blob: a8022fc16235b66c50f04ca2a3043baef9a7527c [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 Tarreau6c1b6672019-02-26 16:43:49 +01007 2019/02/26
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
Willy Tarreauc57f0e22009-05-10 13:12:33 +020054
554. Proxies
564.1. Proxy keywords matrix
574.2. Alphabetically sorted keywords reference
58
Davor Ocelice9ed2812017-12-25 17:49:28 +0100595. Bind and server options
Willy Tarreau086fbf52012-09-24 20:34:51 +0200605.1. Bind options
615.2. Server and default-server options
Baptiste Assmann1fa66662015-04-14 00:28:47 +0200625.3. Server DNS resolution
635.3.1. Global overview
645.3.2. The resolvers section
Willy Tarreauc57f0e22009-05-10 13:12:33 +020065
666. HTTP header manipulation
67
Willy Tarreau74ca5042013-06-11 23:12:07 +0200687. Using ACLs and fetching samples
697.1. ACL basics
707.1.1. Matching booleans
717.1.2. Matching integers
727.1.3. Matching strings
737.1.4. Matching regular expressions (regexes)
747.1.5. Matching arbitrary data blocks
757.1.6. Matching IPv4 and IPv6 addresses
767.2. Using ACLs to form conditions
777.3. Fetching samples
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200787.3.1. Converters
797.3.2. Fetching samples from internal states
807.3.3. Fetching samples at Layer 4
817.3.4. Fetching samples at Layer 5
827.3.5. Fetching samples from buffer contents (Layer 6)
837.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +0200847.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020085
868. Logging
878.1. Log levels
888.2. Log formats
898.2.1. Default log format
908.2.2. TCP log format
918.2.3. HTTP log format
William Lallemand48940402012-01-30 16:47:22 +0100928.2.4. Custom log format
Willy Tarreau5f51e1a2012-12-03 18:40:10 +0100938.2.5. Error log format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200948.3. Advanced logging options
958.3.1. Disabling logging of external tests
968.3.2. Logging before waiting for the session to terminate
978.3.3. Raising log level upon errors
988.3.4. Disabling logging of successful connections
998.4. Timing events
1008.5. Session state at disconnection
1018.6. Non-printable characters
1028.7. Capturing HTTP cookies
1038.8. Capturing HTTP headers
1048.9. Examples of logs
105
Christopher Fauletc3fe5332016-04-07 15:30:10 +02001069. Supported filters
1079.1. Trace
1089.2. HTTP compression
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +02001099.3. Stream Processing Offload Engine (SPOE)
Christopher Faulet99a17a22018-12-11 09:18:27 +01001109.4. Cache
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200111
William Lallemand86d0df02017-11-24 21:36:45 +010011210. Cache
Cyril Bonté7b888f12017-11-26 22:24:31 +010011310.1. Limitation
11410.2. Setup
11510.2.1. Cache section
11610.2.2. Proxy section
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200117
1181. Quick reminder about HTTP
119----------------------------
120
Davor Ocelice9ed2812017-12-25 17:49:28 +0100121When HAProxy is running in HTTP mode, both the request and the response are
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200122fully analyzed and indexed, thus it becomes possible to build matching criteria
123on almost anything found in the contents.
124
125However, it is important to understand how HTTP requests and responses are
126formed, and how HAProxy decomposes them. It will then become easier to write
127correct rules and to debug existing configurations.
128
129
1301.1. The HTTP transaction model
131-------------------------------
132
133The HTTP protocol is transaction-driven. This means that each request will lead
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100134to one and only one response. Traditionally, a TCP connection is established
Davor Ocelice9ed2812017-12-25 17:49:28 +0100135from the client to the server, a request is sent by the client through the
136connection, the server responds, and the connection is closed. A new request
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200137will involve a new connection :
138
139 [CON1] [REQ1] ... [RESP1] [CLO1] [CON2] [REQ2] ... [RESP2] [CLO2] ...
140
141In this mode, called the "HTTP close" mode, there are as many connection
142establishments as there are HTTP transactions. Since the connection is closed
143by the server after the response, the client does not need to know the content
144length.
145
146Due to the transactional nature of the protocol, it was possible to improve it
147to avoid closing a connection between two subsequent transactions. In this mode
148however, it is mandatory that the server indicates the content length for each
149response so that the client does not wait indefinitely. For this, a special
150header is used: "Content-length". This mode is called the "keep-alive" mode :
151
152 [CON] [REQ1] ... [RESP1] [REQ2] ... [RESP2] [CLO] ...
153
154Its advantages are a reduced latency between transactions, and less processing
155power required on the server side. It is generally better than the close mode,
156but not always because the clients often limit their concurrent connections to
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200157a smaller value.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200158
Willy Tarreau95c4e142017-11-26 12:18:55 +0100159Another improvement in the communications is the pipelining mode. It still uses
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200160keep-alive, but the client does not wait for the first response to send the
161second request. This is useful for fetching large number of images composing a
162page :
163
164 [CON] [REQ1] [REQ2] ... [RESP1] [RESP2] [CLO] ...
165
166This can obviously have a tremendous benefit on performance because the network
167latency is eliminated between subsequent requests. Many HTTP agents do not
168correctly support pipelining since there is no way to associate a response with
169the corresponding request in HTTP. For this reason, it is mandatory for the
Cyril Bonté78caf842010-03-10 22:41:43 +0100170server to reply in the exact same order as the requests were received.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200171
Willy Tarreau95c4e142017-11-26 12:18:55 +0100172The next improvement is the multiplexed mode, as implemented in HTTP/2. This
173time, each transaction is assigned a single stream identifier, and all streams
174are multiplexed over an existing connection. Many requests can be sent in
175parallel by the client, and responses can arrive in any order since they also
176carry the stream identifier.
177
Willy Tarreau70dffda2014-01-30 03:07:23 +0100178By default HAProxy operates in keep-alive mode with regards to persistent
179connections: for each connection it processes each request and response, and
180leaves the connection idle on both sides between the end of a response and the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100181start of a new request. When it receives HTTP/2 connections from a client, it
182processes all the requests in parallel and leaves the connection idling,
183waiting for new requests, just as if it was a keep-alive HTTP connection.
Patrick Mezard9ec2ec42010-06-12 17:02:45 +0200184
Christopher Faulet315b39c2018-09-21 16:26:19 +0200185HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +0100186 - keep alive : all requests and responses are processed (default)
187 - tunnel : only the first request and response are processed,
188 everything else is forwarded with no analysis.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100189 - server close : the server-facing connection is closed after the response.
Christopher Faulet315b39c2018-09-21 16:26:19 +0200190 - close : the connection is actively closed after end of response.
Willy Tarreau70dffda2014-01-30 03:07:23 +0100191
Davor Ocelice9ed2812017-12-25 17:49:28 +0100192For HTTP/2, the connection mode resembles more the "server close" mode : given
193the independence of all streams, there is currently no place to hook the idle
Willy Tarreau95c4e142017-11-26 12:18:55 +0100194server connection after a response, so it is closed after the response. HTTP/2
195is only supported for incoming connections, not on connections going to
196servers.
197
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200198
1991.2. HTTP request
200-----------------
201
202First, let's consider this HTTP request :
203
204 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100205 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200206 1 GET /serv/login.php?lang=en&profile=2 HTTP/1.1
207 2 Host: www.mydomain.com
208 3 User-agent: my small browser
209 4 Accept: image/jpeg, image/gif
210 5 Accept: image/png
211
212
2131.2.1. The Request line
214-----------------------
215
216Line 1 is the "request line". It is always composed of 3 fields :
217
218 - a METHOD : GET
219 - a URI : /serv/login.php?lang=en&profile=2
220 - a version tag : HTTP/1.1
221
222All of them are delimited by what the standard calls LWS (linear white spaces),
223which are commonly spaces, but can also be tabs or line feeds/carriage returns
224followed by spaces/tabs. The method itself cannot contain any colon (':') and
225is limited to alphabetic letters. All those various combinations make it
226desirable that HAProxy performs the splitting itself rather than leaving it to
227the user to write a complex or inaccurate regular expression.
228
229The URI itself can have several forms :
230
231 - A "relative URI" :
232
233 /serv/login.php?lang=en&profile=2
234
235 It is a complete URL without the host part. This is generally what is
236 received by servers, reverse proxies and transparent proxies.
237
238 - An "absolute URI", also called a "URL" :
239
240 http://192.168.0.12:8080/serv/login.php?lang=en&profile=2
241
242 It is composed of a "scheme" (the protocol name followed by '://'), a host
243 name or address, optionally a colon (':') followed by a port number, then
244 a relative URI beginning at the first slash ('/') after the address part.
245 This is generally what proxies receive, but a server supporting HTTP/1.1
246 must accept this form too.
247
248 - a star ('*') : this form is only accepted in association with the OPTIONS
249 method and is not relayable. It is used to inquiry a next hop's
250 capabilities.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100251
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200252 - an address:port combination : 192.168.0.12:80
253 This is used with the CONNECT method, which is used to establish TCP
254 tunnels through HTTP proxies, generally for HTTPS, but sometimes for
255 other protocols too.
256
257In a relative URI, two sub-parts are identified. The part before the question
258mark is called the "path". It is typically the relative path to static objects
259on the server. The part after the question mark is called the "query string".
260It is mostly used with GET requests sent to dynamic scripts and is very
261specific to the language, framework or application in use.
262
Willy Tarreau95c4e142017-11-26 12:18:55 +0100263HTTP/2 doesn't convey a version information with the request, so the version is
Davor Ocelice9ed2812017-12-25 17:49:28 +0100264assumed to be the same as the one of the underlying protocol (i.e. "HTTP/2").
Willy Tarreau95c4e142017-11-26 12:18:55 +0100265However, haproxy natively processes HTTP/1.x requests and headers, so requests
266received over an HTTP/2 connection are transcoded to HTTP/1.1 before being
267processed. This explains why they still appear as "HTTP/1.1" in haproxy's logs
268as well as in server logs.
269
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200270
2711.2.2. The request headers
272--------------------------
273
274The headers start at the second line. They are composed of a name at the
275beginning of the line, immediately followed by a colon (':'). Traditionally,
276an LWS is added after the colon but that's not required. Then come the values.
277Multiple identical headers may be folded into one single line, delimiting the
278values with commas, provided that their order is respected. This is commonly
279encountered in the "Cookie:" field. A header may span over multiple lines if
280the subsequent lines begin with an LWS. In the example in 1.2, lines 4 and 5
281define a total of 3 values for the "Accept:" header.
282
Davor Ocelice9ed2812017-12-25 17:49:28 +0100283Contrary to a common misconception, header names are not case-sensitive, and
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200284their values are not either if they refer to other header names (such as the
Willy Tarreau95c4e142017-11-26 12:18:55 +0100285"Connection:" header). In HTTP/2, header names are always sent in lower case,
286as can be seen when running in debug mode.
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200287
288The end of the headers is indicated by the first empty line. People often say
289that it's a double line feed, which is not exact, even if a double line feed
290is one valid form of empty line.
291
292Fortunately, HAProxy takes care of all these complex combinations when indexing
293headers, checking values and counting them, so there is no reason to worry
294about the way they could be written, but it is important not to accuse an
295application of being buggy if it does unusual, valid things.
296
297Important note:
Lukas Tribus23953682017-04-28 13:24:30 +0000298 As suggested by RFC7231, HAProxy normalizes headers by replacing line breaks
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200299 in the middle of headers by LWS in order to join multi-line headers. This
300 is necessary for proper analysis and helps less capable HTTP parsers to work
301 correctly and not to be fooled by such complex constructs.
302
303
3041.3. HTTP response
305------------------
306
307An HTTP response looks very much like an HTTP request. Both are called HTTP
308messages. Let's consider this HTTP response :
309
310 Line Contents
Willy Tarreaud72758d2010-01-12 10:42:19 +0100311 number
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200312 1 HTTP/1.1 200 OK
313 2 Content-length: 350
314 3 Content-Type: text/html
315
Willy Tarreau816b9792009-09-15 21:25:21 +0200316As a special case, HTTP supports so called "Informational responses" as status
317codes 1xx. These messages are special in that they don't convey any part of the
318response, they're just used as sort of a signaling message to ask a client to
Willy Tarreau5843d1a2010-02-01 15:13:32 +0100319continue to post its request for instance. In the case of a status 100 response
320the requested information will be carried by the next non-100 response message
321following the informational one. This implies that multiple responses may be
322sent to a single request, and that this only works when keep-alive is enabled
323(1xx messages are HTTP/1.1 only). HAProxy handles these messages and is able to
324correctly forward and skip them, and only process the next non-100 response. As
325such, these messages are neither logged nor transformed, unless explicitly
326state otherwise. Status 101 messages indicate that the protocol is changing
327over the same connection and that haproxy must switch to tunnel mode, just as
328if a CONNECT had occurred. Then the Upgrade header would contain additional
329information about the type of protocol the connection is switching to.
Willy Tarreau816b9792009-09-15 21:25:21 +0200330
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200331
Davor Ocelice9ed2812017-12-25 17:49:28 +01003321.3.1. The response line
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200333------------------------
334
335Line 1 is the "response line". It is always composed of 3 fields :
336
337 - a version tag : HTTP/1.1
338 - a status code : 200
339 - a reason : OK
340
341The status code is always 3-digit. The first digit indicates a general status :
Davor Ocelice9ed2812017-12-25 17:49:28 +0100342 - 1xx = informational message to be skipped (e.g. 100, 101)
343 - 2xx = OK, content is following (e.g. 200, 206)
344 - 3xx = OK, no content following (e.g. 302, 304)
345 - 4xx = error caused by the client (e.g. 401, 403, 404)
346 - 5xx = error caused by the server (e.g. 500, 502, 503)
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200347
Lukas Tribus23953682017-04-28 13:24:30 +0000348Please refer to RFC7231 for the detailed meaning of all such codes. The
Willy Tarreaud72758d2010-01-12 10:42:19 +0100349"reason" field is just a hint, but is not parsed by clients. Anything can be
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200350found there, but it's a common practice to respect the well-established
351messages. It can be composed of one or multiple words, such as "OK", "Found",
352or "Authentication Required".
353
Davor Ocelice9ed2812017-12-25 17:49:28 +0100354HAProxy may emit the following status codes by itself :
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200355
356 Code When / reason
357 200 access to stats page, and when replying to monitoring requests
358 301 when performing a redirection, depending on the configured code
359 302 when performing a redirection, depending on the configured code
360 303 when performing a redirection, depending on the configured code
Willy Tarreaub67fdc42013-03-29 19:28:11 +0100361 307 when performing a redirection, depending on the configured code
362 308 when performing a redirection, depending on the configured code
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200363 400 for an invalid or too large request
364 401 when an authentication is required to perform the action (when
365 accessing the stats page)
366 403 when a request is forbidden by a "block" ACL or "reqdeny" filter
367 408 when the request timeout strikes before the request is complete
368 500 when haproxy encounters an unrecoverable internal error, such as a
369 memory allocation failure, which should never happen
370 502 when the server returns an empty, invalid or incomplete response, or
371 when an "rspdeny" filter blocks the response.
372 503 when no server was available to handle the request, or in response to
373 monitoring requests which match the "monitor fail" condition
374 504 when the response timeout strikes before the server responds
375
376The error 4xx and 5xx codes above may be customized (see "errorloc" in section
3774.2).
378
379
3801.3.2. The response headers
381---------------------------
382
383Response headers work exactly like request headers, and as such, HAProxy uses
384the same parsing function for both. Please refer to paragraph 1.2.2 for more
385details.
386
387
3882. Configuring HAProxy
389----------------------
390
3912.1. Configuration file format
392------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +0200393
394HAProxy's configuration process involves 3 major sources of parameters :
395
396 - the arguments from the command-line, which always take precedence
397 - the "global" section, which sets process-wide parameters
398 - the proxies sections which can take form of "defaults", "listen",
399 "frontend" and "backend".
400
Willy Tarreau0ba27502007-12-24 16:55:16 +0100401The configuration file syntax consists in lines beginning with a keyword
402referenced in this manual, optionally followed by one or several parameters
William Lallemandf9873ba2015-05-05 17:37:14 +0200403delimited by spaces.
Willy Tarreau0ba27502007-12-24 16:55:16 +0100404
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200405
William Lallemandf9873ba2015-05-05 17:37:14 +02004062.2. Quoting and escaping
407-------------------------
408
409HAProxy's configuration introduces a quoting and escaping system similar to
410many programming languages. The configuration file supports 3 types: escaping
411with a backslash, weak quoting with double quotes, and strong quoting with
412single quotes.
413
414If spaces have to be entered in strings, then they must be escaped by preceding
415them by a backslash ('\') or by quoting them. Backslashes also have to be
416escaped by doubling or strong quoting them.
417
418Escaping is achieved by preceding a special character by a backslash ('\'):
419
420 \ to mark a space and differentiate it from a delimiter
421 \# to mark a hash and differentiate it from a comment
422 \\ to use a backslash
423 \' to use a single quote and differentiate it from strong quoting
424 \" to use a double quote and differentiate it from weak quoting
425
426Weak quoting is achieved by using double quotes (""). Weak quoting prevents
427the interpretation of:
428
429 space as a parameter separator
430 ' single quote as a strong quoting delimiter
431 # hash as a comment start
432
William Lallemandb2f07452015-05-12 14:27:13 +0200433Weak quoting permits the interpretation of variables, if you want to use a non
434-interpreted dollar within a double quoted string, you should escape it with a
435backslash ("\$"), it does not work outside weak quoting.
436
437Interpretation of escaping and special characters are not prevented by weak
William Lallemandf9873ba2015-05-05 17:37:14 +0200438quoting.
439
440Strong quoting is achieved by using single quotes (''). Inside single quotes,
441nothing is interpreted, it's the efficient way to quote regexes.
442
443Quoted and escaped strings are replaced in memory by their interpreted
444equivalent, it allows you to perform concatenation.
445
446 Example:
447 # those are equivalents:
448 log-format %{+Q}o\ %t\ %s\ %{-Q}r
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
454 # those are equivalents:
455 reqrep "^([^\ :]*)\ /static/(.*)" \1\ /\2
456 reqrep "^([^ :]*)\ /static/(.*)" '\1 /\2'
457 reqrep "^([^ :]*)\ /static/(.*)" "\1 /\2"
458 reqrep "^([^ :]*)\ /static/(.*)" "\1\ /\2"
459
460
William Lallemandb2f07452015-05-12 14:27:13 +02004612.3. Environment variables
462--------------------------
463
464HAProxy's configuration supports environment variables. Those variables are
465interpreted only within double quotes. Variables are expanded during the
466configuration parsing. Variable names must be preceded by a dollar ("$") and
467optionally enclosed with braces ("{}") similarly to what is done in Bourne
468shell. Variable names can contain alphanumerical characters or the character
469underscore ("_") but should not start with a digit.
470
471 Example:
472
473 bind "fd@${FD_APP1}"
474
475 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
476
477 user "$HAPROXY_USER"
478
William Lallemanddaf4cd22018-04-17 16:46:13 +0200479A special variable $HAPROXY_LOCALPEER is defined at the startup of the process
480which contains the name of the local peer. (See "-L" in the management guide.)
481
William Lallemandb2f07452015-05-12 14:27:13 +0200482
4832.4. Time format
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200484----------------
485
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +0100486Some parameters involve values representing time, such as timeouts. These
Willy Tarreau0ba27502007-12-24 16:55:16 +0100487values are generally expressed in milliseconds (unless explicitly stated
488otherwise) but may be expressed in any other unit by suffixing the unit to the
489numeric value. It is important to consider this because it will not be repeated
490for every keyword. Supported units are :
491
492 - us : microseconds. 1 microsecond = 1/1000000 second
493 - ms : milliseconds. 1 millisecond = 1/1000 second. This is the default.
494 - s : seconds. 1s = 1000ms
495 - m : minutes. 1m = 60s = 60000ms
496 - h : hours. 1h = 60m = 3600s = 3600000ms
497 - d : days. 1d = 24h = 1440m = 86400s = 86400000ms
498
499
Lukas Tribusaa83a312017-03-21 09:25:09 +00005002.5. Examples
Patrick Mezard35da19c2010-06-12 17:02:47 +0200501-------------
502
503 # Simple configuration for an HTTP proxy listening on port 80 on all
504 # interfaces and forwarding requests to a single backend "servers" with a
505 # single server "server1" listening on 127.0.0.1:8000
506 global
507 daemon
508 maxconn 256
509
510 defaults
511 mode http
512 timeout connect 5000ms
513 timeout client 50000ms
514 timeout server 50000ms
515
516 frontend http-in
517 bind *:80
518 default_backend servers
519
520 backend servers
521 server server1 127.0.0.1:8000 maxconn 32
522
523
524 # The same configuration defined with a single listen block. Shorter but
525 # less expressive, especially in HTTP mode.
526 global
527 daemon
528 maxconn 256
529
530 defaults
531 mode http
532 timeout connect 5000ms
533 timeout client 50000ms
534 timeout server 50000ms
535
536 listen http-in
537 bind *:80
538 server server1 127.0.0.1:8000 maxconn 32
539
540
541Assuming haproxy is in $PATH, test these configurations in a shell with:
542
Willy Tarreauccb289d2010-12-11 20:19:38 +0100543 $ sudo haproxy -f configuration.conf -c
Patrick Mezard35da19c2010-06-12 17:02:47 +0200544
545
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005463. Global parameters
Willy Tarreau6a06a402007-07-15 20:15:28 +0200547--------------------
548
549Parameters in the "global" section are process-wide and often OS-specific. They
550are generally set once for all and do not need being changed once correct. Some
551of them have command-line equivalents.
552
553The following keywords are supported in the "global" section :
554
555 * Process management and security
Emeric Brunc8e8d122012-10-02 18:42:10 +0200556 - ca-base
Willy Tarreau6a06a402007-07-15 20:15:28 +0200557 - chroot
Emeric Brunc8e8d122012-10-02 18:42:10 +0200558 - crt-base
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200559 - cpu-map
Willy Tarreau6a06a402007-07-15 20:15:28 +0200560 - daemon
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200561 - description
562 - deviceatlas-json-file
563 - deviceatlas-log-level
564 - deviceatlas-separator
565 - deviceatlas-properties-cookie
Simon Horman98637e52014-06-20 12:30:16 +0900566 - external-check
Willy Tarreau6a06a402007-07-15 20:15:28 +0200567 - gid
568 - group
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100569 - hard-stop-after
Willy Tarreau6a06a402007-07-15 20:15:28 +0200570 - log
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200571 - log-tag
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100572 - log-send-hostname
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200573 - lua-load
Willy Tarreau6a06a402007-07-15 20:15:28 +0200574 - nbproc
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200575 - nbthread
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200576 - node
Willy Tarreau6a06a402007-07-15 20:15:28 +0200577 - pidfile
Willy Tarreau1d549722016-02-16 12:41:57 +0100578 - presetenv
579 - resetenv
Willy Tarreau6a06a402007-07-15 20:15:28 +0200580 - uid
581 - ulimit-n
582 - user
Willy Tarreau1d549722016-02-16 12:41:57 +0100583 - setenv
Willy Tarreaufbee7132007-10-18 13:53:22 +0200584 - stats
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200585 - ssl-default-bind-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200586 - ssl-default-bind-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200587 - ssl-default-bind-options
588 - ssl-default-server-ciphers
Dirkjan Bussink415150f2018-09-14 11:14:21 +0200589 - ssl-default-server-ciphersuites
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200590 - ssl-default-server-options
591 - ssl-dh-param-file
Emeric Brun850efd52014-01-29 12:24:34 +0100592 - ssl-server-verify
Willy Tarreauceb24bc2010-11-09 12:46:41 +0100593 - unix-bind
Willy Tarreau1d549722016-02-16 12:41:57 +0100594 - unsetenv
Thomas Holmesdb04f192015-05-18 13:21:39 +0100595 - 51degrees-data-file
596 - 51degrees-property-name-list
Dragan Dosen93b38d92015-06-29 16:43:25 +0200597 - 51degrees-property-separator
Dragan Dosenae6d39a2015-06-29 16:43:27 +0200598 - 51degrees-cache-size
Willy Tarreaud72758d2010-01-12 10:42:19 +0100599
Willy Tarreau6a06a402007-07-15 20:15:28 +0200600 * Performance tuning
Willy Tarreau1746eec2014-04-25 10:46:47 +0200601 - max-spread-checks
Willy Tarreau6a06a402007-07-15 20:15:28 +0200602 - maxconn
Willy Tarreau81c25d02011-09-07 15:17:21 +0200603 - maxconnrate
William Lallemandd85f9172012-11-09 17:05:39 +0100604 - maxcomprate
William Lallemand072a2bf2012-11-20 17:01:01 +0100605 - maxcompcpuusage
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100606 - maxpipes
Willy Tarreau93e7c002013-10-07 18:51:07 +0200607 - maxsessrate
Willy Tarreau403edff2012-09-06 11:58:37 +0200608 - maxsslconn
Willy Tarreaue43d5322013-10-07 20:01:52 +0200609 - maxsslrate
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200610 - maxzlibmem
Willy Tarreau6a06a402007-07-15 20:15:28 +0200611 - noepoll
612 - nokqueue
613 - nopoll
Willy Tarreauff4f82d2009-02-06 11:28:13 +0100614 - nosplice
Jarno Huuskonen0e82b922014-04-12 18:22:19 +0300615 - nogetaddrinfo
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +0000616 - noreuseport
Willy Tarreau75c62c22018-11-22 11:02:09 +0100617 - profiling.tasks
Willy Tarreaufe255b72007-10-14 23:09:26 +0200618 - spread-checks
Baptiste Assmann5626f482015-08-23 10:00:10 +0200619 - server-state-base
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +0200620 - server-state-file
Grant Zhang872f9c22017-01-21 01:10:18 +0000621 - ssl-engine
Grant Zhangfa6c7ee2017-01-14 01:42:15 +0000622 - ssl-mode-async
Baptiste Assmann3493d0f2015-10-12 20:21:23 +0200623 - tune.buffers.limit
624 - tune.buffers.reserve
Willy Tarreau27a674e2009-08-17 07:23:33 +0200625 - tune.bufsize
Willy Tarreau43961d52010-10-04 20:39:20 +0200626 - tune.chksize
William Lallemandf3747832012-11-09 12:33:10 +0100627 - tune.comp.maxlevel
Willy Tarreaufe20e5b2017-07-27 11:42:14 +0200628 - tune.h2.header-table-size
Willy Tarreaue6baec02017-07-27 11:45:11 +0200629 - tune.h2.initial-window-size
Willy Tarreau5242ef82017-07-27 11:47:28 +0200630 - tune.h2.max-concurrent-streams
Willy Tarreau193b8c62012-11-22 00:17:38 +0100631 - tune.http.cookielen
Stéphane Cottin23e9e932017-05-18 08:58:41 +0200632 - tune.http.logurilen
Willy Tarreauac1932d2011-10-24 19:14:41 +0200633 - tune.http.maxhdr
Willy Tarreau7e312732014-02-12 16:35:14 +0100634 - tune.idletimer
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100635 - tune.lua.forced-yield
Willy Tarreau32f61e22015-03-18 17:54:59 +0100636 - tune.lua.maxmem
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100637 - tune.lua.session-timeout
638 - tune.lua.task-timeout
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +0200639 - tune.lua.service-timeout
Willy Tarreaua0250ba2008-01-06 11:22:57 +0100640 - tune.maxaccept
641 - tune.maxpollevents
Willy Tarreau27a674e2009-08-17 07:23:33 +0200642 - tune.maxrewrite
Willy Tarreauf3045d22015-04-29 16:24:50 +0200643 - tune.pattern.cache-size
Willy Tarreaubd9a0a72011-10-23 21:14:29 +0200644 - tune.pipesize
Willy Tarreaue803de22010-01-21 17:43:04 +0100645 - tune.rcvbuf.client
646 - tune.rcvbuf.server
Willy Tarreaub22fc302015-12-14 12:04:35 +0100647 - tune.recv_enough
Olivier Houchard1599b802018-05-24 18:59:04 +0200648 - tune.runqueue-depth
Willy Tarreaue803de22010-01-21 17:43:04 +0100649 - tune.sndbuf.client
650 - tune.sndbuf.server
Willy Tarreau6ec58db2012-11-16 16:32:15 +0100651 - tune.ssl.cachesize
Willy Tarreaubfd59462013-02-21 07:46:09 +0100652 - tune.ssl.lifetime
Emeric Brun8dc60392014-05-09 13:52:00 +0200653 - tune.ssl.force-private-cache
Willy Tarreaubfd59462013-02-21 07:46:09 +0100654 - tune.ssl.maxrecord
Remi Gacognef46cd6e2014-06-12 14:58:40 +0200655 - tune.ssl.default-dh-param
Christopher Faulet31af49d2015-06-09 17:29:50 +0200656 - tune.ssl.ssl-ctx-cache-size
Thierry FOURNIER5bf77322017-02-25 12:45:22 +0100657 - tune.ssl.capture-cipherlist-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200658 - tune.vars.global-max-size
Christopher Fauletff2613e2016-11-09 11:36:17 +0100659 - tune.vars.proc-max-size
Thierry FOURNIER4834bc72015-06-06 19:29:07 +0200660 - tune.vars.reqres-max-size
661 - tune.vars.sess-max-size
662 - tune.vars.txn-max-size
William Lallemanda509e4c2012-11-07 16:54:34 +0100663 - tune.zlib.memlevel
664 - tune.zlib.windowsize
Willy Tarreaud72758d2010-01-12 10:42:19 +0100665
Willy Tarreau6a06a402007-07-15 20:15:28 +0200666 * Debugging
667 - debug
668 - quiet
Willy Tarreau6a06a402007-07-15 20:15:28 +0200669
670
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006713.1. Process management and security
Willy Tarreau6a06a402007-07-15 20:15:28 +0200672------------------------------------
673
Emeric Brunc8e8d122012-10-02 18:42:10 +0200674ca-base <dir>
675 Assigns a default directory to fetch SSL CA certificates and CRLs from when a
Emeric Brunfd33a262012-10-11 16:28:27 +0200676 relative path is used with "ca-file" or "crl-file" directives. Absolute
677 locations specified in "ca-file" and "crl-file" prevail and ignore "ca-base".
Emeric Brunc8e8d122012-10-02 18:42:10 +0200678
Willy Tarreau6a06a402007-07-15 20:15:28 +0200679chroot <jail dir>
680 Changes current directory to <jail dir> and performs a chroot() there before
681 dropping privileges. This increases the security level in case an unknown
682 vulnerability would be exploited, since it would make it very hard for the
683 attacker to exploit the system. This only works when the process is started
684 with superuser privileges. It is important to ensure that <jail_dir> is both
Davor Ocelice9ed2812017-12-25 17:49:28 +0100685 empty and non-writable to anyone.
Willy Tarreaud72758d2010-01-12 10:42:19 +0100686
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100687cpu-map [auto:]<process-set>[/<thread-set>] <cpu-set>...
688 On Linux 2.6 and above, it is possible to bind a process or a thread to a
689 specific CPU set. This means that the process or the thread will never run on
690 other CPUs. The "cpu-map" directive specifies CPU sets for process or thread
691 sets. The first argument is a process set, eventually followed by a thread
692 set. These sets have the format
693
694 all | odd | even | number[-[number]]
695
696 <number>> must be a number between 1 and 32 or 64, depending on the machine's
Davor Ocelice9ed2812017-12-25 17:49:28 +0100697 word size. Any process IDs above nbproc and any thread IDs above nbthread are
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100698 ignored. It is possible to specify a range with two such number delimited by
699 a dash ('-'). It also is possible to specify all processes at once using
Christopher Faulet1dcb9cb2017-11-22 10:24:40 +0100700 "all", only odd numbers using "odd" or even numbers using "even", just like
701 with the "bind-process" directive. The second and forthcoming arguments are
Davor Ocelice9ed2812017-12-25 17:49:28 +0100702 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 +0100703 range with two such numbers delimited by a dash ('-'). Multiple CPU numbers
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100704 or ranges may be specified, and the processes or threads will be allowed to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100705 bind to all of them. Obviously, multiple "cpu-map" directives may be
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100706 specified. Each "cpu-map" directive will replace the previous ones when they
707 overlap. A thread will be bound on the intersection of its mapping and the
708 one of the process on which it is attached. If the intersection is null, no
709 specific binding will be set for the thread.
