connections. We accept any connection below 10/s, and
14836 # force excess clients to wait for 100 ms. Since clients are limited to
14837 # 100 max, there cannot be more than 10 incoming mails per second.
14838 frontend mail
14839 bind :25
14840 mode tcp
14841 maxconn 100
14842 acl too_fast fe_sess_rate ge 10
14843 tcp-request inspect-delay 100ms
14844 tcp-request content accept if ! too_fast
14845 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014846
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014847hostname : string
14848 Returns the system hostname.
14849
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014850int(<integer>) : signed integer
14851 Returns a signed integer.
14852
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014853ipv4(<ipv4>) : ipv4
14854 Returns an ipv4.
14855
14856ipv6(<ipv6>) : ipv6
14857 Returns an ipv6.
14858
Willy Tarreau70fe9442018-11-22 16:07:39 +010014859lat_ns_avg : integer
14860 Returns the average number of nanoseconds spent between the moment the task
14861 handling the stream is woken up and the moment it is effectively called. This
14862 number is reset for each new request on the same connections in case of HTTP
14863 keep-alive. This value indicates the overall latency inflicted to the current
14864 request by all other requests being processed in parallel, and is a direct
14865 indicator of perceived performance due to noisy neighbours. In order to keep
14866 the value low, it is possible to reduce the scheduler's run queue depth using
14867 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14868 once using "tune.maxpollevents", to decrease the stream's nice value using
14869 the "nice" option on the "bind" lines or in the frontend, or to look for
14870 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14871 whose processing needs to be adjusted or fixed. Compression of large buffers
14872 could be a culprit, like heavy regex or long lists of regex.
14873 Note: this value is exactly lat_ns_tot divided by cpu_calls.
14874
14875lat_ns_tot : integer
14876 Returns the total number of nanoseconds spent between the moment the task
14877 handling the stream is woken up and the moment it is effectively called. This
14878 number is reset for each new request on the same connections in case of HTTP
14879 keep-alive. This value indicates the overall latency inflicted to the current
14880 request by all other requests being processed in parallel, and is a direct
14881 indicator of perceived performance due to noisy neighbours. In order to keep
14882 the value low, it is possible to reduce the scheduler's run queue depth using
14883 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14884 once using "tune.maxpollevents", to decrease the stream's nice value using
14885 the "nice" option on the "bind" lines or in the frontend, or to look for
14886 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14887 whose processing needs to be adjusted or fixed. Compression of large buffers
14888 could be a culprit, like heavy regex or long lists of regex. Note: while it
14889 may intuitively seem that the total latency adds to a transfer time, it is
14890 almost never true because while a task waits for the CPU, network buffers
14891 continue to fill up and the next call will process more at once. The value
14892 may be artificially high due to a high cpu_calls count, for example when
14893 processing many HTTP chunks, and for this reason it is often preferred to log
14894 lat_ns_avg instead, which is a more relevant performance indicator.
14895
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014896meth(<method>) : method
14897 Returns a method.
14898
Willy Tarreau0f30d262014-11-24 16:02:05 +010014899nbproc : integer
14900 Returns an integer value corresponding to the number of processes that were
14901 started (it equals the global "nbproc" setting). This is useful for logging
14902 and debugging purposes.
14903
Willy Tarreau74ca5042013-06-11 23:12:07 +020014904nbsrv([<backend>]) : integer
14905 Returns an integer value corresponding to the number of usable servers of
14906 either the current backend or the named backend. This is mostly used with
14907 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010014908 switch to an alternate backend when the number of servers is too low to
14909 to handle some load. It is useful to report a failure when combined with
14910 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014911
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040014912prio_class : integer
14913 Returns the priority class of the current session for http mode or connection
14914 for tcp mode. The value will be that set by the last call to "http-request
14915 set-priority-class" or "tcp-request content set-priority-class".
14916
14917prio_offset : integer
14918 Returns the priority offset of the current session for http mode or
14919 connection for tcp mode. The value will be that set by the last call to
14920 "http-request set-priority-offset" or "tcp-request content
14921 set-priority-offset".
14922
Willy Tarreau0f30d262014-11-24 16:02:05 +010014923proc : integer
14924 Returns an integer value corresponding to the position of the process calling
14925 the function, between 1 and global.nbproc. This is useful for logging and
14926 debugging purposes.
14927
Willy Tarreau74ca5042013-06-11 23:12:07 +020014928queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014929 Returns the total number of queued connections of the designated backend,
14930 including all the connections in server queues. If no backend name is
14931 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014932 one. This is useful with ACLs or to pass statistics to backend servers. This
14933 can be used to take actions when queuing goes above a known level, generally
14934 indicating a surge of traffic or a massive slowdown on the servers. One
14935 possible action could be to reject new users but still accept old ones. See
14936 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
14937
Willy Tarreau84310e22014-02-14 11:59:04 +010014938rand([<range>]) : integer
14939 Returns a random integer value within a range of <range> possible values,
14940 starting at zero. If the range is not specified, it defaults to 2^32, which
14941 gives numbers between 0 and 4294967295. It can be useful to pass some values
14942 needed to take some routing decisions for example, or just for debugging
14943 purposes. This random must not be used for security purposes.
14944
Luca Schimweg77306662019-09-10 15:42:52 +020014945uuid([<version>]) : string
14946 Returns a UUID following the RFC4122 standard. If the version is not
14947 specified, a UUID version 4 (fully random) is returned.
14948 Currently, only version 4 is supported.
14949
Willy Tarreau74ca5042013-06-11 23:12:07 +020014950srv_conn([<backend>/]<server>) : integer
14951 Returns an integer value corresponding to the number of currently established
14952 connections on the designated server, possibly including the connection being
14953 evaluated. If <backend> is omitted, then the server is looked up in the
14954 current backend. It can be used to use a specific farm when one server is
14955 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014956 connections with it. See also the "fe_conn", "be_conn", "queue", and
14957 "srv_conn_free" fetch methods.
14958
14959srv_conn_free([<backend>/]<server>) : integer
14960 Returns an integer value corresponding to the number of available connections
14961 on the designated server, possibly including the connection being evaluated.
14962 The value does not include queue slots. If <backend> is omitted, then the
14963 server is looked up in the current backend. It can be used to use a specific
14964 farm when one server is full, or to inform the server about our view of the
14965 number of active connections with it. See also the "be_conn_free" and
14966 "srv_conn" fetch methods.
14967
14968 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
14969 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014970
14971srv_is_up([<backend>/]<server>) : boolean
14972 Returns true when the designated server is UP, and false when it is either
14973 DOWN or in maintenance mode. If <backend> is omitted, then the server is
14974 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010014975 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014976 availability). Another possible use which is more of a hack consists in
14977 using dummy servers as boolean variables that can be enabled or disabled from
14978 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
14979
Willy Tarreauff2b7af2017-10-13 11:46:26 +020014980srv_queue([<backend>/]<server>) : integer
14981 Returns an integer value corresponding to the number of connections currently
14982 pending in the designated server's queue. If <backend> is omitted, then the
14983 server is looked up in the current backend. It can sometimes be used together
14984 with the "use-server" directive to force to use a known faster server when it
14985 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
14986 fetch methods.
14987
Willy Tarreau74ca5042013-06-11 23:12:07 +020014988srv_sess_rate([<backend>/]<server>) : integer
14989 Returns an integer corresponding to the sessions creation rate on the
14990 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014991 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020014992 used with ACLs but can make sense with logs too. This is used to switch to an
14993 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010014994 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014995 overloading servers).
14996
14997 Example :
14998 # Redirect to a separate back
14999 acl srv1_full srv_sess_rate(be1/srv1) gt 50
15000 acl srv2_full srv_sess_rate(be1/srv2) gt 50
15001 use_backend be2 if srv1_full or srv2_full
15002
Willy Tarreau0f30d262014-11-24 16:02:05 +010015003stopping : boolean
15004 Returns TRUE if the process calling the function is currently stopping. This
15005 can be useful for logging, or for relaxing certain checks or helping close
15006 certain connections upon graceful shutdown.
15007
Thierry FOURNIERcc103292015-06-06 19:30:17 +020015008str(<string>) : string
15009 Returns a string.
15010
Willy Tarreau74ca5042013-06-11 23:12:07 +020015011table_avl([<table>]) : integer
15012 Returns the total number of available entries in the current proxy's
15013 stick-table or in the designated stick-table. See also table_cnt.
15014
15015table_cnt([<table>]) : integer
15016 Returns the total number of entries currently in use in the current proxy's
15017 stick-table or in the designated stick-table. See also src_conn_cnt and
15018 table_avl for other entry counting methods.
15019
Christopher Faulet34adb2a2017-11-21 21:45:38 +010015020thread : integer
15021 Returns an integer value corresponding to the position of the thread calling
15022 the function, between 0 and (global.nbthread-1). This is useful for logging
15023 and debugging purposes.
15024
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015025var(<var-name>) : undefined
15026 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010015027 sample fetch fails. The name of the variable starts with an indication
15028 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010015029 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010015030 "sess" : the variable is shared with the whole session
15031 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015032 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010015033 "req" : the variable is shared only during request processing,
15034 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015035 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010015036 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020015037
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200150387.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020015039----------------------------------
15040
15041The layer 4 usually describes just the transport layer which in haproxy is
15042closest to the connection, where no content is yet made available. The fetch
15043methods described here are usable as low as the "tcp-request connection" rule
15044sets unless they require some future information. Those generally include
15045TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015046the incoming connection. For retrieving a value from a sticky counters, the
15047counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020015048"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
15049used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
15050can be specified as the first integer argument when using the "sc_" prefix.
15051Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
15052table may be specified with the "sc*" form, in which case the currently
15053tracked key will be looked up into this alternate table instead of the table
15054currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015055
Jérôme Magnin35e53a62019-01-16 14:38:37 +010015056bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010015057 Returns the backend connection's HTTP major version encoding, which may be 1
15058 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
15059 encoding and not the version present in the request header.
15060
Willy Tarreau74ca5042013-06-11 23:12:07 +020015061be_id : integer
15062 Returns an integer containing the current backend's id. It can be used in
15063 frontends with responses to check which backend processed the request.
15064
Marcin Deranekd2471c22016-12-12 14:08:05 +010015065be_name : string
15066 Returns a string containing the current backend's name. It can be used in
15067 frontends with responses to check which backend processed the request.
15068
Willy Tarreau74ca5042013-06-11 23:12:07 +020015069dst : ip
15070 This is the destination IPv4 address of the connection on the client side,
15071 which is the address the client connected to. It can be useful when running
15072 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
15073 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010015074 RFC 4291. When the incoming connection passed through address translation or
15075 redirection involving connection tracking, the original destination address
15076 before the redirection will be reported. On Linux systems, the source and
15077 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
15078 is set, because a late response may reopen a timed out connection and switch
15079 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015080
15081dst_conn : integer
15082 Returns an integer value corresponding to the number of currently established
15083 connections on the same socket including the one being evaluated. It is
15084 normally used with ACLs but can as well be used to pass the information to
15085 servers in an HTTP header or in logs. It can be used to either return a sorry
15086 page before hard-blocking, or to use a specific backend to drain new requests
15087 when the socket is considered saturated. This offers the ability to assign
15088 different limits to different listening ports or addresses. See also the
15089 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015090
Willy Tarreau16e01562016-08-09 16:46:18 +020015091dst_is_local : boolean
15092 Returns true if the destination address of the incoming connection is local
15093 to the system, or false if the address doesn't exist on the system, meaning
15094 that it was intercepted in transparent mode. It can be useful to apply
15095 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010015096 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020015097 be delivered only on this address, or SSH access could be locally redirected.
15098 Please note that the check involves a few system calls, so it's better to do
15099 it only once per connection.
15100
Willy Tarreau74ca5042013-06-11 23:12:07 +020015101dst_port : integer
15102 Returns an integer value corresponding to the destination TCP port of the
15103 connection on the client side, which is the port the client connected to.
15104 This might be used when running in transparent mode, when assigning dynamic
15105 ports to some clients for a whole application session, to stick all users to
15106 a same server, or to pass the destination port information to a server using
15107 an HTTP header.
15108
Willy Tarreau60ca10a2017-08-18 15:26:54 +020015109fc_http_major : integer
15110 Reports the front connection's HTTP major version encoding, which may be 1
15111 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
15112 encoding and not on the version present in the request header.
15113
Emeric Brun4f603012017-01-05 15:11:44 +010015114fc_rcvd_proxy : boolean
15115 Returns true if the client initiated the connection with a PROXY protocol
15116 header.
15117
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020015118fc_rtt(<unit>) : integer
15119 Returns the Round Trip Time (RTT) measured by the kernel for the client
15120 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
15121 can be set to "ms" for milliseconds or "us" for microseconds. If the server
15122 connection is not established, if the connection is not TCP or if the
15123 operating system does not support TCP_INFO, for example Linux kernels before
15124 2.4, the sample fetch fails.
15125
15126fc_rttvar(<unit>) : integer
15127 Returns the Round Trip Time (RTT) variance measured by the kernel for the
15128 client connection. <unit> is facultative, by default the unit is milliseconds.
15129 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
15130 server connection is not established, if the connection is not TCP or if the
15131 operating system does not support TCP_INFO, for example Linux kernels before
15132 2.4, the sample fetch fails.
15133
Christopher Faulet297df182019-10-17 14:40:48 +020015134fc_unacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015135 Returns the unacked counter measured by the kernel for the client connection.
15136 If the server connection is not established, if the connection is not TCP or
15137 if the operating system does not support TCP_INFO, for example Linux kernels
15138 before 2.4, the sample fetch fails.
15139
Christopher Faulet297df182019-10-17 14:40:48 +020015140fc_sacked : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015141 Returns the sacked counter measured by the kernel for the client connection.
15142 If the server connection is not established, if the connection is not TCP or
15143 if the operating system does not support TCP_INFO, for example Linux kernels
15144 before 2.4, the sample fetch fails.
15145
Christopher Faulet297df182019-10-17 14:40:48 +020015146fc_retrans : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015147 Returns the retransmits counter measured by the kernel for the client
15148 connection. If the server connection is not established, if the connection is
15149 not TCP or if the operating system does not support TCP_INFO, for example
15150 Linux kernels before 2.4, the sample fetch fails.
15151
Christopher Faulet297df182019-10-17 14:40:48 +020015152fc_fackets : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015153 Returns the fack counter measured by the kernel for the client
15154 connection. If the server connection is not established, if the connection is
15155 not TCP or if the operating system does not support TCP_INFO, for example
15156 Linux kernels before 2.4, the sample fetch fails.
15157
Christopher Faulet297df182019-10-17 14:40:48 +020015158fc_lost : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015159 Returns the lost counter measured by the kernel for the client
15160 connection. If the server connection is not established, if the connection is
15161 not TCP or if the operating system does not support TCP_INFO, for example
15162 Linux kernels before 2.4, the sample fetch fails.
15163
Christopher Faulet297df182019-10-17 14:40:48 +020015164fc_reordering : integer
Joe Williams30fcd392016-08-10 07:06:44 -070015165 Returns the reordering counter measured by the kernel for the client
15166 connection. If the server connection is not established, if the connection is
15167 not TCP or if the operating system does not support TCP_INFO, for example
15168 Linux kernels before 2.4, the sample fetch fails.
15169
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020015170fe_defbe : string
15171 Returns a string containing the frontend's default backend name. It can be
15172 used in frontends to check which backend will handle requests by default.
15173
Willy Tarreau74ca5042013-06-11 23:12:07 +020015174fe_id : integer
15175 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010015176 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020015177 coming via a same frontend to the same server.
15178
Marcin Deranekd2471c22016-12-12 14:08:05 +010015179fe_name : string
15180 Returns a string containing the current frontend's name. It can be used in
15181 backends to check from which frontend it was called, or to stick all users
15182 coming via a same frontend to the same server.
15183
Cyril Bonté62ba8702014-04-22 23:52:25 +020015184sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015185sc0_bytes_in_rate([<table>]) : integer
15186sc1_bytes_in_rate([<table>]) : integer
15187sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015188 Returns the average client-to-server bytes rate from the currently tracked
15189 counters, measured in amount of bytes over the period configured in the
15190 table. See also src_bytes_in_rate.
15191
Cyril Bonté62ba8702014-04-22 23:52:25 +020015192sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015193sc0_bytes_out_rate([<table>]) : integer
15194sc1_bytes_out_rate([<table>]) : integer
15195sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015196 Returns the average server-to-client bytes rate from the currently tracked
15197 counters, measured in amount of bytes over the period configured in the
15198 table. See also src_bytes_out_rate.
15199
Cyril Bonté62ba8702014-04-22 23:52:25 +020015200sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015201sc0_clr_gpc0([<table>]) : integer
15202sc1_clr_gpc0([<table>]) : integer
15203sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020015204 Clears the first General Purpose Counter associated to the currently tracked
15205 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010015206 stored value is zero, so first invocation will always return zero. This is
15207 typically used as a second ACL in an expression in order to mark a connection
15208 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020015209
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015210 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020015211 # block if 5 consecutive requests continue to come faster than 10 sess
15212 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015213 acl abuse sc0_http_req_rate gt 10
15214 acl kill sc0_inc_gpc0 gt 5
15215 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020015216 tcp-request connection accept if !abuse save
15217 tcp-request connection reject if abuse kill
15218
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015219sc_clr_gpc1(<ctr>[,<table>]) : integer
15220sc0_clr_gpc1([<table>]) : integer
15221sc1_clr_gpc1([<table>]) : integer
15222sc2_clr_gpc1([<table>]) : integer
15223 Clears the second General Purpose Counter associated to the currently tracked
15224 counters, and returns its previous value. Before the first invocation, the
15225 stored value is zero, so first invocation will always return zero. This is
15226 typically used as a second ACL in an expression in order to mark a connection
15227 when a first ACL was verified.
15228
Cyril Bonté62ba8702014-04-22 23:52:25 +020015229sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015230sc0_conn_cnt([<table>]) : integer
15231sc1_conn_cnt([<table>]) : integer
15232sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015233 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020015234 counters. See also src_conn_cnt.
15235
Cyril Bonté62ba8702014-04-22 23:52:25 +020015236sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015237sc0_conn_cur([<table>]) : integer
15238sc1_conn_cur([<table>]) : integer
15239sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015240 Returns the current amount of concurrent connections tracking the same
15241 tracked counters. This number is automatically incremented when tracking
15242 begins and decremented when tracking stops. See also src_conn_cur.
15243
Cyril Bonté62ba8702014-04-22 23:52:25 +020015244sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015245sc0_conn_rate([<table>]) : integer
15246sc1_conn_rate([<table>]) : integer
15247sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015248 Returns the average connection rate from the currently tracked counters,
15249 measured in amount of connections over the period configured in the table.
15250 See also src_conn_rate.
15251
Cyril Bonté62ba8702014-04-22 23:52:25 +020015252sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015253sc0_get_gpc0([<table>]) : integer
15254sc1_get_gpc0([<table>]) : integer
15255sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015256 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015257 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015258
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015259sc_get_gpc1(<ctr>[,<table>]) : integer
15260sc0_get_gpc1([<table>]) : integer
15261sc1_get_gpc1([<table>]) : integer
15262sc2_get_gpc1([<table>]) : integer
15263 Returns the value of the second General Purpose Counter associated to the
15264 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
15265
Thierry FOURNIER236657b2015-08-19 08:25:14 +020015266sc_get_gpt0(<ctr>[,<table>]) : integer
15267sc0_get_gpt0([<table>]) : integer
15268sc1_get_gpt0([<table>]) : integer
15269sc2_get_gpt0([<table>]) : integer
15270 Returns the value of the first General Purpose Tag associated to the
15271 currently tracked counters. See also src_get_gpt0.