Willy Tarreaufc6c0322012-11-16 16:12:27 +0100710
Christopher Fauletff4121f2017-11-22 16:38:49 +0100711 Ranges can be partially defined. The higher bound can be omitted. In such
712 case, it is replaced by the corresponding maximum value, 32 or 64 depending
713 on the machine's word size.
714
Christopher Faulet26028f62017-11-22 15:01:51 +0100715 The prefix "auto:" can be added before the process set to let HAProxy
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100716 automatically bind a process or a thread to a CPU by incrementing
717 process/thread and CPU sets. To be valid, both sets must have the same
718 size. No matter the declaration order of the CPU sets, it will be bound from
719 the lowest to the highest bound. Having a process and a thread range with the
720 "auto:" prefix is not supported. Only one range is supported, the other one
721 must be a fixed number.
Christopher Faulet26028f62017-11-22 15:01:51 +0100722
723 Examples:
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100724 cpu-map 1-4 0-3 # bind processes 1 to 4 on the first 4 CPUs
725
726 cpu-map 1/all 0-3 # bind all threads of the first process on the
727 # first 4 CPUs
728
729 cpu-map 1- 0- # will be replaced by "cpu-map 1-64 0-63"
730 # or "cpu-map 1-32 0-31" depending on the machine's
731 # word size.
732
Christopher Faulet26028f62017-11-22 15:01:51 +0100733 # all these lines bind the process 1 to the cpu 0, the process 2 to cpu 1
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100734 # and so on.
Christopher Faulet26028f62017-11-22 15:01:51 +0100735 cpu-map auto:1-4 0-3
736 cpu-map auto:1-4 0-1 2-3
737 cpu-map auto:1-4 3 2 1 0
738
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100739 # all these lines bind the thread 1 to the cpu 0, the thread 2 to cpu 1
740 # and so on.
741 cpu-map auto:1/1-4 0-3
742 cpu-map auto:1/1-4 0-1 2-3
743 cpu-map auto:1/1-4 3 2 1 0
744
Davor Ocelice9ed2812017-12-25 17:49:28 +0100745 # bind each process to exactly one CPU using all/odd/even keyword
Christopher Faulet26028f62017-11-22 15:01:51 +0100746 cpu-map auto:all 0-63
747 cpu-map auto:even 0-31
748 cpu-map auto:odd 32-63
749
750 # invalid cpu-map because process and CPU sets have different sizes.
751 cpu-map auto:1-4 0 # invalid
752 cpu-map auto:1 0-3 # invalid
753
Christopher Fauletcb6a9452017-11-22 16:50:41 +0100754 # invalid cpu-map because automatic binding is used with a process range
755 # and a thread range.
756 cpu-map auto:all/all 0 # invalid
757 cpu-map auto:all/1-4 0 # invalid
758 cpu-map auto:1-4/all 0 # invalid
759
Emeric Brunc8e8d122012-10-02 18:42:10 +0200760crt-base <dir>
761 Assigns a default directory to fetch SSL certificates from when a relative
762 path is used with "crtfile" directives. Absolute locations specified after
763 "crtfile" prevail and ignore "crt-base".
764
Willy Tarreau6a06a402007-07-15 20:15:28 +0200765daemon
766 Makes the process fork into background. This is the recommended mode of
767 operation. It is equivalent to the command line "-D" argument. It can be
Lukas Tribusf46bf952017-11-21 12:39:34 +0100768 disabled by the command line "-db" argument. This option is ignored in
769 systemd mode.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200770
David Carlier8167f302015-06-01 13:50:06 +0200771deviceatlas-json-file <path>
772 Sets the path of the DeviceAtlas JSON data file to be loaded by the API.
Davor Ocelice9ed2812017-12-25 17:49:28 +0100773 The path must be a valid JSON data file and accessible by HAProxy process.
David Carlier8167f302015-06-01 13:50:06 +0200774
775deviceatlas-log-level <value>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100776 Sets the level of information returned by the API. This directive is
David Carlier8167f302015-06-01 13:50:06 +0200777 optional and set to 0 by default if not set.
778
779deviceatlas-separator <char>
780 Sets the character separator for the API properties results. This directive
781 is optional and set to | by default if not set.
782
Cyril Bonté0306c4a2015-10-26 22:37:38 +0100783deviceatlas-properties-cookie <name>
Cyril Bonté307ee1e2015-09-28 23:16:06 +0200784 Sets the client cookie's name used for the detection if the DeviceAtlas
785 Client-side component was used during the request. This directive is optional
786 and set to DAPROPS by default if not set.
David Carlier29b3ca32015-09-25 14:09:21 +0100787
Simon Horman98637e52014-06-20 12:30:16 +0900788external-check
789 Allows the use of an external agent to perform health checks.
790 This is disabled by default as a security precaution.
791 See "option external-check".
792
Willy Tarreau6a06a402007-07-15 20:15:28 +0200793gid <number>
794 Changes the process' group ID to <number>. It is recommended that the group
795 ID is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
796 be started with a user belonging to this group, or with superuser privileges.
Michael Schererab012dd2013-01-12 18:35:19 +0100797 Note that if haproxy is started from a user having supplementary groups, it
798 will only be able to drop these groups if started with superuser privileges.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200799 See also "group" and "uid".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100800
Cyril Bonté203ec5a2017-03-23 22:44:13 +0100801hard-stop-after <time>
802 Defines the maximum time allowed to perform a clean soft-stop.
803
804 Arguments :
805 <time> is the maximum time (by default in milliseconds) for which the
806 instance will remain alive when a soft-stop is received via the
807 SIGUSR1 signal.
808
809 This may be used to ensure that the instance will quit even if connections
810 remain opened during a soft-stop (for example with long timeouts for a proxy
811 in tcp mode). It applies both in TCP and HTTP mode.
812
813 Example:
814 global
815 hard-stop-after 30s
816
Willy Tarreau6a06a402007-07-15 20:15:28 +0200817group <group name>
818 Similar to "gid" but uses the GID of group name <group name> from /etc/group.
819 See also "gid" and "user".
Willy Tarreaud72758d2010-01-12 10:42:19 +0100820
Dragan Dosen7ad31542015-09-28 17:16:47 +0200821log <address> [len <length>] [format <format>] <facility> [max level [min level]]
Cyril Bonté3e954872018-03-20 23:30:27 +0100822 Adds a global syslog server. Several global servers can be defined. They
Davor Ocelice9ed2812017-12-25 17:49:28 +0100823 will receive logs for starts and exits, as well as all logs from proxies
Robert Tsai81ae1952007-12-05 10:47:29 +0100824 configured with "log global".
825
826 <address> can be one of:
827
Willy Tarreau2769aa02007-12-27 18:26:09 +0100828 - An IPv4 address optionally followed by a colon and a UDP port. If
Robert Tsai81ae1952007-12-05 10:47:29 +0100829 no port is specified, 514 is used by default (the standard syslog
830 port).
831
David du Colombier24bb5f52011-03-17 10:40:23 +0100832 - An IPv6 address followed by a colon and optionally a UDP port. If
833 no port is specified, 514 is used by default (the standard syslog
834 port).
835
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100836 - A filesystem path to a datagram UNIX domain socket, keeping in mind
Robert Tsai81ae1952007-12-05 10:47:29 +0100837 considerations for chroot (be sure the path is accessible inside
838 the chroot) and uid/gid (be sure the path is appropriately
Davor Ocelice9ed2812017-12-25 17:49:28 +0100839 writable).
Robert Tsai81ae1952007-12-05 10:47:29 +0100840
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100841 - A file descriptor number in the form "fd@<number>", which may point
842 to a pipe, terminal, or socket. In this case unbuffered logs are used
843 and one writev() call per log is performed. This is a bit expensive
844 but acceptable for most workloads. Messages sent this way will not be
845 truncated but may be dropped, in which case the DroppedLogs counter
846 will be incremented. The writev() call is atomic even on pipes for
847 messages up to PIPE_BUF size, which POSIX recommends to be at least
848 512 and which is 4096 bytes on most modern operating systems. Any
849 larger message may be interleaved with messages from other processes.
850 Exceptionally for debugging purposes the file descriptor may also be
851 directed to a file, but doing so will significantly slow haproxy down
852 as non-blocking calls will be ignored. Also there will be no way to
853 purge nor rotate this file without restarting the process. Note that
854 the configured syslog format is preserved, so the output is suitable
Willy Tarreauc1b06452018-11-12 11:57:56 +0100855 for use with a TCP syslog server. See also the "short" and "raw"
856 format below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +0100857
858 - "stdout" / "stderr", which are respectively aliases for "fd@1" and
859 "fd@2", see above.
860
William Lallemandb2f07452015-05-12 14:27:13 +0200861 You may want to reference some environment variables in the address
862 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +0100863
Willy Tarreau18324f52014-06-27 18:10:07 +0200864 <length> is an optional maximum line length. Log lines larger than this value
865 will be truncated before being sent. The reason is that syslog
866 servers act differently on log line length. All servers support the
867 default value of 1024, but some servers simply drop larger lines
868 while others do log them. If a server supports long lines, it may
869 make sense to set this value here in order to avoid truncating long
870 lines. Similarly, if a server drops long lines, it is preferable to
871 truncate them before sending them. Accepted values are 80 to 65535
872 inclusive. The default value of 1024 is generally fine for all
873 standard usages. Some specific cases of long captures or
Davor Ocelice9ed2812017-12-25 17:49:28 +0100874 JSON-formatted logs may require larger values. You may also need to
875 increase "tune.http.logurilen" if your request URIs are truncated.
Willy Tarreau18324f52014-06-27 18:10:07 +0200876
Dragan Dosen7ad31542015-09-28 17:16:47 +0200877 <format> is the log format used when generating syslog messages. It may be
878 one of the following :
879
880 rfc3164 The RFC3164 syslog message format. This is the default.
881 (https://tools.ietf.org/html/rfc3164)
882
883 rfc5424 The RFC5424 syslog message format.
884 (https://tools.ietf.org/html/rfc5424)
885
Willy Tarreaue8746a02018-11-12 08:45:00 +0100886 short A message containing only a level between angle brackets such as
887 '<3>', followed by the text. The PID, date, time, process name
888 and system name are omitted. This is designed to be used with a
889 local log server. This format is compatible with what the systemd
890 logger consumes.
891
Willy Tarreauc1b06452018-11-12 11:57:56 +0100892 raw A message containing only the text. The level, PID, date, time,
893 process name and system name are omitted. This is designed to be
894 used in containers or during development, where the severity only
895 depends on the file descriptor used (stdout/stderr).
896
Robert Tsai81ae1952007-12-05 10:47:29 +0100897 <facility> must be one of the 24 standard syslog facilities :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200898
Willy Tarreaue8746a02018-11-12 08:45:00 +0100899 kern user mail daemon auth syslog lpr news
900 uucp cron auth2 ftp ntp audit alert cron2
901 local0 local1 local2 local3 local4 local5 local6 local7
902
Willy Tarreauc1b06452018-11-12 11:57:56 +0100903 Note that the facility is ignored for the "short" and "raw"
904 formats, but still required as a positional field. It is
905 recommended to use "daemon" in this case to make it clear that
906 it's only supposed to be used locally.
Willy Tarreau6a06a402007-07-15 20:15:28 +0200907
908 An optional level can be specified to filter outgoing messages. By default,
Willy Tarreauf7edefa2009-05-10 17:20:05 +0200909 all messages are sent. If a maximum level is specified, only messages with a
910 severity at least as important as this level will be sent. An optional minimum
911 level can be specified. If it is set, logs emitted with a more severe level
912 than this one will be capped to this level. This is used to avoid sending
913 "emerg" messages on all terminals on some default syslog configurations.
914 Eight levels are known :
Willy Tarreau6a06a402007-07-15 20:15:28 +0200915
Cyril Bontédc4d9032012-04-08 21:57:39 +0200916 emerg alert crit err warning notice info debug
Willy Tarreau6a06a402007-07-15 20:15:28 +0200917
Joe Williamsdf5b38f2010-12-29 17:05:48 +0100918log-send-hostname [<string>]
919 Sets the hostname field in the syslog header. If optional "string" parameter
920 is set the header is set to the string contents, otherwise uses the hostname
921 of the system. Generally used if one is not relaying logs through an
922 intermediate syslog server or for simply customizing the hostname printed in
923 the logs.
924
Kevinm48936af2010-12-22 16:08:21 +0000925log-tag <string>
926 Sets the tag field in the syslog header to this string. It defaults to the
927 program name as launched from the command line, which usually is "haproxy".
928 Sometimes it can be useful to differentiate between multiple processes
Willy Tarreau094af4e2015-01-07 15:03:42 +0100929 running on the same host. See also the per-proxy "log-tag" directive.
Kevinm48936af2010-12-22 16:08:21 +0000930
Thierry FOURNIER90da1912015-03-05 11:17:06 +0100931lua-load <file>
932 This global directive loads and executes a Lua file. This directive can be
933 used multiple times.
934
William Lallemand4cfede82017-11-24 22:02:34 +0100935master-worker [no-exit-on-failure]
William Lallemande202b1e2017-06-01 17:38:56 +0200936 Master-worker mode. It is equivalent to the command line "-W" argument.
937 This mode will launch a "master" which will monitor the "workers". Using
938 this mode, you can reload HAProxy directly by sending a SIGUSR2 signal to
Davor Ocelice9ed2812017-12-25 17:49:28 +0100939 the master. The master-worker mode is compatible either with the foreground
William Lallemande202b1e2017-06-01 17:38:56 +0200940 or daemon mode. It is recommended to use this mode with multiprocess and
941 systemd.
William Lallemand4cfede82017-11-24 22:02:34 +0100942 By default, if a worker exits with a bad return code, in the case of a
943 segfault for example, all workers will be killed, and the master will leave.
944 It is convenient to combine this behavior with Restart=on-failure in a
945 systemd unit file in order to relaunch the whole process. If you don't want
946 this behavior, you must use the keyword "no-exit-on-failure".
William Lallemande202b1e2017-06-01 17:38:56 +0200947
William Lallemand4cfede82017-11-24 22:02:34 +0100948 See also "-W" in the management guide.
William Lallemande202b1e2017-06-01 17:38:56 +0200949
Willy Tarreau6a06a402007-07-15 20:15:28 +0200950nbproc <number>
951 Creates <number> processes when going daemon. This requires the "daemon"
952 mode. By default, only one process is created, which is the recommended mode
953 of operation. For systems limited to small sets of file descriptors per
Willy Tarreau149ab772019-01-26 14:27:06 +0100954 process, it may be needed to fork multiple daemons. When set to a value
955 larger than 1, threads are automatically disabled. USING MULTIPLE PROCESSES
Willy Tarreau1f672a82019-01-26 14:20:55 +0100956 IS HARDER TO DEBUG AND IS REALLY DISCOURAGED. See also "daemon" and
957 "nbthread".
Willy Tarreau6a06a402007-07-15 20:15:28 +0200958
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200959nbthread <number>
960 This setting is only available when support for threads was built in. It
Willy Tarreau26f6ae12019-02-02 12:56:15 +0100961 makes haproxy run on <number> threads. This is exclusive with "nbproc". While
962 "nbproc" historically used to be the only way to use multiple processors, it
963 also involved a number of shortcomings related to the lack of synchronization
964 between processes (health-checks, peers, stick-tables, stats, ...) which do
965 not affect threads. As such, any modern configuration is strongly encouraged
Willy Tarreau149ab772019-01-26 14:27:06 +0100966 to migrate away from "nbproc" to "nbthread". "nbthread" also works when
967 HAProxy is started in foreground. On some platforms supporting CPU affinity,
968 when nbproc is not used, the default "nbthread" value is automatically set to
969 the number of CPUs the process is bound to upon startup. This means that the
970 thread count can easily be adjusted from the calling process using commands
971 like "taskset" or "cpuset". Otherwise, this value defaults to 1. The default
972 value is reported in the output of "haproxy -vv". See also "nbproc".
Christopher Fauletbe0faa22017-08-29 15:37:10 +0200973
Willy Tarreau6a06a402007-07-15 20:15:28 +0200974pidfile <pidfile>
Davor Ocelice9ed2812017-12-25 17:49:28 +0100975 Writes PIDs of all daemons into file <pidfile>. This option is equivalent to
Willy Tarreau6a06a402007-07-15 20:15:28 +0200976 the "-p" command line argument. The file must be accessible to the user
977 starting the process. See also "daemon".
978
Willy Tarreau1d549722016-02-16 12:41:57 +0100979presetenv <name> <value>
980 Sets environment variable <name> to value <value>. If the variable exists, it
981 is NOT overwritten. The changes immediately take effect so that the next line
982 in the configuration file sees the new value. See also "setenv", "resetenv",
983 and "unsetenv".
984
985resetenv [<name> ...]
986 Removes all environment variables except the ones specified in argument. It
987 allows to use a clean controlled environment before setting new values with
988 setenv or unsetenv. Please note that some internal functions may make use of
989 some environment variables, such as time manipulation functions, but also
990 OpenSSL or even external checks. This must be used with extreme care and only
991 after complete validation. The changes immediately take effect so that the
992 next line in the configuration file sees the new environment. See also
993 "setenv", "presetenv", and "unsetenv".
994
Christopher Fauletff4121f2017-11-22 16:38:49 +0100995stats bind-process [ all | odd | even | <process_num>[-[process_num>]] ] ...
Willy Tarreau35b7b162012-10-22 23:17:18 +0200996 Limits the stats socket to a certain set of processes numbers. By default the
997 stats socket is bound to all processes, causing a warning to be emitted when
998 nbproc is greater than 1 because there is no way to select the target process
999 when connecting. However, by using this setting, it becomes possible to pin
1000 the stats socket to a specific set of processes, typically the first one. The
1001 warning will automatically be disabled when this setting is used, whatever
Willy Tarreaua9db57e2013-01-18 11:29:29 +01001002 the number of processes used. The maximum process ID depends on the machine's
Christopher Fauletff4121f2017-11-22 16:38:49 +01001003 word size (32 or 64). Ranges can be partially defined. The higher bound can
1004 be omitted. In such case, it is replaced by the corresponding maximum
1005 value. A better option consists in using the "process" setting of the "stats
1006 socket" line to force the process on each line.
Willy Tarreau35b7b162012-10-22 23:17:18 +02001007
Baptiste Assmann5626f482015-08-23 10:00:10 +02001008server-state-base <directory>
1009 Specifies the directory prefix to be prepended in front of all servers state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001010 file names which do not start with a '/'. See also "server-state-file",
1011 "load-server-state-from-file" and "server-state-file-name".
Baptiste Assmannef1f0fc2015-08-23 10:06:39 +02001012
1013server-state-file <file>
1014 Specifies the path to the file containing state of servers. If the path starts
1015 with a slash ('/'), it is considered absolute, otherwise it is considered
1016 relative to the directory specified using "server-state-base" (if set) or to
1017 the current directory. Before reloading HAProxy, it is possible to save the
1018 servers' current state using the stats command "show servers state". The
1019 output of this command must be written in the file pointed by <file>. When
1020 starting up, before handling traffic, HAProxy will read, load and apply state
1021 for each server found in the file and available in its current running
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02001022 configuration. See also "server-state-base" and "show servers state",
1023 "load-server-state-from-file" and "server-state-file-name"
Baptiste Assmann5626f482015-08-23 10:00:10 +02001024
Willy Tarreau1d549722016-02-16 12:41:57 +01001025setenv <name> <value>
1026 Sets environment variable <name> to value <value>. If the variable exists, it
1027 is overwritten. The changes immediately take effect so that the next line in
1028 the configuration file sees the new value. See also "presetenv", "resetenv",
1029 and "unsetenv".
1030
Willy Tarreau610f04b2014-02-13 11:36:41 +01001031ssl-default-bind-ciphers <ciphers>
1032 This setting is only available when support for OpenSSL was built in. It sets
1033 the default string describing the list of cipher algorithms ("cipher suite")
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001034 that are negotiated during the SSL/TLS handshake up to TLSv1.2 for all
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001035 "bind" lines which do not explicitly define theirs. The format of the string
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001036 is defined in "man 1 ciphers" from OpenSSL man pages. For background
1037 information and recommendations see e.g.
1038 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1039 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
1040 cipher configuration, please check the "ssl-default-bind-ciphersuites" keyword.
1041 Please check the "bind" keyword for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001042
1043ssl-default-bind-ciphersuites <ciphersuites>
1044 This setting is only available when support for OpenSSL was built in and
1045 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default string
1046 describing the list of cipher algorithms ("cipher suite") that are negotiated
1047 during the TLSv1.3 handshake for all "bind" lines which do not explicitly define
1048 theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001049 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1050 cipher configuration for TLSv1.2 and earlier, please check the
1051 "ssl-default-bind-ciphers" keyword. Please check the "bind" keyword for more
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001052 information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001053
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001054ssl-default-bind-options [<option>]...
1055 This setting is only available when support for OpenSSL was built in. It sets
1056 default ssl-options to force on all "bind" lines. Please check the "bind"
1057 keyword to see available options.
1058
1059 Example:
1060 global
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +02001061 ssl-default-bind-options ssl-min-ver TLSv1.0 no-tls-tickets
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001062
Willy Tarreau610f04b2014-02-13 11:36:41 +01001063ssl-default-server-ciphers <ciphers>
1064 This setting is only available when support for OpenSSL was built in. It
1065 sets the default string describing the list of cipher algorithms that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +00001066 negotiated during the SSL/TLS handshake up to TLSv1.2 with the server,
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001067 for all "server" lines which do not explicitly define theirs. The format of
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001068 the string is defined in "man 1 ciphers" from OpenSSL man pages. For background
1069 information and recommendations see e.g.
1070 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
1071 (https://mozilla.github.io/server-side-tls/ssl-config-generator/).
1072 For TLSv1.3 cipher configuration, please check the
1073 "ssl-default-server-ciphersuites" keyword. Please check the "server" keyword
1074 for more information.
Dirkjan Bussink415150f2018-09-14 11:14:21 +02001075
1076ssl-default-server-ciphersuites <ciphersuites>
1077 This setting is only available when support for OpenSSL was built in and
1078 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the default
1079 string describing the list of cipher algorithms that are negotiated during
1080 the TLSv1.3 handshake with the server, for all "server" lines which do not
1081 explicitly define theirs. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +00001082 "man 1 ciphers" from OpenSSL man pages under the section "ciphersuites". For
1083 cipher configuration for TLSv1.2 and earlier, please check the
1084 "ssl-default-server-ciphers" keyword. Please check the "server" keyword for
1085 more information.
Willy Tarreau610f04b2014-02-13 11:36:41 +01001086
Emeric Brun2c86cbf2014-10-30 15:56:50 +01001087ssl-default-server-options [<option>]...
1088 This setting is only available when support for OpenSSL was built in. It sets
1089 default ssl-options to force on all "server" lines. Please check the "server"
1090 keyword to see available options.
1091
Remi Gacogne47783ef2015-05-29 15:53:22 +02001092ssl-dh-param-file <file>
1093 This setting is only available when support for OpenSSL was built in. It sets
1094 the default DH parameters that are used during the SSL/TLS handshake when
1095 ephemeral Diffie-Hellman (DHE) key exchange is used, for all "bind" lines
Davor Ocelice9ed2812017-12-25 17:49:28 +01001096 which do not explicitly define theirs. It will be overridden by custom DH
Remi Gacogne47783ef2015-05-29 15:53:22 +02001097 parameters found in a bind certificate file if any. If custom DH parameters
Cyril Bonté307ee1e2015-09-28 23:16:06 +02001098 are not specified either by using ssl-dh-param-file or by setting them
1099 directly in the certificate file, pre-generated DH parameters of the size
1100 specified by tune.ssl.default-dh-param will be used. Custom parameters are
1101 known to be more secure and therefore their use is recommended.
Remi Gacogne47783ef2015-05-29 15:53:22 +02001102 Custom DH parameters may be generated by using the OpenSSL command
1103 "openssl dhparam <size>", where size should be at least 2048, as 1024-bit DH
1104 parameters should not be considered secure anymore.
1105
Emeric Brun850efd52014-01-29 12:24:34 +01001106ssl-server-verify [none|required]
1107 The default behavior for SSL verify on servers side. If specified to 'none',
1108 servers certificates are not verified. The default is 'required' except if
1109 forced using cmdline option '-dV'.
1110
Willy Tarreauabb175f2012-09-24 12:43:26 +02001111stats socket [<address:port>|<path>] [param*]
1112 Binds a UNIX socket to <path> or a TCPv4/v6 address to <address:port>.
1113 Connections to this socket will return various statistics outputs and even
1114 allow some commands to be issued to change some runtime settings. Please
Willy Tarreau1af20c72017-06-23 16:01:14 +02001115 consult section 9.3 "Unix Socket commands" of Management Guide for more
Kevin Decherf949c7202015-10-13 23:26:44 +02001116 details.
Willy Tarreau6162db22009-10-10 17:13:00 +02001117
Willy Tarreauabb175f2012-09-24 12:43:26 +02001118 All parameters supported by "bind" lines are supported, for instance to
1119 restrict access to some users or their access rights. Please consult
1120 section 5.1 for more information.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001121
1122stats timeout <timeout, in milliseconds>
1123 The default timeout on the stats socket is set to 10 seconds. It is possible
1124 to change this value with "stats timeout". The value must be passed in
Willy Tarreaubefdff12007-12-02 22:27:38 +01001125 milliseconds, or be suffixed by a time unit among { us, ms, s, m, h, d }.
Willy Tarreaufbee7132007-10-18 13:53:22 +02001126
1127stats maxconn <connections>
1128 By default, the stats socket is limited to 10 concurrent connections. It is
1129 possible to change this value with "stats maxconn".
1130
Willy Tarreau6a06a402007-07-15 20:15:28 +02001131uid <number>
1132 Changes the process' user ID to <number>. It is recommended that the user ID
1133 is dedicated to HAProxy or to a small set of similar daemons. HAProxy must
1134 be started with superuser privileges in order to be able to switch to another
1135 one. See also "gid" and "user".
1136
1137ulimit-n <number>
1138 Sets the maximum number of per-process file-descriptors to <number>. By
1139 default, it is automatically computed, so it is recommended not to use this
1140 option.
1141
Willy Tarreauceb24bc2010-11-09 12:46:41 +01001142unix-bind [ prefix <prefix> ] [ mode <mode> ] [ user <user> ] [ uid <uid> ]
1143 [ group <group> ] [ gid <gid> ]
1144
1145 Fixes common settings to UNIX listening sockets declared in "bind" statements.
1146 This is mainly used to simplify declaration of those UNIX sockets and reduce
1147 the risk of errors, since those settings are most commonly required but are
1148 also process-specific. The <prefix> setting can be used to force all socket
1149 path to be relative to that directory. This might be needed to access another
1150 component's chroot. Note that those paths are resolved before haproxy chroots
1151 itself, so they are absolute. The <mode>, <user>, <uid>, <group> and <gid>
1152 all have the same meaning as their homonyms used by the "bind" statement. If
1153 both are specified, the "bind" statement has priority, meaning that the
1154 "unix-bind" settings may be seen as process-wide default settings.
1155
Willy Tarreau1d549722016-02-16 12:41:57 +01001156unsetenv [<name> ...]
1157 Removes environment variables specified in arguments. This can be useful to
1158 hide some sensitive information that are occasionally inherited from the
1159 user's environment during some operations. Variables which did not exist are
1160 silently ignored so that after the operation, it is certain that none of
1161 these variables remain. The changes immediately take effect so that the next
1162 line in the configuration file will not see these variables. See also
1163 "setenv", "presetenv", and "resetenv".
1164
Willy Tarreau6a06a402007-07-15 20:15:28 +02001165user <user name>
1166 Similar to "uid" but uses the UID of user name <user name> from /etc/passwd.
1167 See also "uid" and "group".
1168
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02001169node <name>
1170 Only letters, digits, hyphen and underscore are allowed, like in DNS names.
1171
1172 This statement is useful in HA configurations where two or more processes or
1173 servers share the same IP address. By setting a different node-name on all
1174 nodes, it becomes easy to immediately spot what server is handling the
1175 traffic.
1176
1177description <text>
1178 Add a text that describes the instance.
1179
1180 Please note that it is required to escape certain characters (# for example)
1181 and this text is inserted into a html page so you should avoid using
1182 "<" and ">" characters.
1183
Thomas Holmesdb04f192015-05-18 13:21:39 +0100118451degrees-data-file <file path>
1185 The path of the 51Degrees data file to provide device detection services. The
Davor Ocelice9ed2812017-12-25 17:49:28 +01001186 file should be unzipped and accessible by HAProxy with relevant permissions.
Thomas Holmesdb04f192015-05-18 13:21:39 +01001187
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001188 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001189 compiled with USE_51DEGREES.
1190
Ben Shillitof25e8e52016-12-02 14:25:37 +0000119151degrees-property-name-list [<string> ...]
Thomas Holmesdb04f192015-05-18 13:21:39 +01001192 A list of 51Degrees property names to be load from the dataset. A full list
1193 of names is available on the 51Degrees website:
1194 https://51degrees.com/resources/property-dictionary
1195
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001196 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001197 compiled with USE_51DEGREES.
1198
Dragan Dosen93b38d92015-06-29 16:43:25 +0200119951degrees-property-separator <char>
Thomas Holmesdb04f192015-05-18 13:21:39 +01001200 A char that will be appended to every property value in a response header
1201 containing 51Degrees results. If not set that will be set as ','.
1202
Dragan Dosenae6d39a2015-06-29 16:43:27 +02001203 Please note that this option is only available when haproxy has been
1204 compiled with USE_51DEGREES.
1205
120651degrees-cache-size <number>
1207 Sets the size of the 51Degrees converter cache to <number> entries. This
1208 is an LRU cache which reminds previous device detections and their results.
1209 By default, this cache is disabled.
1210
1211 Please note that this option is only available when haproxy has been
Thomas Holmesdb04f192015-05-18 13:21:39 +01001212 compiled with USE_51DEGREES.
1213
Willy Tarreau6a06a402007-07-15 20:15:28 +02001214
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012153.2. Performance tuning
Willy Tarreau6a06a402007-07-15 20:15:28 +02001216-----------------------
1217
Willy Tarreaubeb859a2018-11-22 18:07:59 +01001218busy-polling
1219 In some situations, especially when dealing with low latency on processors
1220 supporting a variable frequency or when running inside virtual machines, each
1221 time the process waits for an I/O using the poller, the processor goes back
1222 to sleep or is offered to another VM for a long time, and it causes
1223 excessively high latencies. This option provides a solution preventing the
1224 processor from sleeping by always using a null timeout on the pollers. This
1225 results in a significant latency reduction (30 to 100 microseconds observed)
1226 at the expense of a risk to overheat the processor. It may even be used with
1227 threads, in which case improperly bound threads may heavily conflict,
1228 resulting in a worse performance and high values for the CPU stolen fields
1229 in "show info" output, indicating which threads are misconfigured. It is
1230 important not to let the process run on the same processor as the network
1231 interrupts when this option is used. It is also better to avoid using it on
1232 multiple CPU threads sharing the same core. This option is disabled by
1233 default. If it has been enabled, it may still be forcibly disabled by
1234 prefixing it with the "no" keyword. It is ignored by the "select" and
1235 "poll" pollers.
1236
Willy Tarreau1746eec2014-04-25 10:46:47 +02001237max-spread-checks <delay in milliseconds>
1238 By default, haproxy tries to spread the start of health checks across the
1239 smallest health check interval of all the servers in a farm. The principle is
1240 to avoid hammering services running on the same server. But when using large
1241 check intervals (10 seconds or more), the last servers in the farm take some
1242 time before starting to be tested, which can be a problem. This parameter is
1243 used to enforce an upper bound on delay between the first and the last check,
1244 even if the servers' check intervals are larger. When servers run with
1245 shorter intervals, their intervals will be respected though.