15272
Cyril Bonté62ba8702014-04-22 23:52:25 +020015273sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015274sc0_gpc0_rate([<table>]) : integer
15275sc1_gpc0_rate([<table>]) : integer
15276sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015277 Returns the average increment rate of the first General Purpose Counter
15278 associated to the currently tracked counters. It reports the frequency
15279 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015280 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
15281 that the "gpc0_rate" counter must be stored in the stick-table for a value to
15282 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020015283
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015284sc_gpc1_rate(<ctr>[,<table>]) : integer
15285sc0_gpc1_rate([<table>]) : integer
15286sc1_gpc1_rate([<table>]) : integer
15287sc2_gpc1_rate([<table>]) : integer
15288 Returns the average increment rate of the second General Purpose Counter
15289 associated to the currently tracked counters. It reports the frequency
15290 which the gpc1 counter was incremented over the configured period. See also
15291 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
15292 that the "gpc1_rate" counter must be stored in the stick-table for a value to
15293 be returned, as "gpc1" only holds the event count.
15294
Cyril Bonté62ba8702014-04-22 23:52:25 +020015295sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015296sc0_http_err_cnt([<table>]) : integer
15297sc1_http_err_cnt([<table>]) : integer
15298sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015299 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020015300 counters. This includes the both request errors and 4xx error responses.
15301 See also src_http_err_cnt.
15302
Cyril Bonté62ba8702014-04-22 23:52:25 +020015303sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015304sc0_http_err_rate([<table>]) : integer
15305sc1_http_err_rate([<table>]) : integer
15306sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015307 Returns the average rate of HTTP errors from the currently tracked counters,
15308 measured in amount of errors over the period configured in the table. This
15309 includes the both request errors and 4xx error responses. See also
15310 src_http_err_rate.
15311
Cyril Bonté62ba8702014-04-22 23:52:25 +020015312sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015313sc0_http_req_cnt([<table>]) : integer
15314sc1_http_req_cnt([<table>]) : integer
15315sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015316 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020015317 counters. This includes every started request, valid or not. See also
15318 src_http_req_cnt.
15319
Cyril Bonté62ba8702014-04-22 23:52:25 +020015320sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015321sc0_http_req_rate([<table>]) : integer
15322sc1_http_req_rate([<table>]) : integer
15323sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015324 Returns the average rate of HTTP requests from the currently tracked
15325 counters, measured in amount of requests over the period configured in
15326 the table. This includes every started request, valid or not. See also
15327 src_http_req_rate.
15328
Cyril Bonté62ba8702014-04-22 23:52:25 +020015329sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015330sc0_inc_gpc0([<table>]) : integer
15331sc1_inc_gpc0([<table>]) : integer
15332sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015333 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010015334 tracked counters, and returns its new value. Before the first invocation,
15335 the stored value is zero, so first invocation will increase it to 1 and will
15336 return 1. This is typically used as a second ACL in an expression in order
15337 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020015338
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015339 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015340 acl abuse sc0_http_req_rate gt 10
15341 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020015342 tcp-request connection reject if abuse kill
15343
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015344sc_inc_gpc1(<ctr>[,<table>]) : integer
15345sc0_inc_gpc1([<table>]) : integer
15346sc1_inc_gpc1([<table>]) : integer
15347sc2_inc_gpc1([<table>]) : integer
15348 Increments the second General Purpose Counter associated to the currently
15349 tracked counters, and returns its new value. Before the first invocation,
15350 the stored value is zero, so first invocation will increase it to 1 and will
15351 return 1. This is typically used as a second ACL in an expression in order
15352 to mark a connection when a first ACL was verified.
15353
Cyril Bonté62ba8702014-04-22 23:52:25 +020015354sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015355sc0_kbytes_in([<table>]) : integer
15356sc1_kbytes_in([<table>]) : integer
15357sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015358 Returns the total amount of client-to-server data from the currently tracked
15359 counters, measured in kilobytes. The test is currently performed on 32-bit
15360 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020015361
Cyril Bonté62ba8702014-04-22 23:52:25 +020015362sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015363sc0_kbytes_out([<table>]) : integer
15364sc1_kbytes_out([<table>]) : integer
15365sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015366 Returns the total amount of server-to-client data from the currently tracked
15367 counters, measured in kilobytes. The test is currently performed on 32-bit
15368 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020015369
Cyril Bonté62ba8702014-04-22 23:52:25 +020015370sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015371sc0_sess_cnt([<table>]) : integer
15372sc1_sess_cnt([<table>]) : integer
15373sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015374 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020015375 into sessions, which means that they were accepted by a "tcp-request
15376 connection" rule, from the currently tracked counters. A backend may count
15377 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040015378 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020015379 with the client. See also src_sess_cnt.
15380
Cyril Bonté62ba8702014-04-22 23:52:25 +020015381sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015382sc0_sess_rate([<table>]) : integer
15383sc1_sess_rate([<table>]) : integer
15384sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020015385 Returns the average session rate from the currently tracked counters,
15386 measured in amount of sessions over the period configured in the table. A
15387 session is a connection that got past the early "tcp-request connection"
15388 rules. A backend may count more sessions than connections because each
15389 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040015390 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020015391
Cyril Bonté62ba8702014-04-22 23:52:25 +020015392sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020015393sc0_tracked([<table>]) : boolean
15394sc1_tracked([<table>]) : boolean
15395sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020015396 Returns true if the designated session counter is currently being tracked by
15397 the current session. This can be useful when deciding whether or not we want
15398 to set some values in a header passed to the server.
15399
Cyril Bonté62ba8702014-04-22 23:52:25 +020015400sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020015401sc0_trackers([<table>]) : integer
15402sc1_trackers([<table>]) : integer
15403sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010015404 Returns the current amount of concurrent connections tracking the same
15405 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015406 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010015407 that it does not rely on any stored information but on the table's reference
15408 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015409 may sometimes be more suited for layer7 tracking. It can be used to tell a
15410 server how many concurrent connections there are from a given address for
15411 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010015412
Willy Tarreau74ca5042013-06-11 23:12:07 +020015413so_id : integer
15414 Returns an integer containing the current listening socket's id. It is useful
15415 in frontends involving many "bind" lines, or to stick all users coming via a
15416 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015417
Willy Tarreau74ca5042013-06-11 23:12:07 +020015418src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010015419 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020015420 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
15421 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
15422 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010015423 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
15424 directive is used, it can be the address of a client behind another
15425 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010015426 "tcp-request connection" which sees the real address. When the incoming
15427 connection passed through address translation or redirection involving
15428 connection tracking, the original destination address before the redirection
15429 will be reported. On Linux systems, the source and destination may seldom
15430 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
15431 response may reopen a timed out connection and switch what is believed to be
15432 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015433
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010015434 Example:
15435 # add an HTTP header in requests with the originating address' country
15436 http-request set-header X-Country %[src,map_ip(geoip.lst)]
15437
Willy Tarreau74ca5042013-06-11 23:12:07 +020015438src_bytes_in_rate([<table>]) : integer
15439 Returns the average bytes rate from the incoming connection's source address
15440 in the current proxy's stick-table or in the designated stick-table, measured
15441 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015442 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015443
Willy Tarreau74ca5042013-06-11 23:12:07 +020015444src_bytes_out_rate([<table>]) : integer
15445 Returns the average bytes rate to the incoming connection's source address in
15446 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015447 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015448 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015449
Willy Tarreau74ca5042013-06-11 23:12:07 +020015450src_clr_gpc0([<table>]) : integer
15451 Clears the first General Purpose Counter associated to the incoming
15452 connection's source address in the current proxy's stick-table or in the
15453 designated stick-table, and returns its previous value. If the address is not
15454 found, an entry is created and 0 is returned. This is typically used as a
15455 second ACL in an expression in order to mark a connection when a first ACL
15456 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020015457
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015458 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020015459 # block if 5 consecutive requests continue to come faster than 10 sess
15460 # per second, and reset the counter as soon as the traffic slows down.
15461 acl abuse src_http_req_rate gt 10
15462 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010015463 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020015464 tcp-request connection accept if !abuse save
15465 tcp-request connection reject if abuse kill
15466
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015467src_clr_gpc1([<table>]) : integer
15468 Clears the second General Purpose Counter associated to the incoming
15469 connection's source address in the current proxy's stick-table or in the
15470 designated stick-table, and returns its previous value. If the address is not
15471 found, an entry is created and 0 is returned. This is typically used as a
15472 second ACL in an expression in order to mark a connection when a first ACL
15473 was verified.
15474
Willy Tarreau74ca5042013-06-11 23:12:07 +020015475src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015476 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020015477 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015478 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015479 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015480
Willy Tarreau74ca5042013-06-11 23:12:07 +020015481src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020015482 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015483 current incoming connection's source address in the current proxy's
15484 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015485 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015486
Willy Tarreau74ca5042013-06-11 23:12:07 +020015487src_conn_rate([<table>]) : integer
15488 Returns the average connection rate from the incoming connection's source
15489 address in the current proxy's stick-table or in the designated stick-table,
15490 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015491 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015492
Willy Tarreau74ca5042013-06-11 23:12:07 +020015493src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020015494 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015495 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020015496 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015497 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015498
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015499src_get_gpc1([<table>]) : integer
15500 Returns the value of the second General Purpose Counter associated to the
15501 incoming connection's source address in the current proxy's stick-table or in
15502 the designated stick-table. If the address is not found, zero is returned.
15503 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
15504
Thierry FOURNIER236657b2015-08-19 08:25:14 +020015505src_get_gpt0([<table>]) : integer
15506 Returns the value of the first General Purpose Tag associated to the
15507 incoming connection's source address in the current proxy's stick-table or in
15508 the designated stick-table. If the address is not found, zero is returned.
15509 See also sc/sc0/sc1/sc2_get_gpt0.
15510
Willy Tarreau74ca5042013-06-11 23:12:07 +020015511src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015512 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020015513 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015514 stick-table or in the designated stick-table. It reports the frequency
15515 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015516 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
15517 that the "gpc0_rate" counter must be stored in the stick-table for a value to
15518 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020015519
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015520src_gpc1_rate([<table>]) : integer
15521 Returns the average increment rate of the second General Purpose Counter
15522 associated to the incoming connection's source address in the current proxy's
15523 stick-table or in the designated stick-table. It reports the frequency
15524 which the gpc1 counter was incremented over the configured period. See also
15525 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
15526 that the "gpc1_rate" counter must be stored in the stick-table for a value to
15527 be returned, as "gpc1" only holds the event count.
15528
Willy Tarreau74ca5042013-06-11 23:12:07 +020015529src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015530 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020015531 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020015532 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015533 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020015534 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015535
Willy Tarreau74ca5042013-06-11 23:12:07 +020015536src_http_err_rate([<table>]) : integer
15537 Returns the average rate of HTTP errors from the incoming connection's source
15538 address in the current proxy's stick-table or in the designated stick-table,
15539 measured in amount of errors over the period configured in the table. This
15540 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015541 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015542
Willy Tarreau74ca5042013-06-11 23:12:07 +020015543src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015544 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020015545 source address in the current proxy's stick-table or in the designated stick-
15546 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015547 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015548
Willy Tarreau74ca5042013-06-11 23:12:07 +020015549src_http_req_rate([<table>]) : integer
15550 Returns the average rate of HTTP requests from the incoming connection's
15551 source address in the current proxy's stick-table or in the designated stick-
15552 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020015553 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015554 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015555
Willy Tarreau74ca5042013-06-11 23:12:07 +020015556src_inc_gpc0([<table>]) : integer
15557 Increments the first General Purpose Counter associated to the incoming
15558 connection's source address in the current proxy's stick-table or in the
15559 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015560 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015561 This is typically used as a second ACL in an expression in order to mark a
15562 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020015563
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015564 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020015565 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010015566 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020015567 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020015568
Frédéric Lécaille6778b272018-01-29 15:22:53 +010015569src_inc_gpc1([<table>]) : integer
15570 Increments the second General Purpose Counter associated to the incoming
15571 connection's source address in the current proxy's stick-table or in the
15572 designated stick-table, and returns its new value. If the address is not
15573 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
15574 This is typically used as a second ACL in an expression in order to mark a
15575 connection when a first ACL was verified.
15576
Willy Tarreau16e01562016-08-09 16:46:18 +020015577src_is_local : boolean
15578 Returns true if the source address of the incoming connection is local to the
15579 system, or false if the address doesn't exist on the system, meaning that it
15580 comes from a remote machine. Note that UNIX addresses are considered local.
15581 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015582 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020015583 note that the check involves a few system calls, so it's better to do it only
15584 once per connection.
15585
Willy Tarreau74ca5042013-06-11 23:12:07 +020015586src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015587 Returns the total amount of data received from the incoming connection's
15588 source address in the current proxy's stick-table or in the designated
15589 stick-table, measured in kilobytes. If the address is not found, zero is
15590 returned. The test is currently performed on 32-bit integers, which limits
15591 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015592
Willy Tarreau74ca5042013-06-11 23:12:07 +020015593src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015594 Returns the total amount of data sent to the incoming connection's source
15595 address in the current proxy's stick-table or in the designated stick-table,
15596 measured in kilobytes. If the address is not found, zero is returned. The
15597 test is currently performed on 32-bit integers, which limits values to 4
15598 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015599
Willy Tarreau74ca5042013-06-11 23:12:07 +020015600src_port : integer
15601 Returns an integer value corresponding to the TCP source port of the
15602 connection on the client side, which is the port the client connected from.
15603 Usage of this function is very limited as modern protocols do not care much
15604 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010015605
Willy Tarreau74ca5042013-06-11 23:12:07 +020015606src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015607 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020015608 connection's source IPv4 address in the current proxy's stick-table or in the
15609 designated stick-table, that were transformed into sessions, which means that
15610 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015611 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015612
Willy Tarreau74ca5042013-06-11 23:12:07 +020015613src_sess_rate([<table>]) : integer
15614 Returns the average session rate from the incoming connection's source
15615 address in the current proxy's stick-table or in the designated stick-table,
15616 measured in amount of sessions over the period configured in the table. A
15617 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015618 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015619
Willy Tarreau74ca5042013-06-11 23:12:07 +020015620src_updt_conn_cnt([<table>]) : integer
15621 Creates or updates the entry associated to the incoming connection's source
15622 address in the current proxy's stick-table or in the designated stick-table.
15623 This table must be configured to store the "conn_cnt" data type, otherwise
15624 the match will be ignored. The current count is incremented by one, and the
15625 expiration timer refreshed. The updated count is returned, so this match
15626 can't return zero. This was used to reject service abusers based on their
15627 source address. Note: it is recommended to use the more complete "track-sc*"
15628 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015629
15630 Example :
15631 # This frontend limits incoming SSH connections to 3 per 10 second for
15632 # each source address, and rejects excess connections until a 10 second
15633 # silence is observed. At most 20 addresses are tracked.
15634 listen ssh
15635 bind :22
15636 mode tcp
15637 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020015638 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020015639 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015640 server local 127.0.0.1:22
15641
Willy Tarreau74ca5042013-06-11 23:12:07 +020015642srv_id : integer
15643 Returns an integer containing the server's id when processing the response.
15644 While it's almost only used with ACLs, it may be used for logging or
15645 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020015646
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200156477.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020015648----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020015649
Willy Tarreau74ca5042013-06-11 23:12:07 +020015650The layer 5 usually describes just the session layer which in haproxy is
15651closest to the session once all the connection handshakes are finished, but
15652when no content is yet made available. The fetch methods described here are
15653usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015654future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020015655
Ben Shillitof25e8e52016-12-02 14:25:37 +00001565651d.all(<prop>[,<prop>*]) : string
15657 Returns values for the properties requested as a string, where values are
15658 separated by the delimiter specified with "51degrees-property-separator".
15659 The device is identified using all the important HTTP headers from the
15660 request. The function can be passed up to five property names, and if a
15661 property name can't be found, the value "NoData" is returned.
15662
15663 Example :
15664 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
15665 # containing the three properties requested using all relevant headers from
15666 # the request.
15667 frontend http-in
15668 bind *:8081
15669 default_backend servers
15670 http-request set-header X-51D-DeviceTypeMobileTablet \
15671 %[51d.all(DeviceType,IsMobile,IsTablet)]
15672
Emeric Brun645ae792014-04-30 14:21:06 +020015673ssl_bc : boolean
15674 Returns true when the back connection was made via an SSL/TLS transport
15675 layer and is locally deciphered. This means the outgoing connection was made
15676 other a server with the "ssl" option.
15677
15678ssl_bc_alg_keysize : integer
15679 Returns the symmetric cipher key size supported in bits when the outgoing
15680 connection was made over an SSL/TLS transport layer.
15681
Olivier Houchard6b77f492018-11-22 18:18:29 +010015682ssl_bc_alpn : string
15683 This extracts the Application Layer Protocol Negotiation field from an
15684 outgoing connection made via a TLS transport layer.
Michael Prokop4438c602019-05-24 10:25:45 +020015685 The result is a string containing the protocol name negotiated with the
Olivier Houchard6b77f492018-11-22 18:18:29 +010015686 server. The SSL library must have been built with support for TLS
15687 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15688 not advertised unless the "alpn" keyword on the "server" line specifies a
15689 protocol list. Also, nothing forces the server to pick a protocol from this
15690 list, any other one may be requested. The TLS ALPN extension is meant to
15691 replace the TLS NPN extension. See also "ssl_bc_npn".
15692
Emeric Brun645ae792014-04-30 14:21:06 +020015693ssl_bc_cipher : string
15694 Returns the name of the used cipher when the outgoing connection was made
15695 over an SSL/TLS transport layer.
15696
Patrick Hemmer65674662019-06-04 08:13:03 -040015697ssl_bc_client_random : binary
15698 Returns the client random of the back connection when the incoming connection
15699 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15700 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15701
Emeric Brun74f7ffa2018-02-19 16:14:12 +010015702ssl_bc_is_resumed : boolean
15703 Returns true when the back connection was made over an SSL/TLS transport
15704 layer and the newly created SSL session was resumed using a cached
15705 session or a TLS ticket.
15706
Olivier Houchard6b77f492018-11-22 18:18:29 +010015707ssl_bc_npn : string
15708 This extracts the Next Protocol Negotiation field from an outgoing connection
15709 made via a TLS transport layer. The result is a string containing the
Michael Prokop4438c602019-05-24 10:25:45 +020015710 protocol name negotiated with the server . The SSL library must have been
Olivier Houchard6b77f492018-11-22 18:18:29 +010015711 built with support for TLS extensions enabled (check haproxy -vv). Note that
15712 the TLS NPN extension is not advertised unless the "npn" keyword on the
15713 "server" line specifies a protocol list. Also, nothing forces the server to
15714 pick a protocol from this list, any other one may be used. Please note that
15715 the TLS NPN extension was replaced with ALPN.
15716
Emeric Brun645ae792014-04-30 14:21:06 +020015717ssl_bc_protocol : string
15718 Returns the name of the used protocol when the outgoing connection was made
15719 over an SSL/TLS transport layer.
15720
Emeric Brunb73a9b02014-04-30 18:49:19 +020015721ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020015722 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015723 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15724 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020015725
Patrick Hemmer65674662019-06-04 08:13:03 -040015726ssl_bc_server_random : binary
15727 Returns the server random of the back connection when the incoming connection
15728 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15729 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15730
Emeric Brun645ae792014-04-30 14:21:06 +020015731ssl_bc_session_id : binary
15732 Returns the SSL ID of the back connection when the outgoing connection was
15733 made over an SSL/TLS transport layer. It is useful to log if we want to know
15734 if session was reused or not.
15735
Patrick Hemmere0275472018-04-28 19:15:51 -040015736ssl_bc_session_key : binary
15737 Returns the SSL session master key of the back connection when the outgoing
15738 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15739 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15740 BoringSSL.
15741
Emeric Brun645ae792014-04-30 14:21:06 +020015742ssl_bc_use_keysize : integer
15743 Returns the symmetric cipher key size used in bits when the outgoing
15744 connection was made over an SSL/TLS transport layer.