1246
Willy Tarreau6a06a402007-07-15 20:15:28 +02001247maxconn <number>
1248 Sets the maximum per-process number of concurrent connections to <number>. It
1249 is equivalent to the command-line argument "-n". Proxies will stop accepting
1250 connections when this limit is reached. The "ulimit-n" parameter is
Willy Tarreau8274e102014-06-19 15:31:25 +02001251 automatically adjusted according to this value. See also "ulimit-n". Note:
1252 the "select" poller cannot reliably use more than 1024 file descriptors on
1253 some platforms. If your platform only supports select and reports "select
1254 FAILED" on startup, you need to reduce maxconn until it works (slightly
Willy Tarreaub28f3442019-03-04 08:13:43 +01001255 below 500 in general). If this value is not set, it will automatically be
1256 calculated based on the current file descriptors limit reported by the
1257 "ulimit -n" command, possibly reduced to a lower value if a memory limit
1258 is enforced, based on the buffer size, memory allocated to compression, SSL
1259 cache size, and use or not of SSL and the associated maxsslconn (which can
1260 also be automatic).
Willy Tarreau6a06a402007-07-15 20:15:28 +02001261
Willy Tarreau81c25d02011-09-07 15:17:21 +02001262maxconnrate <number>
1263 Sets the maximum per-process number of connections per second to <number>.
1264 Proxies will stop accepting connections when this limit is reached. It can be
1265 used to limit the global capacity regardless of each frontend capacity. It is
1266 important to note that this can only be used as a service protection measure,
1267 as there will not necessarily be a fair share between frontends when the
1268 limit is reached, so it's a good idea to also limit each frontend to some
1269 value close to its expected share. Also, lowering tune.maxaccept can improve
1270 fairness.
1271
William Lallemandd85f9172012-11-09 17:05:39 +01001272maxcomprate <number>
1273 Sets the maximum per-process input compression rate to <number> kilobytes
Davor Ocelice9ed2812017-12-25 17:49:28 +01001274 per second. For each session, if the maximum is reached, the compression
William Lallemandd85f9172012-11-09 17:05:39 +01001275 level will be decreased during the session. If the maximum is reached at the
1276 beginning of a session, the session will not compress at all. If the maximum
1277 is not reached, the compression level will be increased up to
Davor Ocelice9ed2812017-12-25 17:49:28 +01001278 tune.comp.maxlevel. A value of zero means there is no limit, this is the
William Lallemandd85f9172012-11-09 17:05:39 +01001279 default value.
1280
William Lallemand072a2bf2012-11-20 17:01:01 +01001281maxcompcpuusage <number>
1282 Sets the maximum CPU usage HAProxy can reach before stopping the compression
1283 for new requests or decreasing the compression level of current requests.
1284 It works like 'maxcomprate' but measures CPU usage instead of incoming data
1285 bandwidth. The value is expressed in percent of the CPU used by haproxy. In
1286 case of multiple processes (nbproc > 1), each process manages its individual
1287 usage. A value of 100 disable the limit. The default value is 100. Setting
1288 a lower value will prevent the compression work from slowing the whole
1289 process down and from introducing high latencies.
1290
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001291maxpipes <number>
1292 Sets the maximum per-process number of pipes to <number>. Currently, pipes
1293 are only used by kernel-based tcp splicing. Since a pipe contains two file
1294 descriptors, the "ulimit-n" value will be increased accordingly. The default
1295 value is maxconn/4, which seems to be more than enough for most heavy usages.
1296 The splice code dynamically allocates and releases pipes, and can fall back
1297 to standard copy, so setting this value too low may only impact performance.
1298
Willy Tarreau93e7c002013-10-07 18:51:07 +02001299maxsessrate <number>
1300 Sets the maximum per-process number of sessions per second to <number>.
1301 Proxies will stop accepting connections when this limit is reached. It can be
1302 used to limit the global capacity regardless of each frontend capacity. It is
1303 important to note that this can only be used as a service protection measure,
1304 as there will not necessarily be a fair share between frontends when the
1305 limit is reached, so it's a good idea to also limit each frontend to some
1306 value close to its expected share. Also, lowering tune.maxaccept can improve
1307 fairness.
1308
Willy Tarreau403edff2012-09-06 11:58:37 +02001309maxsslconn <number>
1310 Sets the maximum per-process number of concurrent SSL connections to
1311 <number>. By default there is no SSL-specific limit, which means that the
1312 global maxconn setting will apply to all connections. Setting this limit
1313 avoids having openssl use too much memory and crash when malloc returns NULL
1314 (since it unfortunately does not reliably check for such conditions). Note
1315 that the limit applies both to incoming and outgoing connections, so one
1316 connection which is deciphered then ciphered accounts for 2 SSL connections.
Willy Tarreaud0256482015-01-15 21:45:22 +01001317 If this value is not set, but a memory limit is enforced, this value will be
1318 automatically computed based on the memory limit, maxconn, the buffer size,
1319 memory allocated to compression, SSL cache size, and use of SSL in either
1320 frontends, backends or both. If neither maxconn nor maxsslconn are specified
1321 when there is a memory limit, haproxy will automatically adjust these values
1322 so that 100% of the connections can be made over SSL with no risk, and will
1323 consider the sides where it is enabled (frontend, backend, both).
Willy Tarreau403edff2012-09-06 11:58:37 +02001324
Willy Tarreaue43d5322013-10-07 20:01:52 +02001325maxsslrate <number>
1326 Sets the maximum per-process number of SSL sessions per second to <number>.
1327 SSL listeners will stop accepting connections when this limit is reached. It
1328 can be used to limit the global SSL CPU usage regardless of each frontend
1329 capacity. It is important to note that this can only be used as a service
1330 protection measure, as there will not necessarily be a fair share between
1331 frontends when the limit is reached, so it's a good idea to also limit each
1332 frontend to some value close to its expected share. It is also important to
1333 note that the sessions are accounted before they enter the SSL stack and not
1334 after, which also protects the stack against bad handshakes. Also, lowering
1335 tune.maxaccept can improve fairness.
1336
William Lallemand9d5f5482012-11-07 16:12:57 +01001337maxzlibmem <number>
1338 Sets the maximum amount of RAM in megabytes per process usable by the zlib.
1339 When the maximum amount is reached, future sessions will not compress as long
1340 as RAM is unavailable. When sets to 0, there is no limit.
William Lallemande3a7d992012-11-20 11:25:20 +01001341 The default value is 0. The value is available in bytes on the UNIX socket
1342 with "show info" on the line "MaxZlibMemUsage", the memory used by zlib is
1343 "ZlibMemUsage" in bytes.
1344
Willy Tarreau6a06a402007-07-15 20:15:28 +02001345noepoll
1346 Disables the use of the "epoll" event polling system on Linux. It is
1347 equivalent to the command-line argument "-de". The next polling system
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001348 used will generally be "poll". See also "nopoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001349
1350nokqueue
1351 Disables the use of the "kqueue" event polling system on BSD. It is
1352 equivalent to the command-line argument "-dk". The next polling system
1353 used will generally be "poll". See also "nopoll".
1354
1355nopoll
1356 Disables the use of the "poll" event polling system. It is equivalent to the
1357 command-line argument "-dp". The next polling system used will be "select".
Willy Tarreau0ba27502007-12-24 16:55:16 +01001358 It should never be needed to disable "poll" since it's available on all
Willy Tarreaue9f49e72012-11-11 17:42:00 +01001359 platforms supported by HAProxy. See also "nokqueue" and "noepoll".
Willy Tarreau6a06a402007-07-15 20:15:28 +02001360
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001361nosplice
1362 Disables the use of kernel tcp splicing between sockets on Linux. It is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001363 equivalent to the command line argument "-dS". Data will then be copied
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001364 using conventional and more portable recv/send calls. Kernel tcp splicing is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01001365 limited to some very recent instances of kernel 2.6. Most versions between
Willy Tarreauff4f82d2009-02-06 11:28:13 +01001366 2.6.25 and 2.6.28 are buggy and will forward corrupted data, so they must not
1367 be used. This option makes it easier to globally disable kernel splicing in
1368 case of doubt. See also "option splice-auto", "option splice-request" and
1369 "option splice-response".
1370
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001371nogetaddrinfo
1372 Disables the use of getaddrinfo(3) for name resolving. It is equivalent to
1373 the command line argument "-dG". Deprecated gethostbyname(3) will be used.
1374
Lukas Tribusa0bcbdc2016-09-12 21:42:20 +00001375noreuseport
1376 Disables the use of SO_REUSEPORT - see socket(7). It is equivalent to the
1377 command line argument "-dR".
1378
Willy Tarreau75c62c22018-11-22 11:02:09 +01001379profiling.tasks { on | off }
1380 Enables ('on') or disables ('off') per-task CPU profiling. CPU profiling per
1381 task can be very convenient to report where the time is spent and which
1382 requests have what effect on which other request. It is not enabled by
1383 default as it may consume a little bit extra CPU. This requires a system
1384 supporting the clock_gettime(2) syscall with clock identifiers
1385 CLOCK_MONOTONIC and CLOCK_THREAD_CPUTIME_ID, otherwise the reported time will
1386 be zero. This option may be changed at run time using "set profiling" on the
1387 CLI.
1388
Willy Tarreaufe255b72007-10-14 23:09:26 +02001389spread-checks <0..50, in percent>
Simon Hormand60d6912013-11-25 10:46:36 +09001390 Sometimes it is desirable to avoid sending agent and health checks to
1391 servers at exact intervals, for instance when many logical servers are
1392 located on the same physical server. With the help of this parameter, it
1393 becomes possible to add some randomness in the check interval between 0
1394 and +/- 50%. A value between 2 and 5 seems to show good results. The
1395 default value remains at 0.
Willy Tarreaufe255b72007-10-14 23:09:26 +02001396
Davor Ocelice9ed2812017-12-25 17:49:28 +01001397ssl-engine <name> [algo <comma-separated list of algorithms>]
Grant Zhang872f9c22017-01-21 01:10:18 +00001398 Sets the OpenSSL engine to <name>. List of valid values for <name> may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01001399 obtained using the command "openssl engine". This statement may be used
Grant Zhang872f9c22017-01-21 01:10:18 +00001400 multiple times, it will simply enable multiple crypto engines. Referencing an
1401 unsupported engine will prevent haproxy from starting. Note that many engines
1402 will lead to lower HTTPS performance than pure software with recent
1403 processors. The optional command "algo" sets the default algorithms an ENGINE
1404 will supply using the OPENSSL function ENGINE_set_default_string(). A value
Davor Ocelice9ed2812017-12-25 17:49:28 +01001405 of "ALL" uses the engine for all cryptographic operations. If no list of
1406 algo is specified then the value of "ALL" is used. A comma-separated list
Grant Zhang872f9c22017-01-21 01:10:18 +00001407 of different algorithms may be specified, including: RSA, DSA, DH, EC, RAND,
1408 CIPHERS, DIGESTS, PKEY, PKEY_CRYPTO, PKEY_ASN1. This is the same format that
1409 openssl configuration file uses:
1410 https://www.openssl.org/docs/man1.0.2/apps/config.html
1411
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001412ssl-mode-async
1413 Adds SSL_MODE_ASYNC mode to the SSL context. This enables asynchronous TLS
Emeric Brun3854e012017-05-17 20:42:48 +02001414 I/O operations if asynchronous capable SSL engines are used. The current
Emeric Brunb5e42a82017-06-06 12:35:14 +00001415 implementation supports a maximum of 32 engines. The Openssl ASYNC API
1416 doesn't support moving read/write buffers and is not compliant with
1417 haproxy's buffer management. So the asynchronous mode is disabled on
1418 read/write operations (it is only enabled during initial and reneg
1419 handshakes).
Grant Zhangfa6c7ee2017-01-14 01:42:15 +00001420
Willy Tarreau33cb0652014-12-23 22:52:37 +01001421tune.buffers.limit <number>
1422 Sets a hard limit on the number of buffers which may be allocated per process.
1423 The default value is zero which means unlimited. The minimum non-zero value
1424 will always be greater than "tune.buffers.reserve" and should ideally always
1425 be about twice as large. Forcing this value can be particularly useful to
1426 limit the amount of memory a process may take, while retaining a sane
Davor Ocelice9ed2812017-12-25 17:49:28 +01001427 behavior. When this limit is reached, sessions which need a buffer wait for
Willy Tarreau33cb0652014-12-23 22:52:37 +01001428 another one to be released by another session. Since buffers are dynamically
1429 allocated and released, the waiting time is very short and not perceptible
1430 provided that limits remain reasonable. In fact sometimes reducing the limit
1431 may even increase performance by increasing the CPU cache's efficiency. Tests
1432 have shown good results on average HTTP traffic with a limit to 1/10 of the
1433 expected global maxconn setting, which also significantly reduces memory
1434 usage. The memory savings come from the fact that a number of connections
1435 will not allocate 2*tune.bufsize. It is best not to touch this value unless
1436 advised to do so by an haproxy core developer.
1437
Willy Tarreau1058ae72014-12-23 22:40:40 +01001438tune.buffers.reserve <number>
1439 Sets the number of buffers which are pre-allocated and reserved for use only
1440 during memory shortage conditions resulting in failed memory allocations. The
1441 minimum value is 2 and is also the default. There is no reason a user would
1442 want to change this value, it's mostly aimed at haproxy core developers.
1443
Willy Tarreau27a674e2009-08-17 07:23:33 +02001444tune.bufsize <number>
1445 Sets the buffer size to this size (in bytes). Lower values allow more
1446 sessions to coexist in the same amount of RAM, and higher values allow some
1447 applications with very large cookies to work. The default value is 16384 and
1448 can be changed at build time. It is strongly recommended not to change this
1449 from the default value, as very low values will break some services such as
1450 statistics, and values larger than default size will increase memory usage,
1451 possibly causing the system to run out of memory. At least the global maxconn
Willy Tarreau45a66cc2017-11-24 11:28:00 +01001452 parameter should be decreased by the same factor as this one is increased. In
1453 addition, use of HTTP/2 mandates that this value must be 16384 or more. If an
1454 HTTP request is larger than (tune.bufsize - tune.maxrewrite), haproxy will
Dmitry Sivachenkof6f4f7b2012-10-21 18:10:25 +04001455 return HTTP 400 (Bad Request) error. Similarly if an HTTP response is larger
Willy Tarreauc77d3642018-12-12 06:19:42 +01001456 than this size, haproxy will return HTTP 502 (Bad Gateway). Note that the
1457 value set using this parameter will automatically be rounded up to the next
1458 multiple of 8 on 32-bit machines and 16 on 64-bit machines.
Willy Tarreau27a674e2009-08-17 07:23:33 +02001459
Willy Tarreau43961d52010-10-04 20:39:20 +02001460tune.chksize <number>
1461 Sets the check buffer size to this size (in bytes). Higher values may help
1462 find string or regex patterns in very large pages, though doing so may imply
1463 more memory and CPU usage. The default value is 16384 and can be changed at
1464 build time. It is not recommended to change this value, but to use better
1465 checks whenever possible.
1466
William Lallemandf3747832012-11-09 12:33:10 +01001467tune.comp.maxlevel <number>
1468 Sets the maximum compression level. The compression level affects CPU
1469 usage during compression. This value affects CPU usage during compression.
1470 Each session using compression initializes the compression algorithm with
1471 this value. The default value is 1.
1472
Willy Tarreauc299e1e2019-02-27 11:35:12 +01001473tune.fail-alloc
1474 If compiled with DEBUG_FAIL_ALLOC, gives the percentage of chances an
1475 allocation attempt fails. Must be between 0 (no failure) and 100 (no
1476 success). This is useful to debug and make sure memory failures are handled
1477 gracefully.
1478
Willy Tarreaufe20e5b2017-07-27 11:42:14 +02001479tune.h2.header-table-size <number>
1480 Sets the HTTP/2 dynamic header table size. It defaults to 4096 bytes and
1481 cannot be larger than 65536 bytes. A larger value may help certain clients
1482 send more compact requests, depending on their capabilities. This amount of
1483 memory is consumed for each HTTP/2 connection. It is recommended not to
1484 change it.
1485
Willy Tarreaue6baec02017-07-27 11:45:11 +02001486tune.h2.initial-window-size <number>
1487 Sets the HTTP/2 initial window size, which is the number of bytes the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01001488 can upload before waiting for an acknowledgment from haproxy. This setting
1489 only affects payload contents (i.e. the body of POST requests), not headers.
Willy Tarreaue6baec02017-07-27 11:45:11 +02001490 The default value is 65535, which roughly allows up to 5 Mbps of upload
1491 bandwidth per client over a network showing a 100 ms ping time, or 500 Mbps
1492 over a 1-ms local network. It can make sense to increase this value to allow
1493 faster uploads, or to reduce it to increase fairness when dealing with many
1494 clients. It doesn't affect resource usage.
1495
Willy Tarreau5242ef82017-07-27 11:47:28 +02001496tune.h2.max-concurrent-streams <number>
1497 Sets the HTTP/2 maximum number of concurrent streams per connection (ie the
1498 number of outstanding requests on a single connection). The default value is
1499 100. A larger one may slightly improve page load time for complex sites when
1500 visited over high latency networks, but increases the amount of resources a
1501 single client may allocate. A value of zero disables the limit so a single
1502 client may create as many streams as allocatable by haproxy. It is highly
1503 recommended not to change this value.
1504
Willy Tarreaua24b35c2019-02-21 13:24:36 +01001505tune.h2.max-frame-size <number>
1506 Sets the HTTP/2 maximum frame size that haproxy announces it is willing to
1507 receive to its peers. The default value is the largest between 16384 and the
1508 buffer size (tune.bufsize). In any case, haproxy will not announce support
1509 for frame sizes larger than buffers. The main purpose of this setting is to
1510 allow to limit the maximum frame size setting when using large buffers. Too
1511 large frame sizes might have performance impact or cause some peers to
1512 misbehave. It is highly recommended not to change this value.
1513
Willy Tarreau193b8c62012-11-22 00:17:38 +01001514tune.http.cookielen <number>
1515 Sets the maximum length of captured cookies. This is the maximum value that
1516 the "capture cookie xxx len yyy" will be allowed to take, and any upper value
1517 will automatically be truncated to this one. It is important not to set too
1518 high a value because all cookie captures still allocate this size whatever
1519 their configured value (they share a same pool). This value is per request
1520 per response, so the memory allocated is twice this value per connection.
1521 When not specified, the limit is set to 63 characters. It is recommended not
1522 to change this value.
1523
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001524tune.http.logurilen <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001525 Sets the maximum length of request URI in logs. This prevents truncating long
1526 request URIs with valuable query strings in log lines. This is not related
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001527 to syslog limits. If you increase this limit, you may also increase the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001528 'log ... len yyy' parameter. Your syslog daemon may also need specific
Stéphane Cottin23e9e932017-05-18 08:58:41 +02001529 configuration directives too.
1530 The default value is 1024.
1531
Willy Tarreauac1932d2011-10-24 19:14:41 +02001532tune.http.maxhdr <number>
1533 Sets the maximum number of headers in a request. When a request comes with a
1534 number of headers greater than this value (including the first line), it is
1535 rejected with a "400 Bad Request" status code. Similarly, too large responses
1536 are blocked with "502 Bad Gateway". The default value is 101, which is enough
1537 for all usages, considering that the widely deployed Apache server uses the
1538 same limit. It can be useful to push this limit further to temporarily allow
Christopher Faulet50174f32017-06-21 16:31:35 +02001539 a buggy application to work by the time it gets fixed. The accepted range is
1540 1..32767. Keep in mind that each new header consumes 32bits of memory for
1541 each session, so don't push this limit too high.
Willy Tarreauac1932d2011-10-24 19:14:41 +02001542
Willy Tarreau7e312732014-02-12 16:35:14 +01001543tune.idletimer <timeout>
1544 Sets the duration after which haproxy will consider that an empty buffer is
1545 probably associated with an idle stream. This is used to optimally adjust
1546 some packet sizes while forwarding large and small data alternatively. The
1547 decision to use splice() or to send large buffers in SSL is modulated by this
1548 parameter. The value is in milliseconds between 0 and 65535. A value of zero
1549 means that haproxy will not try to detect idle streams. The default is 1000,
Davor Ocelice9ed2812017-12-25 17:49:28 +01001550 which seems to correctly detect end user pauses (e.g. read a page before
Willy Tarreau7e312732014-02-12 16:35:14 +01001551 clicking). There should be not reason for changing this value. Please check
1552 tune.ssl.maxrecord below.
1553
Willy Tarreau7ac908b2019-02-27 12:02:18 +01001554tune.listener.multi-queue { on | off }
1555 Enables ('on') or disables ('off') the listener's multi-queue accept which
1556 spreads the incoming traffic to all threads a "bind" line is allowed to run
1557 on instead of taking them for itself. This provides a smoother traffic
1558 distribution and scales much better, especially in environments where threads
1559 may be unevenly loaded due to external activity (network interrupts colliding
1560 with one thread for example). This option is enabled by default, but it may
1561 be forcefully disabled for troubleshooting or for situations where it is
1562 estimated that the operating system already provides a good enough
1563 distribution and connections are extremely short-lived.
1564
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001565tune.lua.forced-yield <number>
1566 This directive forces the Lua engine to execute a yield each <number> of
Tim Düsterhus4896c442016-11-29 02:15:19 +01001567 instructions executed. This permits interrupting a long script and allows the
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001568 HAProxy scheduler to process other tasks like accepting connections or
1569 forwarding traffic. The default value is 10000 instructions. If HAProxy often
Davor Ocelice9ed2812017-12-25 17:49:28 +01001570 executes some Lua code but more responsiveness is required, this value can be
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001571 lowered. If the Lua code is quite long and its result is absolutely required
1572 to process the data, the <number> can be increased.
1573
Willy Tarreau32f61e22015-03-18 17:54:59 +01001574tune.lua.maxmem
1575 Sets the maximum amount of RAM in megabytes per process usable by Lua. By
1576 default it is zero which means unlimited. It is important to set a limit to
1577 ensure that a bug in a script will not result in the system running out of
1578 memory.
1579
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001580tune.lua.session-timeout <timeout>
1581 This is the execution timeout for the Lua sessions. This is useful for
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001582 preventing infinite loops or spending too much time in Lua. This timeout
1583 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001584 not taken in account. The default timeout is 4s.
Thierry FOURNIER90da1912015-03-05 11:17:06 +01001585
1586tune.lua.task-timeout <timeout>
1587 Purpose is the same as "tune.lua.session-timeout", but this timeout is
1588 dedicated to the tasks. By default, this timeout isn't set because a task may
1589 remain alive during of the lifetime of HAProxy. For example, a task used to
1590 check servers.
1591
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001592tune.lua.service-timeout <timeout>
1593 This is the execution timeout for the Lua services. This is useful for
1594 preventing infinite loops or spending too much time in Lua. This timeout
1595 counts only the pure Lua runtime. If the Lua does a sleep, the sleep is
Davor Ocelice9ed2812017-12-25 17:49:28 +01001596 not taken in account. The default timeout is 4s.
Thierry FOURNIER7dd784b2015-10-01 14:49:33 +02001597
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001598tune.maxaccept <number>
Willy Tarreau16a21472012-11-19 12:39:59 +01001599 Sets the maximum number of consecutive connections a process may accept in a
1600 row before switching to other work. In single process mode, higher numbers
1601 give better performance at high connection rates. However in multi-process
1602 modes, keeping a bit of fairness between processes generally is better to
1603 increase performance. This value applies individually to each listener, so
1604 that the number of processes a listener is bound to is taken into account.
1605 This value defaults to 64. In multi-process mode, it is divided by twice
1606 the number of processes the listener is bound to. Setting this value to -1
1607 completely disables the limitation. It should normally not be needed to tweak
1608 this value.
Willy Tarreaua0250ba2008-01-06 11:22:57 +01001609
1610tune.maxpollevents <number>
1611 Sets the maximum amount of events that can be processed at once in a call to
1612 the polling system. The default value is adapted to the operating system. It
1613 has been noticed that reducing it below 200 tends to slightly decrease
1614 latency at the expense of network bandwidth, and increasing it above 200
1615 tends to trade latency for slightly increased bandwidth.
1616
Willy Tarreau27a674e2009-08-17 07:23:33 +02001617tune.maxrewrite <number>
1618 Sets the reserved buffer space to this size in bytes. The reserved space is
1619 used for header rewriting or appending. The first reads on sockets will never
1620 fill more than bufsize-maxrewrite. Historically it has defaulted to half of
1621 bufsize, though that does not make much sense since there are rarely large
1622 numbers of headers to add. Setting it too high prevents processing of large
1623 requests or responses. Setting it too low prevents addition of new headers
1624 to already large requests or to POST requests. It is generally wise to set it
1625 to about 1024. It is automatically readjusted to half of bufsize if it is
1626 larger than that. This means you don't have to worry about it when changing
1627 bufsize.
1628
Willy Tarreauf3045d22015-04-29 16:24:50 +02001629tune.pattern.cache-size <number>
1630 Sets the size of the pattern lookup cache to <number> entries. This is an LRU
1631 cache which reminds previous lookups and their results. It is used by ACLs
1632 and maps on slow pattern lookups, namely the ones using the "sub", "reg",
1633 "dir", "dom", "end", "bin" match methods as well as the case-insensitive
1634 strings. It applies to pattern expressions which means that it will be able
1635 to memorize the result of a lookup among all the patterns specified on a
1636 configuration line (including all those loaded from files). It automatically
1637 invalidates entries which are updated using HTTP actions or on the CLI. The
1638 default cache size is set to 10000 entries, which limits its footprint to
1639 about 5 MB on 32-bit systems and 8 MB on 64-bit systems. There is a very low
1640 risk of collision in this cache, which is in the order of the size of the
1641 cache divided by 2^64. Typically, at 10000 requests per second with the
1642 default cache size of 10000 entries, there's 1% chance that a brute force
1643 attack could cause a single collision after 60 years, or 0.1% after 6 years.
1644 This is considered much lower than the risk of a memory corruption caused by
1645 aging components. If this is not acceptable, the cache can be disabled by
1646 setting this parameter to 0.
1647
Willy Tarreaubd9a0a72011-10-23 21:14:29 +02001648tune.pipesize <number>
1649 Sets the kernel pipe buffer size to this size (in bytes). By default, pipes
1650 are the default size for the system. But sometimes when using TCP splicing,
1651 it can improve performance to increase pipe sizes, especially if it is
1652 suspected that pipes are not filled and that many calls to splice() are
1653 performed. This has an impact on the kernel's memory footprint, so this must
1654 not be changed if impacts are not understood.
1655
Willy Tarreaue803de22010-01-21 17:43:04 +01001656tune.rcvbuf.client <number>
1657tune.rcvbuf.server <number>
1658 Forces the kernel socket receive buffer size on the client or the server side
1659 to the specified value in bytes. This value applies to all TCP/HTTP frontends
1660 and backends. It should normally never be set, and the default size (0) lets
1661 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001662 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001663 order to save kernel memory by preventing it from buffering too large amounts
1664 of received data. Lower values will significantly increase CPU usage though.
1665
Willy Tarreaub22fc302015-12-14 12:04:35 +01001666tune.recv_enough <number>
Davor Ocelice9ed2812017-12-25 17:49:28 +01001667 HAProxy uses some hints to detect that a short read indicates the end of the
Willy Tarreaub22fc302015-12-14 12:04:35 +01001668 socket buffers. One of them is that a read returns more than <recv_enough>
1669 bytes, which defaults to 10136 (7 segments of 1448 each). This default value
1670 may be changed by this setting to better deal with workloads involving lots
1671 of short messages such as telnet or SSH sessions.
1672
Olivier Houchard1599b802018-05-24 18:59:04 +02001673tune.runqueue-depth <number>
1674 Sets the maxinum amount of task that can be processed at once when running
1675 tasks. The default value is 200. Increasing it may incur latency when
1676 dealing with I/Os, making it too small can incur extra overhead.
1677
Willy Tarreaue803de22010-01-21 17:43:04 +01001678tune.sndbuf.client <number>
1679tune.sndbuf.server <number>
1680 Forces the kernel socket send buffer size on the client or the server side to
1681 the specified value in bytes. This value applies to all TCP/HTTP frontends
1682 and backends. It should normally never be set, and the default size (0) lets
1683 the kernel autotune this value depending on the amount of available memory.
Davor Ocelice9ed2812017-12-25 17:49:28 +01001684 However it can sometimes help to set it to very low values (e.g. 4096) in
Willy Tarreaue803de22010-01-21 17:43:04 +01001685 order to save kernel memory by preventing it from buffering too large amounts
1686 of received data. Lower values will significantly increase CPU usage though.
1687 Another use case is to prevent write timeouts with extremely slow clients due
1688 to the kernel waiting for a large part of the buffer to be read before
1689 notifying haproxy again.
1690
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001691tune.ssl.cachesize <number>
Emeric Brunaf9619d2012-11-28 18:47:52 +01001692 Sets the size of the global SSL session cache, in a number of blocks. A block
1693 is large enough to contain an encoded session without peer certificate.
1694 An encoded session with peer certificate is stored in multiple blocks
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001695 depending on the size of the peer certificate. A block uses approximately
Emeric Brunaf9619d2012-11-28 18:47:52 +01001696 200 bytes of memory. The default value may be forced at build time, otherwise
Davor Ocelice9ed2812017-12-25 17:49:28 +01001697 defaults to 20000. When the cache is full, the most idle entries are purged
Emeric Brunaf9619d2012-11-28 18:47:52 +01001698 and reassigned. Higher values reduce the occurrence of such a purge, hence
1699 the number of CPU-intensive SSL handshakes by ensuring that all users keep
1700 their session as long as possible. All entries are pre-allocated upon startup
Emeric Brun22890a12012-12-28 14:41:32 +01001701 and are shared between all processes if "nbproc" is greater than 1. Setting
1702 this value to 0 disables the SSL session cache.
Willy Tarreau6ec58db2012-11-16 16:32:15 +01001703
Emeric Brun8dc60392014-05-09 13:52:00 +02001704tune.ssl.force-private-cache
Lukas Tribus27935782018-10-01 02:00:16 +02001705 This option disables SSL session cache sharing between all processes. It
Emeric Brun8dc60392014-05-09 13:52:00 +02001706 should normally not be used since it will force many renegotiations due to
1707 clients hitting a random process. But it may be required on some operating
1708 systems where none of the SSL cache synchronization method may be used. In
1709 this case, adding a first layer of hash-based load balancing before the SSL
1710 layer might limit the impact of the lack of session sharing.
1711
Emeric Brun4f65bff2012-11-16 15:11:00 +01001712tune.ssl.lifetime <timeout>
1713 Sets how long a cached SSL session may remain valid. This time is expressed
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001714 in seconds and defaults to 300 (5 min). It is important to understand that it
Emeric Brun4f65bff2012-11-16 15:11:00 +01001715 does not guarantee that sessions will last that long, because if the cache is
1716 full, the longest idle sessions will be purged despite their configured
1717 lifetime. The real usefulness of this setting is to prevent sessions from
1718 being used for too long.
1719
Willy Tarreaubfd59462013-02-21 07:46:09 +01001720tune.ssl.maxrecord <number>
1721 Sets the maximum amount of bytes passed to SSL_write() at a time. Default
1722 value 0 means there is no limit. Over SSL/TLS, the client can decipher the
1723 data only once it has received a full record. With large records, it means
1724 that clients might have to download up to 16kB of data before starting to
1725 process them. Limiting the value can improve page load times on browsers
1726 located over high latency or low bandwidth networks. It is suggested to find
1727 optimal values which fit into 1 or 2 TCP segments (generally 1448 bytes over
1728 Ethernet with TCP timestamps enabled, or 1460 when timestamps are disabled),
1729 keeping in mind that SSL/TLS add some overhead. Typical values of 1419 and
1730 2859 gave good results during tests. Use "strace -e trace=write" to find the
Davor Ocelice9ed2812017-12-25 17:49:28 +01001731 best value. HAProxy will automatically switch to this setting after an idle
Willy Tarreau7e312732014-02-12 16:35:14 +01001732 stream has been detected (see tune.idletimer above).