15745
Willy Tarreau74ca5042013-06-11 23:12:07 +020015746ssl_c_ca_err : integer
15747 When the incoming connection was made over an SSL/TLS transport layer,
15748 returns the ID of the first error detected during verification of the client
15749 certificate at depth > 0, or 0 if no error was encountered during this
15750 verification process. Please refer to your SSL library's documentation to
15751 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020015752
Willy Tarreau74ca5042013-06-11 23:12:07 +020015753ssl_c_ca_err_depth : integer
15754 When the incoming connection was made over an SSL/TLS transport layer,
15755 returns the depth in the CA chain of the first error detected during the
15756 verification of the client certificate. If no error is encountered, 0 is
15757 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015758
Emeric Brun43e79582014-10-29 19:03:26 +010015759ssl_c_der : binary
15760 Returns the DER formatted certificate presented by the client when the
15761 incoming connection was made over an SSL/TLS transport layer. When used for
15762 an ACL, the value(s) to match against can be passed in hexadecimal form.
15763
Willy Tarreau74ca5042013-06-11 23:12:07 +020015764ssl_c_err : integer
15765 When the incoming connection was made over an SSL/TLS transport layer,
15766 returns the ID of the first error detected during verification at depth 0, or
15767 0 if no error was encountered during this verification process. Please refer
15768 to your SSL library's documentation to find the exhaustive list of error
15769 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020015770
Willy Tarreau74ca5042013-06-11 23:12:07 +020015771ssl_c_i_dn([<entry>[,<occ>]]) : string
15772 When the incoming connection was made over an SSL/TLS transport layer,
15773 returns the full distinguished name of the issuer of the certificate
15774 presented by the client when no <entry> is specified, or the value of the
15775 first given entry found from the beginning of the DN. If a positive/negative
15776 occurrence number is specified as the optional second argument, it returns
15777 the value of the nth given entry value from the beginning/end of the DN.
15778 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
15779 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020015780
Willy Tarreau74ca5042013-06-11 23:12:07 +020015781ssl_c_key_alg : string
15782 Returns the name of the algorithm used to generate the key of the certificate
15783 presented by the client when the incoming connection was made over an SSL/TLS
15784 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020015785
Willy Tarreau74ca5042013-06-11 23:12:07 +020015786ssl_c_notafter : string
15787 Returns the end date presented by the client as a formatted string
15788 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15789 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020015790
Willy Tarreau74ca5042013-06-11 23:12:07 +020015791ssl_c_notbefore : string
15792 Returns the start date presented by the client as a formatted string
15793 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15794 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015795
Willy Tarreau74ca5042013-06-11 23:12:07 +020015796ssl_c_s_dn([<entry>[,<occ>]]) : string
15797 When the incoming connection was made over an SSL/TLS transport layer,
15798 returns the full distinguished name of the subject of the certificate
15799 presented by the client when no <entry> is specified, or the value of the
15800 first given entry found from the beginning of the DN. If a positive/negative
15801 occurrence number is specified as the optional second argument, it returns
15802 the value of the nth given entry value from the beginning/end of the DN.
15803 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
15804 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015805
Willy Tarreau74ca5042013-06-11 23:12:07 +020015806ssl_c_serial : binary
15807 Returns the serial of the certificate presented by the client when the
15808 incoming connection was made over an SSL/TLS transport layer. When used for
15809 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015810
Willy Tarreau74ca5042013-06-11 23:12:07 +020015811ssl_c_sha1 : binary
15812 Returns the SHA-1 fingerprint of the certificate presented by the client when
15813 the incoming connection was made over an SSL/TLS transport layer. This can be
15814 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015815 Note that the output is binary, so if you want to pass that signature to the
15816 server, you need to encode it in hex or base64, such as in the example below:
15817
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015818 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015819 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020015820
Willy Tarreau74ca5042013-06-11 23:12:07 +020015821ssl_c_sig_alg : string
15822 Returns the name of the algorithm used to sign the certificate presented by
15823 the client when the incoming connection was made over an SSL/TLS transport
15824 layer.
Emeric Brun87855892012-10-17 17:39:35 +020015825
Willy Tarreau74ca5042013-06-11 23:12:07 +020015826ssl_c_used : boolean
15827 Returns true if current SSL session uses a client certificate even if current
15828 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020015829
Willy Tarreau74ca5042013-06-11 23:12:07 +020015830ssl_c_verify : integer
15831 Returns the verify result error ID when the incoming connection was made over
15832 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
15833 refer to your SSL library's documentation for an exhaustive list of error
15834 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020015835
Willy Tarreau74ca5042013-06-11 23:12:07 +020015836ssl_c_version : integer
15837 Returns the version of the certificate presented by the client when the
15838 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020015839
Emeric Brun43e79582014-10-29 19:03:26 +010015840ssl_f_der : binary
15841 Returns the DER formatted certificate presented by the frontend when the
15842 incoming connection was made over an SSL/TLS transport layer. When used for
15843 an ACL, the value(s) to match against can be passed in hexadecimal form.
15844
Willy Tarreau74ca5042013-06-11 23:12:07 +020015845ssl_f_i_dn([<entry>[,<occ>]]) : string
15846 When the incoming connection was made over an SSL/TLS transport layer,
15847 returns the full distinguished name of the issuer of the certificate
15848 presented by the frontend when no <entry> is specified, or the value of the
15849 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020015850 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020015851 the value of the nth given entry value from the beginning/end of the DN.
15852 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
15853 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020015854
Willy Tarreau74ca5042013-06-11 23:12:07 +020015855ssl_f_key_alg : string
15856 Returns the name of the algorithm used to generate the key of the certificate
15857 presented by the frontend when the incoming connection was made over an
15858 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015859
Willy Tarreau74ca5042013-06-11 23:12:07 +020015860ssl_f_notafter : string
15861 Returns the end date presented by the frontend as a formatted string
15862 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15863 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015864
Willy Tarreau74ca5042013-06-11 23:12:07 +020015865ssl_f_notbefore : string
15866 Returns the start date presented by the frontend as a formatted string
15867 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15868 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020015869
Willy Tarreau74ca5042013-06-11 23:12:07 +020015870ssl_f_s_dn([<entry>[,<occ>]]) : string
15871 When the incoming connection was made over an SSL/TLS transport layer,
15872 returns the full distinguished name of the subject of the certificate
15873 presented by the frontend when no <entry> is specified, or the value of the
15874 first given entry found from the beginning of the DN. If a positive/negative
15875 occurrence number is specified as the optional second argument, it returns
15876 the value of the nth given entry value from the beginning/end of the DN.
15877 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
15878 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020015879
Willy Tarreau74ca5042013-06-11 23:12:07 +020015880ssl_f_serial : binary
15881 Returns the serial of the certificate presented by the frontend when the
15882 incoming connection was made over an SSL/TLS transport layer. When used for
15883 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020015884
Emeric Brun55f4fa82014-04-30 17:11:25 +020015885ssl_f_sha1 : binary
15886 Returns the SHA-1 fingerprint of the certificate presented by the frontend
15887 when the incoming connection was made over an SSL/TLS transport layer. This
15888 can be used to know which certificate was chosen using SNI.
15889
Willy Tarreau74ca5042013-06-11 23:12:07 +020015890ssl_f_sig_alg : string
15891 Returns the name of the algorithm used to sign the certificate presented by
15892 the frontend when the incoming connection was made over an SSL/TLS transport
15893 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015894
Willy Tarreau74ca5042013-06-11 23:12:07 +020015895ssl_f_version : integer
15896 Returns the version of the certificate presented by the frontend when the
15897 incoming connection was made over an SSL/TLS transport layer.
15898
15899ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015900 Returns true when the front connection was made via an SSL/TLS transport
15901 layer and is locally deciphered. This means it has matched a socket declared
15902 with a "bind" line having the "ssl" option.
15903
Willy Tarreau74ca5042013-06-11 23:12:07 +020015904 Example :
15905 # This passes "X-Proto: https" to servers when client connects over SSL
15906 listen http-https
15907 bind :80
15908 bind :443 ssl crt /etc/haproxy.pem
15909 http-request add-header X-Proto https if { ssl_fc }
15910
15911ssl_fc_alg_keysize : integer
15912 Returns the symmetric cipher key size supported in bits when the incoming
15913 connection was made over an SSL/TLS transport layer.
15914
15915ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015916 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015917 incoming connection made via a TLS transport layer and locally deciphered by
15918 haproxy. The result is a string containing the protocol name advertised by
15919 the client. The SSL library must have been built with support for TLS
15920 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15921 not advertised unless the "alpn" keyword on the "bind" line specifies a
15922 protocol list. Also, nothing forces the client to pick a protocol from this
15923 list, any other one may be requested. The TLS ALPN extension is meant to
15924 replace the TLS NPN extension. See also "ssl_fc_npn".
15925
Willy Tarreau74ca5042013-06-11 23:12:07 +020015926ssl_fc_cipher : string
15927 Returns the name of the used cipher when the incoming connection was made
15928 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020015929
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015930ssl_fc_cipherlist_bin : binary
15931 Returns the binary form of the client hello cipher list. The maximum returned
15932 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015933 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015934
15935ssl_fc_cipherlist_hex : string
15936 Returns the binary form of the client hello cipher list encoded as
15937 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015938 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015939
15940ssl_fc_cipherlist_str : string
15941 Returns the decoded text form of the client hello cipher list. The maximum
15942 number of ciphers returned is according with the value of
15943 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010015944 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020015945 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015946
15947ssl_fc_cipherlist_xxh : integer
15948 Returns a xxh64 of the cipher list. This hash can be return only is the value
15949 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015950 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015951
Patrick Hemmer65674662019-06-04 08:13:03 -040015952ssl_fc_client_random : binary
15953 Returns the client random of the front connection when the incoming connection
15954 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
15955 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
15956
Willy Tarreau74ca5042013-06-11 23:12:07 +020015957ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015958 Returns true if a client certificate is present in an incoming connection over
15959 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010015960 Note: on SSL session resumption with Session ID or TLS ticket, client
15961 certificate is not present in the current connection but may be retrieved
15962 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
15963 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015964
Olivier Houchardccaa7de2017-10-02 11:51:03 +020015965ssl_fc_has_early : boolean
15966 Returns true if early data were sent, and the handshake didn't happen yet. As
15967 it has security implications, it is useful to be able to refuse those, or
15968 wait until the handshake happened.
15969
Willy Tarreau74ca5042013-06-11 23:12:07 +020015970ssl_fc_has_sni : boolean
15971 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015972 in an incoming connection was made over an SSL/TLS transport layer. Returns
15973 true when the incoming connection presents a TLS SNI field. This requires
John Roesler27c754c2019-07-10 15:45:51 -050015974 that the SSL library is built with support for TLS extensions enabled (check
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015975 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020015976
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020015977ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015978 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010015979 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
15980 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015981
Willy Tarreau74ca5042013-06-11 23:12:07 +020015982ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015983 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020015984 made via a TLS transport layer and locally deciphered by haproxy. The result
15985 is a string containing the protocol name advertised by the client. The SSL
15986 library must have been built with support for TLS extensions enabled (check
15987 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
15988 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
15989 forces the client to pick a protocol from this list, any other one may be
15990 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020015991
Willy Tarreau74ca5042013-06-11 23:12:07 +020015992ssl_fc_protocol : string
15993 Returns the name of the used protocol when the incoming connection was made
15994 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020015995
Emeric Brunb73a9b02014-04-30 18:49:19 +020015996ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040015997 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015998 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15999 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040016000
Patrick Hemmer65674662019-06-04 08:13:03 -040016001ssl_fc_server_random : binary
16002 Returns the server random of the front connection when the incoming connection
16003 was made over an SSL/TLS transport layer. It is useful to to decrypt traffic
16004 sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or BoringSSL.
16005
Willy Tarreau74ca5042013-06-11 23:12:07 +020016006ssl_fc_session_id : binary
16007 Returns the SSL ID of the front connection when the incoming connection was
16008 made over an SSL/TLS transport layer. It is useful to stick a given client to
16009 a server. It is important to note that some browsers refresh their session ID
16010 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020016011
Patrick Hemmere0275472018-04-28 19:15:51 -040016012ssl_fc_session_key : binary
16013 Returns the SSL session master key of the front connection when the incoming
16014 connection was made over an SSL/TLS transport layer. It is useful to decrypt
16015 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
16016 BoringSSL.
16017
16018
Willy Tarreau74ca5042013-06-11 23:12:07 +020016019ssl_fc_sni : string
16020 This extracts the Server Name Indication TLS extension (SNI) field from an
16021 incoming connection made via an SSL/TLS transport layer and locally
16022 deciphered by haproxy. The result (when present) typically is a string
16023 matching the HTTPS host name (253 chars or less). The SSL library must have
16024 been built with support for TLS extensions enabled (check haproxy -vv).
16025
16026 This fetch is different from "req_ssl_sni" above in that it applies to the
16027 connection being deciphered by haproxy and not to SSL contents being blindly
16028 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
John Roesler27c754c2019-07-10 15:45:51 -050016029 requires that the SSL library is built with support for TLS extensions
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020016030 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020016031
Willy Tarreau74ca5042013-06-11 23:12:07 +020016032 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020016033 ssl_fc_sni_end : suffix match
16034 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020016035
Willy Tarreau74ca5042013-06-11 23:12:07 +020016036ssl_fc_use_keysize : integer
16037 Returns the symmetric cipher key size used in bits when the incoming
16038 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020016039
Willy Tarreaub6fb4202008-07-20 11:18:28 +020016040
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200160417.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020016042------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020016043
Willy Tarreau74ca5042013-06-11 23:12:07 +020016044Fetching samples from buffer contents is a bit different from the previous
16045sample fetches above because the sampled data are ephemeral. These data can
16046only be used when they're available and will be lost when they're forwarded.
16047For this reason, samples fetched from buffer contents during a request cannot
16048be used in a response for example. Even while the data are being fetched, they
16049can change. Sometimes it is necessary to set some delays or combine multiple
16050sample fetch methods to ensure that the expected data are complete and usable,
16051for example through TCP request content inspection. Please see the "tcp-request
16052content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020016053
Willy Tarreau74ca5042013-06-11 23:12:07 +020016054payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010016055 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016056 "stick on", "stick match"), and for "res.payload" when used in the context of
16057 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010016058
Willy Tarreau74ca5042013-06-11 23:12:07 +020016059payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
16060 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010016061 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020016062 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010016063
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020016064req.hdrs : string
16065 Returns the current request headers as string including the last empty line
16066 separating headers from the request body. The last empty line can be used to
16067 detect a truncated header block. This sample fetch is useful for some SPOE
16068 headers analyzers and for advanced logging.
16069
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020016070req.hdrs_bin : binary
16071 Returns the current request headers contained in preparsed binary form. This
16072 is useful for offloading some processing with SPOE. Each string is described
16073 by a length followed by the number of bytes indicated in the length. The
16074 length is represented using the variable integer encoding detailed in the
16075 SPOE documentation. The end of the list is marked by a couple of empty header
16076 names and values (length of 0 for both).
16077
16078 *(<str:header-name><str:header-value>)<empty string><empty string>
16079
16080 int: refer to the SPOE documentation for the encoding
16081 str: <int:length><bytes>
16082
Willy Tarreau74ca5042013-06-11 23:12:07 +020016083req.len : integer
16084req_len : integer (deprecated)
16085 Returns an integer value corresponding to the number of bytes present in the
16086 request buffer. This is mostly used in ACL. It is important to understand
16087 that this test does not return false as long as the buffer is changing. This
16088 means that a check with equality to zero will almost always immediately match
16089 at the beginning of the session, while a test for more data will wait for
16090 that data to come in and return false only when haproxy is certain that no
16091 more data will come in. This test was designed to be used with TCP request
16092 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016093
Willy Tarreau74ca5042013-06-11 23:12:07 +020016094req.payload(<offset>,<length>) : binary
16095 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020016096 in the request buffer. As a special case, if the <length> argument is zero,
16097 the the whole buffer from <offset> to the end is extracted. This can be used
16098 with ACLs in order to check for the presence of some content in a buffer at
16099 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016100
Willy Tarreau74ca5042013-06-11 23:12:07 +020016101 ACL alternatives :
16102 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016103
Willy Tarreau74ca5042013-06-11 23:12:07 +020016104req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
16105 This extracts a binary block whose size is specified at <offset1> for <length>
16106 bytes, and which starts at <offset2> if specified or just after the length in
16107 the request buffer. The <offset2> parameter also supports relative offsets if
16108 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016109
Willy Tarreau74ca5042013-06-11 23:12:07 +020016110 ACL alternatives :
16111 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016112
Willy Tarreau74ca5042013-06-11 23:12:07 +020016113 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016114
Willy Tarreau74ca5042013-06-11 23:12:07 +020016115req.proto_http : boolean
16116req_proto_http : boolean (deprecated)
16117 Returns true when data in the request buffer look like HTTP and correctly
16118 parses as such. It is the same parser as the common HTTP request parser which
16119 is used so there should be no surprises. The test does not match until the
16120 request is complete, failed or timed out. This test may be used to report the
16121 protocol in TCP logs, but the biggest use is to block TCP request analysis
16122 until a complete HTTP request is present in the buffer, for example to track
16123 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016124
Willy Tarreau74ca5042013-06-11 23:12:07 +020016125 Example:
16126 # track request counts per "base" (concatenation of Host+URL)
16127 tcp-request inspect-delay 10s
16128 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020016129 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020016130
Willy Tarreau74ca5042013-06-11 23:12:07 +020016131req.rdp_cookie([<name>]) : string
16132rdp_cookie([<name>]) : string (deprecated)
16133 When the request buffer looks like the RDP protocol, extracts the RDP cookie
16134 <name>, or any cookie if unspecified. The parser only checks for the first
16135 cookie, as illustrated in the RDP protocol specification. The cookie name is
16136 case insensitive. Generally the "MSTS" cookie name will be used, as it can
16137 contain the user name of the client connecting to the server if properly
16138 configured on the client. The "MSTSHASH" cookie is often used as well for
16139 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016140
Willy Tarreau74ca5042013-06-11 23:12:07 +020016141 This differs from "balance rdp-cookie" in that any balancing algorithm may be
16142 used and thus the distribution of clients to backend servers is not linked to
16143 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
16144 such as "balance roundrobin" or "balance leastconn" will lead to a more even
16145 distribution of clients to backend servers than the hash used by "balance
16146 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016147
Willy Tarreau74ca5042013-06-11 23:12:07 +020016148 ACL derivatives :
16149 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016150
Willy Tarreau74ca5042013-06-11 23:12:07 +020016151 Example :
16152 listen tse-farm
16153 bind 0.0.0.0:3389
16154 # wait up to 5s for an RDP cookie in the request
16155 tcp-request inspect-delay 5s
16156 tcp-request content accept if RDP_COOKIE
16157 # apply RDP cookie persistence
16158 persist rdp-cookie
16159 # Persist based on the mstshash cookie
16160 # This is only useful makes sense if
16161 # balance rdp-cookie is not used
16162 stick-table type string size 204800
16163 stick on req.rdp_cookie(mstshash)
16164 server srv1 1.1.1.1:3389
16165 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016166
Willy Tarreau74ca5042013-06-11 23:12:07 +020016167 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
16168 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016169
Willy Tarreau74ca5042013-06-11 23:12:07 +020016170req.rdp_cookie_cnt([name]) : integer
16171rdp_cookie_cnt([name]) : integer (deprecated)
16172 Tries to parse the request buffer as RDP protocol, then returns an integer
16173 corresponding to the number of RDP cookies found. If an optional cookie name
16174 is passed, only cookies matching this name are considered. This is mostly
16175 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016176
Willy Tarreau74ca5042013-06-11 23:12:07 +020016177 ACL derivatives :
16178 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016179
Alex Zorin4afdd132018-12-30 13:56:28 +110016180req.ssl_alpn : string
16181 Returns a string containing the values of the Application-Layer Protocol
16182 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
16183 ClientHello message. Note that this only applies to raw contents found in the
16184 request buffer and not to the contents deciphered via an SSL data layer, so
16185 this will not work with "bind" lines having the "ssl" option. This is useful
16186 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020016187 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110016188
16189 Examples :
16190 # Wait for a client hello for at most 5 seconds
16191 tcp-request inspect-delay 5s
16192 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020016193 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110016194 default_backend bk_default
16195
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020016196req.ssl_ec_ext : boolean
16197 Returns a boolean identifying if client sent the Supported Elliptic Curves
16198 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020016199 message. This can be used to present ECC compatible clients with EC
16200 certificate and to use RSA for all others, on the same IP address. Note that
16201 this only applies to raw contents found in the request buffer and not to
16202 contents deciphered via an SSL data layer, so this will not work with "bind"
16203 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020016204
Willy Tarreau74ca5042013-06-11 23:12:07 +020016205req.ssl_hello_type : integer
16206req_ssl_hello_type : integer (deprecated)
16207 Returns an integer value containing the type of the SSL hello message found
16208 in the request buffer if the buffer contains data that parse as a complete
16209 SSL (v3 or superior) client hello message. Note that this only applies to raw
16210 contents found in the request buffer and not to contents deciphered via an
16211 SSL data layer, so this will not work with "bind" lines having the "ssl"
16212 option. This is mostly used in ACL to detect presence of an SSL hello message
16213 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016214
Willy Tarreau74ca5042013-06-11 23:12:07 +020016215req.ssl_sni : string
16216req_ssl_sni : string (deprecated)
16217 Returns a string containing the value of the Server Name TLS extension sent
16218 by a client in a TLS stream passing through the request buffer if the buffer
16219 contains data that parse as a complete SSL (v3 or superior) client hello
16220 message. Note that this only applies to raw contents found in the request
16221 buffer and not to contents deciphered via an SSL data layer, so this will not
16222 work with "bind" lines having the "ssl" option. SNI normally contains the
16223 name of the host the client tries to connect to (for recent browsers). SNI is
16224 useful for allowing or denying access to certain hosts when SSL/TLS is used
16225 by the client. This test was designed to be used with TCP request content
16226 inspection. If content switching is needed, it is recommended to first wait
16227 for a complete client hello (type 1), like in the example below. See also
16228 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016229
Willy Tarreau74ca5042013-06-11 23:12:07 +020016230 ACL derivatives :
16231 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016232
Willy Tarreau74ca5042013-06-11 23:12:07 +020016233 Examples :
16234 # Wait for a client hello for at most 5 seconds
16235 tcp-request inspect-delay 5s
16236 tcp-request content accept if { req_ssl_hello_type 1 }
16237 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
16238 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020016239
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053016240req.ssl_st_ext : integer
16241 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
16242 Returns 1 if the client sent SessionTicket TLS Extension
16243 Returns 2 if the client also sent non-zero length TLS SessionTicket
16244 Note that this only applies to raw contents found in the request buffer and
16245 not to contents deciphered via an SSL data layer, so this will not work with
16246 "bind" lines having the "ssl" option. This can for example be used to detect
16247 whether the client sent a SessionTicket or not and stick it accordingly, if
16248 no SessionTicket then stick on SessionID or don't stick as there's no server
16249 side state is there when SessionTickets are in use.