Willy Tarreaubfd59462013-02-21 07:46:09 +01001733
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001734tune.ssl.default-dh-param <number>
1735 Sets the maximum size of the Diffie-Hellman parameters used for generating
1736 the ephemeral/temporary Diffie-Hellman key in case of DHE key exchange. The
1737 final size will try to match the size of the server's RSA (or DSA) key (e.g,
1738 a 2048 bits temporary DH key for a 2048 bits RSA key), but will not exceed
1739 this maximum value. Default value if 1024. Only 1024 or higher values are
1740 allowed. Higher values will increase the CPU load, and values greater than
1741 1024 bits are not supported by Java 7 and earlier clients. This value is not
Remi Gacogne47783ef2015-05-29 15:53:22 +02001742 used if static Diffie-Hellman parameters are supplied either directly
1743 in the certificate file or by using the ssl-dh-param-file parameter.
Remi Gacognef46cd6e2014-06-12 14:58:40 +02001744
Christopher Faulet31af49d2015-06-09 17:29:50 +02001745tune.ssl.ssl-ctx-cache-size <number>
1746 Sets the size of the cache used to store generated certificates to <number>
1747 entries. This is a LRU cache. Because generating a SSL certificate
1748 dynamically is expensive, they are cached. The default cache size is set to
1749 1000 entries.
1750
Thierry FOURNIER5bf77322017-02-25 12:45:22 +01001751tune.ssl.capture-cipherlist-size <number>
1752 Sets the maximum size of the buffer used for capturing client-hello cipher
1753 list. If the value is 0 (default value) the capture is disabled, otherwise
1754 a buffer is allocated for each SSL/TLS connection.
1755
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001756tune.vars.global-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001757tune.vars.proc-max-size <size>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001758tune.vars.reqres-max-size <size>
1759tune.vars.sess-max-size <size>
1760tune.vars.txn-max-size <size>
Christopher Fauletff2613e2016-11-09 11:36:17 +01001761 These five tunes help to manage the maximum amount of memory used by the
1762 variables system. "global" limits the overall amount of memory available for
1763 all scopes. "proc" limits the memory for the process scope, "sess" limits the
1764 memory for the session scope, "txn" for the transaction scope, and "reqres"
1765 limits the memory for each request or response processing.
1766 Memory accounting is hierarchical, meaning more coarse grained limits include
1767 the finer grained ones: "proc" includes "sess", "sess" includes "txn", and
1768 "txn" includes "reqres".
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001769
Daniel Schneller0b547052016-03-21 20:46:57 +01001770 For example, when "tune.vars.sess-max-size" is limited to 100,
1771 "tune.vars.txn-max-size" and "tune.vars.reqres-max-size" cannot exceed
1772 100 either. If we create a variable "txn.var" that contains 100 bytes,
1773 all available space is consumed.
1774 Notice that exceeding the limits at runtime will not result in an error
1775 message, but values might be cut off or corrupted. So make sure to accurately
1776 plan for the amount of space needed to store all your variables.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02001777
William Lallemanda509e4c2012-11-07 16:54:34 +01001778tune.zlib.memlevel <number>
1779 Sets the memLevel parameter in zlib initialization for each session. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03001780 defines how much memory should be allocated for the internal compression
William Lallemanda509e4c2012-11-07 16:54:34 +01001781 state. A value of 1 uses minimum memory but is slow and reduces compression
Davor Ocelice9ed2812017-12-25 17:49:28 +01001782 ratio, a value of 9 uses maximum memory for optimal speed. Can be a value
William Lallemanda509e4c2012-11-07 16:54:34 +01001783 between 1 and 9. The default value is 8.
1784
1785tune.zlib.windowsize <number>
1786 Sets the window size (the size of the history buffer) as a parameter of the
1787 zlib initialization for each session. Larger values of this parameter result
Davor Ocelice9ed2812017-12-25 17:49:28 +01001788 in better compression at the expense of memory usage. Can be a value between
1789 8 and 15. The default value is 15.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001790
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017913.3. Debugging
1792--------------
Willy Tarreau6a06a402007-07-15 20:15:28 +02001793
1794debug
1795 Enables debug mode which dumps to stdout all exchanges, and disables forking
1796 into background. It is the equivalent of the command-line argument "-d". It
1797 should never be used in a production configuration since it may prevent full
1798 system startup.
1799
1800quiet
1801 Do not display any message during startup. It is equivalent to the command-
1802 line argument "-q".
1803
Emeric Brunf099e792010-09-27 12:05:28 +02001804
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010018053.4. Userlists
1806--------------
1807It is possible to control access to frontend/backend/listen sections or to
1808http stats by allowing only authenticated and authorized users. To do this,
1809it is required to create at least one userlist and to define users.
1810
1811userlist <listname>
Cyril Bonté78caf842010-03-10 22:41:43 +01001812 Creates new userlist with name <listname>. Many independent userlists can be
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001813 used to store authentication & authorization data for independent customers.
1814
1815group <groupname> [users <user>,<user>,(...)]
Cyril Bonté78caf842010-03-10 22:41:43 +01001816 Adds group <groupname> to the current userlist. It is also possible to
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001817 attach users to this group by using a comma separated list of names
1818 proceeded by "users" keyword.
1819
Cyril Bontéf0c60612010-02-06 14:44:47 +01001820user <username> [password|insecure-password <password>]
1821 [groups <group>,<group>,(...)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001822 Adds user <username> to the current userlist. Both secure (encrypted) and
1823 insecure (unencrypted) passwords can be used. Encrypted passwords are
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001824 evaluated using the crypt(3) function, so depending on the system's
1825 capabilities, different algorithms are supported. For example, modern Glibc
1826 based Linux systems support MD5, SHA-256, SHA-512, and, of course, the
1827 classic DES-based method of encrypting passwords.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001828
Daniel Schnellerd06f31c2017-11-06 16:51:04 +01001829 Attention: Be aware that using encrypted passwords might cause significantly
1830 increased CPU usage, depending on the number of requests, and the algorithm
1831 used. For any of the hashed variants, the password for each request must
1832 be processed through the chosen algorithm, before it can be compared to the
1833 value specified in the config file. Most current algorithms are deliberately
1834 designed to be expensive to compute to achieve resistance against brute
1835 force attacks. They do not simply salt/hash the clear text password once,
1836 but thousands of times. This can quickly become a major factor in haproxy's
1837 overall CPU consumption!
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001838
1839 Example:
Cyril Bontéf0c60612010-02-06 14:44:47 +01001840 userlist L1
1841 group G1 users tiger,scott
1842 group G2 users xdb,scott
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001843
Cyril Bontéf0c60612010-02-06 14:44:47 +01001844 user tiger password $6$k6y3o.eP$JlKBx9za9667qe4(...)xHSwRv6J.C0/D7cV91
1845 user scott insecure-password elgato
1846 user xdb insecure-password hello
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001847
Cyril Bontéf0c60612010-02-06 14:44:47 +01001848 userlist L2
1849 group G1
1850 group G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001851
Cyril Bontéf0c60612010-02-06 14:44:47 +01001852 user tiger password $6$k6y3o.eP$JlKBx(...)xHSwRv6J.C0/D7cV91 groups G1
1853 user scott insecure-password elgato groups G1,G2
1854 user xdb insecure-password hello groups G2
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01001855
1856 Please note that both lists are functionally identical.
Willy Tarreau6a06a402007-07-15 20:15:28 +02001857
Emeric Brunf099e792010-09-27 12:05:28 +02001858
18593.5. Peers
Cyril Bontédc4d9032012-04-08 21:57:39 +02001860----------
Emeric Brun94900952015-06-11 18:25:54 +02001861It is possible to propagate entries of any data-types in stick-tables between
1862several haproxy instances over TCP connections in a multi-master fashion. Each
1863instance pushes its local updates and insertions to remote peers. The pushed
1864values overwrite remote ones without aggregation. Interrupted exchanges are
1865automatically detected and recovered from the last known point.
1866In addition, during a soft restart, the old process connects to the new one
1867using such a TCP connection to push all its entries before the new process
1868tries to connect to other peers. That ensures very fast replication during a
1869reload, it typically takes a fraction of a second even for large tables.
1870Note that Server IDs are used to identify servers remotely, so it is important
1871that configurations look similar or at least that the same IDs are forced on
1872each server on all participants.
Emeric Brunf099e792010-09-27 12:05:28 +02001873
1874peers <peersect>
Jamie Gloudon801a0a32012-08-25 00:18:33 -04001875 Creates a new peer list with name <peersect>. It is an independent section,
Emeric Brunf099e792010-09-27 12:05:28 +02001876 which is referenced by one or more stick-tables.
1877
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01001878bind [<address>]:<port_range> [, ...] [param*]
1879 Defines the binding parameters of the local peer of this "peers" section.
1880 Such lines are not supported with "peer" line in the same "peers" section.
1881
Willy Tarreau77e4bd12015-05-01 20:02:17 +02001882disabled
1883 Disables a peers section. It disables both listening and any synchronization
1884 related to this section. This is provided to disable synchronization of stick
1885 tables without having to comment out all "peers" references.
1886
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01001887default-bind [param*]
1888 Defines the binding parameters for the local peer, excepted its address.
1889
1890default-server [param*]
1891 Change default options for a server in a "peers" section.
1892
1893 Arguments:
1894 <param*> is a list of parameters for this server. The "default-server"
1895 keyword accepts an important number of options and has a complete
1896 section dedicated to it. Please refer to section 5 for more
1897 details.
1898
1899
1900 See also: "server" and section 5 about server options
1901
Willy Tarreau77e4bd12015-05-01 20:02:17 +02001902enable
1903 This re-enables a disabled peers section which was previously disabled.
1904
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01001905peer <peername> <ip>:<port> [param*]
Emeric Brunf099e792010-09-27 12:05:28 +02001906 Defines a peer inside a peers section.
1907 If <peername> is set to the local peer name (by default hostname, or forced
1908 using "-L" command line option), haproxy will listen for incoming remote peer
1909 connection on <ip>:<port>. Otherwise, <ip>:<port> defines where to connect to
1910 to join the remote peer, and <peername> is used at the protocol level to
1911 identify and validate the remote peer on the server side.
1912
1913 During a soft restart, local peer <ip>:<port> is used by the old instance to
1914 connect the new one and initiate a complete replication (teaching process).
1915
1916 It is strongly recommended to have the exact same peers declaration on all
1917 peers and to only rely on the "-L" command line argument to change the local
1918 peer name. This makes it easier to maintain coherent configuration files
1919 across all peers.
1920
William Lallemandb2f07452015-05-12 14:27:13 +02001921 You may want to reference some environment variables in the address
1922 parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01001923
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01001924 Note: "peer" keyword may transparently be replaced by "server" keyword (see
1925 "server" keyword explanation below).
1926
1927server <peername> [<ip>:<port>] [param*]
1928 As previously mentionned, "peer" keyword may be replaced by "server" keyword
1929 with a support for all "server" parameters found in 5.2 paragraph.
1930 If the underlying peer is local, <ip>:<port> parameters must not be present.
1931 These parameters must be provided on a "bind" line (see "bind" keyword
1932 of this "peers" section).
1933 Some of these parameters are irrelevant for "peers" sections.
1934
1935
Cyril Bontédc4d9032012-04-08 21:57:39 +02001936 Example:
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01001937 # The old way.
Emeric Brunf099e792010-09-27 12:05:28 +02001938 peers mypeers
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001939 peer haproxy1 192.168.0.1:1024
1940 peer haproxy2 192.168.0.2:1024
1941 peer haproxy3 10.2.0.1:1024
Emeric Brunf099e792010-09-27 12:05:28 +02001942
1943 backend mybackend
1944 mode tcp
1945 balance roundrobin
1946 stick-table type ip size 20k peers mypeers
1947 stick on src
1948
Willy Tarreauf7b30a92010-12-06 22:59:17 +01001949 server srv1 192.168.0.30:80
1950 server srv2 192.168.0.31:80
Emeric Brunf099e792010-09-27 12:05:28 +02001951
Frédéric Lécaille2f167b32019-01-11 14:13:54 +01001952 Example:
1953 peers mypeers
1954 bind 127.0.0.11:10001 ssl crt mycerts/pem
1955 default-server ssl verify none
1956 server hostA 127.0.0.10:10000
1957 server hostB #local peer
Emeric Brunf099e792010-09-27 12:05:28 +02001958
Simon Horman51a1cf62015-02-03 13:00:44 +090019593.6. Mailers
1960------------
1961It is possible to send email alerts when the state of servers changes.
1962If configured email alerts are sent to each mailer that is configured
1963in a mailers section. Email is sent to mailers using SMTP.
1964
Pieter Baauw386a1272015-08-16 15:26:24 +02001965mailers <mailersect>
Simon Horman51a1cf62015-02-03 13:00:44 +09001966 Creates a new mailer list with the name <mailersect>. It is an
1967 independent section which is referenced by one or more proxies.
1968
1969mailer <mailername> <ip>:<port>
1970 Defines a mailer inside a mailers section.
1971
1972 Example:
1973 mailers mymailers
1974 mailer smtp1 192.168.0.1:587
1975 mailer smtp2 192.168.0.2:587
1976
1977 backend mybackend
1978 mode tcp
1979 balance roundrobin
1980
1981 email-alert mailers mymailers
1982 email-alert from test1@horms.org
1983 email-alert to test2@horms.org
1984
1985 server srv1 192.168.0.30:80
1986 server srv2 192.168.0.31:80
1987
Pieter Baauw235fcfc2016-02-13 15:33:40 +01001988timeout mail <time>
1989 Defines the time available for a mail/connection to be made and send to
1990 the mail-server. If not defined the default value is 10 seconds. To allow
1991 for at least two SYN-ACK packets to be send during initial TCP handshake it
1992 is advised to keep this value above 4 seconds.
1993
1994 Example:
1995 mailers mymailers
1996 timeout mail 20s
1997 mailer smtp1 192.168.0.1:587
Simon Horman51a1cf62015-02-03 13:00:44 +09001998
Willy Tarreauc57f0e22009-05-10 13:12:33 +020019994. Proxies
Willy Tarreau6a06a402007-07-15 20:15:28 +02002000----------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002001
Willy Tarreau6a06a402007-07-15 20:15:28 +02002002Proxy configuration can be located in a set of sections :
William Lallemand6e62fb62015-04-28 16:55:23 +02002003 - defaults [<name>]
Willy Tarreau6a06a402007-07-15 20:15:28 +02002004 - frontend <name>
2005 - backend <name>
2006 - listen <name>
2007
2008A "defaults" section sets default parameters for all other sections following
2009its declaration. Those default parameters are reset by the next "defaults"
2010section. See below for the list of parameters which can be set in a "defaults"
Willy Tarreau0ba27502007-12-24 16:55:16 +01002011section. The name is optional but its use is encouraged for better readability.
Willy Tarreau6a06a402007-07-15 20:15:28 +02002012
2013A "frontend" section describes a set of listening sockets accepting client
2014connections.
2015
2016A "backend" section describes a set of servers to which the proxy will connect
2017to forward incoming connections.
2018
2019A "listen" section defines a complete proxy with its frontend and backend
2020parts combined in one section. It is generally useful for TCP-only traffic.
2021
Willy Tarreau0ba27502007-12-24 16:55:16 +01002022All proxy names must be formed from upper and lower case letters, digits,
2023'-' (dash), '_' (underscore) , '.' (dot) and ':' (colon). ACL names are
2024case-sensitive, which means that "www" and "WWW" are two different proxies.
2025
2026Historically, all proxy names could overlap, it just caused troubles in the
2027logs. Since the introduction of content switching, it is mandatory that two
2028proxies with overlapping capabilities (frontend/backend) have different names.
2029However, it is still permitted that a frontend and a backend share the same
2030name, as this configuration seems to be commonly encountered.
2031
2032Right now, two major proxy modes are supported : "tcp", also known as layer 4,
2033and "http", also known as layer 7. In layer 4 mode, HAProxy simply forwards
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002034bidirectional traffic between two sides. In layer 7 mode, HAProxy analyzes the
Willy Tarreau0ba27502007-12-24 16:55:16 +01002035protocol, and can interact with it by allowing, blocking, switching, adding,
2036modifying, or removing arbitrary contents in requests or responses, based on
2037arbitrary criteria.
2038
Willy Tarreau70dffda2014-01-30 03:07:23 +01002039In HTTP mode, the processing applied to requests and responses flowing over
2040a connection depends in the combination of the frontend's HTTP options and
Christopher Faulet315b39c2018-09-21 16:26:19 +02002041the backend's. HAProxy supports 4 connection modes :
Willy Tarreau70dffda2014-01-30 03:07:23 +01002042
2043 - KAL : keep alive ("option http-keep-alive") which is the default mode : all
2044 requests and responses are processed, and connections remain open but idle
2045 between responses and new requests.
2046
2047 - TUN: tunnel ("option http-tunnel") : this was the default mode for versions
2048 1.0 to 1.5-dev21 : only the first request and response are processed, and
2049 everything else is forwarded with no analysis at all. This mode should not
Christopher Faulet4212a302018-09-21 10:42:19 +02002050 be used as it creates lots of trouble with logging and HTTP processing. It
2051 is supported only on frontends.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002052
Willy Tarreau70dffda2014-01-30 03:07:23 +01002053 - SCL: server close ("option http-server-close") : the server-facing
2054 connection is closed after the end of the response is received, but the
2055 client-facing connection remains open.
2056
Christopher Faulet315b39c2018-09-21 16:26:19 +02002057 - CLO: close ("option httpclose"): the connection is closed after the end of
2058 the response and "Connection: close" appended in both directions.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002059
2060The effective mode that will be applied to a connection passing through a
2061frontend and a backend can be determined by both proxy modes according to the
2062following matrix, but in short, the modes are symmetric, keep-alive is the
Christopher Faulet315b39c2018-09-21 16:26:19 +02002063weakest option and close is the strongest.
Willy Tarreau70dffda2014-01-30 03:07:23 +01002064
Christopher Faulet315b39c2018-09-21 16:26:19 +02002065 Backend mode
Willy Tarreau70dffda2014-01-30 03:07:23 +01002066
Christopher Faulet315b39c2018-09-21 16:26:19 +02002067 | KAL | SCL | CLO
2068 ----+-----+-----+----
2069 KAL | KAL | SCL | CLO
2070 ----+-----+-----+----
2071 TUN | TUN | SCL | CLO
2072 Frontend ----+-----+-----+----
2073 mode SCL | SCL | SCL | CLO
2074 ----+-----+-----+----
2075 CLO | CLO | CLO | CLO
Willy Tarreau70dffda2014-01-30 03:07:23 +01002076
Willy Tarreau0ba27502007-12-24 16:55:16 +01002077
Willy Tarreau70dffda2014-01-30 03:07:23 +01002078
Willy Tarreauc57f0e22009-05-10 13:12:33 +020020794.1. Proxy keywords matrix
2080--------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002081
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002082The following list of keywords is supported. Most of them may only be used in a
2083limited set of section types. Some of them are marked as "deprecated" because
2084they are inherited from an old syntax which may be confusing or functionally
2085limited, and there are new recommended keywords to replace them. Keywords
Davor Ocelice9ed2812017-12-25 17:49:28 +01002086marked with "(*)" can be optionally inverted using the "no" prefix, e.g. "no
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002087option contstats". This makes sense when the option has been enabled by default
Willy Tarreau3842f002009-06-14 11:39:52 +02002088and must be disabled for a specific instance. Such options may also be prefixed
2089with "default" in order to restore default settings regardless of what has been
2090specified in a previous "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002091
Willy Tarreau6a06a402007-07-15 20:15:28 +02002092
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002093 keyword defaults frontend listen backend
2094------------------------------------+----------+----------+---------+---------
2095acl - X X X
Willy Tarreau294d0f02015-08-10 19:40:12 +02002096appsession - - - -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002097backlog X X X -
2098balance X - X X
2099bind - X X -
2100bind-process X X X X
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002101block (deprecated) - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002102capture cookie - X X -
2103capture request header - X X -
2104capture response header - X X -
2105clitimeout (deprecated) X X X -
William Lallemand82fe75c2012-10-23 10:25:10 +02002106compression X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002107contimeout (deprecated) X - X X
2108cookie X - X X
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02002109declare capture - X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002110default-server X - X X
2111default_backend X X X -
2112description - X X X
2113disabled X X X X
2114dispatch - - X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002115email-alert from X X X X
Simon Horman64e34162015-02-06 11:11:57 +09002116email-alert level X X X X
Simon Horman51a1cf62015-02-03 13:00:44 +09002117email-alert mailers X X X X
2118email-alert myhostname X X X X
2119email-alert to X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002120enabled X X X X
2121errorfile X X X X
2122errorloc X X X X
2123errorloc302 X X X X
2124-- keyword -------------------------- defaults - frontend - listen -- backend -
2125errorloc303 X X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002126force-persist - - X X
Christopher Fauletc3fe5332016-04-07 15:30:10 +02002127filter - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002128fullconn X - X X
2129grace X X X X
2130hash-type X - X X
2131http-check disable-on-404 X - X X
Willy Tarreaubd741542010-03-16 18:46:54 +01002132http-check expect - - X X
Willy Tarreau7ab6aff2010-10-12 06:30:16 +02002133http-check send-state X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002134http-request - X X X
Willy Tarreaue365c0b2013-06-11 16:06:12 +02002135http-response - X X X
Willy Tarreau30631952015-08-06 15:05:24 +02002136http-reuse X - X X
Baptiste Assmann2c42ef52013-10-09 21:57:02 +02002137http-send-name-header - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002138id - X X X
Cyril Bonté4288c5a2018-03-12 22:02:59 +01002139ignore-persist - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002140load-server-state-from-file X - X X
William Lallemand0f99e342011-10-12 17:50:54 +02002141log (*) X X X X
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01002142log-format X X X -
Dragan Dosen7ad31542015-09-28 17:16:47 +02002143log-format-sd X X X -
Willy Tarreau094af4e2015-01-07 15:03:42 +01002144log-tag X X X X
Willy Tarreauc35362a2014-04-25 13:58:37 +02002145max-keep-alive-queue X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002146maxconn X X X -
2147mode X X X X
2148monitor fail - X X -
2149monitor-net X X X -
2150monitor-uri X X X -
2151option abortonclose (*) X - X X
2152option accept-invalid-http-request (*) X X X -
2153option accept-invalid-http-response (*) X - X X
2154option allbackups (*) X - X X
2155option checkcache (*) X - X X
2156option clitcpka (*) X X X -
2157option contstats (*) X X X -
2158option dontlog-normal (*) X X X -
2159option dontlognull (*) X X X -
Christopher Faulet315b39c2018-09-21 16:26:19 +02002160option forceclose (deprectated) (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002161-- keyword -------------------------- defaults - frontend - listen -- backend -
2162option forwardfor X X X X
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02002163option http-buffer-request (*) X X X X
Willy Tarreau82649f92015-05-01 22:40:51 +02002164option http-ignore-probes (*) X X X -
Willy Tarreau16bfb022010-01-16 19:48:41 +01002165option http-keep-alive (*) X X X X
Willy Tarreau96e31212011-05-30 18:10:30 +02002166option http-no-delay (*) X X X X
Christopher Faulet98db9762018-09-21 10:25:19 +02002167option http-pretend-keepalive (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002168option http-server-close (*) X X X X
Christopher Faulet4212a302018-09-21 10:42:19 +02002169option http-tunnel (*) X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002170option http-use-proxy-header (*) X X X -
Willy Tarreau68ad3a42018-10-22 11:49:15 +02002171option http-use-htx (*) X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002172option httpchk X - X X
2173option httpclose (*) X X X X
2174option httplog X X X X
2175option http_proxy (*) X X X X
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002176option independent-streams (*) X X X X
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02002177option ldap-check X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002178option external-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002179option log-health-checks (*) X - X X
2180option log-separate-errors (*) X X X -
2181option logasap (*) X X X -
2182option mysql-check X - X X
2183option nolinger (*) X X X X
2184option originalto X X X X
2185option persist (*) X - X X
Baptiste Assmann809e22a2015-10-12 20:22:55 +02002186option pgsql-check X - X X
2187option prefer-last-server (*) X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002188option redispatch (*) X - X X
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02002189option redis-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002190option smtpchk X - X X
2191option socket-stats (*) X X X -
2192option splice-auto (*) X X X X
2193option splice-request (*) X X X X
2194option splice-response (*) X X X X
Christopher Fauletba7bc162016-11-07 21:07:38 +01002195option spop-check - - - X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002196option srvtcpka (*) X - X X
2197option ssl-hello-chk X - X X
2198-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreaued179852013-12-16 01:07:00 +01002199option tcp-check X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002200option tcp-smart-accept (*) X X X -
2201option tcp-smart-connect (*) X - X X
2202option tcpka X X X X
2203option tcplog X X X X
2204option transparent (*) X - X X
Simon Horman98637e52014-06-20 12:30:16 +09002205external-check command X - X X
2206external-check path X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002207persist rdp-cookie X - X X
2208rate-limit sessions X X X -
2209redirect - X X X
2210redisp (deprecated) X - X X
2211redispatch (deprecated) X - X X
2212reqadd - X X X
2213reqallow - X X X
2214reqdel - X X X
2215reqdeny - X X X
2216reqiallow - X X X
2217reqidel - X X X
2218reqideny - X X X
2219reqipass - X X X
2220reqirep - X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002221reqitarpit - X X X
2222reqpass - X X X
2223reqrep - X X X
2224-- keyword -------------------------- defaults - frontend - listen -- backend -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002225reqtarpit - X X X
2226retries X - X X
2227rspadd - X X X
2228rspdel - X X X
2229rspdeny - X X X
2230rspidel - X X X
2231rspideny - X X X
2232rspirep - X X X
2233rsprep - X X X
2234server - - X X
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02002235server-state-file-name X - X X
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02002236server-template - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002237source X - X X
2238srvtimeout (deprecated) X - X X
Baptiste Assmann5a549212015-10-12 20:30:24 +02002239stats admin - X X X
2240stats auth X X X X
2241stats enable X X X X
2242stats hide-version X X X X
2243stats http-request - X X X
2244stats realm X X X X
2245stats refresh X X X X
2246stats scope X X X X
2247stats show-desc X X X X
2248stats show-legends X X X X
2249stats show-node X X X X
2250stats uri X X X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002251-- keyword -------------------------- defaults - frontend - listen -- backend -
2252stick match - - X X
2253stick on - - X X
2254stick store-request - - X X
Willy Tarreaud8dc99f2011-07-01 11:33:25 +02002255stick store-response - - X X
Adam Spiers68af3c12017-04-06 16:31:39 +01002256stick-table - X X X
Willy Tarreau938c7fe2014-04-25 14:21:39 +02002257tcp-check connect - - X X
2258tcp-check expect - - X X
2259tcp-check send - - X X
2260tcp-check send-binary - - X X
Willy Tarreaue9656522010-08-17 15:40:09 +02002261tcp-request connection - X X -
2262tcp-request content - X X X
Willy Tarreaua56235c2010-09-14 11:31:36 +02002263tcp-request inspect-delay - X X X
Willy Tarreau4f614292016-10-21 17:49:36 +02002264tcp-request session - X X -
Emeric Brun0a3b67f2010-09-24 15:34:53 +02002265tcp-response content - - X X
2266tcp-response inspect-delay - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002267timeout check X - X X
2268timeout client X X X -
Willy Tarreau05cdd962014-05-10 14:30:07 +02002269timeout client-fin X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002270timeout clitimeout (deprecated) X X X -
2271timeout connect X - X X
2272timeout contimeout (deprecated) X - X X
2273timeout http-keep-alive X X X X
2274timeout http-request X X X X
2275timeout queue X - X X
2276timeout server X - X X
Willy Tarreau05cdd962014-05-10 14:30:07 +02002277timeout server-fin X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002278timeout srvtimeout (deprecated) X - X X
2279timeout tarpit X X X X
Willy Tarreauce887fd2012-05-12 12:50:00 +02002280timeout tunnel X - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002281transparent (deprecated) X - X X
William Lallemanda73203e2012-03-12 12:48:57 +01002282unique-id-format X X X -
2283unique-id-header X X X -
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002284use_backend - X X -
Willy Tarreau4a5cade2012-04-05 21:09:48 +02002285use-server - - X X
Willy Tarreau5c6f7b32010-02-26 13:34:29 +01002286------------------------------------+----------+----------+---------+---------
2287 keyword defaults frontend listen backend
Willy Tarreau6a06a402007-07-15 20:15:28 +02002288
Willy Tarreau0ba27502007-12-24 16:55:16 +01002289
Willy Tarreauc57f0e22009-05-10 13:12:33 +020022904.2. Alphabetically sorted keywords reference
2291---------------------------------------------
Willy Tarreau0ba27502007-12-24 16:55:16 +01002292
2293This section provides a description of each keyword and its usage.
2294
2295
2296acl <aclname> <criterion> [flags] [operator] <value> ...
2297 Declare or complete an access list.
2298 May be used in sections : defaults | frontend | listen | backend
2299 no | yes | yes | yes
2300 Example:
2301 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2302 acl invalid_src src_port 0:1023
2303 acl local_dst hdr(host) -i localhost
2304
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002305 See section 7 about ACL usage.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002306
2307
Cyril Bontéb21570a2009-11-29 20:04:48 +01002308appsession <cookie> len <length> timeout <holdtime>
2309 [request-learn] [prefix] [mode <path-parameters|query-string>]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002310 Define session stickiness on an existing application cookie.
2311 May be used in sections : defaults | frontend | listen | backend
2312 no | no | yes | yes
2313 Arguments :
2314 <cookie> this is the name of the cookie used by the application and which
2315 HAProxy will have to learn for each new session.
2316
Cyril Bontéb21570a2009-11-29 20:04:48 +01002317 <length> this is the max number of characters that will be memorized and
Willy Tarreau0ba27502007-12-24 16:55:16 +01002318 checked in each cookie value.
2319
2320 <holdtime> this is the time after which the cookie will be removed from
2321 memory if unused. If no unit is specified, this time is in
2322 milliseconds.
2323
Cyril Bontébf47aeb2009-10-15 00:15:40 +02002324 request-learn
2325 If this option is specified, then haproxy will be able to learn
2326 the cookie found in the request in case the server does not
2327 specify any in response. This is typically what happens with
2328 PHPSESSID cookies, or when haproxy's session expires before
2329 the application's session and the correct server is selected.
2330 It is recommended to specify this option to improve reliability.
2331
Cyril Bontéb21570a2009-11-29 20:04:48 +01002332 prefix When this option is specified, haproxy will match on the cookie
2333 prefix (or URL parameter prefix). The appsession value is the
2334 data following this prefix.
2335
2336 Example :
2337 appsession ASPSESSIONID len 64 timeout 3h prefix
2338
Davor Ocelice9ed2812017-12-25 17:49:28 +01002339 This will match the cookie ASPSESSIONIDXXX=XXXX,
2340 the appsession value will be XXX=XXXX.
Cyril Bontéb21570a2009-11-29 20:04:48 +01002341
2342 mode This option allows to change the URL parser mode.
2343 2 modes are currently supported :
2344 - path-parameters :
2345 The parser looks for the appsession in the path parameters
2346 part (each parameter is separated by a semi-colon), which is
2347 convenient for JSESSIONID for example.
2348 This is the default mode if the option is not set.
2349 - query-string :
2350 In this mode, the parser will look for the appsession in the
2351 query string.
2352
Willy Tarreau294d0f02015-08-10 19:40:12 +02002353 As of version 1.6, appsessions was removed. It is more flexible and more
2354 convenient to use stick-tables instead, and stick-tables support multi-master
2355 replication and data conservation across reloads, which appsessions did not.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002356
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01002357 See also : "cookie", "capture cookie", "balance", "stick", "stick-table",
2358 "ignore-persist", "nbproc" and "bind-process".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002359
2360
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002361backlog <conns>
2362 Give hints to the system about the approximate listen backlog desired size
2363 May be used in sections : defaults | frontend | listen | backend
2364 yes | yes | yes | no
2365 Arguments :
2366 <conns> is the number of pending connections. Depending on the operating
2367 system, it may represent the number of already acknowledged
Cyril Bontédc4d9032012-04-08 21:57:39 +02002368 connections, of non-acknowledged ones, or both.