16250
Willy Tarreau74ca5042013-06-11 23:12:07 +020016251req.ssl_ver : integer
16252req_ssl_ver : integer (deprecated)
16253 Returns an integer value containing the version of the SSL/TLS protocol of a
16254 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
16255 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
16256 composed of the major version multiplied by 65536, added to the minor
16257 version. Note that this only applies to raw contents found in the request
16258 buffer and not to contents deciphered via an SSL data layer, so this will not
16259 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010016260 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020016261 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016262
Willy Tarreau74ca5042013-06-11 23:12:07 +020016263 ACL derivatives :
16264 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010016265
Willy Tarreau47e8eba2013-09-11 23:28:46 +020016266res.len : integer
16267 Returns an integer value corresponding to the number of bytes present in the
16268 response buffer. This is mostly used in ACL. It is important to understand
16269 that this test does not return false as long as the buffer is changing. This
16270 means that a check with equality to zero will almost always immediately match
16271 at the beginning of the session, while a test for more data will wait for
16272 that data to come in and return false only when haproxy is certain that no
16273 more data will come in. This test was designed to be used with TCP response
16274 content inspection.
16275
Willy Tarreau74ca5042013-06-11 23:12:07 +020016276res.payload(<offset>,<length>) : binary
16277 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020016278 in the response buffer. As a special case, if the <length> argument is zero,
16279 the the whole buffer from <offset> to the end is extracted. This can be used
16280 with ACLs in order to check for the presence of some content in a buffer at
16281 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016282
Willy Tarreau74ca5042013-06-11 23:12:07 +020016283res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
16284 This extracts a binary block whose size is specified at <offset1> for <length>
16285 bytes, and which starts at <offset2> if specified or just after the length in
16286 the response buffer. The <offset2> parameter also supports relative offsets
16287 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016288
Willy Tarreau74ca5042013-06-11 23:12:07 +020016289 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016290
Willy Tarreau971f7b62015-09-29 14:06:59 +020016291res.ssl_hello_type : integer
16292rep_ssl_hello_type : integer (deprecated)
16293 Returns an integer value containing the type of the SSL hello message found
16294 in the response buffer if the buffer contains data that parses as a complete
16295 SSL (v3 or superior) hello message. Note that this only applies to raw
16296 contents found in the response buffer and not to contents deciphered via an
16297 SSL data layer, so this will not work with "server" lines having the "ssl"
16298 option. This is mostly used in ACL to detect presence of an SSL hello message
16299 that is supposed to contain an SSL session ID usable for stickiness.
16300
Willy Tarreau74ca5042013-06-11 23:12:07 +020016301wait_end : boolean
16302 This fetch either returns true when the inspection period is over, or does
16303 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010016304 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020016305 actions, such as a delayed reject for some special addresses. Since it either
16306 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010016307 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020016308 "WAIT_END" is always usable without prior declaration. This test was designed
16309 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016310
Willy Tarreau74ca5042013-06-11 23:12:07 +020016311 Examples :
16312 # delay every incoming request by 2 seconds
16313 tcp-request inspect-delay 2s
16314 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010016315
Willy Tarreau74ca5042013-06-11 23:12:07 +020016316 # don't immediately tell bad guys they are rejected
16317 tcp-request inspect-delay 10s
16318 acl goodguys src 10.0.0.0/24
16319 acl badguys src 10.0.1.0/24
16320 tcp-request content accept if goodguys
16321 tcp-request content reject if badguys WAIT_END
16322 tcp-request content reject
16323
16324
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200163257.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020016326--------------------------------------
16327
16328It is possible to fetch samples from HTTP contents, requests and responses.
16329This application layer is also called layer 7. It is only possible to fetch the
16330data in this section when a full HTTP request or response has been parsed from
16331its respective request or response buffer. This is always the case with all
16332HTTP specific rules and for sections running with "mode http". When using TCP
16333content inspection, it may be necessary to support an inspection delay in order
16334to let the request or response come in first. These fetches may require a bit
16335more CPU resources than the layer 4 ones, but not much since the request and
16336response are indexed.
16337
16338base : string
16339 This returns the concatenation of the first Host header and the path part of
16340 the request, which starts at the first slash and ends before the question
16341 mark. It can be useful in virtual hosted environments to detect URL abuses as
16342 well as to improve shared caches efficiency. Using this with a limited size
16343 stick table also allows one to collect statistics about most commonly
16344 requested objects by host/path. With ACLs it can allow simple content
16345 switching rules involving the host and the path at the same time, such as
16346 "www.example.com/favicon.ico". See also "path" and "uri".
16347
16348 ACL derivatives :
16349 base : exact string match
16350 base_beg : prefix match
16351 base_dir : subdir match
16352 base_dom : domain match
16353 base_end : suffix match
16354 base_len : length match
16355 base_reg : regex match
16356 base_sub : substring match
16357
16358base32 : integer
16359 This returns a 32-bit hash of the value returned by the "base" fetch method
16360 above. This is useful to track per-URL activity on high traffic sites without
16361 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020016362 memory. The output type is an unsigned integer. The hash function used is
16363 SDBM with full avalanche on the output. Technically, base32 is exactly equal
16364 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020016365
16366base32+src : binary
16367 This returns the concatenation of the base32 fetch above and the src fetch
16368 below. The resulting type is of type binary, with a size of 8 or 20 bytes
16369 depending on the source address family. This can be used to track per-IP,
16370 per-URL counters.
16371
William Lallemand65ad6e12014-01-31 15:08:02 +010016372capture.req.hdr(<idx>) : string
16373 This extracts the content of the header captured by the "capture request
16374 header", idx is the position of the capture keyword in the configuration.
16375 The first entry is an index of 0. See also: "capture request header".
16376
16377capture.req.method : string
16378 This extracts the METHOD of an HTTP request. It can be used in both request
16379 and response. Unlike "method", it can be used in both request and response
16380 because it's allocated.
16381
16382capture.req.uri : string
16383 This extracts the request's URI, which starts at the first slash and ends
16384 before the first space in the request (without the host part). Unlike "path"
16385 and "url", it can be used in both request and response because it's
16386 allocated.
16387
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020016388capture.req.ver : string
16389 This extracts the request's HTTP version and returns either "HTTP/1.0" or
16390 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
16391 logs because it relies on a persistent flag.
16392
William Lallemand65ad6e12014-01-31 15:08:02 +010016393capture.res.hdr(<idx>) : string
16394 This extracts the content of the header captured by the "capture response
16395 header", idx is the position of the capture keyword in the configuration.
16396 The first entry is an index of 0.
16397 See also: "capture response header"
16398
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020016399capture.res.ver : string
16400 This extracts the response's HTTP version and returns either "HTTP/1.0" or
16401 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
16402 persistent flag.
16403
Willy Tarreaua5910cc2015-05-02 00:46:08 +020016404req.body : binary
16405 This returns the HTTP request's available body as a block of data. It
16406 requires that the request body has been buffered made available using
16407 "option http-buffer-request". In case of chunked-encoded body, currently only
16408 the first chunk is analyzed.
16409
Thierry FOURNIER9826c772015-05-20 15:50:54 +020016410req.body_param([<name>) : string
16411 This fetch assumes that the body of the POST request is url-encoded. The user
16412 can check if the "content-type" contains the value
16413 "application/x-www-form-urlencoded". This extracts the first occurrence of the
16414 parameter <name> in the body, which ends before '&'. The parameter name is
16415 case-sensitive. If no name is given, any parameter will match, and the first
16416 one will be returned. The result is a string corresponding to the value of the
16417 parameter <name> as presented in the request body (no URL decoding is
16418 performed). Note that the ACL version of this fetch iterates over multiple
16419 parameters and will iteratively report all parameters values if no name is
16420 given.
16421
Willy Tarreaua5910cc2015-05-02 00:46:08 +020016422req.body_len : integer
16423 This returns the length of the HTTP request's available body in bytes. It may
16424 be lower than the advertised length if the body is larger than the buffer. It
16425 requires that the request body has been buffered made available using
16426 "option http-buffer-request".
16427
16428req.body_size : integer
16429 This returns the advertised length of the HTTP request's body in bytes. It
16430 will represent the advertised Content-Length header, or the size of the first
16431 chunk in case of chunked encoding. In order to parse the chunks, it requires
16432 that the request body has been buffered made available using
16433 "option http-buffer-request".
16434
Willy Tarreau74ca5042013-06-11 23:12:07 +020016435req.cook([<name>]) : string
16436cook([<name>]) : string (deprecated)
16437 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16438 header line from the request, and returns its value as string. If no name is
16439 specified, the first cookie value is returned. When used with ACLs, all
16440 matching cookies are evaluated. Spaces around the name and the value are
16441 ignored as requested by the Cookie header specification (RFC6265). The cookie
16442 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
16443 well return an empty value if it is present. Use the "found" match to detect
16444 presence. Use the res.cook() variant for response cookies sent by the server.
16445
16446 ACL derivatives :
16447 cook([<name>]) : exact string match
16448 cook_beg([<name>]) : prefix match
16449 cook_dir([<name>]) : subdir match
16450 cook_dom([<name>]) : domain match
16451 cook_end([<name>]) : suffix match
16452 cook_len([<name>]) : length match
16453 cook_reg([<name>]) : regex match
16454 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010016455
Willy Tarreau74ca5042013-06-11 23:12:07 +020016456req.cook_cnt([<name>]) : integer
16457cook_cnt([<name>]) : integer (deprecated)
16458 Returns an integer value representing the number of occurrences of the cookie
16459 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016460
Willy Tarreau74ca5042013-06-11 23:12:07 +020016461req.cook_val([<name>]) : integer
16462cook_val([<name>]) : integer (deprecated)
16463 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16464 header line from the request, and converts its value to an integer which is
16465 returned. If no name is specified, the first cookie value is returned. When
16466 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020016467
Willy Tarreau74ca5042013-06-11 23:12:07 +020016468cookie([<name>]) : string (deprecated)
16469 This extracts the last occurrence of the cookie name <name> on a "Cookie"
16470 header line from the request, or a "Set-Cookie" header from the response, and
16471 returns its value as a string. A typical use is to get multiple clients
16472 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020016473 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020016474 multi-peer synchronization and state keeping across restarts. If no name is
16475 specified, the first cookie value is returned. This fetch should not be used
16476 anymore and should be replaced by req.cook() or res.cook() instead as it
16477 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016478
Willy Tarreau74ca5042013-06-11 23:12:07 +020016479hdr([<name>[,<occ>]]) : string
16480 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
16481 used on responses. Please refer to these respective fetches for more details.
16482 In case of doubt about the fetch direction, please use the explicit ones.
16483 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016484 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016485
Willy Tarreau74ca5042013-06-11 23:12:07 +020016486req.fhdr(<name>[,<occ>]) : string
16487 This extracts the last occurrence of header <name> in an HTTP request. When
16488 used from an ACL, all occurrences are iterated over until a match is found.
16489 Optionally, a specific occurrence might be specified as a position number.
16490 Positive values indicate a position from the first occurrence, with 1 being
16491 the first one. Negative values indicate positions relative to the last one,
16492 with -1 being the last one. It differs from req.hdr() in that any commas
16493 present in the value are returned and are not used as delimiters. This is
16494 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016495
Willy Tarreau74ca5042013-06-11 23:12:07 +020016496req.fhdr_cnt([<name>]) : integer
16497 Returns an integer value representing the number of occurrences of request
16498 header field name <name>, or the total number of header fields if <name> is
16499 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
16500 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016501
Willy Tarreau74ca5042013-06-11 23:12:07 +020016502req.hdr([<name>[,<occ>]]) : string
16503 This extracts the last occurrence of header <name> in an HTTP request. When
16504 used from an ACL, all occurrences are iterated over until a match is found.
16505 Optionally, a specific occurrence might be specified as a position number.
16506 Positive values indicate a position from the first occurrence, with 1 being
16507 the first one. Negative values indicate positions relative to the last one,
16508 with -1 being the last one. A typical use is with the X-Forwarded-For header
16509 once converted to IP, associated with an IP stick-table. The function
16510 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000016511 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020016512 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010016513 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010016514
Willy Tarreau74ca5042013-06-11 23:12:07 +020016515 ACL derivatives :
16516 hdr([<name>[,<occ>]]) : exact string match
16517 hdr_beg([<name>[,<occ>]]) : prefix match
16518 hdr_dir([<name>[,<occ>]]) : subdir match
16519 hdr_dom([<name>[,<occ>]]) : domain match
16520 hdr_end([<name>[,<occ>]]) : suffix match
16521 hdr_len([<name>[,<occ>]]) : length match
16522 hdr_reg([<name>[,<occ>]]) : regex match
16523 hdr_sub([<name>[,<occ>]]) : substring match
16524
16525req.hdr_cnt([<name>]) : integer
16526hdr_cnt([<header>]) : integer (deprecated)
16527 Returns an integer value representing the number of occurrences of request
16528 header field name <name>, or the total number of header field values if
16529 <name> is not specified. It is important to remember that one header line may
16530 count as several headers if it has several values. The function considers any
16531 comma as a delimiter for distinct values. If full-line headers are desired
16532 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
16533 detect presence, absence or abuse of a specific header, as well as to block
16534 request smuggling attacks by rejecting requests which contain more than one
16535 of certain headers. See "req.hdr" for more information on header matching.
16536
16537req.hdr_ip([<name>[,<occ>]]) : ip
16538hdr_ip([<name>[,<occ>]]) : ip (deprecated)
16539 This extracts the last occurrence of header <name> in an HTTP request,
16540 converts it to an IPv4 or IPv6 address and returns this address. When used
16541 with ACLs, all occurrences are checked, and if <name> is omitted, every value
16542 of every header is checked. Optionally, a specific occurrence might be
16543 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010016544 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020016545 positions relative to the last one, with -1 being the last one. A typical use
16546 is with the X-Forwarded-For and X-Client-IP headers.
16547
16548req.hdr_val([<name>[,<occ>]]) : integer
16549hdr_val([<name>[,<occ>]]) : integer (deprecated)
16550 This extracts the last occurrence of header <name> in an HTTP request, and
16551 converts it to an integer value. When used with ACLs, all occurrences are
16552 checked, and if <name> is omitted, every value of every header is checked.
16553 Optionally, a specific occurrence might be specified as a position number.
16554 Positive values indicate a position from the first occurrence, with 1 being
16555 the first one. Negative values indicate positions relative to the last one,
16556 with -1 being the last one. A typical use is with the X-Forwarded-For header.
16557
Frédéric Lécailleec891192019-02-26 15:02:35 +010016558
16559
Willy Tarreau74ca5042013-06-11 23:12:07 +020016560http_auth(<userlist>) : boolean
16561 Returns a boolean indicating whether the authentication data received from
16562 the client match a username & password stored in the specified userlist. This
16563 fetch function is not really useful outside of ACLs. Currently only http
16564 basic auth is supported.
16565
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010016566http_auth_group(<userlist>) : string
16567 Returns a string corresponding to the user name found in the authentication
16568 data received from the client if both the user name and password are valid
16569 according to the specified userlist. The main purpose is to use it in ACLs
16570 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016571 This fetch function is not really useful outside of ACLs. Currently only http
16572 basic auth is supported.
16573
16574 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010016575 http_auth_group(<userlist>) : group ...
16576 Returns true when the user extracted from the request and whose password is
16577 valid according to the specified userlist belongs to at least one of the
16578 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020016579
16580http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020016581 Returns true when the request being processed is the first one of the
16582 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020016583 from some requests when a request is not the first one, or to help grouping
16584 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020016585
Willy Tarreau74ca5042013-06-11 23:12:07 +020016586method : integer + string
16587 Returns an integer value corresponding to the method in the HTTP request. For
16588 example, "GET" equals 1 (check sources to establish the matching). Value 9
16589 means "other method" and may be converted to a string extracted from the
16590 stream. This should not be used directly as a sample, this is only meant to
16591 be used from ACLs, which transparently convert methods from patterns to these
16592 integer + string values. Some predefined ACL already check for most common
16593 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016594
Willy Tarreau74ca5042013-06-11 23:12:07 +020016595 ACL derivatives :
16596 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016597
Willy Tarreau74ca5042013-06-11 23:12:07 +020016598 Example :
16599 # only accept GET and HEAD requests
16600 acl valid_method method GET HEAD
16601 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020016602
Willy Tarreau74ca5042013-06-11 23:12:07 +020016603path : string
16604 This extracts the request's URL path, which starts at the first slash and
16605 ends before the question mark (without the host part). A typical use is with
16606 prefetch-capable caches, and with portals which need to aggregate multiple
16607 information from databases and keep them in caches. Note that with outgoing
16608 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010016609 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020016610 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016611
Willy Tarreau74ca5042013-06-11 23:12:07 +020016612 ACL derivatives :
16613 path : exact string match
16614 path_beg : prefix match
16615 path_dir : subdir match
16616 path_dom : domain match
16617 path_end : suffix match
16618 path_len : length match
16619 path_reg : regex match
16620 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016621
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016622query : string
16623 This extracts the request's query string, which starts after the first
16624 question mark. If no question mark is present, this fetch returns nothing. If
16625 a question mark is present but nothing follows, it returns an empty string.
16626 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010016627 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016628 which stops before the question mark.
16629
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016630req.hdr_names([<delim>]) : string
16631 This builds a string made from the concatenation of all header names as they
16632 appear in the request when the rule is evaluated. The default delimiter is
16633 the comma (',') but it may be overridden as an optional argument <delim>. In
16634 this case, only the first character of <delim> is considered.
16635
Willy Tarreau74ca5042013-06-11 23:12:07 +020016636req.ver : string
16637req_ver : string (deprecated)
16638 Returns the version string from the HTTP request, for example "1.1". This can
16639 be useful for logs, but is mostly there for ACL. Some predefined ACL already
16640 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016641
Willy Tarreau74ca5042013-06-11 23:12:07 +020016642 ACL derivatives :
16643 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020016644
Willy Tarreau74ca5042013-06-11 23:12:07 +020016645res.comp : boolean
16646 Returns the boolean "true" value if the response has been compressed by
16647 HAProxy, otherwise returns boolean "false". This may be used to add
16648 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016649
Willy Tarreau74ca5042013-06-11 23:12:07 +020016650res.comp_algo : string
16651 Returns a string containing the name of the algorithm used if the response
16652 was compressed by HAProxy, for example : "deflate". This may be used to add
16653 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016654
Willy Tarreau74ca5042013-06-11 23:12:07 +020016655res.cook([<name>]) : string
16656scook([<name>]) : string (deprecated)
16657 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16658 header line from the response, and returns its value as string. If no name is
16659 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016660
Willy Tarreau74ca5042013-06-11 23:12:07 +020016661 ACL derivatives :
16662 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016663
Willy Tarreau74ca5042013-06-11 23:12:07 +020016664res.cook_cnt([<name>]) : integer
16665scook_cnt([<name>]) : integer (deprecated)
16666 Returns an integer value representing the number of occurrences of the cookie
16667 <name> in the response, or all cookies if <name> is not specified. This is
16668 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016669
Willy Tarreau74ca5042013-06-11 23:12:07 +020016670res.cook_val([<name>]) : integer
16671scook_val([<name>]) : integer (deprecated)
16672 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16673 header line from the response, and converts its value to an integer which is
16674 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016675
Willy Tarreau74ca5042013-06-11 23:12:07 +020016676res.fhdr([<name>[,<occ>]]) : string
16677 This extracts the last occurrence of header <name> in an HTTP response, or of
16678 the last header if no <name> is specified. Optionally, a specific occurrence
16679 might be specified as a position number. Positive values indicate a position
16680 from the first occurrence, with 1 being the first one. Negative values
16681 indicate positions relative to the last one, with -1 being the last one. It
16682 differs from res.hdr() in that any commas present in the value are returned
16683 and are not used as delimiters. If this is not desired, the res.hdr() fetch
16684 should be used instead. This is sometimes useful with headers such as Date or
16685 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016686
Willy Tarreau74ca5042013-06-11 23:12:07 +020016687res.fhdr_cnt([<name>]) : integer
16688 Returns an integer value representing the number of occurrences of response
16689 header field name <name>, or the total number of header fields if <name> is
16690 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
16691 the number of full line headers and does not stop on commas. If this is not
16692 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016693
Willy Tarreau74ca5042013-06-11 23:12:07 +020016694res.hdr([<name>[,<occ>]]) : string
16695shdr([<name>[,<occ>]]) : string (deprecated)
16696 This extracts the last occurrence of header <name> in an HTTP response, or of
16697 the last header if no <name> is specified. Optionally, a specific occurrence
16698 might be specified as a position number. Positive values indicate a position
16699 from the first occurrence, with 1 being the first one. Negative values
16700 indicate positions relative to the last one, with -1 being the last one. This
16701 can be useful to learn some data into a stick-table. The function considers
16702 any comma as a delimiter for distinct values. If this is not desired, the
16703 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016704
Willy Tarreau74ca5042013-06-11 23:12:07 +020016705 ACL derivatives :
16706 shdr([<name>[,<occ>]]) : exact string match
16707 shdr_beg([<name>[,<occ>]]) : prefix match
16708 shdr_dir([<name>[,<occ>]]) : subdir match
16709 shdr_dom([<name>[,<occ>]]) : domain match
16710 shdr_end([<name>[,<occ>]]) : suffix match
16711 shdr_len([<name>[,<occ>]]) : length match
16712 shdr_reg([<name>[,<occ>]]) : regex match
16713 shdr_sub([<name>[,<occ>]]) : substring match
16714
16715res.hdr_cnt([<name>]) : integer
16716shdr_cnt([<name>]) : integer (deprecated)
16717 Returns an integer value representing the number of occurrences of response
16718 header field name <name>, or the total number of header fields if <name> is
16719 not specified. The function considers any comma as a delimiter for distinct
16720 values. If this is not desired, the res.fhdr_cnt() fetch should be used
16721 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016722
Willy Tarreau74ca5042013-06-11 23:12:07 +020016723res.hdr_ip([<name>[,<occ>]]) : ip
16724shdr_ip([<name>[,<occ>]]) : ip (deprecated)
16725 This extracts the last occurrence of header <name> in an HTTP response,
16726 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
16727 specific occurrence might be specified as a position number. Positive values
16728 indicate a position from the first occurrence, with 1 being the first one.
16729 Negative values indicate positions relative to the last one, with -1 being
16730 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016731
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016732res.hdr_names([<delim>]) : string
16733 This builds a string made from the concatenation of all header names as they
16734 appear in the response when the rule is evaluated. The default delimiter is
16735 the comma (',') but it may be overridden as an optional argument <delim>. In
16736 this case, only the first character of <delim> is considered.
16737
Willy Tarreau74ca5042013-06-11 23:12:07 +020016738res.hdr_val([<name>[,<occ>]]) : integer
16739shdr_val([<name>[,<occ>]]) : integer (deprecated)
16740 This extracts the last occurrence of header <name> in an HTTP response, and
16741 converts it to an integer value. Optionally, a specific occurrence might be
16742 specified as a position number. Positive values indicate a position from the
16743 first occurrence, with 1 being the first one. Negative values indicate
16744 positions relative to the last one, with -1 being the last one. This can be
16745 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016746
Willy Tarreau74ca5042013-06-11 23:12:07 +020016747res.ver : string
16748resp_ver : string (deprecated)
16749 Returns the version string from the HTTP response, for example "1.1". This
16750 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020016751
Willy Tarreau74ca5042013-06-11 23:12:07 +020016752 ACL derivatives :
16753 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016754
Willy Tarreau74ca5042013-06-11 23:12:07 +020016755set-cookie([<name>]) : string (deprecated)
16756 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16757 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020016758 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020016759 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016760
Willy Tarreau74ca5042013-06-11 23:12:07 +020016761 This fetch function is deprecated and has been superseded by the "res.cook"
16762 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016763
Willy Tarreau74ca5042013-06-11 23:12:07 +020016764status : integer
16765 Returns an integer containing the HTTP status code in the HTTP response, for
16766 example, 302. It is mostly used within ACLs and integer ranges, for example,
16767 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016768
Thierry Fournier0e00dca2016-04-07 15:47:40 +020016769unique-id : string
16770 Returns the unique-id attached to the request. The directive
16771 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
16772 fails. Note that the unique-id is usually used with HTTP requests, however this
16773 sample fetch can be used with other protocols. Obviously, if it is used with
16774 other protocols than HTTP, the unique-id-format directive must not contain
16775 HTTP parts. See: unique-id-format and unique-id-header
16776
Willy Tarreau74ca5042013-06-11 23:12:07 +020016777url : string
16778 This extracts the request's URL as presented in the request. A typical use is
16779 with prefetch-capable caches, and with portals which need to aggregate
16780 multiple information from databases and keep them in caches. With ACLs, using
16781 "path" is preferred over using "url", because clients may send a full URL as
16782 is normally done with proxies. The only real use is to match "*" which does
16783 not match in "path", and for which there is already a predefined ACL. See
16784 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016785
Willy Tarreau74ca5042013-06-11 23:12:07 +020016786 ACL derivatives :
16787 url : exact string match
16788 url_beg : prefix match
16789 url_dir : subdir match
16790 url_dom : domain match
16791 url_end : suffix match
16792 url_len : length match
16793 url_reg : regex match
16794 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016795
Willy Tarreau74ca5042013-06-11 23:12:07 +020016796url_ip : ip
16797 This extracts the IP address from the request's URL when the host part is
16798 presented as an IP address. Its use is very limited. For instance, a
16799 monitoring system might use this field as an alternative for the source IP in
16800 order to test what path a given source address would follow, or to force an
16801 entry in a table for a given source address. With ACLs it can be used to
16802 restrict access to certain systems through a proxy, for example when combined
16803 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016804
Willy Tarreau74ca5042013-06-11 23:12:07 +020016805url_port : integer
16806 This extracts the port part from the request's URL. Note that if the port is
16807 not specified in the request, port 80 is assumed. With ACLs it can be used to
16808 restrict access to certain systems through a proxy, for example when combined
16809 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016810
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016811urlp([<name>[,<delim>]]) : string
16812url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020016813 This extracts the first occurrence of the parameter <name> in the query
16814 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016815 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
16816 any parameter will match, and the first one will be returned. The result is
16817 a string corresponding to the value of the parameter <name> as presented in
16818 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020016819 stickiness based on a client ID, to extract an application cookie passed as a
16820 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016821 this fetch iterates over multiple parameters and will iteratively report all
16822 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016823
Willy Tarreau74ca5042013-06-11 23:12:07 +020016824 ACL derivatives :
16825 urlp(<name>[,<delim>]) : exact string match
16826 urlp_beg(<name>[,<delim>]) : prefix match
16827 urlp_dir(<name>[,<delim>]) : subdir match
16828 urlp_dom(<name>[,<delim>]) : domain match
16829 urlp_end(<name>[,<delim>]) : suffix match
16830 urlp_len(<name>[,<delim>]) : length match
16831 urlp_reg(<name>[,<delim>]) : regex match
16832 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016833
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016834
Willy Tarreau74ca5042013-06-11 23:12:07 +020016835 Example :
16836 # match http://example.com/foo?PHPSESSIONID=some_id
16837 stick on urlp(PHPSESSIONID)
16838 # match http://example.com/foo;JSESSIONID=some_id
16839 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016840
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016841urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020016842 See "urlp" above. This one extracts the URL parameter <name> in the request
16843 and converts it to an integer value. This can be used for session stickiness
16844 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020016845
Dragan Dosen0070cd52016-06-16 12:19:49 +020016846url32 : integer
16847 This returns a 32-bit hash of the value obtained by concatenating the first
16848 Host header and the whole URL including parameters (not only the path part of
16849 the request, as in the "base32" fetch above). This is useful to track per-URL
16850 activity. A shorter hash is stored, saving a lot of memory. The output type
16851 is an unsigned integer.
16852
16853url32+src : binary
16854 This returns the concatenation of the "url32" fetch and the "src" fetch. The
16855 resulting type is of type binary, with a size of 8 or 20 bytes depending on
16856 the source address family. This can be used to track per-IP, per-URL counters.
16857
Willy Tarreau198a7442008-01-17 12:05:32 +010016858
Willy Tarreau74ca5042013-06-11 23:12:07 +0200168597.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016860---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016861
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016862Some predefined ACLs are hard-coded so that they do not have to be declared in
16863every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020016864order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010016865
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016866ACL name Equivalent to Usage
16867---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016868FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020016869HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016870HTTP_1.0 req_ver 1.0 match HTTP version 1.0
16871HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010016872HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
16873HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
16874HTTP_URL_SLASH url_beg / match URL beginning with "/"
16875HTTP_URL_STAR url * match URL equal to "*"
16876LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016877METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016878METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016879METH_GET method GET HEAD match HTTP GET or HEAD method
16880METH_HEAD method HEAD match HTTP HEAD method
16881METH_OPTIONS method OPTIONS match HTTP OPTIONS method
16882METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016883METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016884METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020016885RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016886REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016887TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016888WAIT_END wait_end wait for end of content analysis
16889---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016890
Willy Tarreaub937b7e2010-01-12 15:27:54 +010016891
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200168928. Logging
16893----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010016894
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016895One of HAProxy's strong points certainly lies is its precise logs. It probably
16896provides the finest level of information available for such a product, which is
16897very important for troubleshooting complex environments. Standard information
16898provided in logs include client ports, TCP/HTTP state timers, precise session
16899state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016900to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016901headers.
16902
16903In order to improve administrators reactivity, it offers a great transparency
16904about encountered problems, both internal and external, and it is possible to
16905send logs to different sources at the same time with different level filters :
16906
16907 - global process-level logs (system errors, start/stop, etc..)
16908 - per-instance system and internal errors (lack of resource, bugs, ...)
16909 - per-instance external troubles (servers up/down, max connections)
16910 - per-instance activity (client connections), either at the establishment or
16911 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010016912 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060016913 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016914
16915The ability to distribute different levels of logs to different log servers
16916allow several production teams to interact and to fix their problems as soon
16917as possible. For example, the system team might monitor system-wide errors,
16918while the application team might be monitoring the up/down for their servers in
16919real time, and the security team might analyze the activity logs with one hour
16920delay.
16921
16922
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169238.1. Log levels
16924---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016925
Simon Hormandf791f52011-05-29 15:01:10 +090016926TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016927source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090016928HTTP request, HTTP return code, number of bytes transmitted, conditions
16929in which the session ended, and even exchanged cookies values. For example
16930track a particular user's problems. All messages may be sent to up to two
16931syslog servers. Check the "log" keyword in section 4.2 for more information
16932about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016933
16934
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169358.2. Log formats
16936----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016937
William Lallemand48940402012-01-30 16:47:22 +010016938HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090016939and will be detailed in the following sections. A few of them may vary
16940slightly with the configuration, due to indicators specific to certain
16941options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016942
16943 - the default format, which is very basic and very rarely used. It only
16944 provides very basic information about the incoming connection at the moment
16945 it is accepted : source IP:port, destination IP:port, and frontend-name.
16946 This mode will eventually disappear so it will not be described to great
16947 extents.
16948
16949 - the TCP format, which is more advanced. This format is enabled when "option
16950 tcplog" is set on the frontend. HAProxy will then usually wait for the
16951 connection to terminate before logging. This format provides much richer
16952 information, such as timers, connection counts, queue size, etc... This
16953 format is recommended for pure TCP proxies.
16954
16955 - the HTTP format, which is the most advanced for HTTP proxying. This format
16956 is enabled when "option httplog" is set on the frontend. It provides the
16957 same information as the TCP format with some HTTP-specific fields such as
16958 the request, the status code, and captures of headers and cookies. This
16959 format is recommended for HTTP proxies.
16960
Emeric Brun3a058f32009-06-30 18:26:00 +020016961 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
16962 fields arranged in the same order as the CLF format. In this mode, all
16963 timers, captures, flags, etc... appear one per field after the end of the
16964 common fields, in the same order they appear in the standard HTTP format.
16965
William Lallemand48940402012-01-30 16:47:22 +010016966 - the custom log format, allows you to make your own log line.
16967
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016968Next sections will go deeper into details for each of these formats. Format
16969specification will be performed on a "field" basis. Unless stated otherwise, a
16970field is a portion of text delimited by any number of spaces. Since syslog
16971servers are susceptible of inserting fields at the beginning of a line, it is
16972always assumed that the first field is the one containing the process name and
16973identifier.
16974
16975Note : Since log lines may be quite long, the log examples in sections below
16976 might be broken into multiple lines. The example log lines will be
16977 prefixed with 3 closing angle brackets ('>>>') and each time a log is
16978 broken into multiple lines, each non-final line will end with a
16979 backslash ('\') and the next line will start indented by two characters.
16980
16981
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169828.2.1. Default log format
16983-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016984
16985This format is used when no specific option is set. The log is emitted as soon
16986as the connection is accepted. One should note that this currently is the only
16987format which logs the request's destination IP and ports.
16988
16989 Example :
16990 listen www
16991 mode http
16992 log global
16993 server srv1 127.0.0.1:8000
16994
16995 >>> Feb 6 12:12:09 localhost \
16996 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
16997 (www/HTTP)
16998
16999 Field Format Extract from the example above
17000 1 process_name '[' pid ']:' haproxy[14385]:
17001 2 'Connect from' Connect from
17002 3 source_ip ':' source_port 10.0.1.2:33312
17003 4 'to' to
17004 5 destination_ip ':' destination_port 10.0.3.31:8012
17005 6 '(' frontend_name '/' mode ')' (www/HTTP)
17006
17007Detailed fields description :
17008 - "source_ip" is the IP address of the client which initiated the connection.
17009 - "source_port" is the TCP port of the client which initiated the connection.
17010 - "destination_ip" is the IP address the client connected to.
17011 - "destination_port" is the TCP port the client connected to.
17012 - "frontend_name" is the name of the frontend (or listener) which received
17013 and processed the connection.
17014 - "mode is the mode the frontend is operating (TCP or HTTP).
17015
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017016In case of a UNIX socket, the source and destination addresses are marked as
17017"unix:" and the ports reflect the internal ID of the socket which accepted the
17018connection (the same ID as reported in the stats).
17019
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017020It is advised not to use this deprecated format for newer installations as it
17021will eventually disappear.
17022
17023
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170248.2.2. TCP log format
17025---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017026
17027The TCP format is used when "option tcplog" is specified in the frontend, and
17028is the recommended format for pure TCP proxies. It provides a lot of precious
17029information for troubleshooting. Since this format includes timers and byte
17030counts, the log is normally emitted at the end of the session. It can be
17031emitted earlier if "option logasap" is specified, which makes sense in most
17032environments with long sessions such as remote terminals. Sessions which match
17033the "monitor" rules are never logged. It is also possible not to emit logs for
17034sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017035specifying "option dontlognull" in the frontend. Successful connections will
17036not be logged if "option dontlog-normal" is specified in the frontend. A few
17037fields may slightly vary depending on some configuration options, those are
17038marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017039
17040 Example :
17041 frontend fnt
17042 mode tcp
17043 option tcplog
17044 log global
17045 default_backend bck
17046
17047 backend bck
17048 server srv1 127.0.0.1:8000
17049
17050 >>> Feb 6 12:12:56 localhost \
17051 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
17052 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
17053
17054 Field Format Extract from the example above
17055 1 process_name '[' pid ']:' haproxy[14387]:
17056 2 client_ip ':' client_port 10.0.1.2:33313
17057 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
17058 4 frontend_name fnt
17059 5 backend_name '/' server_name bck/srv1
17060 6 Tw '/' Tc '/' Tt* 0/0/5007
17061 7 bytes_read* 212
17062 8 termination_state --
17063 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
17064 10 srv_queue '/' backend_queue 0/0
17065
17066Detailed fields description :
17067 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017068 connection to haproxy. If the connection was accepted on a UNIX socket
17069 instead, the IP address would be replaced with the word "unix". Note that
17070 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017071 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010017072 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017073 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017074
17075 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017076 If the connection was accepted on a UNIX socket instead, the port would be
17077 replaced with the ID of the accepting socket, which is also reported in the
17078 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017079
17080 - "accept_date" is the exact date when the connection was received by haproxy
17081 (which might be very slightly different from the date observed on the
17082 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020017083 the same date which may appear in any upstream firewall's log. When used in
17084 HTTP mode, the accept_date field will be reset to the first moment the
17085 connection is ready to receive a new request (end of previous response for
17086 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017087
17088 - "frontend_name" is the name of the frontend (or listener) which received
17089 and processed the connection.