Willy Tarreauc73ce2b2008-01-06 10:55:10 +01002369
2370 In order to protect against SYN flood attacks, one solution is to increase
2371 the system's SYN backlog size. Depending on the system, sometimes it is just
2372 tunable via a system parameter, sometimes it is not adjustable at all, and
2373 sometimes the system relies on hints given by the application at the time of
2374 the listen() syscall. By default, HAProxy passes the frontend's maxconn value
2375 to the listen() syscall. On systems which can make use of this value, it can
2376 sometimes be useful to be able to specify a different value, hence this
2377 backlog parameter.
2378
2379 On Linux 2.4, the parameter is ignored by the system. On Linux 2.6, it is
2380 used as a hint and the system accepts up to the smallest greater power of
2381 two, and never more than some limits (usually 32768).
2382
2383 See also : "maxconn" and the target operating system's tuning guide.
2384
2385
Willy Tarreau0ba27502007-12-24 16:55:16 +01002386balance <algorithm> [ <arguments> ]
Willy Tarreau226071e2014-04-10 11:55:45 +02002387balance url_param <param> [check_post]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002388 Define the load balancing algorithm to be used in a backend.
2389 May be used in sections : defaults | frontend | listen | backend
2390 yes | no | yes | yes
2391 Arguments :
2392 <algorithm> is the algorithm used to select a server when doing load
2393 balancing. This only applies when no persistence information
2394 is available, or when a connection is redispatched to another
2395 server. <algorithm> may be one of the following :
2396
2397 roundrobin Each server is used in turns, according to their weights.
2398 This is the smoothest and fairest algorithm when the server's
2399 processing time remains equally distributed. This algorithm
2400 is dynamic, which means that server weights may be adjusted
Willy Tarreau9757a382009-10-03 12:56:50 +02002401 on the fly for slow starts for instance. It is limited by
Godbacha34bdc02013-07-22 07:44:53 +08002402 design to 4095 active servers per backend. Note that in some
Willy Tarreau9757a382009-10-03 12:56:50 +02002403 large farms, when a server becomes up after having been down
2404 for a very short time, it may sometimes take a few hundreds
2405 requests for it to be re-integrated into the farm and start
2406 receiving traffic. This is normal, though very rare. It is
2407 indicated here in case you would have the chance to observe
2408 it, so that you don't worry.
2409
2410 static-rr Each server is used in turns, according to their weights.
2411 This algorithm is as similar to roundrobin except that it is
2412 static, which means that changing a server's weight on the
2413 fly will have no effect. On the other hand, it has no design
2414 limitation on the number of servers, and when a server goes
2415 up, it is always immediately reintroduced into the farm, once
2416 the full map is recomputed. It also uses slightly less CPU to
2417 run (around -1%).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002418
Willy Tarreau2d2a7f82008-03-17 12:07:56 +01002419 leastconn The server with the lowest number of connections receives the
2420 connection. Round-robin is performed within groups of servers
2421 of the same load to ensure that all servers will be used. Use
2422 of this algorithm is recommended where very long sessions are
2423 expected, such as LDAP, SQL, TSE, etc... but is not very well
2424 suited for protocols using short sessions such as HTTP. This
2425 algorithm is dynamic, which means that server weights may be
2426 adjusted on the fly for slow starts for instance.
2427
Willy Tarreauf09c6602012-02-13 17:12:08 +01002428 first The first server with available connection slots receives the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03002429 connection. The servers are chosen from the lowest numeric
Willy Tarreauf09c6602012-02-13 17:12:08 +01002430 identifier to the highest (see server parameter "id"), which
2431 defaults to the server's position in the farm. Once a server
Willy Tarreau64559c52012-04-07 09:08:45 +02002432 reaches its maxconn value, the next server is used. It does
Willy Tarreauf09c6602012-02-13 17:12:08 +01002433 not make sense to use this algorithm without setting maxconn.
2434 The purpose of this algorithm is to always use the smallest
2435 number of servers so that extra servers can be powered off
2436 during non-intensive hours. This algorithm ignores the server
2437 weight, and brings more benefit to long session such as RDP
Willy Tarreau64559c52012-04-07 09:08:45 +02002438 or IMAP than HTTP, though it can be useful there too. In
2439 order to use this algorithm efficiently, it is recommended
2440 that a cloud controller regularly checks server usage to turn
2441 them off when unused, and regularly checks backend queue to
2442 turn new servers on when the queue inflates. Alternatively,
2443 using "http-check send-state" may inform servers on the load.
Willy Tarreauf09c6602012-02-13 17:12:08 +01002444
Willy Tarreau0ba27502007-12-24 16:55:16 +01002445 source The source IP address is hashed and divided by the total
2446 weight of the running servers to designate which server will
2447 receive the request. This ensures that the same client IP
2448 address will always reach the same server as long as no
2449 server goes down or up. If the hash result changes due to the
2450 number of running servers changing, many clients will be
2451 directed to a different server. This algorithm is generally
2452 used in TCP mode where no cookie may be inserted. It may also
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002453 be used on the Internet to provide a best-effort stickiness
Willy Tarreau0ba27502007-12-24 16:55:16 +01002454 to clients which refuse session cookies. This algorithm is
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002455 static by default, which means that changing a server's
2456 weight on the fly will have no effect, but this can be
2457 changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002458
Oskar Stolc8dc41842012-05-19 10:19:54 +01002459 uri This algorithm hashes either the left part of the URI (before
2460 the question mark) or the whole URI (if the "whole" parameter
2461 is present) and divides the hash value by the total weight of
2462 the running servers. The result designates which server will
2463 receive the request. This ensures that the same URI will
2464 always be directed to the same server as long as no server
2465 goes up or down. This is used with proxy caches and
2466 anti-virus proxies in order to maximize the cache hit rate.
2467 Note that this algorithm may only be used in an HTTP backend.
2468 This algorithm is static by default, which means that
2469 changing a server's weight on the fly will have no effect,
2470 but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002471
Oskar Stolc8dc41842012-05-19 10:19:54 +01002472 This algorithm supports two optional parameters "len" and
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002473 "depth", both followed by a positive integer number. These
2474 options may be helpful when it is needed to balance servers
2475 based on the beginning of the URI only. The "len" parameter
2476 indicates that the algorithm should only consider that many
2477 characters at the beginning of the URI to compute the hash.
2478 Note that having "len" set to 1 rarely makes sense since most
2479 URIs start with a leading "/".
2480
2481 The "depth" parameter indicates the maximum directory depth
2482 to be used to compute the hash. One level is counted for each
2483 slash in the request. If both parameters are specified, the
2484 evaluation stops when either is reached.
2485
Willy Tarreau0ba27502007-12-24 16:55:16 +01002486 url_param The URL parameter specified in argument will be looked up in
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002487 the query string of each HTTP GET request.
2488
2489 If the modifier "check_post" is used, then an HTTP POST
Cyril Bontédc4d9032012-04-08 21:57:39 +02002490 request entity will be searched for the parameter argument,
2491 when it is not found in a query string after a question mark
Willy Tarreau226071e2014-04-10 11:55:45 +02002492 ('?') in the URL. The message body will only start to be
2493 analyzed once either the advertised amount of data has been
2494 received or the request buffer is full. In the unlikely event
2495 that chunked encoding is used, only the first chunk is
Cyril Bontédc4d9032012-04-08 21:57:39 +02002496 scanned. Parameter values separated by a chunk boundary, may
Willy Tarreau226071e2014-04-10 11:55:45 +02002497 be randomly balanced if at all. This keyword used to support
2498 an optional <max_wait> parameter which is now ignored.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002499
2500 If the parameter is found followed by an equal sign ('=') and
2501 a value, then the value is hashed and divided by the total
2502 weight of the running servers. The result designates which
2503 server will receive the request.
2504
2505 This is used to track user identifiers in requests and ensure
2506 that a same user ID will always be sent to the same server as
2507 long as no server goes up or down. If no value is found or if
2508 the parameter is not found, then a round robin algorithm is
2509 applied. Note that this algorithm may only be used in an HTTP
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002510 backend. This algorithm is static by default, which means
2511 that changing a server's weight on the fly will have no
2512 effect, but this can be changed using "hash-type".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002513
Cyril Bontédc4d9032012-04-08 21:57:39 +02002514 hdr(<name>) The HTTP header <name> will be looked up in each HTTP
2515 request. Just as with the equivalent ACL 'hdr()' function,
2516 the header name in parenthesis is not case sensitive. If the
2517 header is absent or if it does not contain any value, the
2518 roundrobin algorithm is applied instead.
Benoitaffb4812009-03-25 13:02:10 +01002519
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002520 An optional 'use_domain_only' parameter is available, for
Benoitaffb4812009-03-25 13:02:10 +01002521 reducing the hash algorithm to the main domain part with some
2522 specific headers such as 'Host'. For instance, in the Host
2523 value "haproxy.1wt.eu", only "1wt" will be considered.
2524
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002525 This algorithm is static by default, which means that
2526 changing a server's weight on the fly will have no effect,
2527 but this can be changed using "hash-type".
2528
Willy Tarreau21c741a2019-01-14 18:14:27 +01002529 random
2530 random(<draws>)
2531 A random number will be used as the key for the consistent
Willy Tarreau760e81d2018-05-03 07:20:40 +02002532 hashing function. This means that the servers' weights are
2533 respected, dynamic weight changes immediately take effect, as
2534 well as new server additions. Random load balancing can be
2535 useful with large farms or when servers are frequently added
Willy Tarreau21c741a2019-01-14 18:14:27 +01002536 or removed as it may avoid the hammering effect that could
2537 result from roundrobin or leastconn in this situation. The
2538 hash-balance-factor directive can be used to further improve
2539 fairness of the load balancing, especially in situations
2540 where servers show highly variable response times. When an
2541 argument <draws> is present, it must be an integer value one
2542 or greater, indicating the number of draws before selecting
2543 the least loaded of these servers. It was indeed demonstrated
2544 that picking the least loaded of two servers is enough to
2545 significantly improve the fairness of the algorithm, by
2546 always avoiding to pick the most loaded server within a farm
2547 and getting rid of any bias that could be induced by the
2548 unfair distribution of the consistent list. Higher values N
2549 will take away N-1 of the highest loaded servers at the
2550 expense of performance. With very high values, the algorithm
2551 will converge towards the leastconn's result but much slower.
2552 The default value is 2, which generally shows very good
2553 distribution and performance. This algorithm is also known as
2554 the Power of Two Random Choices and is described here :
2555 http://www.eecs.harvard.edu/~michaelm/postscripts/handbook2001.pdf
Willy Tarreau760e81d2018-05-03 07:20:40 +02002556
Emeric Brun736aa232009-06-30 17:56:00 +02002557 rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02002558 rdp-cookie(<name>)
Emeric Brun736aa232009-06-30 17:56:00 +02002559 The RDP cookie <name> (or "mstshash" if omitted) will be
2560 looked up and hashed for each incoming TCP request. Just as
2561 with the equivalent ACL 'req_rdp_cookie()' function, the name
2562 is not case-sensitive. This mechanism is useful as a degraded
2563 persistence mode, as it makes it possible to always send the
2564 same user (or the same session ID) to the same server. If the
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002565 cookie is not found, the normal roundrobin algorithm is
Emeric Brun736aa232009-06-30 17:56:00 +02002566 used instead.
2567
2568 Note that for this to work, the frontend must ensure that an
2569 RDP cookie is already present in the request buffer. For this
2570 you must use 'tcp-request content accept' rule combined with
2571 a 'req_rdp_cookie_cnt' ACL.
2572
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02002573 This algorithm is static by default, which means that
2574 changing a server's weight on the fly will have no effect,
2575 but this can be changed using "hash-type".
2576
Cyril Bontédc4d9032012-04-08 21:57:39 +02002577 See also the rdp_cookie pattern fetch function.
Simon Hormanab814e02011-06-24 14:50:20 +09002578
Willy Tarreau0ba27502007-12-24 16:55:16 +01002579 <arguments> is an optional list of arguments which may be needed by some
Marek Majkowski9c30fc12008-04-27 23:25:55 +02002580 algorithms. Right now, only "url_param" and "uri" support an
2581 optional argument.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002582
Willy Tarreau3cd9af22009-03-15 14:06:41 +01002583 The load balancing algorithm of a backend is set to roundrobin when no other
2584 algorithm, mode nor option have been set. The algorithm may only be set once
2585 for each backend.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002586
Lukas Tribus80512b12018-10-27 20:07:40 +02002587 With authentication schemes that require the same connection like NTLM, URI
2588 based alghoritms must not be used, as they would cause subsequent requests
2589 to be routed to different backend servers, breaking the invalid assumptions
2590 NTLM relies on.
2591
Willy Tarreau0ba27502007-12-24 16:55:16 +01002592 Examples :
2593 balance roundrobin
2594 balance url_param userid
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002595 balance url_param session_id check_post 64
Benoitaffb4812009-03-25 13:02:10 +01002596 balance hdr(User-Agent)
2597 balance hdr(host)
2598 balance hdr(Host) use_domain_only
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002599
2600 Note: the following caveats and limitations on using the "check_post"
2601 extension with "url_param" must be considered :
2602
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002603 - all POST requests are eligible for consideration, because there is no way
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002604 to determine if the parameters will be found in the body or entity which
2605 may contain binary data. Therefore another method may be required to
2606 restrict consideration of POST requests that have no URL parameters in
2607 the body. (see acl reqideny http_end)
2608
2609 - using a <max_wait> value larger than the request buffer size does not
2610 make sense and is useless. The buffer size is set at build time, and
2611 defaults to 16 kB.
2612
2613 - Content-Encoding is not supported, the parameter search will probably
2614 fail; and load balancing will fall back to Round Robin.
2615
2616 - Expect: 100-continue is not supported, load balancing will fall back to
2617 Round Robin.
2618
Lukas Tribus23953682017-04-28 13:24:30 +00002619 - Transfer-Encoding (RFC7230 3.3.1) is only supported in the first chunk.
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002620 If the entire parameter value is not present in the first chunk, the
2621 selection of server is undefined (actually, defined by how little
2622 actually appeared in the first chunk).
2623
2624 - This feature does not support generation of a 100, 411 or 501 response.
2625
2626 - In some cases, requesting "check_post" MAY attempt to scan the entire
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002627 contents of a message body. Scanning normally terminates when linear
matt.farnsworth@nokia.com1c2ab962008-04-14 20:47:37 +02002628 white space or control characters are found, indicating the end of what
2629 might be a URL parameter list. This is probably not a concern with SGML
2630 type message bodies.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002631
Willy Tarreau294d0f02015-08-10 19:40:12 +02002632 See also : "dispatch", "cookie", "transparent", "hash-type" and "http_proxy".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002633
2634
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002635bind [<address>]:<port_range> [, ...] [param*]
2636bind /<path> [, ...] [param*]
Willy Tarreau0ba27502007-12-24 16:55:16 +01002637 Define one or several listening addresses and/or ports in a frontend.
2638 May be used in sections : defaults | frontend | listen | backend
2639 no | yes | yes | no
2640 Arguments :
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002641 <address> is optional and can be a host name, an IPv4 address, an IPv6
2642 address, or '*'. It designates the address the frontend will
2643 listen on. If unset, all IPv4 addresses of the system will be
2644 listened on. The same will apply for '*' or the system's
David du Colombier9c938da2011-03-17 10:40:27 +01002645 special address "0.0.0.0". The IPv6 equivalent is '::'.
Willy Tarreau24709282013-03-10 21:32:12 +01002646 Optionally, an address family prefix may be used before the
2647 address to force the family regardless of the address format,
2648 which can be useful to specify a path to a unix socket with
2649 no slash ('/'). Currently supported prefixes are :
2650 - 'ipv4@' -> address is always IPv4
2651 - 'ipv6@' -> address is always IPv6
2652 - 'unix@' -> address is a path to a local unix socket
Willy Tarreau70f72e02014-07-08 00:37:50 +02002653 - 'abns@' -> address is in abstract namespace (Linux only).
2654 Note: since abstract sockets are not "rebindable", they
2655 do not cope well with multi-process mode during
2656 soft-restart, so it is better to avoid them if
2657 nbproc is greater than 1. The effect is that if the
2658 new process fails to start, only one of the old ones
2659 will be able to rebind to the socket.
Willy Tarreau40aa0702013-03-10 23:51:38 +01002660 - 'fd@<n>' -> use file descriptor <n> inherited from the
2661 parent. The fd must be bound and may or may not already
2662 be listening.
William Lallemand2fe7dd02018-09-11 16:51:29 +02002663 - 'sockpair@<n>'-> like fd@ but you must use the fd of a
2664 connected unix socket or of a socketpair. The bind waits
2665 to receive a FD over the unix socket and uses it as if it
2666 was the FD of an accept(). Should be used carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02002667 You may want to reference some environment variables in the
2668 address parameter, see section 2.3 about environment
2669 variables.
Willy Tarreaub1e52e82008-01-13 14:49:51 +01002670
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002671 <port_range> is either a unique TCP port, or a port range for which the
2672 proxy will accept connections for the IP address specified
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002673 above. The port is mandatory for TCP listeners. Note that in
2674 the case of an IPv6 address, the port is always the number
2675 after the last colon (':'). A range can either be :
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002676 - a numerical port (ex: '80')
2677 - a dash-delimited ports range explicitly stating the lower
2678 and upper bounds (ex: '2000-2100') which are included in
2679 the range.
2680
2681 Particular care must be taken against port ranges, because
2682 every <address:port> couple consumes one socket (= a file
2683 descriptor), so it's easy to consume lots of descriptors
2684 with a simple range, and to run out of sockets. Also, each
2685 <address:port> couple must be used only once among all
2686 instances running on a same system. Please note that binding
2687 to ports lower than 1024 generally require particular
Jamie Gloudon801a0a32012-08-25 00:18:33 -04002688 privileges to start the program, which are independent of
Willy Tarreauc5011ca2010-03-22 11:53:56 +01002689 the 'uid' parameter.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002690
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002691 <path> is a UNIX socket path beginning with a slash ('/'). This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01002692 alternative to the TCP listening port. HAProxy will then
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002693 receive UNIX connections on the socket located at this place.
2694 The path must begin with a slash and by default is absolute.
2695 It can be relative to the prefix defined by "unix-bind" in
2696 the global section. Note that the total length of the prefix
2697 followed by the socket path cannot exceed some system limits
2698 for UNIX sockets, which commonly are set to 107 characters.
2699
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002700 <param*> is a list of parameters common to all sockets declared on the
2701 same line. These numerous parameters depend on OS and build
2702 options and have a complete section dedicated to them. Please
2703 refer to section 5 to for more details.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002704
Willy Tarreau0ba27502007-12-24 16:55:16 +01002705 It is possible to specify a list of address:port combinations delimited by
2706 commas. The frontend will then listen on all of these addresses. There is no
2707 fixed limit to the number of addresses and ports which can be listened on in
2708 a frontend, as well as there is no limit to the number of "bind" statements
2709 in a frontend.
2710
2711 Example :
2712 listen http_proxy
2713 bind :80,:443
2714 bind 10.0.0.1:10080,10.0.0.1:10443
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002715 bind /var/run/ssl-frontend.sock user root mode 600 accept-proxy
Willy Tarreau0ba27502007-12-24 16:55:16 +01002716
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002717 listen http_https_proxy
2718 bind :80
Cyril Bonté0d44fc62012-10-09 22:45:33 +02002719 bind :443 ssl crt /etc/haproxy/site.pem
Willy Tarreaua0ee1d02012-09-10 09:01:23 +02002720
Willy Tarreau24709282013-03-10 21:32:12 +01002721 listen http_https_proxy_explicit
2722 bind ipv6@:80
2723 bind ipv4@public_ssl:443 ssl crt /etc/haproxy/site.pem
2724 bind unix@ssl-frontend.sock user root mode 600 accept-proxy
2725
Willy Tarreaudad36a32013-03-11 01:20:04 +01002726 listen external_bind_app1
William Lallemandb2f07452015-05-12 14:27:13 +02002727 bind "fd@${FD_APP1}"
Willy Tarreaudad36a32013-03-11 01:20:04 +01002728
Willy Tarreau55dcaf62015-09-27 15:03:15 +02002729 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
2730 sun_path length is used for the address length. Some other programs
2731 such as socat use the string length only by default. Pass the option
2732 ",unix-tightsocklen=0" to any abstract socket definition in socat to
2733 make it compatible with HAProxy's.
2734
Willy Tarreauceb24bc2010-11-09 12:46:41 +01002735 See also : "source", "option forwardfor", "unix-bind" and the PROXY protocol
Willy Tarreaub6205fd2012-09-24 12:27:33 +02002736 documentation, and section 5 about bind options.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002737
2738
Christopher Fauletff4121f2017-11-22 16:38:49 +01002739bind-process [ all | odd | even | <process_num>[-[<process_num>]] ] ...
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002740 Limit visibility of an instance to a certain set of processes numbers.
2741 May be used in sections : defaults | frontend | listen | backend
2742 yes | yes | yes | yes
2743 Arguments :
2744 all All process will see this instance. This is the default. It
2745 may be used to override a default value.
2746
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002747 odd This instance will be enabled on processes 1,3,5,...63. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002748 option may be combined with other numbers.
2749
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002750 even This instance will be enabled on processes 2,4,6,...64. This
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002751 option may be combined with other numbers. Do not use it
2752 with less than 2 processes otherwise some instances might be
2753 missing from all processes.
2754
Christopher Fauletff4121f2017-11-22 16:38:49 +01002755 process_num The instance will be enabled on this process number or range,
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002756 whose values must all be between 1 and 32 or 64 depending on
Christopher Fauletff4121f2017-11-22 16:38:49 +01002757 the machine's word size. Ranges can be partially defined. The
2758 higher bound can be omitted. In such case, it is replaced by
2759 the corresponding maximum value. If a proxy is bound to
2760 process numbers greater than the configured global.nbproc, it
2761 will either be forced to process #1 if a single process was
Willy Tarreau102df612014-05-07 23:56:38 +02002762 specified, or to all processes otherwise.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002763
2764 This keyword limits binding of certain instances to certain processes. This
2765 is useful in order not to have too many processes listening to the same
2766 ports. For instance, on a dual-core machine, it might make sense to set
2767 'nbproc 2' in the global section, then distributes the listeners among 'odd'
2768 and 'even' instances.
2769
Willy Tarreaua9db57e2013-01-18 11:29:29 +01002770 At the moment, it is not possible to reference more than 32 or 64 processes
2771 using this keyword, but this should be more than enough for most setups.
2772 Please note that 'all' really means all processes regardless of the machine's
2773 word size, and is not limited to the first 32 or 64.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002774
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002775 Each "bind" line may further be limited to a subset of the proxy's processes,
2776 please consult the "process" bind keyword in section 5.1.
2777
Willy Tarreaub369a042014-09-16 13:21:03 +02002778 When a frontend has no explicit "bind-process" line, it tries to bind to all
2779 the processes referenced by its "bind" lines. That means that frontends can
2780 easily adapt to their listeners' processes.
2781
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002782 If some backends are referenced by frontends bound to other processes, the
2783 backend automatically inherits the frontend's processes.
2784
2785 Example :
2786 listen app_ip1
2787 bind 10.0.0.1:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002788 bind-process odd
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002789
2790 listen app_ip2
2791 bind 10.0.0.2:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002792 bind-process even
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002793
2794 listen management
2795 bind 10.0.0.3:80
Willy Tarreaubfcd3112010-10-23 11:22:08 +02002796 bind-process 1 2 3 4
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002797
Willy Tarreau110ecc12012-11-15 17:50:01 +01002798 listen management
2799 bind 10.0.0.4:80
2800 bind-process 1-4
2801
Willy Tarreau6ae1ba62014-05-07 19:01:58 +02002802 See also : "nbproc" in global section, and "process" in section 5.1.
Willy Tarreau0b9c02c2009-02-04 22:05:05 +01002803
2804
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002805block { if | unless } <condition> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002806 Block a layer 7 request if/unless a condition is matched
2807 May be used in sections : defaults | frontend | listen | backend
2808 no | yes | yes | yes
2809
2810 The HTTP request will be blocked very early in the layer 7 processing
2811 if/unless <condition> is matched. A 403 error will be returned if the request
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002812 is blocked. The condition has to reference ACLs (see section 7). This is
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02002813 typically used to deny access to certain sensitive resources if some
Willy Tarreau0ba27502007-12-24 16:55:16 +01002814 conditions are met or not met. There is no fixed limit to the number of
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002815 "block" statements per instance. To block connections at layer 4 (without
2816 sending a 403 error) see "tcp-request connection reject" and
2817 "tcp-request content reject" rules.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002818
Jarno Huuskonen8c8c3492016-12-28 18:50:29 +02002819 This form is deprecated, do not use it in any new configuration, use the new
2820 "http-request deny" instead.
2821
Willy Tarreau0ba27502007-12-24 16:55:16 +01002822 Example:
2823 acl invalid_src src 0.0.0.0/7 224.0.0.0/3
2824 acl invalid_src src_port 0:1023
2825 acl local_dst hdr(host) -i localhost
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +03002826 # block is deprecated. Use http-request deny instead:
2827 #block if invalid_src || local_dst
2828 http-request deny if invalid_src || local_dst
Willy Tarreau0ba27502007-12-24 16:55:16 +01002829
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03002830 See also : section 7 about ACL usage, "http-request deny",
2831 "http-response deny", "tcp-request connection reject" and
2832 "tcp-request content reject".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002833
2834capture cookie <name> len <length>
2835 Capture and log a cookie in the request and in the response.
2836 May be used in sections : defaults | frontend | listen | backend
2837 no | yes | yes | no
2838 Arguments :
2839 <name> is the beginning of the name of the cookie to capture. In order
2840 to match the exact name, simply suffix the name with an equal
2841 sign ('='). The full name will appear in the logs, which is
2842 useful with application servers which adjust both the cookie name
Davor Ocelice9ed2812017-12-25 17:49:28 +01002843 and value (e.g. ASPSESSIONXXX).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002844
2845 <length> is the maximum number of characters to report in the logs, which
2846 include the cookie name, the equal sign and the value, all in the
2847 standard "name=value" form. The string will be truncated on the
2848 right if it exceeds <length>.
2849
2850 Only the first cookie is captured. Both the "cookie" request headers and the
2851 "set-cookie" response headers are monitored. This is particularly useful to
2852 check for application bugs causing session crossing or stealing between
2853 users, because generally the user's cookies can only change on a login page.
2854
2855 When the cookie was not presented by the client, the associated log column
2856 will report "-". When a request does not cause a cookie to be assigned by the
2857 server, a "-" is reported in the response column.
2858
2859 The capture is performed in the frontend only because it is necessary that
2860 the log format does not change for a given frontend depending on the
2861 backends. This may change in the future. Note that there can be only one
Willy Tarreau193b8c62012-11-22 00:17:38 +01002862 "capture cookie" statement in a frontend. The maximum capture length is set
2863 by the global "tune.http.cookielen" setting and defaults to 63 characters. It
2864 is not possible to specify a capture in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002865
2866 Example:
2867 capture cookie ASPSESSION len 32
2868
2869 See also : "capture request header", "capture response header" as well as
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002870 section 8 about logging.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002871
2872
2873capture request header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002874 Capture and log the last occurrence of the specified request header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002875 May be used in sections : defaults | frontend | listen | backend
2876 no | yes | yes | no
2877 Arguments :
2878 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002879 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002880 appear in the requests, with the first letter of each word in
2881 upper case. The header name will not appear in the logs, only the
2882 value is reported, but the position in the logs is respected.
2883
2884 <length> is the maximum number of characters to extract from the value and
2885 report in the logs. The string will be truncated on the right if
2886 it exceeds <length>.
2887
Willy Tarreau4460d032012-11-21 23:37:37 +01002888 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002889 value will be added to the logs between braces ('{}'). If multiple headers
2890 are captured, they will be delimited by a vertical bar ('|') and will appear
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002891 in the same order they were declared in the configuration. Non-existent
2892 headers will be logged just as an empty string. Common uses for request
2893 header captures include the "Host" field in virtual hosting environments, the
2894 "Content-length" when uploads are supported, "User-agent" to quickly
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01002895 differentiate between real users and robots, and "X-Forwarded-For" in proxied
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002896 environments to find where the request came from.
2897
2898 Note that when capturing headers such as "User-agent", some spaces may be
2899 logged, making the log analysis more difficult. Thus be careful about what
2900 you log if you know your log parser is not smart enough to rely on the
2901 braces.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002902
Willy Tarreau0900abb2012-11-22 00:21:46 +01002903 There is no limit to the number of captured request headers nor to their
2904 length, though it is wise to keep them low to limit memory usage per session.
2905 In order to keep log format consistent for a same frontend, header captures
2906 can only be declared in a frontend. It is not possible to specify a capture
2907 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002908
2909 Example:
2910 capture request header Host len 15
2911 capture request header X-Forwarded-For len 15
Cyril Bontéd1b0f7c2015-10-26 22:37:39 +01002912 capture request header Referer len 15
Willy Tarreau0ba27502007-12-24 16:55:16 +01002913
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002914 See also : "capture cookie", "capture response header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002915 about logging.
2916
2917
2918capture response header <name> len <length>
Willy Tarreau4460d032012-11-21 23:37:37 +01002919 Capture and log the last occurrence of the specified response header.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002920 May be used in sections : defaults | frontend | listen | backend
2921 no | yes | yes | no
2922 Arguments :
2923 <name> is the name of the header to capture. The header names are not
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002924 case-sensitive, but it is a common practice to write them as they
Willy Tarreau0ba27502007-12-24 16:55:16 +01002925 appear in the response, with the first letter of each word in
2926 upper case. The header name will not appear in the logs, only the
2927 value is reported, but the position in the logs is respected.
2928
2929 <length> is the maximum number of characters to extract from the value and
2930 report in the logs. The string will be truncated on the right if
2931 it exceeds <length>.
2932
Willy Tarreau4460d032012-11-21 23:37:37 +01002933 The complete value of the last occurrence of the header is captured. The
Willy Tarreau0ba27502007-12-24 16:55:16 +01002934 result will be added to the logs between braces ('{}') after the captured
2935 request headers. If multiple headers are captured, they will be delimited by
2936 a vertical bar ('|') and will appear in the same order they were declared in
Willy Tarreaucc6c8912009-02-22 10:53:55 +01002937 the configuration. Non-existent headers will be logged just as an empty
2938 string. Common uses for response header captures include the "Content-length"
2939 header which indicates how many bytes are expected to be returned, the
2940 "Location" header to track redirections.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002941
Willy Tarreau0900abb2012-11-22 00:21:46 +01002942 There is no limit to the number of captured response headers nor to their
2943 length, though it is wise to keep them low to limit memory usage per session.
2944 In order to keep log format consistent for a same frontend, header captures
2945 can only be declared in a frontend. It is not possible to specify a capture
2946 in a "defaults" section.
Willy Tarreau0ba27502007-12-24 16:55:16 +01002947
2948 Example:
2949 capture response header Content-length len 9
2950 capture response header Location len 15
2951
Willy Tarreauc57f0e22009-05-10 13:12:33 +02002952 See also : "capture cookie", "capture request header" as well as section 8
Willy Tarreau0ba27502007-12-24 16:55:16 +01002953 about logging.
2954
2955
Cyril Bontéf0c60612010-02-06 14:44:47 +01002956clitimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01002957 Set the maximum inactivity time on the client side.
2958 May be used in sections : defaults | frontend | listen | backend
2959 yes | yes | yes | no
2960 Arguments :
2961 <timeout> is the timeout value is specified in milliseconds by default, but
2962 can be in any other unit if the number is suffixed by the unit,
2963 as explained at the top of this document.
2964
2965 The inactivity timeout applies when the client is expected to acknowledge or
2966 send data. In HTTP mode, this timeout is particularly important to consider
2967 during the first phase, when the client sends the request, and during the
2968 response while it is reading data sent by the server. The value is specified
2969 in milliseconds by default, but can be in any other unit if the number is
2970 suffixed by the unit, as specified at the top of this document. In TCP mode
2971 (and to a lesser extent, in HTTP mode), it is highly recommended that the
2972 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01002973 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +01002974 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +01002975 (e.g. 4 or 5 seconds).