17090
17091 - "backend_name" is the name of the backend (or listener) which was selected
17092 to manage the connection to the server. This will be the same as the
17093 frontend if no switching rule has been applied, which is common for TCP
17094 applications.
17095
17096 - "server_name" is the name of the last server to which the connection was
17097 sent, which might differ from the first one if there were connection errors
17098 and a redispatch occurred. Note that this server belongs to the backend
17099 which processed the request. If the connection was aborted before reaching
17100 a server, "<NOSRV>" is indicated instead of a server name.
17101
17102 - "Tw" is the total time in milliseconds spent waiting in the various queues.
17103 It can be "-1" if the connection was aborted before reaching the queue.
17104 See "Timers" below for more details.
17105
17106 - "Tc" is the total time in milliseconds spent waiting for the connection to
17107 establish to the final server, including retries. It can be "-1" if the
17108 connection was aborted before a connection could be established. See
17109 "Timers" below for more details.
17110
17111 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017112 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017113 "option logasap" was specified, then the time counting stops at the moment
17114 the log is emitted. In this case, a '+' sign is prepended before the value,
17115 indicating that the final one will be larger. See "Timers" below for more
17116 details.
17117
17118 - "bytes_read" is the total number of bytes transmitted from the server to
17119 the client when the log is emitted. If "option logasap" is specified, the
17120 this value will be prefixed with a '+' sign indicating that the final one
17121 may be larger. Please note that this value is a 64-bit counter, so log
17122 analysis tools must be able to handle it without overflowing.
17123
17124 - "termination_state" is the condition the session was in when the session
17125 ended. This indicates the session state, which side caused the end of
17126 session to happen, and for what reason (timeout, error, ...). The normal
17127 flags should be "--", indicating the session was closed by either end with
17128 no data remaining in buffers. See below "Session state at disconnection"
17129 for more details.
17130
17131 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017132 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017133 limits have been reached. For instance, if actconn is close to 512 when
17134 multiple connection errors occur, chances are high that the system limits
17135 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017136 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017137
17138 - "feconn" is the total number of concurrent connections on the frontend when
17139 the session was logged. It is useful to estimate the amount of resource
17140 required to sustain high loads, and to detect when the frontend's "maxconn"
17141 has been reached. Most often when this value increases by huge jumps, it is
17142 because there is congestion on the backend servers, but sometimes it can be
17143 caused by a denial of service attack.
17144
17145 - "beconn" is the total number of concurrent connections handled by the
17146 backend when the session was logged. It includes the total number of
17147 concurrent connections active on servers as well as the number of
17148 connections pending in queues. It is useful to estimate the amount of
17149 additional servers needed to support high loads for a given application.
17150 Most often when this value increases by huge jumps, it is because there is
17151 congestion on the backend servers, but sometimes it can be caused by a
17152 denial of service attack.
17153
17154 - "srv_conn" is the total number of concurrent connections still active on
17155 the server when the session was logged. It can never exceed the server's
17156 configured "maxconn" parameter. If this value is very often close or equal
17157 to the server's "maxconn", it means that traffic regulation is involved a
17158 lot, meaning that either the server's maxconn value is too low, or that
17159 there aren't enough servers to process the load with an optimal response
17160 time. When only one of the server's "srv_conn" is high, it usually means
17161 that this server has some trouble causing the connections to take longer to
17162 be processed than on other servers.
17163
17164 - "retries" is the number of connection retries experienced by this session
17165 when trying to connect to the server. It must normally be zero, unless a
17166 server is being stopped at the same moment the connection was attempted.
17167 Frequent retries generally indicate either a network problem between
17168 haproxy and the server, or a misconfigured system backlog on the server
17169 preventing new connections from being queued. This field may optionally be
17170 prefixed with a '+' sign, indicating that the session has experienced a
17171 redispatch after the maximal retry count has been reached on the initial
17172 server. In this case, the server name appearing in the log is the one the
17173 connection was redispatched to, and not the first one, though both may
17174 sometimes be the same in case of hashing for instance. So as a general rule
17175 of thumb, when a '+' is present in front of the retry count, this count
17176 should not be attributed to the logged server.
17177
17178 - "srv_queue" is the total number of requests which were processed before
17179 this one in the server queue. It is zero when the request has not gone
17180 through the server queue. It makes it possible to estimate the approximate
17181 server's response time by dividing the time spent in queue by the number of
17182 requests in the queue. It is worth noting that if a session experiences a
17183 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010017184 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017185 backend queue unless a redispatch occurs.
17186
17187 - "backend_queue" is the total number of requests which were processed before
17188 this one in the backend's global queue. It is zero when the request has not
17189 gone through the global queue. It makes it possible to estimate the average
17190 queue length, which easily translates into a number of missing servers when
17191 divided by a server's "maxconn" parameter. It is worth noting that if a
17192 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010017193 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017194 through both the server queue and the backend queue unless a redispatch
17195 occurs.
17196
17197
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171988.2.3. HTTP log format
17199----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017200
17201The HTTP format is the most complete and the best suited for HTTP proxies. It
17202is enabled by when "option httplog" is specified in the frontend. It provides
17203the same level of information as the TCP format with additional features which
17204are specific to the HTTP protocol. Just like the TCP format, the log is usually
17205emitted at the end of the session, unless "option logasap" is specified, which
17206generally only makes sense for download sites. A session which matches the
17207"monitor" rules will never logged. It is also possible not to log sessions for
17208which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017209frontend. Successful connections will not be logged if "option dontlog-normal"
17210is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017211
17212Most fields are shared with the TCP log, some being different. A few fields may
17213slightly vary depending on some configuration options. Those ones are marked
17214with a star ('*') after the field name below.
17215
17216 Example :
17217 frontend http-in
17218 mode http
17219 option httplog
17220 log global
17221 default_backend bck
17222
17223 backend static
17224 server srv1 127.0.0.1:8000
17225
17226 >>> Feb 6 12:14:14 localhost \
17227 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
17228 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 +010017229 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017230
17231 Field Format Extract from the example above
17232 1 process_name '[' pid ']:' haproxy[14389]:
17233 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017234 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017235 4 frontend_name http-in
17236 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017237 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017238 7 status_code 200
17239 8 bytes_read* 2750
17240 9 captured_request_cookie -
17241 10 captured_response_cookie -
17242 11 termination_state ----
17243 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
17244 13 srv_queue '/' backend_queue 0/0
17245 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
17246 15 '{' captured_response_headers* '}' {}
17247 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010017248
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017249Detailed fields description :
17250 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017251 connection to haproxy. If the connection was accepted on a UNIX socket
17252 instead, the IP address would be replaced with the word "unix". Note that
17253 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017254 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010017255 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010017256 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017257
17258 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010017259 If the connection was accepted on a UNIX socket instead, the port would be
17260 replaced with the ID of the accepting socket, which is also reported in the
17261 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017262
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017263 - "request_date" is the exact date when the first byte of the HTTP request
17264 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017265
17266 - "frontend_name" is the name of the frontend (or listener) which received
17267 and processed the connection.
17268
17269 - "backend_name" is the name of the backend (or listener) which was selected
17270 to manage the connection to the server. This will be the same as the
17271 frontend if no switching rule has been applied.
17272
17273 - "server_name" is the name of the last server to which the connection was
17274 sent, which might differ from the first one if there were connection errors
17275 and a redispatch occurred. Note that this server belongs to the backend
17276 which processed the request. If the request was aborted before reaching a
17277 server, "<NOSRV>" is indicated instead of a server name. If the request was
17278 intercepted by the stats subsystem, "<STATS>" is indicated instead.
17279
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017280 - "TR" is the total time in milliseconds spent waiting for a full HTTP
17281 request from the client (not counting body) after the first byte was
17282 received. It can be "-1" if the connection was aborted before a complete
John Roesler27c754c2019-07-10 15:45:51 -050017283 request could be received or a bad request was received. It should
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017284 always be very small because a request generally fits in one single packet.
17285 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020017286 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
17287 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017288
17289 - "Tw" is the total time in milliseconds spent waiting in the various queues.
17290 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020017291 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017292
17293 - "Tc" is the total time in milliseconds spent waiting for the connection to
17294 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020017295 request was aborted before a connection could be established. See section
17296 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017297
17298 - "Tr" is the total time in milliseconds spent waiting for the server to send
17299 a full HTTP response, not counting data. It can be "-1" if the request was
17300 aborted before a complete response could be received. It generally matches
17301 the server's processing time for the request, though it may be altered by
17302 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020017303 "GET" requests generally indicate an overloaded server. See section 8.4
17304 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017305
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017306 - "Ta" is the time the request remained active in haproxy, which is the total
17307 time in milliseconds elapsed between the first byte of the request was
17308 received and the last byte of response was sent. It covers all possible
17309 processing except the handshake (see Th) and idle time (see Ti). There is
17310 one exception, if "option logasap" was specified, then the time counting
17311 stops at the moment the log is emitted. In this case, a '+' sign is
17312 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020017313 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017314
17315 - "status_code" is the HTTP status code returned to the client. This status
17316 is generally set by the server, but it might also be set by haproxy when
17317 the server cannot be reached or when its response is blocked by haproxy.
17318
17319 - "bytes_read" is the total number of bytes transmitted to the client when
17320 the log is emitted. This does include HTTP headers. If "option logasap" is
John Roesler27c754c2019-07-10 15:45:51 -050017321 specified, this value will be prefixed with a '+' sign indicating that
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017322 the final one may be larger. Please note that this value is a 64-bit
17323 counter, so log analysis tools must be able to handle it without
17324 overflowing.
17325
17326 - "captured_request_cookie" is an optional "name=value" entry indicating that
17327 the client had this cookie in the request. The cookie name and its maximum
17328 length are defined by the "capture cookie" statement in the frontend
17329 configuration. The field is a single dash ('-') when the option is not
17330 set. Only one cookie may be captured, it is generally used to track session
17331 ID exchanges between a client and a server to detect session crossing
17332 between clients due to application bugs. For more details, please consult
17333 the section "Capturing HTTP headers and cookies" below.
17334
17335 - "captured_response_cookie" is an optional "name=value" entry indicating
17336 that the server has returned a cookie with its response. The cookie name
17337 and its maximum length are defined by the "capture cookie" statement in the
17338 frontend configuration. The field is a single dash ('-') when the option is
17339 not set. Only one cookie may be captured, it is generally used to track
17340 session ID exchanges between a client and a server to detect session
17341 crossing between clients due to application bugs. For more details, please
17342 consult the section "Capturing HTTP headers and cookies" below.
17343
17344 - "termination_state" is the condition the session was in when the session
17345 ended. This indicates the session state, which side caused the end of
17346 session to happen, for what reason (timeout, error, ...), just like in TCP
17347 logs, and information about persistence operations on cookies in the last
17348 two characters. The normal flags should begin with "--", indicating the
17349 session was closed by either end with no data remaining in buffers. See
17350 below "Session state at disconnection" for more details.
17351
17352 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017353 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017354 limits have been reached. For instance, if actconn is close to 512 or 1024
17355 when multiple connection errors occur, chances are high that the system
17356 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017357 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017358 system.
17359
17360 - "feconn" is the total number of concurrent connections on the frontend when
17361 the session was logged. It is useful to estimate the amount of resource
17362 required to sustain high loads, and to detect when the frontend's "maxconn"
17363 has been reached. Most often when this value increases by huge jumps, it is
17364 because there is congestion on the backend servers, but sometimes it can be
17365 caused by a denial of service attack.
17366
17367 - "beconn" is the total number of concurrent connections handled by the
17368 backend when the session was logged. It includes the total number of
17369 concurrent connections active on servers as well as the number of
17370 connections pending in queues. It is useful to estimate the amount of
17371 additional servers needed to support high loads for a given application.
17372 Most often when this value increases by huge jumps, it is because there is
17373 congestion on the backend servers, but sometimes it can be caused by a
17374 denial of service attack.
17375
17376 - "srv_conn" is the total number of concurrent connections still active on
17377 the server when the session was logged. It can never exceed the server's
17378 configured "maxconn" parameter. If this value is very often close or equal
17379 to the server's "maxconn", it means that traffic regulation is involved a
17380 lot, meaning that either the server's maxconn value is too low, or that
17381 there aren't enough servers to process the load with an optimal response
17382 time. When only one of the server's "srv_conn" is high, it usually means
17383 that this server has some trouble causing the requests to take longer to be
17384 processed than on other servers.
17385
17386 - "retries" is the number of connection retries experienced by this session
17387 when trying to connect to the server. It must normally be zero, unless a
17388 server is being stopped at the same moment the connection was attempted.
17389 Frequent retries generally indicate either a network problem between
17390 haproxy and the server, or a misconfigured system backlog on the server
17391 preventing new connections from being queued. This field may optionally be
17392 prefixed with a '+' sign, indicating that the session has experienced a
17393 redispatch after the maximal retry count has been reached on the initial
17394 server. In this case, the server name appearing in the log is the one the
17395 connection was redispatched to, and not the first one, though both may
17396 sometimes be the same in case of hashing for instance. So as a general rule
17397 of thumb, when a '+' is present in front of the retry count, this count
17398 should not be attributed to the logged server.
17399
17400 - "srv_queue" is the total number of requests which were processed before
17401 this one in the server queue. It is zero when the request has not gone
17402 through the server queue. It makes it possible to estimate the approximate
17403 server's response time by dividing the time spent in queue by the number of
17404 requests in the queue. It is worth noting that if a session experiences a
17405 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010017406 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017407 backend queue unless a redispatch occurs.
17408
17409 - "backend_queue" is the total number of requests which were processed before
17410 this one in the backend's global queue. It is zero when the request has not
17411 gone through the global queue. It makes it possible to estimate the average
17412 queue length, which easily translates into a number of missing servers when
17413 divided by a server's "maxconn" parameter. It is worth noting that if a
17414 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010017415 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017416 through both the server queue and the backend queue unless a redispatch
17417 occurs.
17418
17419 - "captured_request_headers" is a list of headers captured in the request due
17420 to the presence of the "capture request header" statement in the frontend.
17421 Multiple headers can be captured, they will be delimited by a vertical bar
17422 ('|'). When no capture is enabled, the braces do not appear, causing a
17423 shift of remaining fields. It is important to note that this field may
17424 contain spaces, and that using it requires a smarter log parser than when
17425 it's not used. Please consult the section "Capturing HTTP headers and
17426 cookies" below for more details.
17427
17428 - "captured_response_headers" is a list of headers captured in the response
17429 due to the presence of the "capture response header" statement in the
17430 frontend. Multiple headers can be captured, they will be delimited by a
17431 vertical bar ('|'). When no capture is enabled, the braces do not appear,
17432 causing a shift of remaining fields. It is important to note that this
17433 field may contain spaces, and that using it requires a smarter log parser
17434 than when it's not used. Please consult the section "Capturing HTTP headers
17435 and cookies" below for more details.
17436
17437 - "http_request" is the complete HTTP request line, including the method,
17438 request and HTTP version string. Non-printable characters are encoded (see
17439 below the section "Non-printable characters"). This is always the last
17440 field, and it is always delimited by quotes and is the only one which can
17441 contain quotes. If new fields are added to the log format, they will be
17442 added before this field. This field might be truncated if the request is
17443 huge and does not fit in the standard syslog buffer (1024 characters). This
17444 is the reason why this field must always remain the last one.
17445
17446
Cyril Bontédc4d9032012-04-08 21:57:39 +0200174478.2.4. Custom log format
17448------------------------
William Lallemand48940402012-01-30 16:47:22 +010017449
Willy Tarreau2beef582012-12-20 17:22:52 +010017450The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010017451mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010017452
Davor Ocelice9ed2812017-12-25 17:49:28 +010017453HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010017454Variables can take arguments using braces ('{}'), and multiple arguments are
17455separated by commas within the braces. Flags may be added or removed by
17456prefixing them with a '+' or '-' sign.
17457
17458Special variable "%o" may be used to propagate its flags to all other
17459variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010017460("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010017461
Willy Tarreauc8368452012-12-21 00:09:23 +010017462If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020017463as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010017464less common information such as the client's SSL certificate's DN, or to log
17465the key that would be used to store an entry into a stick table.
17466
William Lallemand48940402012-01-30 16:47:22 +010017467Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017468In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010017469in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010017470
Dragan Dosen835b9212016-02-12 13:23:03 +010017471Note: when using the RFC5424 syslog message format, the characters '"',
17472'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
17473https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
17474such cases, the use of the flag "E" should be considered.
17475
William Lallemand48940402012-01-30 16:47:22 +010017476Flags are :
17477 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017478 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010017479 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
17480 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010017481
17482 Example:
17483
17484 log-format %T\ %t\ Some\ Text
17485 log-format %{+Q}o\ %t\ %s\ %{-Q}r
17486
Dragan Dosen835b9212016-02-12 13:23:03 +010017487 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
17488
William Lallemand48940402012-01-30 16:47:22 +010017489At the moment, the default HTTP format is defined this way :
17490
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017491 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
17492 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010017493
William Lallemandbddd4fd2012-02-27 11:23:10 +010017494the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010017495
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017496 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
17497 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
17498 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010017499
William Lallemandbddd4fd2012-02-27 11:23:10 +010017500and the default TCP format is defined this way :
17501
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017502 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
17503 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010017504
William Lallemand48940402012-01-30 16:47:22 +010017505Please refer to the table below for currently defined variables :
17506
William Lallemandbddd4fd2012-02-27 11:23:10 +010017507 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017508 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017509 +---+------+-----------------------------------------------+-------------+
17510 | | %o | special variable, apply flags on all next var | |
17511 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010017512 | | %B | bytes_read (from server to client) | numeric |
17513 | H | %CC | captured_request_cookie | string |
17514 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020017515 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000017516 | H | %HM | HTTP method (ex: POST) | string |
17517 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000017518 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000017519 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
17520 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010017521 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020017522 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017523 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017524 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017525 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020017526 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080017527 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017528 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
17529 | H | %Ti | idle time before the HTTP request | numeric |
17530 | H | %Tq | Th + Ti + TR | numeric |
17531 | H | %TR | time to receive the full request from 1st byte| numeric |
17532 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017533 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017534 | | %Tt | Tt | numeric |
17535 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010017536 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017537 | | %ac | actconn | numeric |
17538 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017539 | | %bc | beconn (backend concurrent connections) | numeric |
17540 | | %bi | backend_source_ip (connecting address) | IP |
17541 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017542 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010017543 | | %ci | client_ip (accepted address) | IP |
17544 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017545 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017546 | | %fc | feconn (frontend concurrent connections) | numeric |
17547 | | %fi | frontend_ip (accepting address) | IP |
17548 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020017549 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020017550 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017551 | | %hr | captured_request_headers default style | string |
17552 | | %hrl | captured_request_headers CLF style | string list |
17553 | | %hs | captured_response_headers default style | string |
17554 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020017555 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020017556 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017557 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017558 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010017559 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017560 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017561 | | %sc | srv_conn (server concurrent connections) | numeric |
17562 | | %si | server_IP (target address) | IP |
17563 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017564 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017565 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
17566 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010017567 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017568 | H | %tr | date_time of HTTP request | date |
17569 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020017570 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017571 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017572 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010017573 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010017574
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020017575 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010017576
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010017577
175788.2.5. Error log format
17579-----------------------
17580
17581When an incoming connection fails due to an SSL handshake or an invalid PROXY
17582protocol header, haproxy will log the event using a shorter, fixed line format.
17583By default, logs are emitted at the LOG_INFO level, unless the option
17584"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017585will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010017586logged if the "dontlognull" option is set.