Willy Tarreau0ba27502007-12-24 16:55:16 +01002976
2977 This parameter is specific to frontends, but can be specified once for all in
2978 "defaults" sections. This is in fact one of the easiest solutions not to
2979 forget about it. An unspecified timeout results in an infinite timeout, which
2980 is not recommended. Such a usage is accepted and works but reports a warning
2981 during startup because it may results in accumulation of expired sessions in
2982 the system if the system's timeouts are not configured either.
2983
2984 This parameter is provided for compatibility but is currently deprecated.
2985 Please use "timeout client" instead.
2986
Willy Tarreau036fae02008-01-06 13:24:40 +01002987 See also : "timeout client", "timeout http-request", "timeout server", and
2988 "srvtimeout".
Willy Tarreau0ba27502007-12-24 16:55:16 +01002989
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002990compression algo <algorithm> ...
2991compression type <mime type> ...
Willy Tarreau70737d12012-10-27 00:34:28 +02002992compression offload
William Lallemand82fe75c2012-10-23 10:25:10 +02002993 Enable HTTP compression.
2994 May be used in sections : defaults | frontend | listen | backend
2995 yes | yes | yes | yes
2996 Arguments :
Cyril Bonté316a8cf2012-11-11 13:38:27 +01002997 algo is followed by the list of supported compression algorithms.
2998 type is followed by the list of MIME types that will be compressed.
2999 offload makes haproxy work as a compression offloader only (see notes).
3000
3001 The currently supported algorithms are :
Willy Tarreauc91840a2015-03-28 17:00:39 +01003002 identity this is mostly for debugging, and it was useful for developing
3003 the compression feature. Identity does not apply any change on
3004 data.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003005
Willy Tarreauc91840a2015-03-28 17:00:39 +01003006 gzip applies gzip compression. This setting is only available when
Baptiste Assmannf085d632015-12-21 17:57:32 +01003007 support for zlib or libslz was built in.
Willy Tarreauc91840a2015-03-28 17:00:39 +01003008
3009 deflate same as "gzip", but with deflate algorithm and zlib format.
3010 Note that this algorithm has ambiguous support on many
3011 browsers and no support at all from recent ones. It is
3012 strongly recommended not to use it for anything else than
3013 experimentation. This setting is only available when support
Baptiste Assmannf085d632015-12-21 17:57:32 +01003014 for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003015
Willy Tarreauc91840a2015-03-28 17:00:39 +01003016 raw-deflate same as "deflate" without the zlib wrapper, and used as an
3017 alternative when the browser wants "deflate". All major
3018 browsers understand it and despite violating the standards,
3019 it is known to work better than "deflate", at least on MSIE
3020 and some versions of Safari. Do not use it in conjunction
3021 with "deflate", use either one or the other since both react
3022 to the same Accept-Encoding token. This setting is only
Baptiste Assmannf085d632015-12-21 17:57:32 +01003023 available when support for zlib or libslz was built in.
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003024
Dmitry Sivachenko87c208b2012-11-22 20:03:26 +04003025 Compression will be activated depending on the Accept-Encoding request
Cyril Bonté316a8cf2012-11-11 13:38:27 +01003026 header. With identity, it does not take care of that header.
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003027 If backend servers support HTTP compression, these directives
3028 will be no-op: haproxy will see the compressed response and will not
3029 compress again. If backend servers do not support HTTP compression and
3030 there is Accept-Encoding header in request, haproxy will compress the
3031 matching response.
Willy Tarreau70737d12012-10-27 00:34:28 +02003032
3033 The "offload" setting makes haproxy remove the Accept-Encoding header to
3034 prevent backend servers from compressing responses. It is strongly
3035 recommended not to do this because this means that all the compression work
3036 will be done on the single point where haproxy is located. However in some
3037 deployment scenarios, haproxy may be installed in front of a buggy gateway
Dmitry Sivachenkoc9f3b452012-11-28 17:47:11 +04003038 with broken HTTP compression implementation which can't be turned off.
3039 In that case haproxy can be used to prevent that gateway from emitting
3040 invalid payloads. In this case, simply removing the header in the
3041 configuration does not work because it applies before the header is parsed,
3042 so that prevents haproxy from compressing. The "offload" setting should
Willy Tarreauffea9fd2014-07-12 16:37:02 +02003043 then be used for such scenarios. Note: for now, the "offload" setting is
3044 ignored when set in a defaults section.
William Lallemand82fe75c2012-10-23 10:25:10 +02003045
William Lallemand05097442012-11-20 12:14:28 +01003046 Compression is disabled when:
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003047 * the request does not advertise a supported compression algorithm in the
3048 "Accept-Encoding" header
3049 * the response message is not HTTP/1.1
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003050 * HTTP status code is not one of 200, 201, 202, or 203
Baptiste Assmann650d53d2013-01-05 15:44:44 +01003051 * response contain neither a "Content-Length" header nor a
3052 "Transfer-Encoding" whose last value is "chunked"
3053 * response contains a "Content-Type" header whose first value starts with
3054 "multipart"
3055 * the response contains the "no-transform" value in the "Cache-control"
3056 header
3057 * User-Agent matches "Mozilla/4" unless it is MSIE 6 with XP SP2, or MSIE 7
3058 and later
3059 * The response contains a "Content-Encoding" header, indicating that the
3060 response is already compressed (see compression offload)
Tim Duesterhusbb48c9a2019-01-30 23:46:04 +01003061 * The response contains an invalid "ETag" header or multiple ETag headers
William Lallemand05097442012-11-20 12:14:28 +01003062
Tim Duesterhusb229f012019-01-29 16:38:56 +01003063 Note: The compression does not emit the Warning header.
William Lallemand05097442012-11-20 12:14:28 +01003064
William Lallemand82fe75c2012-10-23 10:25:10 +02003065 Examples :
3066 compression algo gzip
3067 compression type text/html text/plain
Willy Tarreau0ba27502007-12-24 16:55:16 +01003068
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003069
Cyril Bontéf0c60612010-02-06 14:44:47 +01003070contimeout <timeout> (deprecated)
Willy Tarreau0ba27502007-12-24 16:55:16 +01003071 Set the maximum time to wait for a connection attempt to a server to succeed.
3072 May be used in sections : defaults | frontend | listen | backend
3073 yes | no | yes | yes
3074 Arguments :
3075 <timeout> is the timeout value is specified in milliseconds by default, but
3076 can be in any other unit if the number is suffixed by the unit,
3077 as explained at the top of this document.
3078
3079 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003080 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +01003081 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +01003082 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003083 connect timeout also presets the queue timeout to the same value if this one
3084 has not been specified. Historically, the contimeout was also used to set the
3085 tarpit timeout in a listen section, which is not possible in a pure frontend.
3086
3087 This parameter is specific to backends, but can be specified once for all in
3088 "defaults" sections. This is in fact one of the easiest solutions not to
3089 forget about it. An unspecified timeout results in an infinite timeout, which
3090 is not recommended. Such a usage is accepted and works but reports a warning
3091 during startup because it may results in accumulation of failed sessions in
3092 the system if the system's timeouts are not configured either.
3093
3094 This parameter is provided for backwards compatibility but is currently
3095 deprecated. Please use "timeout connect", "timeout queue" or "timeout tarpit"
3096 instead.
3097
3098 See also : "timeout connect", "timeout queue", "timeout tarpit",
3099 "timeout server", "contimeout".
3100
3101
Willy Tarreau55165fe2009-05-10 12:02:55 +02003102cookie <name> [ rewrite | insert | prefix ] [ indirect ] [ nocache ]
Willy Tarreau4992dd22012-05-31 21:02:17 +02003103 [ postonly ] [ preserve ] [ httponly ] [ secure ]
3104 [ domain <domain> ]* [ maxidle <idle> ] [ maxlife <life> ]
Olivier Houchard4e694042017-03-14 20:01:29 +01003105 [ dynamic ]
Willy Tarreau0ba27502007-12-24 16:55:16 +01003106 Enable cookie-based persistence in a backend.
3107 May be used in sections : defaults | frontend | listen | backend
3108 yes | no | yes | yes
3109 Arguments :
3110 <name> is the name of the cookie which will be monitored, modified or
3111 inserted in order to bring persistence. This cookie is sent to
3112 the client via a "Set-Cookie" header in the response, and is
3113 brought back by the client in a "Cookie" header in all requests.
3114 Special care should be taken to choose a name which does not
3115 conflict with any likely application cookie. Also, if the same
Davor Ocelice9ed2812017-12-25 17:49:28 +01003116 backends are subject to be used by the same clients (e.g.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003117 HTTP/HTTPS), care should be taken to use different cookie names
3118 between all backends if persistence between them is not desired.
3119
3120 rewrite This keyword indicates that the cookie will be provided by the
3121 server and that haproxy will have to modify its value to set the
3122 server's identifier in it. This mode is handy when the management
3123 of complex combinations of "Set-cookie" and "Cache-control"
3124 headers is left to the application. The application can then
3125 decide whether or not it is appropriate to emit a persistence
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003126 cookie. Since all responses should be monitored, this mode
3127 doesn't work in HTTP tunnel mode. Unless the application
Davor Ocelice9ed2812017-12-25 17:49:28 +01003128 behavior is very complex and/or broken, it is advised not to
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003129 start with this mode for new deployments. This keyword is
3130 incompatible with "insert" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003131
3132 insert This keyword indicates that the persistence cookie will have to
Willy Tarreaua79094d2010-08-31 22:54:15 +02003133 be inserted by haproxy in server responses if the client did not
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003134
Willy Tarreaua79094d2010-08-31 22:54:15 +02003135 already have a cookie that would have permitted it to access this
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003136 server. When used without the "preserve" option, if the server
3137 emits a cookie with the same name, it will be remove before
Davor Ocelice9ed2812017-12-25 17:49:28 +01003138 processing. For this reason, this mode can be used to upgrade
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003139 existing configurations running in the "rewrite" mode. The cookie
3140 will only be a session cookie and will not be stored on the
3141 client's disk. By default, unless the "indirect" option is added,
3142 the server will see the cookies emitted by the client. Due to
3143 caching effects, it is generally wise to add the "nocache" or
3144 "postonly" keywords (see below). The "insert" keyword is not
3145 compatible with "rewrite" and "prefix".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003146
3147 prefix This keyword indicates that instead of relying on a dedicated
3148 cookie for the persistence, an existing one will be completed.
3149 This may be needed in some specific environments where the client
3150 does not support more than one single cookie and the application
3151 already needs it. In this case, whenever the server sets a cookie
3152 named <name>, it will be prefixed with the server's identifier
3153 and a delimiter. The prefix will be removed from all client
3154 requests so that the server still finds the cookie it emitted.
3155 Since all requests and responses are subject to being modified,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01003156 this mode doesn't work with tunnel mode. The "prefix" keyword is
Willy Tarreau37229df2011-10-17 12:24:55 +02003157 not compatible with "rewrite" and "insert". Note: it is highly
3158 recommended not to use "indirect" with "prefix", otherwise server
3159 cookie updates would not be sent to clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003160
Willy Tarreaua79094d2010-08-31 22:54:15 +02003161 indirect When this option is specified, no cookie will be emitted to a
3162 client which already has a valid one for the server which has
3163 processed the request. If the server sets such a cookie itself,
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003164 it will be removed, unless the "preserve" option is also set. In
3165 "insert" mode, this will additionally remove cookies from the
3166 requests transmitted to the server, making the persistence
3167 mechanism totally transparent from an application point of view.
Willy Tarreau37229df2011-10-17 12:24:55 +02003168 Note: it is highly recommended not to use "indirect" with
3169 "prefix", otherwise server cookie updates would not be sent to
3170 clients.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003171
3172 nocache This option is recommended in conjunction with the insert mode
3173 when there is a cache between the client and HAProxy, as it
3174 ensures that a cacheable response will be tagged non-cacheable if
3175 a cookie needs to be inserted. This is important because if all
3176 persistence cookies are added on a cacheable home page for
3177 instance, then all customers will then fetch the page from an
3178 outer cache and will all share the same persistence cookie,
3179 leading to one server receiving much more traffic than others.
3180 See also the "insert" and "postonly" options.
3181
3182 postonly This option ensures that cookie insertion will only be performed
3183 on responses to POST requests. It is an alternative to the
3184 "nocache" option, because POST responses are not cacheable, so
3185 this ensures that the persistence cookie will never get cached.
3186 Since most sites do not need any sort of persistence before the
3187 first POST which generally is a login request, this is a very
3188 efficient method to optimize caching without risking to find a
3189 persistence cookie in the cache.
3190 See also the "insert" and "nocache" options.
3191
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003192 preserve This option may only be used with "insert" and/or "indirect". It
3193 allows the server to emit the persistence cookie itself. In this
3194 case, if a cookie is found in the response, haproxy will leave it
3195 untouched. This is useful in order to end persistence after a
3196 logout request for instance. For this, the server just has to
Davor Ocelice9ed2812017-12-25 17:49:28 +01003197 emit a cookie with an invalid value (e.g. empty) or with a date in
Willy Tarreauba4c5be2010-10-23 12:46:42 +02003198 the past. By combining this mechanism with the "disable-on-404"
3199 check option, it is possible to perform a completely graceful
3200 shutdown because users will definitely leave the server after
3201 they logout.
3202
Willy Tarreau4992dd22012-05-31 21:02:17 +02003203 httponly This option tells haproxy to add an "HttpOnly" cookie attribute
3204 when a cookie is inserted. This attribute is used so that a
3205 user agent doesn't share the cookie with non-HTTP components.
3206 Please check RFC6265 for more information on this attribute.
3207
3208 secure This option tells haproxy to add a "Secure" cookie attribute when
3209 a cookie is inserted. This attribute is used so that a user agent
3210 never emits this cookie over non-secure channels, which means
3211 that a cookie learned with this flag will be presented only over
3212 SSL/TLS connections. Please check RFC6265 for more information on
3213 this attribute.
3214
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003215 domain This option allows to specify the domain at which a cookie is
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003216 inserted. It requires exactly one parameter: a valid domain
Willy Tarreau68a897b2009-12-03 23:28:34 +01003217 name. If the domain begins with a dot, the browser is allowed to
3218 use it for any host ending with that name. It is also possible to
3219 specify several domain names by invoking this option multiple
3220 times. Some browsers might have small limits on the number of
3221 domains, so be careful when doing that. For the record, sending
3222 10 domains to MSIE 6 or Firefox 2 works as expected.
Krzysztof Piotr Oledzkiefe3b6f2008-05-23 23:49:32 +02003223
Willy Tarreau996a92c2010-10-13 19:30:47 +02003224 maxidle This option allows inserted cookies to be ignored after some idle
3225 time. It only works with insert-mode cookies. When a cookie is
3226 sent to the client, the date this cookie was emitted is sent too.
3227 Upon further presentations of this cookie, if the date is older
3228 than the delay indicated by the parameter (in seconds), it will
3229 be ignored. Otherwise, it will be refreshed if needed when the
3230 response is sent to the client. This is particularly useful to
3231 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003232 too long on the same server (e.g. after a farm size change). When
Willy Tarreau996a92c2010-10-13 19:30:47 +02003233 this option is set and a cookie has no date, it is always
3234 accepted, but gets refreshed in the response. This maintains the
3235 ability for admins to access their sites. Cookies that have a
3236 date in the future further than 24 hours are ignored. Doing so
3237 lets admins fix timezone issues without risking kicking users off
3238 the site.
3239
3240 maxlife This option allows inserted cookies to be ignored after some life
3241 time, whether they're in use or not. It only works with insert
3242 mode cookies. When a cookie is first sent to the client, the date
3243 this cookie was emitted is sent too. Upon further presentations
3244 of this cookie, if the date is older than the delay indicated by
3245 the parameter (in seconds), it will be ignored. If the cookie in
3246 the request has no date, it is accepted and a date will be set.
3247 Cookies that have a date in the future further than 24 hours are
3248 ignored. Doing so lets admins fix timezone issues without risking
3249 kicking users off the site. Contrary to maxidle, this value is
3250 not refreshed, only the first visit date counts. Both maxidle and
3251 maxlife may be used at the time. This is particularly useful to
3252 prevent users who never close their browsers from remaining for
Davor Ocelice9ed2812017-12-25 17:49:28 +01003253 too long on the same server (e.g. after a farm size change). This
Willy Tarreau996a92c2010-10-13 19:30:47 +02003254 is stronger than the maxidle method in that it forces a
3255 redispatch after some absolute delay.
3256
Olivier Houchard4e694042017-03-14 20:01:29 +01003257 dynamic Activate dynamic cookies. When used, a session cookie is
3258 dynamically created for each server, based on the IP and port
3259 of the server, and a secret key, specified in the
3260 "dynamic-cookie-key" backend directive.
3261 The cookie will be regenerated each time the IP address change,
3262 and is only generated for IPv4/IPv6.
3263
Willy Tarreau0ba27502007-12-24 16:55:16 +01003264 There can be only one persistence cookie per HTTP backend, and it can be
3265 declared in a defaults section. The value of the cookie will be the value
3266 indicated after the "cookie" keyword in a "server" statement. If no cookie
3267 is declared for a given server, the cookie is not set.
Willy Tarreau6a06a402007-07-15 20:15:28 +02003268
Willy Tarreau0ba27502007-12-24 16:55:16 +01003269 Examples :
3270 cookie JSESSIONID prefix
3271 cookie SRV insert indirect nocache
3272 cookie SRV insert postonly indirect
Willy Tarreau996a92c2010-10-13 19:30:47 +02003273 cookie SRV insert indirect nocache maxidle 30m maxlife 8h
Willy Tarreau0ba27502007-12-24 16:55:16 +01003274
Willy Tarreau294d0f02015-08-10 19:40:12 +02003275 See also : "balance source", "capture cookie", "server" and "ignore-persist".
Willy Tarreau0ba27502007-12-24 16:55:16 +01003276
Willy Tarreau983e01e2010-01-11 18:42:06 +01003277
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003278declare capture [ request | response ] len <length>
3279 Declares a capture slot.
3280 May be used in sections : defaults | frontend | listen | backend
3281 no | yes | yes | no
3282 Arguments:
3283 <length> is the length allowed for the capture.
3284
3285 This declaration is only available in the frontend or listen section, but the
3286 reserved slot can be used in the backends. The "request" keyword allocates a
3287 capture slot for use in the request, and "response" allocates a capture slot
3288 for use in the response.
3289
3290 See also: "capture-req", "capture-res" (sample converters),
Baptiste Assmann5ac425c2015-10-21 23:13:46 +02003291 "capture.req.hdr", "capture.res.hdr" (sample fetches),
Thierry FOURNIERa0a1b752015-05-26 17:44:32 +02003292 "http-request capture" and "http-response capture".
3293
3294
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003295default-server [param*]
3296 Change default options for a server in a backend
3297 May be used in sections : defaults | frontend | listen | backend
3298 yes | no | yes | yes
3299 Arguments:
Willy Tarreau983e01e2010-01-11 18:42:06 +01003300 <param*> is a list of parameters for this server. The "default-server"
3301 keyword accepts an important number of options and has a complete
3302 section dedicated to it. Please refer to section 5 for more
3303 details.
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003304
Willy Tarreau983e01e2010-01-11 18:42:06 +01003305 Example :
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +01003306 default-server inter 1000 weight 13
3307
3308 See also: "server" and section 5 about server options
Willy Tarreau0ba27502007-12-24 16:55:16 +01003309
Willy Tarreau983e01e2010-01-11 18:42:06 +01003310
Willy Tarreau0ba27502007-12-24 16:55:16 +01003311default_backend <backend>
3312 Specify the backend to use when no "use_backend" rule has been matched.
3313 May be used in sections : defaults | frontend | listen | backend
3314 yes | yes | yes | no
3315 Arguments :
3316 <backend> is the name of the backend to use.
3317
3318 When doing content-switching between frontend and backends using the
3319 "use_backend" keyword, it is often useful to indicate which backend will be
3320 used when no rule has matched. It generally is the dynamic backend which
3321 will catch all undetermined requests.
3322
Willy Tarreau0ba27502007-12-24 16:55:16 +01003323 Example :
3324
3325 use_backend dynamic if url_dyn
3326 use_backend static if url_css url_img extension_img
3327 default_backend dynamic
3328
Willy Tarreau98d04852015-05-26 12:18:29 +02003329 See also : "use_backend"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003330
Willy Tarreau0ba27502007-12-24 16:55:16 +01003331
Baptiste Assmann27f51342013-10-09 06:51:49 +02003332description <string>
3333 Describe a listen, frontend or backend.
3334 May be used in sections : defaults | frontend | listen | backend
3335 no | yes | yes | yes
3336 Arguments : string
3337
3338 Allows to add a sentence to describe the related object in the HAProxy HTML
3339 stats page. The description will be printed on the right of the object name
3340 it describes.
3341 No need to backslash spaces in the <string> arguments.
3342
3343
Willy Tarreau0ba27502007-12-24 16:55:16 +01003344disabled
3345 Disable a proxy, frontend or backend.
3346 May be used in sections : defaults | frontend | listen | backend
3347 yes | yes | yes | yes
3348 Arguments : none
3349
3350 The "disabled" keyword is used to disable an instance, mainly in order to
3351 liberate a listening port or to temporarily disable a service. The instance
3352 will still be created and its configuration will be checked, but it will be
3353 created in the "stopped" state and will appear as such in the statistics. It
3354 will not receive any traffic nor will it send any health-checks or logs. It
3355 is possible to disable many instances at once by adding the "disabled"
3356 keyword in a "defaults" section.
3357
3358 See also : "enabled"
3359
3360
Willy Tarreau5ce94572010-06-07 14:35:41 +02003361dispatch <address>:<port>
3362 Set a default server address
3363 May be used in sections : defaults | frontend | listen | backend
3364 no | no | yes | yes
Cyril Bonté108cf6e2012-04-21 23:30:29 +02003365 Arguments :
Willy Tarreau5ce94572010-06-07 14:35:41 +02003366
3367 <address> is the IPv4 address of the default server. Alternatively, a
3368 resolvable hostname is supported, but this name will be resolved
3369 during start-up.
3370
3371 <ports> is a mandatory port specification. All connections will be sent
3372 to this port, and it is not permitted to use port offsets as is
3373 possible with normal servers.
3374
Willy Tarreau787aed52011-04-15 06:45:37 +02003375 The "dispatch" keyword designates a default server for use when no other
Willy Tarreau5ce94572010-06-07 14:35:41 +02003376 server can take the connection. In the past it was used to forward non
3377 persistent connections to an auxiliary load balancer. Due to its simple
3378 syntax, it has also been used for simple TCP relays. It is recommended not to
3379 use it for more clarity, and to use the "server" directive instead.
3380
3381 See also : "server"
3382
Olivier Houchard4e694042017-03-14 20:01:29 +01003383
3384dynamic-cookie-key <string>
3385 Set the dynamic cookie secret key for a backend.
3386 May be used in sections : defaults | frontend | listen | backend
3387 yes | no | yes | yes
3388 Arguments : The secret key to be used.
3389
3390 When dynamic cookies are enabled (see the "dynamic" directive for cookie),
Davor Ocelice9ed2812017-12-25 17:49:28 +01003391 a dynamic cookie is created for each server (unless one is explicitly
Olivier Houchard4e694042017-03-14 20:01:29 +01003392 specified on the "server" line), using a hash of the IP address of the
3393 server, the TCP port, and the secret key.
Davor Ocelice9ed2812017-12-25 17:49:28 +01003394 That way, we can ensure session persistence across multiple load-balancers,
Olivier Houchard4e694042017-03-14 20:01:29 +01003395 even if servers are dynamically added or removed.
Willy Tarreau5ce94572010-06-07 14:35:41 +02003396
Willy Tarreau0ba27502007-12-24 16:55:16 +01003397enabled
3398 Enable a proxy, frontend or backend.
3399 May be used in sections : defaults | frontend | listen | backend
3400 yes | yes | yes | yes
3401 Arguments : none
3402
3403 The "enabled" keyword is used to explicitly enable an instance, when the
3404 defaults has been set to "disabled". This is very rarely used.
3405
3406 See also : "disabled"
3407
3408
3409errorfile <code> <file>
3410 Return a file contents instead of errors generated by HAProxy
3411 May be used in sections : defaults | frontend | listen | backend
3412 yes | yes | yes | yes
3413 Arguments :
3414 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003415 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3416 503, and 504.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003417
3418 <file> designates a file containing the full HTTP response. It is
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003419 recommended to follow the common practice of appending ".http" to
Willy Tarreau0ba27502007-12-24 16:55:16 +01003420 the filename so that people do not confuse the response with HTML
Willy Tarreau59140a22009-02-22 12:02:30 +01003421 error pages, and to use absolute paths, since files are read
3422 before any chroot is performed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01003423
3424 It is important to understand that this keyword is not meant to rewrite
3425 errors returned by the server, but errors detected and returned by HAProxy.
3426 This is why the list of supported errors is limited to a small set.
3427
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003428 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3429
Willy Tarreau0ba27502007-12-24 16:55:16 +01003430 The files are returned verbatim on the TCP socket. This allows any trick such
3431 as redirections to another URL or site, as well as tricks to clean cookies,
3432 force enable or disable caching, etc... The package provides default error
3433 files returning the same contents as default errors.
3434
Willy Tarreau59140a22009-02-22 12:02:30 +01003435 The files should not exceed the configured buffer size (BUFSIZE), which
3436 generally is 8 or 16 kB, otherwise they will be truncated. It is also wise
Davor Ocelice9ed2812017-12-25 17:49:28 +01003437 not to put any reference to local contents (e.g. images) in order to avoid
Willy Tarreau59140a22009-02-22 12:02:30 +01003438 loops between the client and HAProxy when all servers are down, causing an
3439 error to be returned instead of an image. For better HTTP compliance, it is
3440 recommended that all header lines end with CR-LF and not LF alone.
3441
Willy Tarreau0ba27502007-12-24 16:55:16 +01003442 The files are read at the same time as the configuration and kept in memory.
3443 For this reason, the errors continue to be returned even when the process is
3444 chrooted, and no file change is considered while the process is running. A
Willy Tarreauc27debf2008-01-06 08:57:02 +01003445 simple method for developing those files consists in associating them to the
Willy Tarreau0ba27502007-12-24 16:55:16 +01003446 403 status code and interrogating a blocked URL.
3447
3448 See also : "errorloc", "errorloc302", "errorloc303"
3449
Willy Tarreau59140a22009-02-22 12:02:30 +01003450 Example :
3451 errorfile 400 /etc/haproxy/errorfiles/400badreq.http
Willy Tarreau989222a2016-01-15 10:26:26 +01003452 errorfile 408 /dev/null # work around Chrome pre-connect bug
Willy Tarreau59140a22009-02-22 12:02:30 +01003453 errorfile 403 /etc/haproxy/errorfiles/403forbid.http
3454 errorfile 503 /etc/haproxy/errorfiles/503sorry.http
3455
Willy Tarreau2769aa02007-12-27 18:26:09 +01003456
3457errorloc <code> <url>
3458errorloc302 <code> <url>
3459 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3460 May be used in sections : defaults | frontend | listen | backend
3461 yes | yes | yes | yes
3462 Arguments :
3463 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003464 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3465 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003466
3467 <url> it is the exact contents of the "Location" header. It may contain
3468 either a relative URI to an error page hosted on the same site,
3469 or an absolute URI designating an error page on another site.
3470 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003471 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003472
3473 It is important to understand that this keyword is not meant to rewrite
3474 errors returned by the server, but errors detected and returned by HAProxy.
3475 This is why the list of supported errors is limited to a small set.
3476
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003477 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3478
Willy Tarreau2769aa02007-12-27 18:26:09 +01003479 Note that both keyword return the HTTP 302 status code, which tells the
3480 client to fetch the designated URL using the same HTTP method. This can be
3481 quite problematic in case of non-GET methods such as POST, because the URL
3482 sent to the client might not be allowed for something other than GET. To
Willy Tarreau989222a2016-01-15 10:26:26 +01003483 work around this problem, please use "errorloc303" which send the HTTP 303
Willy Tarreau2769aa02007-12-27 18:26:09 +01003484 status code, indicating to the client that the URL must be fetched with a GET
3485 request.
3486
3487 See also : "errorfile", "errorloc303"
3488
3489
3490errorloc303 <code> <url>
3491 Return an HTTP redirection to a URL instead of errors generated by HAProxy
3492 May be used in sections : defaults | frontend | listen | backend
3493 yes | yes | yes | yes
3494 Arguments :
3495 <code> is the HTTP status code. Currently, HAProxy is capable of
Olivier Houchard51a76d82017-10-02 16:12:07 +02003496 generating codes 200, 400, 403, 405, 408, 425, 429, 500, 502,
3497 503, and 504.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003498
3499 <url> it is the exact contents of the "Location" header. It may contain
3500 either a relative URI to an error page hosted on the same site,
3501 or an absolute URI designating an error page on another site.
3502 Special care should be given to relative URIs to avoid redirect
Davor Ocelice9ed2812017-12-25 17:49:28 +01003503 loops if the URI itself may generate the same error (e.g. 500).
Willy Tarreau2769aa02007-12-27 18:26:09 +01003504
3505 It is important to understand that this keyword is not meant to rewrite
3506 errors returned by the server, but errors detected and returned by HAProxy.
3507 This is why the list of supported errors is limited to a small set.
3508
Willy Tarreauae94d4d2011-05-11 16:28:49 +02003509 Code 200 is emitted in response to requests matching a "monitor-uri" rule.
3510
Willy Tarreau2769aa02007-12-27 18:26:09 +01003511 Note that both keyword return the HTTP 303 status code, which tells the
3512 client to fetch the designated URL using the same HTTP GET method. This
3513 solves the usual problems associated with "errorloc" and the 302 code. It is
3514 possible that some very old browsers designed before HTTP/1.1 do not support
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01003515 it, but no such problem has been reported till now.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003516
3517 See also : "errorfile", "errorloc", "errorloc302"
3518
3519
Simon Horman51a1cf62015-02-03 13:00:44 +09003520email-alert from <emailaddr>
3521 Declare the from email address to be used in both the envelope and header
Davor Ocelice9ed2812017-12-25 17:49:28 +01003522 of email alerts. This is the address that email alerts are sent from.
Simon Horman51a1cf62015-02-03 13:00:44 +09003523 May be used in sections: defaults | frontend | listen | backend
3524 yes | yes | yes | yes
3525
3526 Arguments :
3527
3528 <emailaddr> is the from email address to use when sending email alerts
3529
3530 Also requires "email-alert mailers" and "email-alert to" to be set
3531 and if so sending email alerts is enabled for the proxy.
3532
Simon Horman64e34162015-02-06 11:11:57 +09003533 See also : "email-alert level", "email-alert mailers",
Cyril Bonté307ee1e2015-09-28 23:16:06 +02003534 "email-alert myhostname", "email-alert to", section 3.6 about
3535 mailers.
Simon Horman64e34162015-02-06 11:11:57 +09003536
3537
3538email-alert level <level>
3539 Declare the maximum log level of messages for which email alerts will be
3540 sent. This acts as a filter on the sending of email alerts.
3541 May be used in sections: defaults | frontend | listen | backend
3542 yes | yes | yes | yes
3543
3544 Arguments :
3545
3546 <level> One of the 8 syslog levels:
3547 emerg alert crit err warning notice info debug
3548 The above syslog levels are ordered from lowest to highest.
3549
3550 By default level is alert
3551
3552 Also requires "email-alert from", "email-alert mailers" and
3553 "email-alert to" to be set and if so sending email alerts is enabled
3554 for the proxy.
3555
Simon Horman1421e212015-04-30 13:10:35 +09003556 Alerts are sent when :
3557
3558 * An un-paused server is marked as down and <level> is alert or lower
3559 * A paused server is marked as down and <level> is notice or lower
3560 * A server is marked as up or enters the drain state and <level>
3561 is notice or lower
3562 * "option log-health-checks" is enabled, <level> is info or lower,
3563 and a health check status update occurs
3564
Simon Horman64e34162015-02-06 11:11:57 +09003565 See also : "email-alert from", "email-alert mailers",
3566 "email-alert myhostname", "email-alert to",
Simon Horman51a1cf62015-02-03 13:00:44 +09003567 section 3.6 about mailers.
3568
3569
3570email-alert mailers <mailersect>
3571 Declare the mailers to be used when sending email alerts
3572 May be used in sections: defaults | frontend | listen | backend
3573 yes | yes | yes | yes
3574
3575 Arguments :
3576
3577 <mailersect> is the name of the mailers section to send email alerts.
3578
3579 Also requires "email-alert from" and "email-alert to" to be set
3580 and if so sending email alerts is enabled for the proxy.
3581
Simon Horman64e34162015-02-06 11:11:57 +09003582 See also : "email-alert from", "email-alert level", "email-alert myhostname",
3583 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003584
3585
3586email-alert myhostname <hostname>
3587 Declare the to hostname address to be used when communicating with
3588 mailers.
3589 May be used in sections: defaults | frontend | listen | backend
3590 yes | yes | yes | yes
3591
3592 Arguments :
3593
Baptiste Assmann738bad92015-12-21 15:27:53 +01003594 <hostname> is the hostname to use when communicating with mailers
Simon Horman51a1cf62015-02-03 13:00:44 +09003595
3596 By default the systems hostname is used.
3597
3598 Also requires "email-alert from", "email-alert mailers" and
3599 "email-alert to" to be set and if so sending email alerts is enabled
3600 for the proxy.
3601
Simon Horman64e34162015-02-06 11:11:57 +09003602 See also : "email-alert from", "email-alert level", "email-alert mailers",
3603 "email-alert to", section 3.6 about mailers.
Simon Horman51a1cf62015-02-03 13:00:44 +09003604
3605
3606email-alert to <emailaddr>
Davor Ocelice9ed2812017-12-25 17:49:28 +01003607 Declare both the recipient address in the envelope and to address in the
Simon Horman51a1cf62015-02-03 13:00:44 +09003608 header of email alerts. This is the address that email alerts are sent to.
3609 May be used in sections: defaults | frontend | listen | backend
3610 yes | yes | yes | yes
3611
3612 Arguments :
3613
3614 <emailaddr> is the to email address to use when sending email alerts
3615
3616 Also requires "email-alert mailers" and "email-alert to" to be set
3617 and if so sending email alerts is enabled for the proxy.
3618
Simon Horman64e34162015-02-06 11:11:57 +09003619 See also : "email-alert from", "email-alert level", "email-alert mailers",
Simon Horman51a1cf62015-02-03 13:00:44 +09003620 "email-alert myhostname", section 3.6 about mailers.
3621
3622
Willy Tarreau4de91492010-01-22 19:10:05 +01003623force-persist { if | unless } <condition>
3624 Declare a condition to force persistence on down servers
3625 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01003626 no | no | yes | yes
Willy Tarreau4de91492010-01-22 19:10:05 +01003627
3628 By default, requests are not dispatched to down servers. It is possible to
3629 force this using "option persist", but it is unconditional and redispatches
3630 to a valid server if "option redispatch" is set. That leaves with very little
3631 possibilities to force some requests to reach a server which is artificially
3632 marked down for maintenance operations.
3633
3634 The "force-persist" statement allows one to declare various ACL-based
3635 conditions which, when met, will cause a request to ignore the down status of
3636 a server and still try to connect to it. That makes it possible to start a
3637 server, still replying an error to the health checks, and run a specially
3638 configured browser to test the service. Among the handy methods, one could
3639 use a specific source IP address, or a specific cookie. The cookie also has
3640 the advantage that it can easily be added/removed on the browser from a test
3641 page. Once the service is validated, it is then possible to open the service
3642 to the world by returning a valid response to health checks.
3643
3644 The forced persistence is enabled when an "if" condition is met, or unless an
3645 "unless" condition is met. The final redispatch is always disabled when this
3646 is used.
3647
Cyril Bonté0d4bf012010-04-25 23:21:46 +02003648 See also : "option redispatch", "ignore-persist", "persist",
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +02003649 and section 7 about ACL usage.
Willy Tarreau4de91492010-01-22 19:10:05 +01003650
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003651
3652filter <name> [param*]
3653 Add the filter <name> in the filter list attached to the proxy.
3654 May be used in sections : defaults | frontend | listen | backend
3655 no | yes | yes | yes
3656 Arguments :
3657 <name> is the name of the filter. Officially supported filters are
3658 referenced in section 9.
3659
Tim Düsterhus4896c442016-11-29 02:15:19 +01003660 <param*> is a list of parameters accepted by the filter <name>. The
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003661 parsing of these parameters are the responsibility of the
Tim Düsterhus4896c442016-11-29 02:15:19 +01003662 filter. Please refer to the documentation of the corresponding
3663 filter (section 9) for all details on the supported parameters.
Christopher Fauletc3fe5332016-04-07 15:30:10 +02003664
3665 Multiple occurrences of the filter line can be used for the same proxy. The
3666 same filter can be referenced many times if needed.
3667
3668 Example:
3669 listen
3670 bind *:80
3671
3672 filter trace name BEFORE-HTTP-COMP
3673 filter compression
3674 filter trace name AFTER-HTTP-COMP
3675
3676 compression algo gzip
3677 compression offload
3678
3679 server srv1 192.168.0.1:80
3680
3681 See also : section 9.
3682
Willy Tarreau4de91492010-01-22 19:10:05 +01003683
Willy Tarreau2769aa02007-12-27 18:26:09 +01003684fullconn <conns>
3685 Specify at what backend load the servers will reach their maxconn
3686 May be used in sections : defaults | frontend | listen | backend
3687 yes | no | yes | yes
3688 Arguments :
3689 <conns> is the number of connections on the backend which will make the
3690 servers use the maximal number of connections.
3691
Willy Tarreau198a7442008-01-17 12:05:32 +01003692 When a server has a "maxconn" parameter specified, it means that its number
Willy Tarreau2769aa02007-12-27 18:26:09 +01003693 of concurrent connections will never go higher. Additionally, if it has a
Willy Tarreau198a7442008-01-17 12:05:32 +01003694 "minconn" parameter, it indicates a dynamic limit following the backend's
Willy Tarreau2769aa02007-12-27 18:26:09 +01003695 load. The server will then always accept at least <minconn> connections,
3696 never more than <maxconn>, and the limit will be on the ramp between both
3697 values when the backend has less than <conns> concurrent connections. This
3698 makes it possible to limit the load on the servers during normal loads, but
3699 push it further for important loads without overloading the servers during
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003700 exceptional loads.
Willy Tarreau2769aa02007-12-27 18:26:09 +01003701
Willy Tarreaufbb78422011-06-05 15:38:35 +02003702 Since it's hard to get this value right, haproxy automatically sets it to
3703 10% of the sum of the maxconns of all frontends that may branch to this
Bertrand Jacquin702d44f2013-11-19 11:43:06 +01003704 backend (based on "use_backend" and "default_backend" rules). That way it's
3705 safe to leave it unset. However, "use_backend" involving dynamic names are
3706 not counted since there is no way to know if they could match or not.
Willy Tarreaufbb78422011-06-05 15:38:35 +02003707
Willy Tarreau2769aa02007-12-27 18:26:09 +01003708 Example :
3709 # The servers will accept between 100 and 1000 concurrent connections each
3710 # and the maximum of 1000 will be reached when the backend reaches 10000
3711 # connections.
3712 backend dynamic
3713 fullconn 10000
3714 server srv1 dyn1:80 minconn 100 maxconn 1000
3715 server srv2 dyn2:80 minconn 100 maxconn 1000
3716
3717 See also : "maxconn", "server"
3718
3719
3720grace <time>
3721 Maintain a proxy operational for some time after a soft stop
3722 May be used in sections : defaults | frontend | listen | backend
Cyril Bonté99ed3272010-01-24 23:29:44 +01003723 yes | yes | yes | yes
Willy Tarreau2769aa02007-12-27 18:26:09 +01003724 Arguments :
3725 <time> is the time (by default in milliseconds) for which the instance
3726 will remain operational with the frontend sockets still listening
3727 when a soft-stop is received via the SIGUSR1 signal.
3728
3729 This may be used to ensure that the services disappear in a certain order.
3730 This was designed so that frontends which are dedicated to monitoring by an
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01003731 external equipment fail immediately while other ones remain up for the time
Willy Tarreau2769aa02007-12-27 18:26:09 +01003732 needed by the equipment to detect the failure.
3733
3734 Note that currently, there is very little benefit in using this parameter,
3735 and it may in fact complicate the soft-reconfiguration process more than
3736 simplify it.
3737
Willy Tarreau0ba27502007-12-24 16:55:16 +01003738
Andrew Rodland17be45e2016-10-25 17:04:12 -04003739hash-balance-factor <factor>
3740 Specify the balancing factor for bounded-load consistent hashing
3741 May be used in sections : defaults | frontend | listen | backend
3742 yes | no | no | yes
3743 Arguments :
3744 <factor> is the control for the maximum number of concurrent requests to
3745 send to a server, expressed as a percentage of the average number
Frédéric Lécaille93d33162019-03-06 09:35:59 +01003746 of concurrent requests across all of the active servers.
Andrew Rodland17be45e2016-10-25 17:04:12 -04003747
3748 Specifying a "hash-balance-factor" for a server with "hash-type consistent"
3749 enables an algorithm that prevents any one server from getting too many
3750 requests at once, even if some hash buckets receive many more requests than
3751 others. Setting <factor> to 0 (the default) disables the feature. Otherwise,
3752 <factor> is a percentage greater than 100. For example, if <factor> is 150,
3753 then no server will be allowed to have a load more than 1.5 times the average.
3754 If server weights are used, they will be respected.
3755
3756 If the first-choice server is disqualified, the algorithm will choose another
3757 server based on the request hash, until a server with additional capacity is
3758 found. A higher <factor> allows more imbalance between the servers, while a
3759 lower <factor> means that more servers will be checked on average, affecting
3760 performance. Reasonable values are from 125 to 200.
3761
Willy Tarreau760e81d2018-05-03 07:20:40 +02003762 This setting is also used by "balance random" which internally relies on the
3763 consistent hashing mechanism.
3764
Andrew Rodland17be45e2016-10-25 17:04:12 -04003765 See also : "balance" and "hash-type".
3766
3767
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003768hash-type <method> <function> <modifier>
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003769 Specify a method to use for mapping hashes to servers
3770 May be used in sections : defaults | frontend | listen | backend
3771 yes | no | yes | yes
3772 Arguments :
Bhaskar98634f02013-10-29 23:30:51 -04003773 <method> is the method used to select a server from the hash computed by
3774 the <function> :
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003775
Bhaskar98634f02013-10-29 23:30:51 -04003776 map-based the hash table is a static array containing all alive servers.
3777 The hashes will be very smooth, will consider weights, but
3778 will be static in that weight changes while a server is up
3779 will be ignored. This means that there will be no slow start.
3780 Also, since a server is selected by its position in the array,
3781 most mappings are changed when the server count changes. This
3782 means that when a server goes up or down, or when a server is
3783 added to a farm, most connections will be redistributed to
3784 different servers. This can be inconvenient with caches for
3785 instance.
Willy Tarreau798a39c2010-11-24 15:04:29 +01003786
Bhaskar98634f02013-10-29 23:30:51 -04003787 consistent the hash table is a tree filled with many occurrences of each
3788 server. The hash key is looked up in the tree and the closest
3789 server is chosen. This hash is dynamic, it supports changing
3790 weights while the servers are up, so it is compatible with the
3791 slow start feature. It has the advantage that when a server
3792 goes up or down, only its associations are moved. When a
3793 server is added to the farm, only a few part of the mappings
3794 are redistributed, making it an ideal method for caches.
3795 However, due to its principle, the distribution will never be
3796 very smooth and it may sometimes be necessary to adjust a
3797 server's weight or its ID to get a more balanced distribution.
3798 In order to get the same distribution on multiple load
3799 balancers, it is important that all servers have the exact
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003800 same IDs. Note: consistent hash uses sdbm and avalanche if no
3801 hash function is specified.
Bhaskar98634f02013-10-29 23:30:51 -04003802
3803 <function> is the hash function to be used :
3804
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03003805 sdbm this function was created initially for sdbm (a public-domain
Bhaskar98634f02013-10-29 23:30:51 -04003806 reimplementation of ndbm) database library. It was found to do
3807 well in scrambling bits, causing better distribution of the keys
3808 and fewer splits. It also happens to be a good general hashing
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003809 function with good distribution, unless the total server weight
3810 is a multiple of 64, in which case applying the avalanche
3811 modifier may help.
Bhaskar98634f02013-10-29 23:30:51 -04003812
3813 djb2 this function was first proposed by Dan Bernstein many years ago
3814 on comp.lang.c. Studies have shown that for certain workload this
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003815 function provides a better distribution than sdbm. It generally
3816 works well with text-based inputs though it can perform extremely
3817 poorly with numeric-only input or when the total server weight is
3818 a multiple of 33, unless the avalanche modifier is also used.
3819
Willy Tarreaua0f42712013-11-14 14:30:35 +01003820 wt6 this function was designed for haproxy while testing other
3821 functions in the past. It is not as smooth as the other ones, but
3822 is much less sensible to the input data set or to the number of
3823 servers. It can make sense as an alternative to sdbm+avalanche or
3824 djb2+avalanche for consistent hashing or when hashing on numeric
3825 data such as a source IP address or a visitor identifier in a URL
3826 parameter.
3827
Willy Tarreau324f07f2015-01-20 19:44:50 +01003828 crc32 this is the most common CRC32 implementation as used in Ethernet,
3829 gzip, PNG, etc. It is slower than the other ones but may provide
3830 a better distribution or less predictable results especially when
3831 used on strings.
3832
Bhaskar Maddalab6c0ac92013-11-05 11:54:02 -05003833 <modifier> indicates an optional method applied after hashing the key :
3834
3835 avalanche This directive indicates that the result from the hash
3836 function above should not be used in its raw form but that
3837 a 4-byte full avalanche hash must be applied first. The
3838 purpose of this step is to mix the resulting bits from the
3839 previous hash in order to avoid any undesired effect when
3840 the input contains some limited values or when the number of
3841 servers is a multiple of one of the hash's components (64
3842 for SDBM, 33 for DJB2). Enabling avalanche tends to make the
3843 result less predictable, but it's also not as smooth as when
3844 using the original function. Some testing might be needed
3845 with some workloads. This hash is one of the many proposed
3846 by Bob Jenkins.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003847
Bhaskar98634f02013-10-29 23:30:51 -04003848 The default hash type is "map-based" and is recommended for most usages. The
3849 default function is "sdbm", the selection of a function should be based on
3850 the range of the values being hashed.
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003851
Andrew Rodland17be45e2016-10-25 17:04:12 -04003852 See also : "balance", "hash-balance-factor", "server"
Willy Tarreau6b2e11b2009-10-01 07:52:15 +02003853
3854
Willy Tarreau0ba27502007-12-24 16:55:16 +01003855http-check disable-on-404
3856 Enable a maintenance mode upon HTTP/404 response to health-checks
3857 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau2769aa02007-12-27 18:26:09 +01003858 yes | no | yes | yes
Willy Tarreau0ba27502007-12-24 16:55:16 +01003859 Arguments : none
3860
3861 When this option is set, a server which returns an HTTP code 404 will be
3862 excluded from further load-balancing, but will still receive persistent
3863 connections. This provides a very convenient method for Web administrators
3864 to perform a graceful shutdown of their servers. It is also important to note
3865 that a server which is detected as failed while it was in this mode will not
3866 generate an alert, just a notice. If the server responds 2xx or 3xx again, it
3867 will immediately be reinserted into the farm. The status on the stats page
3868 reports "NOLB" for a server in this mode. It is important to note that this
Willy Tarreaubd741542010-03-16 18:46:54 +01003869 option only works in conjunction with the "httpchk" option. If this option
3870 is used with "http-check expect", then it has precedence over it so that 404
3871 responses will still be considered as soft-stop.
3872
3873 See also : "option httpchk", "http-check expect"
3874
3875
3876http-check expect [!] <match> <pattern>
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003877 Make HTTP health checks consider response contents or specific status codes
Willy Tarreaubd741542010-03-16 18:46:54 +01003878 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau1ee51a62011-08-19 20:04:17 +02003879 yes | no | yes | yes
Willy Tarreaubd741542010-03-16 18:46:54 +01003880 Arguments :
3881 <match> is a keyword indicating how to look for a specific pattern in the
3882 response. The keyword may be one of "status", "rstatus",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003883 "string", or "rstring". The keyword may be preceded by an
Willy Tarreaubd741542010-03-16 18:46:54 +01003884 exclamation mark ("!") to negate the match. Spaces are allowed
3885 between the exclamation mark and the keyword. See below for more
3886 details on the supported keywords.
3887
3888 <pattern> is the pattern to look for. It may be a string or a regular
3889 expression. If the pattern contains spaces, they must be escaped
3890 with the usual backslash ('\').
3891
3892 By default, "option httpchk" considers that response statuses 2xx and 3xx
3893 are valid, and that others are invalid. When "http-check expect" is used,
3894 it defines what is considered valid or invalid. Only one "http-check"
3895 statement is supported in a backend. If a server fails to respond or times
3896 out, the check obviously fails. The available matches are :
3897
3898 status <string> : test the exact string match for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003899 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003900 response's status code is exactly this string. If the
3901 "status" keyword is prefixed with "!", then the response
3902 will be considered invalid if the status code matches.
3903
3904 rstatus <regex> : test a regular expression for the HTTP status code.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003905 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003906 response's status code matches the expression. If the
3907 "rstatus" keyword is prefixed with "!", then the response
3908 will be considered invalid if the status code matches.
3909 This is mostly used to check for multiple codes.
3910
3911 string <string> : test the exact string match in the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003912 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003913 response's body contains this exact string. If the
3914 "string" keyword is prefixed with "!", then the response
3915 will be considered invalid if the body contains this
3916 string. This can be used to look for a mandatory word at
3917 the end of a dynamic page, or to detect a failure when a
Davor Ocelice9ed2812017-12-25 17:49:28 +01003918 specific error appears on the check page (e.g. a stack
Willy Tarreaubd741542010-03-16 18:46:54 +01003919 trace).
3920
3921 rstring <regex> : test a regular expression on the HTTP response body.
Jamie Gloudonaaa21002012-08-25 00:18:33 -04003922 A health check response will be considered valid if the
Willy Tarreaubd741542010-03-16 18:46:54 +01003923 response's body matches this expression. If the "rstring"
3924 keyword is prefixed with "!", then the response will be
3925 considered invalid if the body matches the expression.
3926 This can be used to look for a mandatory word at the end
3927 of a dynamic page, or to detect a failure when a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01003928 error appears on the check page (e.g. a stack trace).
Willy Tarreaubd741542010-03-16 18:46:54 +01003929
3930 It is important to note that the responses will be limited to a certain size
3931 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
3932 Thus, too large responses may not contain the mandatory pattern when using
3933 "string" or "rstring". If a large response is absolutely required, it is
3934 possible to change the default max size by setting the global variable.
3935 However, it is worth keeping in mind that parsing very large responses can
3936 waste some CPU cycles, especially when regular expressions are used, and that
3937 it is always better to focus the checks on smaller resources.
3938
Cyril Bonté32602d22015-01-30 00:07:07 +01003939 Also "http-check expect" doesn't support HTTP keep-alive. Keep in mind that it
3940 will automatically append a "Connection: close" header, meaning that this
3941 header should not be present in the request provided by "option httpchk".
3942
Willy Tarreaubd741542010-03-16 18:46:54 +01003943 Last, if "http-check expect" is combined with "http-check disable-on-404",
3944 then this last one has precedence when the server responds with 404.
3945
3946 Examples :
3947 # only accept status 200 as valid
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003948 http-check expect status 200
Willy Tarreaubd741542010-03-16 18:46:54 +01003949
3950 # consider SQL errors as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003951 http-check expect ! string SQL\ Error
Willy Tarreaubd741542010-03-16 18:46:54 +01003952
3953 # consider status 5xx only as errors
Willy Tarreau8f2a1e72011-01-06 16:36:10 +01003954 http-check expect ! rstatus ^5
Willy Tarreaubd741542010-03-16 18:46:54 +01003955
3956 # check that we have a correct hexadecimal tag before /html
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03003957 http-check expect rstring <!--tag:[0-9a-f]*--></html>
Willy Tarreau0ba27502007-12-24 16:55:16 +01003958
Willy Tarreaubd741542010-03-16 18:46:54 +01003959 See also : "option httpchk", "http-check disable-on-404"
Willy Tarreau2769aa02007-12-27 18:26:09 +01003960
3961
Willy Tarreauef781042010-01-27 11:53:01 +01003962http-check send-state
3963 Enable emission of a state header with HTTP health checks
3964 May be used in sections : defaults | frontend | listen | backend
3965 yes | no | yes | yes
3966 Arguments : none
3967
3968 When this option is set, haproxy will systematically send a special header
3969 "X-Haproxy-Server-State" with a list of parameters indicating to each server
3970 how they are seen by haproxy. This can be used for instance when a server is
3971 manipulated without access to haproxy and the operator needs to know whether
3972 haproxy still sees it up or not, or if the server is the last one in a farm.
3973
3974 The header is composed of fields delimited by semi-colons, the first of which
3975 is a word ("UP", "DOWN", "NOLB"), possibly followed by a number of valid
3976 checks on the total number before transition, just as appears in the stats
3977 interface. Next headers are in the form "<variable>=<value>", indicating in
3978 no specific order some values available in the stats interface :
Joseph Lynch514061c2015-01-15 17:52:59 -08003979 - a variable "address", containing the address of the backend server.
3980 This corresponds to the <address> field in the server declaration. For
3981 unix domain sockets, it will read "unix".
3982
3983 - a variable "port", containing the port of the backend server. This
3984 corresponds to the <port> field in the server declaration. For unix
3985 domain sockets, it will read "unix".
3986
Willy Tarreauef781042010-01-27 11:53:01 +01003987 - a variable "name", containing the name of the backend followed by a slash
3988 ("/") then the name of the server. This can be used when a server is
3989 checked in multiple backends.
3990
3991 - a variable "node" containing the name of the haproxy node, as set in the
3992 global "node" variable, otherwise the system's hostname if unspecified.
3993
3994 - a variable "weight" indicating the weight of the server, a slash ("/")
3995 and the total weight of the farm (just counting usable servers). This
3996 helps to know if other servers are available to handle the load when this
3997 one fails.
3998
3999 - a variable "scur" indicating the current number of concurrent connections
4000 on the server, followed by a slash ("/") then the total number of
4001 connections on all servers of the same backend.
4002
4003 - a variable "qcur" indicating the current number of requests in the
4004 server's queue.
4005
4006 Example of a header received by the application server :
4007 >>> X-Haproxy-Server-State: UP 2/3; name=bck/srv2; node=lb1; weight=1/2; \
4008 scur=13/22; qcur=0
4009
4010 See also : "option httpchk", "http-check disable-on-404"
4011
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004012
4013http-request <action> [options...] [ { if | unless } <condition> ]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004014 Access control for Layer 7 requests
4015
4016 May be used in sections: defaults | frontend | listen | backend
4017 no | yes | yes | yes
4018
Willy Tarreau20b0de52012-12-24 15:45:22 +01004019 The http-request statement defines a set of rules which apply to layer 7
4020 processing. The rules are evaluated in their declaration order when they are
4021 met in a frontend, listen or backend section. Any rule may optionally be
4022 followed by an ACL-based condition, in which case it will only be evaluated
4023 if the condition is true.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004024
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004025 The first keyword is the rule's action. The supported actions are described
4026 below.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004027
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004028 There is no limit to the number of http-request statements per instance.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004029
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004030 It is important to know that http-request rules are processed very early in
4031 the HTTP processing, just after "block" rules and before "reqdel" or "reqrep"
4032 or "reqadd" rules. That way, headers added by "add-header"/"set-header" are
4033 visible by almost all further ACL rules.
Willy Tarreau53275e82017-11-24 07:52:01 +01004034
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004035 Using "reqadd"/"reqdel"/"reqrep" to manipulate request headers is discouraged
4036 in newer versions (>= 1.5). But if you need to use regular expression to
4037 delete headers, you can still use "reqdel". Also please use
4038 "http-request deny/allow/tarpit" instead of "reqdeny"/"reqpass"/"reqtarpit".
Willy Tarreauccbcc372012-12-27 12:37:57 +01004039
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004040 Example:
4041 acl nagios src 192.168.129.3
4042 acl local_net src 192.168.0.0/16
4043 acl auth_ok http_auth(L1)
Willy Tarreau20b0de52012-12-24 15:45:22 +01004044
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004045 http-request allow if nagios
4046 http-request allow if local_net auth_ok
4047 http-request auth realm Gimme if local_net auth_ok
4048 http-request deny
Willy Tarreau81499eb2012-12-27 12:19:02 +01004049
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004050 Example:
4051 acl key req.hdr(X-Add-Acl-Key) -m found
4052 acl add path /addacl
4053 acl del path /delacl
Willy Tarreau20b0de52012-12-24 15:45:22 +01004054
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004055 acl myhost hdr(Host) -f myhost.lst
Willy Tarreau20b0de52012-12-24 15:45:22 +01004056
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004057 http-request add-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key add
4058 http-request del-acl(myhost.lst) %[req.hdr(X-Add-Acl-Key)] if key del
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004059
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004060 Example:
4061 acl value req.hdr(X-Value) -m found
4062 acl setmap path /setmap
4063 acl delmap path /delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004064
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004065 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004066
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004067 http-request set-map(map.lst) %[src] %[req.hdr(X-Value)] if setmap value
4068 http-request del-map(map.lst) %[src] if delmap
Sasha Pachev218f0642014-06-16 12:05:59 -06004069
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004070 See also : "stats http-request", section 3.4 about userlists and section 7
4071 about ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004072
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004073http-request add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004074
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004075 This is used to add a new entry into an ACL. The ACL must be loaded from a
4076 file (even a dummy empty file). The file name of the ACL to be updated is
4077 passed between parentheses. It takes one argument: <key fmt>, which follows
4078 log-format rules, to collect content of the new entry. It performs a lookup
4079 in the ACL before insertion, to avoid duplicated (or more) values. This
4080 lookup is done by a linear search and can be expensive with large lists!
4081 It is the equivalent of the "add acl" command from the stats socket, but can
4082 be triggered by an HTTP request.
Sasha Pachev218f0642014-06-16 12:05:59 -06004083
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004084http-request add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004085
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004086 This appends an HTTP header field whose name is specified in <name> and
4087 whose value is defined by <fmt> which follows the log-format rules (see
4088 Custom Log Format in section 8.2.4). This is particularly useful to pass
4089 connection-specific information to the server (e.g. the client's SSL
4090 certificate), or to combine several headers into one. This rule is not
4091 final, so it is possible to add other similar rules. Note that header
4092 addition is performed immediately, so one rule might reuse the resulting
4093 header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004094
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004095http-request allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004096
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004097 This stops the evaluation of the rules and lets the request pass the check.
4098 No further "http-request" rules are evaluated.
Sasha Pachev218f0642014-06-16 12:05:59 -06004099
Sasha Pachev218f0642014-06-16 12:05:59 -06004100
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004101http-request auth [realm <realm>] [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004102
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004103 This stops the evaluation of the rules and immediately responds with an
4104 HTTP 401 or 407 error code to invite the user to present a valid user name
4105 and password. No further "http-request" rules are evaluated. An optional
4106 "realm" parameter is supported, it sets the authentication realm that is
4107 returned with the response (typically the application's name).
Sasha Pachev218f0642014-06-16 12:05:59 -06004108
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004109 Example:
4110 acl auth_ok http_auth_group(L1) G1
4111 http-request auth unless auth_ok
Sasha Pachev218f0642014-06-16 12:05:59 -06004112
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004113http-request cache-use <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004114
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004115 See section 10.2 about cache setup.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004116
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004117http-request capture <sample> [ len <length> | id <id> ]
4118 [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004119
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004120 This captures sample expression <sample> from the request buffer, and
4121 converts it to a string of at most <len> characters. The resulting string is
4122 stored into the next request "capture" slot, so it will possibly appear next
4123 to some captured HTTP headers. It will then automatically appear in the logs,
4124 and it will be possible to extract it using sample fetch rules to feed it
4125 into headers or anything. The length should be limited given that this size
4126 will be allocated for each capture during the whole session life.
4127 Please check section 7.3 (Fetching samples) and "capture request header" for
4128 more information.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004129
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004130 If the keyword "id" is used instead of "len", the action tries to store the
4131 captured string in a previously declared capture slot. This is useful to run
4132 captures in backends. The slot id can be declared by a previous directive
4133 "http-request capture" or with the "declare capture" keyword. If the slot
4134 <id> doesn't exist, then HAProxy fails parsing the configuration to prevent
4135 unexpected behavior at run time.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004136
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004137http-request del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004138
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004139 This is used to delete an entry from an ACL. The ACL must be loaded from a
4140 file (even a dummy empty file). The file name of the ACL to be updated is
4141 passed between parentheses. It takes one argument: <key fmt>, which follows
4142 log-format rules, to collect content of the entry to delete.
4143 It is the equivalent of the "del acl" command from the stats socket, but can
4144 be triggered by an HTTP request.
Willy Tarreaua0dc23f2015-01-22 20:46:11 +01004145
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004146http-request del-header <name> [ { if | unless } <condition> ]
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004147
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004148 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004149
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004150http-request del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004151
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004152 This is used to delete an entry from a MAP. The MAP must be loaded from a
4153 file (even a dummy empty file). The file name of the MAP to be updated is
4154 passed between parentheses. It takes one argument: <key fmt>, which follows
4155 log-format rules, to collect content of the entry to delete.
4156 It takes one argument: "file name" It is the equivalent of the "del map"
4157 command from the stats socket, but can be triggered by an HTTP request.
Willy Tarreau51347ed2013-06-11 19:34:13 +02004158
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004159http-request deny [deny_status <status>] [ { if | unless } <condition> ]
Patrick Hemmer268a7072018-05-11 12:52:31 -04004160
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004161 This stops the evaluation of the rules and immediately rejects the request
4162 and emits an HTTP 403 error, or optionally the status code specified as an
4163 argument to "deny_status". The list of permitted status codes is limited to
4164 those that can be overridden by the "errorfile" directive.
4165 No further "http-request" rules are evaluated.
Patrick Hemmer268a7072018-05-11 12:52:31 -04004166
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004167http-request early-hint <name> <fmt> [ { if | unless } <condition> ]
4168
4169 This is used to build an HTTP 103 Early Hints response prior to any other one.
4170 This appends an HTTP header field to this response whose name is specified in
4171 <name> and whose value is defined by <fmt> which follows the log-format rules
4172 (see Custom Log Format in section 8.2.4). This is particularly useful to pass
Frédéric Lécaille3aac1062018-11-13 09:42:13 +01004173 to the client some Link headers to preload resources required to render the
4174 HTML documents.
Frédéric Lécaille06f5b642018-11-12 11:01:10 +01004175
4176 See RFC 8297 for more information.
4177
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004178http-request redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004179
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004180 This performs an HTTP redirection based on a redirect rule. This is exactly
4181 the same as the "redirect" statement except that it inserts a redirect rule
4182 which can be processed in the middle of other "http-request" rules and that
4183 these rules use the "log-format" strings. See the "redirect" keyword for the
4184 rule's syntax.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004185
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004186http-request reject [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004187
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004188 This stops the evaluation of the rules and immediately closes the connection
4189 without sending any response. It acts similarly to the
4190 "tcp-request content reject" rules. It can be useful to force an immediate
4191 connection closure on HTTP/2 connections.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004192
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004193http-request replace-header <name> <match-regex> <replace-fmt>
4194 [ { if | unless } <condition> ]
Willy Tarreaua9083d02015-05-08 15:27:59 +02004195
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004196 This matches the regular expression in all occurrences of header field
4197 <name> according to <match-regex>, and replaces them with the
4198 <replace-fmt> argument. Format characters are allowed in replace-fmt and
4199 work like in <fmt> arguments in "http-request add-header". The match is
4200 only case-sensitive. It is important to understand that this action only
4201 considers whole header lines, regardless of the number of values they may
4202 contain. This usage is suited to headers naturally containing commas in
4203 their value, such as If-Modified-Since and so on.