17587
17588The format looks like this :
17589
17590 >>> Dec 3 18:27:14 localhost \
17591 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
17592 Connection error during SSL handshake
17593
17594 Field Format Extract from the example above
17595 1 process_name '[' pid ']:' haproxy[6103]:
17596 2 client_ip ':' client_port 127.0.0.1:56059
17597 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
17598 4 frontend_name "/" bind_name ":" frt/f1:
17599 5 message Connection error during SSL handshake
17600
17601These fields just provide minimal information to help debugging connection
17602failures.
17603
17604
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176058.3. Advanced logging options
17606-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017607
17608Some advanced logging options are often looked for but are not easy to find out
17609just by looking at the various options. Here is an entry point for the few
17610options which can enable better logging. Please refer to the keywords reference
17611for more information about their usage.
17612
17613
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176148.3.1. Disabling logging of external tests
17615------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017616
17617It is quite common to have some monitoring tools perform health checks on
17618haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
17619commercial load-balancer, and sometimes it will simply be a more complete
17620monitoring system such as Nagios. When the tests are very frequent, users often
17621ask how to disable logging for those checks. There are three possibilities :
17622
17623 - if connections come from everywhere and are just TCP probes, it is often
17624 desired to simply disable logging of connections without data exchange, by
17625 setting "option dontlognull" in the frontend. It also disables logging of
17626 port scans, which may or may not be desired.
17627
17628 - if the connection come from a known source network, use "monitor-net" to
17629 declare this network as monitoring only. Any host in this network will then
17630 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017631 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017632 such as other load-balancers.
17633
17634 - if the tests are performed on a known URI, use "monitor-uri" to declare
17635 this URI as dedicated to monitoring. Any host sending this request will
17636 only get the result of a health-check, and the request will not be logged.
17637
17638
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176398.3.2. Logging before waiting for the session to terminate
17640----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017641
17642The problem with logging at end of connection is that you have no clue about
17643what is happening during very long sessions, such as remote terminal sessions
17644or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010017645"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017646just before data transfer begins. This means that in case of TCP, it will still
17647log the connection status to the server, and in case of HTTP, it will log just
17648after processing the server headers. In this case, the number of bytes reported
17649is the number of header bytes sent to the client. In order to avoid confusion
17650with normal logs, the total time field and the number of bytes are prefixed
17651with a '+' sign which means that real numbers are certainly larger.
17652
17653
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176548.3.3. Raising log level upon errors
17655------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017656
17657Sometimes it is more convenient to separate normal traffic from errors logs,
17658for instance in order to ease error monitoring from log files. When the option
17659"log-separate-errors" is used, connections which experience errors, timeouts,
17660retries, redispatches or HTTP status codes 5xx will see their syslog level
17661raised from "info" to "err". This will help a syslog daemon store the log in
17662a separate file. It is very important to keep the errors in the normal traffic
17663file too, so that log ordering is not altered. You should also be careful if
17664you already have configured your syslog daemon to store all logs higher than
17665"notice" in an "admin" file, because the "err" level is higher than "notice".
17666
17667
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176688.3.4. Disabling logging of successful connections
17669--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017670
17671Although this may sound strange at first, some large sites have to deal with
17672multiple thousands of logs per second and are experiencing difficulties keeping
17673them intact for a long time or detecting errors within them. If the option
17674"dontlog-normal" is set on the frontend, all normal connections will not be
17675logged. In this regard, a normal connection is defined as one without any
17676error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
17677and a response with a status 5xx is not considered normal and will be logged
17678too. Of course, doing is is really discouraged as it will remove most of the
17679useful information from the logs. Do this only if you have no other
17680alternative.
17681
17682
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176838.4. Timing events
17684------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017685
17686Timers provide a great help in troubleshooting network problems. All values are
17687reported in milliseconds (ms). These timers should be used in conjunction with
17688the session termination flags. In TCP mode with "option tcplog" set on the
17689frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017690mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
17691addition, three other measures are provided, "Th", "Ti", and "Tq".
17692
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010017693Timings events in HTTP mode:
17694
17695 first request 2nd request
17696 |<-------------------------------->|<-------------- ...
17697 t tr t tr ...
17698 ---|----|----|----|----|----|----|----|----|--
17699 : Th Ti TR Tw Tc Tr Td : Ti ...
17700 :<---- Tq ---->: :
17701 :<-------------- Tt -------------->:
17702 :<--------- Ta --------->:
17703
17704Timings events in TCP mode:
17705
17706 TCP session
17707 |<----------------->|
17708 t t
17709 ---|----|----|----|----|---
17710 | Th Tw Tc Td |
17711 |<------ Tt ------->|
17712
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017713 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017714 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017715 only happen once during the whole connection's lifetime. A large time here
17716 may indicate that the client only pre-established the connection without
17717 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010017718 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020017719 SSL handshake was very expensive to compute. Please note that this time is
17720 reported only before the first request, so it is safe to average it over
17721 all request to calculate the amortized value. The second and subsequent
17722 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017723
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017724 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
17725 counts between the end of the handshakes and the first byte of the HTTP
17726 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020017727 to count after the end of the transmission the previous response. When a
17728 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
17729 after the previous request. Some browsers pre-establish connections to a
17730 server in order to reduce the latency of a future request, and keep them
17731 pending until they need it. This delay will be reported as the idle time. A
17732 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017733
17734 - TR: total time to get the client request (HTTP mode only). It's the time
17735 elapsed between the first bytes received and the moment the proxy received
17736 the empty line marking the end of the HTTP headers. The value "-1"
17737 indicates that the end of headers has never been seen. This happens when
17738 the client closes prematurely or times out. This time is usually very short
17739 since most requests fit in a single packet. A large time may indicate a
17740 request typed by hand during a test.
17741
17742 - Tq: total time to get the client request from the accept date or since the
17743 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010017744 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 +020017745 returns -1 as well. This timer used to be very useful before the arrival of
17746 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
17747 it in favor of TR nowadays, as the idle time adds a lot of noise to the
17748 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017749
17750 - Tw: total time spent in the queues waiting for a connection slot. It
17751 accounts for backend queue as well as the server queues, and depends on the
17752 queue size, and the time needed for the server to complete previous
17753 requests. The value "-1" means that the request was killed before reaching
17754 the queue, which is generally what happens with invalid or denied requests.
17755
17756 - Tc: total time to establish the TCP connection to the server. It's the time
17757 elapsed between the moment the proxy sent the connection request, and the
17758 moment it was acknowledged by the server, or between the TCP SYN packet and
17759 the matching SYN/ACK packet in return. The value "-1" means that the
17760 connection never established.
17761
17762 - Tr: server response time (HTTP mode only). It's the time elapsed between
17763 the moment the TCP connection was established to the server and the moment
17764 the server sent its complete response headers. It purely shows its request
17765 processing time, without the network overhead due to the data transmission.
17766 It is worth noting that when the client has data to send to the server, for
17767 instance during a POST request, the time already runs, and this can distort
17768 apparent response time. For this reason, it's generally wise not to trust
17769 too much this field for POST requests initiated from clients behind an
17770 untrusted network. A value of "-1" here means that the last the response
17771 header (empty line) was never seen, most likely because the server timeout
17772 stroke before the server managed to process the request.
17773
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017774 - Ta: total active time for the HTTP request, between the moment the proxy
17775 received the first byte of the request header and the emission of the last
17776 byte of the response body. The exception is when the "logasap" option is
17777 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
17778 a '+' sign. From this field, we can deduce "Td", the data transmission time,
17779 by subtracting other timers when valid :
17780
17781 Td = Ta - (TR + Tw + Tc + Tr)
17782
17783 Timers with "-1" values have to be excluded from this equation. Note that
17784 "Ta" can never be negative.
17785
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017786 - Tt: total session duration time, between the moment the proxy accepted it
17787 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017788 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
17789 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017790 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017791
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017792 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017793
17794 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017795 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
17796 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017797
17798These timers provide precious indications on trouble causes. Since the TCP
17799protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
17800that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017801due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
17802"Tt" is close to a timeout value specified in the configuration, it often means
17803that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017804
17805Most common cases :
17806
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017807 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
17808 the client and the proxy. This is very rare on local networks but might
17809 happen when clients are on far remote networks and send large requests. It
17810 may happen that values larger than usual appear here without any network
17811 cause. Sometimes, during an attack or just after a resource starvation has
17812 ended, haproxy may accept thousands of connections in a few milliseconds.
17813 The time spent accepting these connections will inevitably slightly delay
17814 processing of other connections, and it can happen that request times in the
17815 order of a few tens of milliseconds are measured after a few thousands of
17816 new connections have been accepted at once. Using one of the keep-alive
17817 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020017818 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017819
17820 - If "Tc" is close to 3000, a packet has probably been lost between the
17821 server and the proxy during the server connection phase. This value should
17822 always be very low, such as 1 ms on local networks and less than a few tens
17823 of ms on remote networks.
17824
Willy Tarreau55165fe2009-05-10 12:02:55 +020017825 - If "Tr" is nearly always lower than 3000 except some rare values which seem
17826 to be the average majored by 3000, there are probably some packets lost
17827 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017828
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017829 - If "Ta" is large even for small byte counts, it generally is because
17830 neither the client nor the server decides to close the connection while
17831 haproxy is running in tunnel mode and both have agreed on a keep-alive
17832 connection mode. In order to solve this issue, it will be needed to specify
17833 one of the HTTP options to manipulate keep-alive or close options on either
17834 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
17835 important when connection regulation is used with the "maxconn" option on
17836 the servers, since no new connection will be sent to the server until
17837 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017838
17839Other noticeable HTTP log cases ('xx' means any value to be ignored) :
17840
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017841 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017842 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017843 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017844
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017845 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017846 or it aborted too early. Check the session termination flags
17847 then "timeout http-request" and "timeout client" settings.
17848
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017849 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017850 servers were out of order, because the request was invalid
17851 or forbidden by ACL rules. Check the session termination
17852 flags.
17853
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017854 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
17855 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017856 Check the session termination flags, then check the
17857 "timeout connect" setting. Note that the tarpit action might
17858 return similar-looking patterns, with "Tw" equal to the time
17859 the client connection was maintained open.
17860
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017861 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017862 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017863 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017864 termination flags, then check the "timeout server" setting.
17865
17866
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200178678.5. Session state at disconnection
17868-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017869
17870TCP and HTTP logs provide a session termination indicator in the
17871"termination_state" field, just before the number of active connections. It is
178722-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
17873each of which has a special meaning :
17874
17875 - On the first character, a code reporting the first event which caused the
17876 session to terminate :
17877
17878 C : the TCP session was unexpectedly aborted by the client.
17879
17880 S : the TCP session was unexpectedly aborted by the server, or the
17881 server explicitly refused it.
17882
17883 P : the session was prematurely aborted by the proxy, because of a
17884 connection limit enforcement, because a DENY filter was matched,
17885 because of a security check which detected and blocked a dangerous
17886 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010017887 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020017888
17889 L : the session was locally processed by haproxy and was not passed to
17890 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017891
17892 R : a resource on the proxy has been exhausted (memory, sockets, source
17893 ports, ...). Usually, this appears during the connection phase, and
17894 system logs should contain a copy of the precise error. If this
17895 happens, it must be considered as a very serious anomaly which
17896 should be fixed as soon as possible by any means.
17897
17898 I : an internal error was identified by the proxy during a self-check.
17899 This should NEVER happen, and you are encouraged to report any log
17900 containing this, because this would almost certainly be a bug. It
17901 would be wise to preventively restart the process after such an
17902 event too, in case it would be caused by memory corruption.
17903
Simon Horman752dc4a2011-06-21 14:34:59 +090017904 D : the session was killed by haproxy because the server was detected
17905 as down and was configured to kill all connections when going down.
17906
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070017907 U : the session was killed by haproxy on this backup server because an
17908 active server was detected as up and was configured to kill all
17909 backup connections when going up.
17910
Willy Tarreaua2a64e92011-09-07 23:01:56 +020017911 K : the session was actively killed by an admin operating on haproxy.
17912
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017913 c : the client-side timeout expired while waiting for the client to
17914 send or receive data.
17915
17916 s : the server-side timeout expired while waiting for the server to
17917 send or receive data.
17918
17919 - : normal session completion, both the client and the server closed
17920 with nothing left in the buffers.
17921
17922 - on the second character, the TCP or HTTP session state when it was closed :
17923
Willy Tarreauf7b30a92010-12-06 22:59:17 +010017924 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017925 (HTTP mode only). Nothing was sent to any server.
17926
17927 Q : the proxy was waiting in the QUEUE for a connection slot. This can
17928 only happen when servers have a 'maxconn' parameter set. It can
17929 also happen in the global queue after a redispatch consecutive to
17930 a failed attempt to connect to a dying server. If no redispatch is
17931 reported, then no connection attempt was made to any server.
17932
17933 C : the proxy was waiting for the CONNECTION to establish on the
17934 server. The server might at most have noticed a connection attempt.
17935
17936 H : the proxy was waiting for complete, valid response HEADERS from the
17937 server (HTTP only).
17938
17939 D : the session was in the DATA phase.
17940
17941 L : the proxy was still transmitting LAST data to the client while the
17942 server had already finished. This one is very rare as it can only
17943 happen when the client dies while receiving the last packets.
17944
17945 T : the request was tarpitted. It has been held open with the client
17946 during the whole "timeout tarpit" duration or until the client
17947 closed, both of which will be reported in the "Tw" timer.
17948
17949 - : normal session completion after end of data transfer.
17950
17951 - the third character tells whether the persistence cookie was provided by
17952 the client (only in HTTP mode) :
17953
17954 N : the client provided NO cookie. This is usually the case for new
17955 visitors, so counting the number of occurrences of this flag in the
17956 logs generally indicate a valid trend for the site frequentation.
17957
17958 I : the client provided an INVALID cookie matching no known server.
17959 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020017960 cookies between HTTP/HTTPS sites, persistence conditionally
17961 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017962
17963 D : the client provided a cookie designating a server which was DOWN,
17964 so either "option persist" was used and the client was sent to
17965 this server, or it was not set and the client was redispatched to
17966 another server.
17967
Willy Tarreau996a92c2010-10-13 19:30:47 +020017968 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017969 server.
17970
Willy Tarreau996a92c2010-10-13 19:30:47 +020017971 E : the client provided a valid cookie, but with a last date which was
17972 older than what is allowed by the "maxidle" cookie parameter, so
17973 the cookie is consider EXPIRED and is ignored. The request will be
17974 redispatched just as if there was no cookie.
17975
17976 O : the client provided a valid cookie, but with a first date which was
17977 older than what is allowed by the "maxlife" cookie parameter, so
17978 the cookie is consider too OLD and is ignored. The request will be
17979 redispatched just as if there was no cookie.
17980
Willy Tarreauc89ccb62012-04-05 21:18:22 +020017981 U : a cookie was present but was not used to select the server because
17982 some other server selection mechanism was used instead (typically a
17983 "use-server" rule).
17984
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017985 - : does not apply (no cookie set in configuration).
17986
17987 - the last character reports what operations were performed on the persistence
17988 cookie returned by the server (only in HTTP mode) :
17989
17990 N : NO cookie was provided by the server, and none was inserted either.
17991
17992 I : no cookie was provided by the server, and the proxy INSERTED one.
17993 Note that in "cookie insert" mode, if the server provides a cookie,
17994 it will still be overwritten and reported as "I" here.
17995
Willy Tarreau996a92c2010-10-13 19:30:47 +020017996 U : the proxy UPDATED the last date in the cookie that was presented by
17997 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017998 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020017999 date indicated in the cookie. If any other change happens, such as
18000 a redispatch, then the cookie will be marked as inserted instead.
18001
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018002 P : a cookie was PROVIDED by the server and transmitted as-is.
18003
18004 R : the cookie provided by the server was REWRITTEN by the proxy, which
18005 happens in "cookie rewrite" or "cookie prefix" modes.
18006
18007 D : the cookie provided by the server was DELETED by the proxy.
18008
18009 - : does not apply (no cookie set in configuration).
18010
Willy Tarreau996a92c2010-10-13 19:30:47 +020018011The combination of the two first flags gives a lot of information about what
18012was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018013helpful to detect server saturation, network troubles, local system resource
18014starvation, attacks, etc...
18015
18016The most common termination flags combinations are indicated below. They are
18017alphabetically sorted, with the lowercase set just after the upper case for
18018easier finding and understanding.
18019
18020 Flags Reason
18021
18022 -- Normal termination.
18023
18024 CC The client aborted before the connection could be established to the
18025 server. This can happen when haproxy tries to connect to a recently
18026 dead (or unchecked) server, and the client aborts while haproxy is
18027 waiting for the server to respond or for "timeout connect" to expire.
18028
18029 CD The client unexpectedly aborted during data transfer. This can be
18030 caused by a browser crash, by an intermediate equipment between the
18031 client and haproxy which decided to actively break the connection,
18032 by network routing issues between the client and haproxy, or by a
18033 keep-alive session between the server and the client terminated first
18034 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010018035
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018036 cD The client did not send nor acknowledge any data for as long as the
18037 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020018038 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018039
18040 CH The client aborted while waiting for the server to start responding.
18041 It might be the server taking too long to respond or the client
18042 clicking the 'Stop' button too fast.
18043
18044 cH The "timeout client" stroke while waiting for client data during a
18045 POST request. This is sometimes caused by too large TCP MSS values
18046 for PPPoE networks which cannot transport full-sized packets. It can
18047 also happen when client timeout is smaller than server timeout and
18048 the server takes too long to respond.
18049
18050 CQ The client aborted while its session was queued, waiting for a server
18051 with enough empty slots to accept it. It might be that either all the
18052 servers were saturated or that the assigned server was taking too
18053 long a time to respond.
18054
18055 CR The client aborted before sending a full HTTP request. Most likely
18056 the request was typed by hand using a telnet client, and aborted
18057 too early. The HTTP status code is likely a 400 here. Sometimes this
18058 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020018059 and the client. "option http-ignore-probes" can be used to ignore
18060 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018061
18062 cR The "timeout http-request" stroke before the client sent a full HTTP
18063 request. This is sometimes caused by too large TCP MSS values on the
18064 client side for PPPoE networks which cannot transport full-sized
18065 packets, or by clients sending requests by hand and not typing fast
18066 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020018067 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020018068 some browsers started to implement a "pre-connect" feature consisting
18069 in speculatively connecting to some recently visited web sites just
18070 in case the user would like to visit them. This results in many
18071 connections being established to web sites, which end up in 408
18072 Request Timeout if the timeout strikes first, or 400 Bad Request when
18073 the browser decides to close them first. These ones pollute the log
18074 and feed the error counters. Some versions of some browsers have even
18075 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010018076 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020018077 http-ignore-probes" in the frontend, resulting in connections with
18078 zero data transfer to be totally ignored. This will definitely hide
18079 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018080
18081 CT The client aborted while its session was tarpitted. It is important to
18082 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020018083 wrong tarpit rules have been written. If a lot of them happen, it
18084 might make sense to lower the "timeout tarpit" value to something
18085 closer to the average reported "Tw" timer, in order not to consume
18086 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018087
Willy Tarreau570f2212013-06-10 16:42:09 +020018088 LR The request was intercepted and locally handled by haproxy. Generally
18089 it means that this was a redirect or a stats request.
18090
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018091 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018092 the TCP connection (the proxy received a TCP RST or an ICMP message
18093 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010018094 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018095 or no ARP response on local network). When this happens in HTTP mode,
18096 the status code is likely a 502 or 503 here.
18097
18098 sC The "timeout connect" stroke before a connection to the server could
18099 complete. When this happens in HTTP mode, the status code is likely a
18100 503 or 504 here.
18101
18102 SD The connection to the server died with an error during the data
18103 transfer. This usually means that haproxy has received an RST from
18104 the server or an ICMP message from an intermediate equipment while
18105 exchanging data with the server. This can be caused by a server crash
18106 or by a network issue on an intermediate equipment.
18107
18108 sD The server did not send nor acknowledge any data for as long as the
18109 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010018110 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018111 load-balancers, ...), as well as keep-alive sessions maintained
18112 between the client and the server expiring first on haproxy.
18113
18114 SH The server aborted before sending its full HTTP response headers, or
18115 it crashed while processing the request. Since a server aborting at
18116 this moment is very rare, it would be wise to inspect its logs to
18117 control whether it crashed and why. The logged request may indicate a
18118 small set of faulty requests, demonstrating bugs in the application.
18119 Sometimes this might also be caused by an IDS killing the connection
18120 between haproxy and the server.
18121
18122 sH The "timeout server" stroke before the server could return its
18123 response headers. This is the most common anomaly, indicating too
18124 long transactions, probably caused by server or database saturation.
18125 The immediate workaround consists in increasing the "timeout server"
18126 setting, but it is important to keep in mind that the user experience
18127 will suffer from these long response times. The only long term
18128 solution is to fix the application.
18129
18130 sQ The session spent too much time in queue and has been expired. See
18131 the "timeout queue" and "timeout connect" settings to find out how to
18132 fix this if it happens too often. If it often happens massively in
18133 short periods, it may indicate general problems on the affected
18134 servers due to I/O or database congestion, or saturation caused by
18135 external attacks.
18136
18137 PC The proxy refused to establish a connection to the server because the
18138 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020018139 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018140 so that it does not happen anymore. This status is very rare and
18141 might happen when the global "ulimit-n" parameter is forced by hand.
18142
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010018143 PD The proxy blocked an incorrectly formatted chunked encoded message in
18144 a request or a response, after the server has emitted its headers. In
18145 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010018146 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020018147 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010018148
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018149 PH The proxy blocked the server's response, because it was invalid,
18150 incomplete, dangerous (cache control), or matched a security filter.
18151 In any case, an HTTP 502 error is sent to the client. One possible
18152 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010018153 containing unauthorized characters. It is also possible but quite
18154 rare, that the proxy blocked a chunked-encoding request from the
18155 client due to an invalid syntax, before the server responded. In this
18156 case, an HTTP 400 error is sent to the client and reported in the
18157 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018158
18159 PR The proxy blocked the client's HTTP request, either because of an
18160 invalid HTTP syntax, in which case it returned an HTTP 400 error to
18161 the client, or because a deny filter matched, in which case it
18162 returned an HTTP 403 error.
18163
18164 PT The proxy blocked the client's request and has tarpitted its
18165 connection before returning it a 500 server error. Nothing was sent
18166 to the server. The connection was maintained open for as long as
18167 reported by the "Tw" timer field.
18168
18169 RC A local resource has been exhausted (memory, sockets, source ports)
18170 preventing the connection to the server from establishing. The error
18171 logs will tell precisely what was missing. This is very rare and can
18172 only be solved by proper system tuning.
18173
Willy Tarreau996a92c2010-10-13 19:30:47 +020018174The combination of the two last flags gives a lot of information about how
18175persistence was handled by the client, the server and by haproxy. This is very
18176important to troubleshoot disconnections, when users complain they have to
18177re-authenticate. The commonly encountered flags are :
18178
18179 -- Persistence cookie is not enabled.
18180
18181 NN No cookie was provided by the client, none was inserted in the
18182 response. For instance, this can be in insert mode with "postonly"
18183 set on a GET request.
18184
18185 II A cookie designating an invalid server was provided by the client,
18186 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040018187 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020018188 value can be presented by a client when no other server knows it.
18189
18190 NI No cookie was provided by the client, one was inserted in the
18191 response. This typically happens for first requests from every user
18192 in "insert" mode, which makes it an easy way to count real users.
18193
18194 VN A cookie was provided by the client, none was inserted in the
18195 response. This happens for most responses for which the client has
18196 already got a cookie.
18197
18198 VU A cookie was provided by the client, with a last visit date which is
18199 not completely up-to-date, so an updated cookie was provided in
18200 response. This can also happen if there was no date at all, or if
18201 there was a date but the "maxidle" parameter was not set, so that the
18202 cookie can be switched to unlimited time.
18203
18204 EI A cookie was provided by the client, with a last visit date which is
18205 too old for the "maxidle" parameter, so the cookie was ignored and a
18206 new cookie was inserted in the response.
18207
18208 OI A cookie was provided by the client, with a first visit date which is
18209 too old for the "maxlife" parameter, so the cookie was ignored and a
18210 new cookie was inserted in the response.
18211
18212 DI The server designated by the cookie was down, a new server was
18213 selected and a new cookie was emitted in the response.
18214
18215 VI The server designated by the cookie was not marked dead but could not
18216 be reached. A redispatch happened and selected another one, which was
18217 then advertised in the response.
18218
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018219
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200182208.6. Non-printable characters
18221-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018222
18223In order not to cause trouble to log analysis tools or terminals during log
18224consulting, non-printable characters are not sent as-is into log files, but are
18225converted to the two-digits hexadecimal representation of their ASCII code,
18226prefixed by the character '#'. The only characters that can be logged without
18227being escaped are comprised between 32 and 126 (inclusive). Obviously, the
18228escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
18229is the same for the character '"' which becomes "#22", as well as '{', '|' and
18230'}' when logging headers.
18231
18232Note that the space character (' ') is not encoded in headers, which can cause
18233issues for tools relying on space count to locate fields. A typical header
18234containing spaces is "User-Agent".
18235
18236Last, it has been observed that some syslog daemons such as syslog-ng escape
18237the quote ('"') with a backslash ('\'). The reverse operation can safely be
18238performed since no quote may appear anywhere else in the logs.
18239
18240
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200182418.7. Capturing HTTP cookies
18242---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018243
18244Cookie capture simplifies the tracking a complete user session. This can be
18245achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018246section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018247cookie will simultaneously be checked in the request ("Cookie:" header) and in
18248the response ("Set-Cookie:" header). The respective values will be reported in
18249the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018250locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018251not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
18252user switches to a new session for example, because the server will reassign it
18253a new cookie. It is also possible to detect if a server unexpectedly sets a
18254wrong cookie to a client, leading to session crossing.
18255
18256 Examples :
18257 # capture the first cookie whose name starts with "ASPSESSION"
18258 capture cookie ASPSESSION len 32
18259
18260 # capture the first cookie whose name is exactly "vgnvisitor"
18261 capture cookie vgnvisitor= len 32
18262
18263
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200182648.8. Capturing HTTP headers
18265---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018266
18267Header captures are useful to track unique request identifiers set by an upper
18268proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
18269the response, one can search for information about the response length, how the
18270server asked the cache to behave, or an object location during a redirection.
18271
18272Header captures are performed using the "capture request header" and "capture
18273response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018274section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018275
18276It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018277time. Non-existent headers are logged as empty strings, and if one header
18278appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018279are grouped within braces '{' and '}' in the same order as they were declared,
18280and delimited with a vertical bar '|' without any space. Response headers
18281follow the same representation, but are displayed after a space following the
18282request headers block. These blocks are displayed just before the HTTP request
18283in the logs.
18284
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020018285As a special case, it is possible to specify an HTTP header capture in a TCP
18286frontend. The purpose is to enable logging of headers which will be parsed in
18287an HTTP backend if the request is then switched to this HTTP backend.
18288
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018289 Example :
18290 # This instance chains to the outgoing proxy
18291 listen proxy-out
18292 mode http
18293 option httplog
18294 option logasap
18295 log global
18296 server cache1 192.168.1.1:3128
18297
18298 # log the name of the virtual server
18299 capture request header Host len 20
18300
18301 # log the amount of data uploaded during a POST
18302 capture request header Content-Length len 10
18303
18304 # log the beginning of the referrer
18305 capture request header Referer len 20
18306
18307 # server name (useful for outgoing proxies only)
18308 capture response header Server len 20
18309
18310 # logging the content-length is useful with "option logasap"
18311 capture response header Content-Length len 10
18312
Davor Ocelice9ed2812017-12-25 17:49:28 +010018313 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018314 capture response header Cache-Control len 8
18315
18316 # the Via header will report the next proxy's name
18317 capture response header Via len 20
18318
18319 # log the URL location during a redirection
18320 capture response header Location len 20
18321
18322 >>> Aug 9 20:26:09 localhost \
18323 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
18324 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
18325 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
18326 "GET http://fr.adserver.yahoo.com/"
18327
18328 >>> Aug 9 20:30:46 localhost \
18329 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
18330 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
18331 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018332 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018333
18334 >>> Aug 9 20:30:46 localhost \
18335 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
18336 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
18337 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
18338 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018339 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018340
18341
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200183428.9. Examples of logs
18343---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018344
18345These are real-world examples of logs accompanied with an explanation. Some of
18346them have been made up by hand. The syslog part has been removed for better
18347reading. Their sole purpose is to explain how to decipher them.
18348
18349 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
18350 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
18351 "HEAD / HTTP/1.0"
18352
18353 => long request (6.5s) entered by hand through 'telnet'. The server replied
18354 in 147 ms, and the session ended normally ('----')
18355
18356 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
18357 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
18358 0/9 "HEAD / HTTP/1.0"
18359
18360 => Idem, but the request was queued in the global queue behind 9 other
18361 requests, and waited there for 1230 ms.
18362
18363 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
18364 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
18365 "GET /image.iso HTTP/1.0"
18366
18367 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010018368 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018369 14 ms, 243 bytes of headers were sent to the client, and total time from
18370 accept to first data byte is 30 ms.
18371
18372 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
18373 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
18374 "GET /cgi-bin/bug.cgi? HTTP/1.0"
18375
18376 => the proxy blocked a server response either because of an "rspdeny" or
18377 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020018378 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018379 risked being cached. In this case, the response is replaced with a "502
18380 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
18381 to return the 502 and not the server.
18382
18383 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018384 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 +010018385
18386 => the client never completed its request and aborted itself ("C---") after
18387 8.5s, while the proxy was waiting for the request headers ("-R--").
18388 Nothing was sent to any server.
18389
18390 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
18391 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
18392
18393 => The client never completed its request, which was aborted by the
18394 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010018395 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018396 send a 408 return code to the client.
18397
18398 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
18399 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
18400
18401 => This log was produced with "option tcplog". The client timed out after
18402 5 seconds ("c----").
18403
18404 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
18405 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010018406 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018407
18408 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020018409 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010018410 (config says 'retries 3'), and no redispatch (otherwise we would have
18411 seen "/+3"). Status code 503 was returned to the client. There were 115
18412 connections on this server, 202 connections on this proxy, and 205 on
18413 the global process. It is possible that the server refused the
18414 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010018415
Willy Tarreau52b2d222011-09-07 23:48:48 +020018416
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200184179. Supported filters
18418--------------------
18419
18420Here are listed officially supported filters with the list of parameters they
18421accept. Depending on compile options, some of these filters might be
18422unavailable. The list of available filters is reported in haproxy -vv.
18423
18424See also : "filter"
18425
184269.1. Trace
18427----------
18428
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010018429filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018430
18431 Arguments:
18432 <name> is an arbitrary name that will be reported in
18433 messages. If no name is provided, "TRACE" is used.
18434
18435 <random-parsing> enables the random parsing of data exchanged between
18436 the client and the server. By default, this filter
18437 parses all available data. With this parameter, it
18438 only parses a random amount of the available data.
18439
Davor Ocelice9ed2812017-12-25 17:49:28 +010018440 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018441 default, this filter forwards all previously parsed
18442 data. With this parameter, it only forwards a random
18443 amount of the parsed data.
18444
Davor Ocelice9ed2812017-12-25 17:49:28 +010018445 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010018446
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018447This filter can be used as a base to develop new filters. It defines all
18448callbacks and print a message on the standard error stream (stderr) with useful
18449information for all of them. It may be useful to debug the activity of other
18450filters or, quite simply, HAProxy's activity.
18451
18452Using <random-parsing> and/or <random-forwarding> parameters is a good way to
18453tests the behavior of a filter that parses data exchanged between a client and
18454a server by adding some latencies in the processing.
18455
18456
184579.2. HTTP compression
18458---------------------
18459
18460filter compression
18461
18462The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
18463keyword must still be used to enable and configure the HTTP compression. And
Christopher Faulet27d93c32018-12-15 22:32:02 +010018464when no other filter is used, it is enough. When used with the cache enabled,
18465it is also enough. In this case, the compression is always done after the
18466response is stored in the cache. But it is mandatory to explicitly use a filter
18467line to enable the HTTP compression when at least one filter other than the
18468cache is used for the same listener/frontend/backend. This is important to know
18469the filters evaluation order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018470
Christopher Faulet27d93c32018-12-15 22:32:02 +010018471See also : "compression" and section 9.4 about the cache filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020018472
18473
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200184749.3. Stream Processing Offload Engine (SPOE)
18475--------------------------------------------
18476
18477filter spoe [engine <name>] config <file>
18478
18479 Arguments :
18480
18481 <name> is the engine name that will be used to find the right scope in
18482 the configuration file. If not provided, all the file will be
18483 parsed.
18484
18485 <file> is the path of the engine configuration file. This file can
18486 contain configuration of several engines. In this case, each
18487 part must be placed in its own scope.
18488
18489The Stream Processing Offload Engine (SPOE) is a filter communicating with
18490external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010018491streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020018492exchanged with them are configured in dedicated files, for the main part. It
18493also requires dedicated backends, defined in HAProxy configuration.
18494
18495SPOE communicates with external components using an in-house binary protocol,
18496the Stream Processing Offload Protocol (SPOP).
18497
Tim Düsterhus4896c442016-11-29 02:15:19 +010018498For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020018499"doc/SPOE.txt".
18500
18501Important note:
18502 The SPOE filter is highly experimental for now and was not heavily
18503 tested. It is really not production ready. So use it carefully.
18504
Christopher Faulet99a17a22018-12-11 09:18:27 +0100185059.4. Cache
18506----------
18507
18508filter cache <name>
18509
18510 Arguments :
18511
18512 <name> is name of the cache section this filter will use.
18513
18514The cache uses a filter to store cacheable responses. The HTTP rules
18515"cache-store" and "cache-use" must be used to define how and when to use a
John Roesler27c754c2019-07-10 15:45:51 -050018516cache. By default the corresponding filter is implicitly defined. And when no
Christopher Faulet27d93c32018-12-15 22:32:02 +010018517other filters than cache or compression are used, it is enough. In such case,
18518the compression filter is always evaluated after the cache filter. But it is
18519mandatory to explicitly use a filter line to use a cache when at least one
18520filter other than the compression is used for the same
18521listener/frontend/backend. This is important to know the filters evaluation
18522order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010018523
Christopher Faulet27d93c32018-12-15 22:32:02 +010018524See also : section 9.2 about the compression filter and section 10 about cache.
Christopher Faulet99a17a22018-12-11 09:18:27 +010018525
William Lallemand86d0df02017-11-24 21:36:45 +01001852610. Cache
18527---------
18528
18529HAProxy provides a cache, which was designed to perform cache on small objects
18530(favicon, css...). This is a minimalist low-maintenance cache which runs in
18531RAM.
18532
18533The cache is based on a memory which is shared between processes and threads,
Cyril Bonté7b888f12017-11-26 22:24:31 +010018534this memory is split in blocks of 1k.
William Lallemand86d0df02017-11-24 21:36:45 +010018535
18536If an object is not used anymore, it can be deleted to store a new object
18537independently of its expiration date. The oldest objects are deleted first
18538when we try to allocate a new one.
18539
Cyril Bonté7b888f12017-11-26 22:24:31 +010018540The cache uses a hash of the host header and the URI as the key.
William Lallemand86d0df02017-11-24 21:36:45 +010018541
18542It's possible to view the status of a cache using the Unix socket command
18543"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
18544for more details.
18545
18546When an object is delivered from the cache, the server name in the log is
18547replaced by "<CACHE>".
18548
Cyril Bonté7b888f12017-11-26 22:24:31 +01001854910.1. Limitation
18550----------------
William Lallemand86d0df02017-11-24 21:36:45 +010018551
18552The cache won't store and won't deliver objects in these cases:
18553
18554- If the response is not a 200
18555- If the response contains a Vary header
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020018556- If the Content-Length + the headers size is greater than "max-object-size"
William Lallemand86d0df02017-11-24 21:36:45 +010018557- If the response is not cacheable
18558
18559- If the request is not a GET
18560- If the HTTP version of the request is smaller than 1.1
William Lallemand8a16fe02018-05-22 11:04:33 +020018561- If the request contains an Authorization header
William Lallemand86d0df02017-11-24 21:36:45 +010018562
Christopher Faulet99a17a22018-12-11 09:18:27 +010018563Caution!: For HAProxy version prior to 1.9, due to the limitation of the
18564filters, it is not recommended to use the cache with other filters. Using them
18565can cause undefined behavior if they modify the response (compression for
18566example). For HAProxy 1.9 and greater, it is safe, for HTX proxies only (see
18567"option http-use-htx" for details).
William Lallemand86d0df02017-11-24 21:36:45 +010018568
Cyril Bonté7b888f12017-11-26 22:24:31 +01001856910.2. Setup
18570-----------
William Lallemand86d0df02017-11-24 21:36:45 +010018571
18572To setup a cache, you must define a cache section and use it in a proxy with
18573the corresponding http-request and response actions.
18574
Cyril Bonté7b888f12017-11-26 22:24:31 +01001857510.2.1. Cache section
18576---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010018577
18578cache <name>
18579 Declare a cache section, allocate a shared cache memory named <name>, the
18580 size of cache is mandatory.
18581
18582total-max-size <megabytes>
Davor Ocelice9ed2812017-12-25 17:49:28 +010018583 Define the size in RAM of the cache in megabytes. This size is split in
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020018584 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
William Lallemand86d0df02017-11-24 21:36:45 +010018585
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020018586max-object-size <bytes>
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020018587 Define the maximum size of the objects to be cached. Must not be greater than
18588 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
18589 All objects with sizes larger than "max-object-size" will not be cached.
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020018590
William Lallemand86d0df02017-11-24 21:36:45 +010018591max-age <seconds>
18592 Define the maximum expiration duration. The expiration is set has the lowest
18593 value between the s-maxage or max-age (in this order) directive in the
18594 Cache-Control response header and this value. The default value is 60
18595 seconds, which means that you can't cache an object more than 60 seconds by
18596 default.
18597
Cyril Bonté7b888f12017-11-26 22:24:31 +01001859810.2.2. Proxy section
18599---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010018600
Jarno Huuskonen251a6b72019-01-04 14:05:02 +020018601http-request cache-use <name> [ { if | unless } <condition> ]
William Lallemand86d0df02017-11-24 21:36:45 +010018602 Try to deliver a cached object from the cache <name>. This directive is also
18603 mandatory to store the cache as it calculates the cache hash. If you want to
18604 use a condition for both storage and delivering that's a good idea to put it
18605 after this one.
18606
Jarno Huuskonen251a6b72019-01-04 14:05:02 +020018607http-response cache-store <name> [ { if | unless } <condition> ]
William Lallemand86d0df02017-11-24 21:36:45 +010018608 Store an http-response within the cache. The storage of the response headers
18609 is done at this step, which means you can use others http-response actions
18610 to modify headers before or after the storage of the response. This action
18611 is responsible for the setup of the cache storage filter.
18612
18613
18614Example:
18615
18616 backend bck1
18617 mode http
18618
18619 http-request cache-use foobar
18620 http-response cache-store foobar
18621 server srv1 127.0.0.1:80
18622
18623 cache foobar
18624 total-max-size 4
18625 max-age 240
18626
Willy Tarreau0ba27502007-12-24 16:55:16 +010018627/*
18628 * Local variables:
18629 * fill-column: 79
18630 * End:
18631 */