Thierry FOURNIER82bf70d2015-05-26 17:58:29 +02004204
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004205 Example:
4206 http-request replace-header Cookie foo=([^;]*);(.*) foo=\1;ip=%bi;\2
William Lallemand86d0df02017-11-24 21:36:45 +01004207
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004208 # applied to:
4209 Cookie: foo=foobar; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004210
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004211 # outputs:
4212 Cookie: foo=foobar;ip=192.168.1.20; expires=Tue, 14-Jun-2016 01:40:45 GMT;
Willy Tarreau09448f72014-06-25 18:12:15 +02004213
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004214 # assuming the backend IP is 192.168.1.20
Willy Tarreau09448f72014-06-25 18:12:15 +02004215
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004216http-request replace-value <name> <match-regex> <replace-fmt>
4217 [ { if | unless } <condition> ]
Willy Tarreau09448f72014-06-25 18:12:15 +02004218
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004219 This works like "replace-header" except that it matches the regex against
4220 every comma-delimited value of the header field <name> instead of the
4221 entire header. This is suited for all headers which are allowed to carry
4222 more than one value. An example could be the Accept header.
Willy Tarreau09448f72014-06-25 18:12:15 +02004223
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004224 Example:
4225 http-request replace-value X-Forwarded-For ^192\.168\.(.*)$ 172.16.\1
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004226
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004227 # applied to:
4228 X-Forwarded-For: 192.168.10.1, 192.168.13.24, 10.0.0.37
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004229
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004230 # outputs:
4231 X-Forwarded-For: 172.16.10.1, 172.16.13.24, 10.0.0.37
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004232
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004233http-request sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4234http-request sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004235
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004236 This actions increments the GPC0 or GPC1 counter according with the sticky
4237 counter designated by <sc-id>. If an error occurs, this action silently fails
4238 and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004239
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004240http-request sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004241
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004242 This action sets the GPT0 tag according to the sticky counter designated by
4243 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4244 occurs, this action silently fails and the actions evaluation continues.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004245
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004246http-request set-dst <expr> [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004247
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004248 This is used to set the destination IP address to the value of specified
4249 expression. Useful when a proxy in front of HAProxy rewrites destination IP,
4250 but provides the correct IP in a HTTP header; or you want to mask the IP for
4251 privacy. If you want to connect to the new address/port, use '0.0.0.0:0' as a
4252 server address in the backend.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004253
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004254 Arguments:
4255 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4256 by some converters.
Christopher Faulet85d79c92016-11-09 16:54:56 +01004257
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004258 Example:
4259 http-request set-dst hdr(x-dst)
4260 http-request set-dst dst,ipmask(24)
Christopher Faulet85d79c92016-11-09 16:54:56 +01004261
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004262 When possible, set-dst preserves the original destination port as long as the
4263 address family allows it, otherwise the destination port is set to 0.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004264
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004265http-request set-dst-port <expr> [ { if | unless } <condition> ]
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004266
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004267 This is used to set the destination port address to the value of specified
4268 expression. If you want to connect to the new address/port, use '0.0.0.0:0'
4269 as a server address in the backend.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004270
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004271 Arguments:
4272 <expr> Is a standard HAProxy expression formed by a sample-fetch
4273 followed by some converters.
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004274
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004275 Example:
4276 http-request set-dst-port hdr(x-port)
4277 http-request set-dst-port int(4000)
Adis Nezirovic2fbcafc2015-07-06 15:44:30 +02004278
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004279 When possible, set-dst-port preserves the original destination address as
4280 long as the address family supports a port, otherwise it forces the
4281 destination address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02004282
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004283http-request set-header <name> <fmt> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004284
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004285 This does the same as "http-request add-header" except that the header name
4286 is first removed if it existed. This is useful when passing security
4287 information to the server, where the header must not be manipulated by
4288 external users. Note that the new value is computed before the removal so it
4289 is possible to concatenate a value to an existing header.
William Lallemand44be6402016-05-25 01:51:35 +02004290
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004291 Example:
4292 http-request set-header X-Haproxy-Current-Date %T
4293 http-request set-header X-SSL %[ssl_fc]
4294 http-request set-header X-SSL-Session_ID %[ssl_fc_session_id,hex]
4295 http-request set-header X-SSL-Client-Verify %[ssl_c_verify]
4296 http-request set-header X-SSL-Client-DN %{+Q}[ssl_c_s_dn]
4297 http-request set-header X-SSL-Client-CN %{+Q}[ssl_c_s_dn(cn)]
4298 http-request set-header X-SSL-Issuer %{+Q}[ssl_c_i_dn]
4299 http-request set-header X-SSL-Client-NotBefore %{+Q}[ssl_c_notbefore]
4300 http-request set-header X-SSL-Client-NotAfter %{+Q}[ssl_c_notafter]
William Lallemand44be6402016-05-25 01:51:35 +02004301
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004302http-request set-log-level <level> [ { if | unless } <condition> ]
William Lallemand44be6402016-05-25 01:51:35 +02004303
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004304 This is used to change the log level of the current request when a certain
4305 condition is met. Valid levels are the 8 syslog levels (see the "log"
4306 keyword) plus the special level "silent" which disables logging for this
4307 request. This rule is not final so the last matching rule wins. This rule
4308 can be useful to disable health checks coming from another equipment.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004309
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004310http-request set-map(<file-name>) <key fmt> <value fmt>
4311 [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004312
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004313 This is used to add a new entry into a MAP. The MAP must be loaded from a
4314 file (even a dummy empty file). The file name of the MAP to be updated is
4315 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4316 log-format rules, used to collect MAP key, and <value fmt>, which follows
4317 log-format rules, used to collect content for the new entry.
4318 It performs a lookup in the MAP before insertion, to avoid duplicated (or
4319 more) values. This lookup is done by a linear search and can be expensive
4320 with large lists! It is the equivalent of the "set map" command from the
4321 stats socket, but can be triggered by an HTTP request.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004322
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004323http-request set-mark <mark> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004324
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004325 This is used to set the Netfilter MARK on all packets sent to the client to
4326 the value passed in <mark> on platforms which support it. This value is an
4327 unsigned 32 bit value which can be matched by netfilter and by the routing
4328 table. It can be expressed both in decimal or hexadecimal format (prefixed by
4329 "0x"). This can be useful to force certain packets to take a different route
4330 (for example a cheaper network path for bulk downloads). This works on Linux
4331 kernels 2.6.32 and above and requires admin privileges.
Willy Tarreau00005ce2016-10-21 15:07:45 +02004332
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004333http-request set-method <fmt> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004334
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004335 This rewrites the request method with the result of the evaluation of format
4336 string <fmt>. There should be very few valid reasons for having to do so as
4337 this is more likely to break something than to fix it.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004338
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004339http-request set-nice <nice> [ { if | unless } <condition> ]
William Lallemand13e9b0c2016-05-25 02:34:07 +02004340
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004341 This sets the "nice" factor of the current request being processed. It only
4342 has effect against the other requests being processed at the same time.
4343 The default value is 0, unless altered by the "nice" setting on the "bind"
4344 line. The accepted range is -1024..1024. The higher the value, the nicest
4345 the request will be. Lower values will make the request more important than
4346 other ones. This can be useful to improve the speed of some requests, or
4347 lower the priority of non-important requests. Using this setting without
4348 prior experimentation can cause some major slowdown.
William Lallemand13e9b0c2016-05-25 02:34:07 +02004349
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004350http-request set-path <fmt> [ { if | unless } <condition> ]
Willy Tarreau00005ce2016-10-21 15:07:45 +02004351
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004352 This rewrites the request path with the result of the evaluation of format
4353 string <fmt>. The query string, if any, is left intact. If a scheme and
4354 authority is found before the path, they are left intact as well. If the
4355 request doesn't have a path ("*"), this one is replaced with the format.
4356 This can be used to prepend a directory component in front of a path for
4357 example. See also "http-request set-query" and "http-request set-uri".
Willy Tarreau2d392c22015-08-24 01:43:45 +02004358
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004359 Example :
4360 # prepend the host name before the path
4361 http-request set-path /%[hdr(host)]%[path]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004362
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004363http-request set-priority-class <expr> [ { if | unless } <condition> ]
Olivier Houchardccaa7de2017-10-02 11:51:03 +02004364
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004365 This is used to set the queue priority class of the current request.
4366 The value must be a sample expression which converts to an integer in the
4367 range -2047..2047. Results outside this range will be truncated.
4368 The priority class determines the order in which queued requests are
4369 processed. Lower values have higher priority.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004370
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004371http-request set-priority-offset <expr> [ { if | unless } <condition> ]
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004372
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004373 This is used to set the queue priority timestamp offset of the current
4374 request. The value must be a sample expression which converts to an integer
4375 in the range -524287..524287. Results outside this range will be truncated.
4376 When a request is queued, it is ordered first by the priority class, then by
4377 the current timestamp adjusted by the given offset in milliseconds. Lower
4378 values have higher priority.
4379 Note that the resulting timestamp is is only tracked with enough precision
4380 for 524,287ms (8m44s287ms). If the request is queued long enough to where the
4381 adjusted timestamp exceeds this value, it will be misidentified as highest
4382 priority. Thus it is important to set "timeout queue" to a value, where when
4383 combined with the offset, does not exceed this limit.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004384
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004385http-request set-query <fmt> [ { if | unless } <condition> ]
Willy Tarreau20b0de52012-12-24 15:45:22 +01004386
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004387 This rewrites the request's query string which appears after the first
4388 question mark ("?") with the result of the evaluation of format string <fmt>.
4389 The part prior to the question mark is left intact. If the request doesn't
4390 contain a question mark and the new value is not empty, then one is added at
4391 the end of the URI, followed by the new value. If a question mark was
4392 present, it will never be removed even if the value is empty. This can be
4393 used to add or remove parameters from the query string.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004394
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004395 See also "http-request set-query" and "http-request set-uri".
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004396
4397 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004398 # replace "%3D" with "=" in the query string
4399 http-request set-query %[query,regsub(%3D,=,g)]
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004400
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004401http-request set-src <expr> [ { if | unless } <condition> ]
4402 This is used to set the source IP address to the value of specified
4403 expression. Useful when a proxy in front of HAProxy rewrites source IP, but
4404 provides the correct IP in a HTTP header; or you want to mask source IP for
4405 privacy.
4406
4407 Arguments :
4408 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4409 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004410
Cyril Bonté78caf842010-03-10 22:41:43 +01004411 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004412 http-request set-src hdr(x-forwarded-for)
4413 http-request set-src src,ipmask(24)
4414
4415 When possible, set-src preserves the original source port as long as the
4416 address family allows it, otherwise the source port is set to 0.
4417
4418http-request set-src-port <expr> [ { if | unless } <condition> ]
4419
4420 This is used to set the source port address to the value of specified
4421 expression.
4422
4423 Arguments:
4424 <expr> Is a standard HAProxy expression formed by a sample-fetch followed
4425 by some converters.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +01004426
Willy Tarreau20b0de52012-12-24 15:45:22 +01004427 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004428 http-request set-src-port hdr(x-port)
4429 http-request set-src-port int(4000)
4430
4431 When possible, set-src-port preserves the original source address as long as
4432 the address family supports a port, otherwise it forces the source address to
4433 IPv4 "0.0.0.0" before rewriting the port.
4434
4435http-request set-tos <tos> [ { if | unless } <condition> ]
4436
4437 This is used to set the TOS or DSCP field value of packets sent to the client
4438 to the value passed in <tos> on platforms which support this. This value
4439 represents the whole 8 bits of the IP TOS field, and can be expressed both in
4440 decimal or hexadecimal format (prefixed by "0x"). Note that only the 6 higher
4441 bits are used in DSCP or TOS, and the two lower bits are always 0. This can
4442 be used to adjust some routing behavior on border routers based on some
4443 information from the request.
4444
4445 See RFC 2474, 2597, 3260 and 4594 for more information.
4446
4447http-request set-uri <fmt> [ { if | unless } <condition> ]
4448
4449 This rewrites the request URI with the result of the evaluation of format
4450 string <fmt>. The scheme, authority, path and query string are all replaced
4451 at once. This can be used to rewrite hosts in front of proxies, or to
4452 perform complex modifications to the URI such as moving parts between the
4453 path and the query string.
4454 See also "http-request set-path" and "http-request set-query".
4455
4456http-request set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4457
4458 This is used to set the contents of a variable. The variable is declared
4459 inline.
4460
4461 Arguments:
4462 <var-name> The name of the variable starts with an indication about its
4463 scope. The scopes allowed are:
4464 "proc" : the variable is shared with the whole process
4465 "sess" : the variable is shared with the whole session
4466 "txn" : the variable is shared with the transaction
4467 (request and response)
4468 "req" : the variable is shared only during request
4469 processing
4470 "res" : the variable is shared only during response
4471 processing
4472 This prefix is followed by a name. The separator is a '.'.
4473 The name may only contain characters 'a-z', 'A-Z', '0-9'
4474 and '_'.
4475
4476 <expr> Is a standard HAProxy expression formed by a sample-fetch
4477 followed by some converters.
Willy Tarreau20b0de52012-12-24 15:45:22 +01004478
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004479 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004480 http-request set-var(req.my_var) req.fhdr(user-agent),lower
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004481
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004482http-request send-spoe-group <engine-name> <group-name>
4483 [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004484
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004485 This action is used to trigger sending of a group of SPOE messages. To do so,
4486 the SPOE engine used to send messages must be defined, as well as the SPOE
4487 group to send. Of course, the SPOE engine must refer to an existing SPOE
4488 filter. If not engine name is provided on the SPOE filter line, the SPOE
4489 agent name must be used.
4490
4491 Arguments:
4492 <engine-name> The SPOE engine name.
4493
4494 <group-name> The SPOE group name as specified in the engine
4495 configuration.
4496
4497http-request silent-drop [ { if | unless } <condition> ]
4498
4499 This stops the evaluation of the rules and makes the client-facing connection
4500 suddenly disappear using a system-dependent way that tries to prevent the
4501 client from being notified. The effect it then that the client still sees an
4502 established connection while there's none on HAProxy. The purpose is to
4503 achieve a comparable effect to "tarpit" except that it doesn't use any local
4504 resource at all on the machine running HAProxy. It can resist much higher
4505 loads than "tarpit", and slow down stronger attackers. It is important to
4506 understand the impact of using this mechanism. All stateful equipment placed
4507 between the client and HAProxy (firewalls, proxies, load balancers) will also
4508 keep the established connection for a long time and may suffer from this
4509 action.
4510 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4511 option is used to block the emission of a TCP reset. On other systems, the
4512 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4513 router, though it's still delivered to local networks. Do not use it unless
4514 you fully understand how it works.
4515
4516http-request tarpit [deny_status <status>] [ { if | unless } <condition> ]
4517
4518 This stops the evaluation of the rules and immediately blocks the request
4519 without responding for a delay specified by "timeout tarpit" or
4520 "timeout connect" if the former is not set. After that delay, if the client
4521 is still connected, an HTTP error 500 (or optionally the status code
4522 specified as an argument to "deny_status") is returned so that the client
4523 does not suspect it has been tarpitted. Logs will report the flags "PT".
4524 The goal of the tarpit rule is to slow down robots during an attack when
4525 they're limited on the number of concurrent requests. It can be very
4526 efficient against very dumb robots, and will significantly reduce the load
4527 on firewalls compared to a "deny" rule. But when facing "correctly"
4528 developed robots, it can make things worse by forcing haproxy and the front
4529 firewall to support insane number of concurrent connections.
4530 See also the "silent-drop" action.
4531
4532http-request track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4533http-request track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4534http-request track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
4535
4536 This enables tracking of sticky counters from current request. These rules do
4537 not stop evaluation and do not change default action. The number of counters
4538 that may be simultaneously tracked by the same connection is set in
4539 MAX_SESS_STKCTR at build time (reported in haproxy -vv) which defaults to 3,
4540 so the track-sc number is between 0 and (MAX_SESS_STCKTR-1). The first
4541 "track-sc0" rule executed enables tracking of the counters of the specified
4542 table as the first set. The first "track-sc1" rule executed enables tracking
4543 of the counters of the specified table as the second set. The first
4544 "track-sc2" rule executed enables tracking of the counters of the specified
4545 table as the third set. It is a recommended practice to use the first set of
4546 counters for the per-frontend counters and the second set for the per-backend
4547 ones. But this is just a guideline, all may be used everywhere.
4548
4549 Arguments :
4550 <key> is mandatory, and is a sample expression rule as described in
4551 section 7.3. It describes what elements of the incoming request or
4552 connection will be analyzed, extracted, combined, and used to
4553 select which table entry to update the counters.
4554
4555 <table> is an optional table to be used instead of the default one, which
4556 is the stick-table declared in the current proxy. All the counters
4557 for the matches and updates for the key will then be performed in
4558 that table until the session ends.
4559
4560 Once a "track-sc*" rule is executed, the key is looked up in the table and if
4561 it is not found, an entry is allocated for it. Then a pointer to that entry
4562 is kept during all the session's life, and this entry's counters are updated
4563 as often as possible, every time the session's counters are updated, and also
4564 systematically when the session ends. Counters are only updated for events
4565 that happen after the tracking has been started. As an exception, connection
4566 counters and request counters are systematically updated so that they reflect
4567 useful information.
4568
4569 If the entry tracks concurrent connection counters, one connection is counted
4570 for as long as the entry is tracked, and the entry will not expire during
4571 that time. Tracking counters also provides a performance advantage over just
4572 checking the keys, because only one table lookup is performed for all ACL
4573 checks that make use of it.
4574
4575http-request unset-var(<var-name>) [ { if | unless } <condition> ]
4576
4577 This is used to unset a variable. See above for details about <var-name>.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004578
4579 Example:
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004580 http-request unset-var(req.my_var)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004581
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004582http-request wait-for-handshake [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004583
Cyril Bontéc6ad23b2018-10-17 00:14:50 +02004584 This will delay the processing of the request until the SSL handshake
4585 happened. This is mostly useful to delay processing early data until we're
4586 sure they are valid.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004587
Willy Tarreauef781042010-01-27 11:53:01 +01004588
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004589http-response <action> <options...> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004590 Access control for Layer 7 responses
4591
4592 May be used in sections: defaults | frontend | listen | backend
4593 no | yes | yes | yes
4594
4595 The http-response statement defines a set of rules which apply to layer 7
4596 processing. The rules are evaluated in their declaration order when they are
4597 met in a frontend, listen or backend section. Any rule may optionally be
4598 followed by an ACL-based condition, in which case it will only be evaluated
4599 if the condition is true. Since these rules apply on responses, the backend
4600 rules are applied first, followed by the frontend's rules.
4601
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004602 The first keyword is the rule's action. The supported actions are described
4603 below.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004604
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004605 There is no limit to the number of http-response statements per instance.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004606
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004607 It is important to know that http-response rules are processed very early in
4608 the HTTP processing, before "rspdel" or "rsprep" or "rspadd" rules. That way,
4609 headers added by "add-header"/"set-header" are visible by almost all further
4610 ACL rules.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004611
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004612 Using "rspadd"/"rspdel"/"rsprep" to manipulate request headers is discouraged
4613 in newer versions (>= 1.5). But if you need to use regular expression to
4614 delete headers, you can still use "rspdel". Also please use
4615 "http-response deny" instead of "rspdeny".
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004616
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004617 Example:
4618 acl key_acl res.hdr(X-Acl-Key) -m found
Thierry FOURNIERdad3d1d2014-04-22 18:07:25 +02004619
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004620 acl myhost hdr(Host) -f myhost.lst
Sasha Pachev218f0642014-06-16 12:05:59 -06004621
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004622 http-response add-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
4623 http-response del-acl(myhost.lst) %[res.hdr(X-Acl-Key)] if key_acl
Sasha Pachev218f0642014-06-16 12:05:59 -06004624
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004625 Example:
4626 acl value res.hdr(X-Value) -m found
Sasha Pachev218f0642014-06-16 12:05:59 -06004627
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004628 use_backend bk_appli if { hdr(Host),map_str(map.lst) -m found }
Sasha Pachev218f0642014-06-16 12:05:59 -06004629
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004630 http-response set-map(map.lst) %[src] %[res.hdr(X-Value)] if value
4631 http-response del-map(map.lst) %[src] if ! value
Sasha Pachev218f0642014-06-16 12:05:59 -06004632
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004633 See also : "http-request", section 3.4 about userlists and section 7 about
4634 ACL usage.
Sasha Pachev218f0642014-06-16 12:05:59 -06004635
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004636http-response add-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004637
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004638 This is used to add a new entry into an ACL. The ACL must be loaded from a
4639 file (even a dummy empty file). The file name of the ACL to be updated is
4640 passed between parentheses. It takes one argument: <key fmt>, which follows
4641 log-format rules, to collect content of the new entry. It performs a lookup
4642 in the ACL before insertion, to avoid duplicated (or more) values.
4643 This lookup is done by a linear search and can be expensive with large lists!
4644 It is the equivalent of the "add acl" command from the stats socket, but can
4645 be triggered by an HTTP response.
Sasha Pachev218f0642014-06-16 12:05:59 -06004646
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004647http-response add-header <name> <fmt> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004648
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004649 This appends an HTTP header field whose name is specified in <name> and whose
4650 value is defined by <fmt> which follows the log-format rules (see Custom Log
4651 Format in section 8.2.4). This may be used to send a cookie to a client for
4652 example, or to pass some internal information.
4653 This rule is not final, so it is possible to add other similar rules.
4654 Note that header addition is performed immediately, so one rule might reuse
4655 the resulting header from a previous rule.
Sasha Pachev218f0642014-06-16 12:05:59 -06004656
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004657http-response allow [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004658
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004659 This stops the evaluation of the rules and lets the response pass the check.
4660 No further "http-response" rules are evaluated for the current section.
Sasha Pachev218f0642014-06-16 12:05:59 -06004661
Jarno Huuskonen251a6b72019-01-04 14:05:02 +02004662http-response cache-store <name> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004663
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004664 See section 10.2 about cache setup.
Sasha Pachev218f0642014-06-16 12:05:59 -06004665
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004666http-response capture <sample> id <id> [ { if | unless } <condition> ]
Sasha Pachev218f0642014-06-16 12:05:59 -06004667
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004668 This captures sample expression <sample> from the response buffer, and
4669 converts it to a string. The resulting string is stored into the next request
4670 "capture" slot, so it will possibly appear next to some captured HTTP
4671 headers. It will then automatically appear in the logs, and it will be
4672 possible to extract it using sample fetch rules to feed it into headers or
4673 anything. Please check section 7.3 (Fetching samples) and
4674 "capture response header" for more information.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004675
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004676 The keyword "id" is the id of the capture slot which is used for storing the
4677 string. The capture slot must be defined in an associated frontend.
4678 This is useful to run captures in backends. The slot id can be declared by a
4679 previous directive "http-response capture" or with the "declare capture"
4680 keyword.
4681 If the slot <id> doesn't exist, then HAProxy fails parsing the configuration
4682 to prevent unexpected behavior at run time.
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004683
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004684http-response del-acl(<file-name>) <key fmt> [ { if | unless } <condition> ]
Thierry FOURNIER35d70ef2015-08-26 16:21:56 +02004685
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004686 This is used to delete an entry from an ACL. The ACL must be loaded from a
4687 file (even a dummy empty file). The file name of the ACL to be updated is
4688 passed between parentheses. It takes one argument: <key fmt>, which follows
4689 log-format rules, to collect content of the entry to delete.
4690 It is the equivalent of the "del acl" command from the stats socket, but can
4691 be triggered by an HTTP response.
Willy Tarreauf4c43c12013-06-11 17:01:13 +02004692
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004693http-response del-header <name> [ { if | unless } <condition> ]
Willy Tarreau9a355ec2013-06-11 17:45:46 +02004694
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004695 This removes all HTTP header fields whose name is specified in <name>.
Willy Tarreau42cf39e2013-06-11 18:51:32 +02004696
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004697http-response del-map(<file-name>) <key fmt> [ { if | unless } <condition> ]
Willy Tarreau51347ed2013-06-11 19:34:13 +02004698
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004699 This is used to delete an entry from a MAP. The MAP must be loaded from a
4700 file (even a dummy empty file). The file name of the MAP to be updated is
4701 passed between parentheses. It takes one argument: <key fmt>, which follows
4702 log-format rules, to collect content of the entry to delete.
4703 It takes one argument: "file name" It is the equivalent of the "del map"
4704 command from the stats socket, but can be triggered by an HTTP response.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004705
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004706http-response deny [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004707
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004708 This stops the evaluation of the rules and immediately rejects the response
4709 and emits an HTTP 502 error. No further "http-response" rules are evaluated.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004710
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004711http-response redirect <rule> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004712
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004713 This performs an HTTP redirection based on a redirect rule.
4714 This supports a format string similarly to "http-request redirect" rules,
4715 with the exception that only the "location" type of redirect is possible on
4716 the response. See the "redirect" keyword for the rule's syntax. When a
4717 redirect rule is applied during a response, connections to the server are
4718 closed so that no data can be forwarded from the server to the client.
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004719
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004720http-response replace-header <name> <regex-match> <replace-fmt>
4721 [ { if | unless } <condition> ]
Thierry FOURNIERe80fada2015-05-26 18:06:31 +02004722
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004723 This matches the regular expression in all occurrences of header field <name>
4724 according to <match-regex>, and replaces them with the <replace-fmt> argument.
4725 Format characters are allowed in replace-fmt and work like in <fmt> arguments
4726 in "add-header". The match is only case-sensitive. It is important to
4727 understand that this action only considers whole header lines, regardless of
4728 the number of values they may contain. This usage is suited to headers
4729 naturally containing commas in their value, such as Set-Cookie, Expires and
4730 so on.
William Lallemand86d0df02017-11-24 21:36:45 +01004731
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004732 Example:
4733 http-response replace-header Set-Cookie (C=[^;]*);(.*) \1;ip=%bi;\2
Willy Tarreau51d861a2015-05-22 17:30:48 +02004734
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004735 # applied to:
4736 Set-Cookie: C=1; expires=Tue, 14-Jun-2016 01:40:45 GMT
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004737
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004738 # outputs:
4739 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 +02004740
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004741 # assuming the backend IP is 192.168.1.20.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004742
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004743http-response replace-value <name> <regex-match> <replace-fmt>
4744 [ { if | unless } <condition> ]
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004745
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004746 This works like "replace-header" except that it matches the regex against
4747 every comma-delimited value of the header field <name> instead of the entire
4748 header. This is suited for all headers which are allowed to carry more than
4749 one value. An example could be the Accept header.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02004750
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004751 Example:
4752 http-response replace-value Cache-control ^public$ private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004753
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004754 # applied to:
4755 Cache-Control: max-age=3600, public
Christopher Faulet85d79c92016-11-09 16:54:56 +01004756
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004757 # outputs:
4758 Cache-Control: max-age=3600, private
Christopher Faulet85d79c92016-11-09 16:54:56 +01004759
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004760http-response sc-inc-gpc0(<sc-id>) [ { if | unless } <condition> ]
4761http-response sc-inc-gpc1(<sc-id>) [ { if | unless } <condition> ]
Ruoshan Huange4edc6b2016-07-14 15:07:45 +08004762
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004763 This action increments the GPC0 or GPC1 counter according with the sticky
4764 counter designated by <sc-id>. If an error occurs, this action silently fails
4765 and the actions evaluation continues.
Thierry FOURNIER236657b2015-08-19 08:25:14 +02004766
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004767http-response sc-set-gpt0(<sc-id>) <int> [ { if | unless } <condition> ]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02004768
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004769 This action sets the GPT0 tag according to the sticky counter designated by
4770 <sc-id> and the value of <int>. The expected result is a boolean. If an error
4771 occurs, this action silently fails and the actions evaluation continues.
Frédéric Lécaille6778b272018-01-29 15:22:53 +01004772
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004773http-response send-spoe-group [ { if | unless } <condition> ]
Willy Tarreau2d392c22015-08-24 01:43:45 +02004774
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004775 This action is used to trigger sending of a group of SPOE messages. To do so,
4776 the SPOE engine used to send messages must be defined, as well as the SPOE
4777 group to send. Of course, the SPOE engine must refer to an existing SPOE
4778 filter. If not engine name is provided on the SPOE filter line, the SPOE
4779 agent name must be used.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004780
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004781 Arguments:
4782 <engine-name> The SPOE engine name.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004783
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004784 <group-name> The SPOE group name as specified in the engine
4785 configuration.
Christopher Faulet76c09ef2017-09-21 11:03:52 +02004786
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004787http-response set-header <name> <fmt> [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004788
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004789 This does the same as "add-header" except that the header name is first
4790 removed if it existed. This is useful when passing security information to
4791 the server, where the header must not be manipulated by external users.
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004792
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004793http-response set-log-level <level> [ { if | unless } <condition> ]
4794
4795 This is used to change the log level of the current request when a certain
4796 condition is met. Valid levels are the 8 syslog levels (see the "log"
4797 keyword) plus the special level "silent" which disables logging for this
4798 request. This rule is not final so the last matching rule wins. This rule can
4799 be useful to disable health checks coming from another equipment.
4800
4801http-response set-map(<file-name>) <key fmt> <value fmt>
4802
4803 This is used to add a new entry into a MAP. The MAP must be loaded from a
4804 file (even a dummy empty file). The file name of the MAP to be updated is
4805 passed between parentheses. It takes 2 arguments: <key fmt>, which follows
4806 log-format rules, used to collect MAP key, and <value fmt>, which follows
4807 log-format rules, used to collect content for the new entry. It performs a
4808 lookup in the MAP before insertion, to avoid duplicated (or more) values.
4809 This lookup is done by a linear search and can be expensive with large lists!
4810 It is the equivalent of the "set map" command from the stats socket, but can
4811 be triggered by an HTTP response.
4812
4813http-response set-mark <mark> [ { if | unless } <condition> ]
4814
4815 This is used to set the Netfilter MARK on all packets sent to the client to
4816 the value passed in <mark> on platforms which support it. This value is an
4817 unsigned 32 bit value which can be matched by netfilter and by the routing
4818 table. It can be expressed both in decimal or hexadecimal format (prefixed
4819 by "0x"). This can be useful to force certain packets to take a different
4820 route (for example a cheaper network path for bulk downloads). This works on
4821 Linux kernels 2.6.32 and above and requires admin privileges.
4822
4823http-response set-nice <nice> [ { if | unless } <condition> ]
4824
4825 This sets the "nice" factor of the current request being processed.
4826 It only has effect against the other requests being processed at the same
4827 time. The default value is 0, unless altered by the "nice" setting on the
4828 "bind" line. The accepted range is -1024..1024. The higher the value, the
4829 nicest the request will be. Lower values will make the request more important
4830 than other ones. This can be useful to improve the speed of some requests, or
4831 lower the priority of non-important requests. Using this setting without
4832 prior experimentation can cause some major slowdown.
4833
4834http-response set-status <status> [reason <str>]
4835 [ { if | unless } <condition> ]
4836
4837 This replaces the response status code with <status> which must be an integer
4838 between 100 and 999. Optionally, a custom reason text can be provided defined
4839 by <str>, or the default reason for the specified code will be used as a
4840 fallback.
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08004841
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004842 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004843 # return "431 Request Header Fields Too Large"
4844 http-response set-status 431
4845 # return "503 Slow Down", custom reason
4846 http-response set-status 503 reason "Slow Down".
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004847
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004848http-response set-tos <tos> [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004849
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004850 This is used to set the TOS or DSCP field value of packets sent to the client
4851 to the value passed in <tos> on platforms which support this.
4852 This value represents the whole 8 bits of the IP TOS field, and can be
4853 expressed both in decimal or hexadecimal format (prefixed by "0x"). Note that
4854 only the 6 higher bits are used in DSCP or TOS, and the two lower bits are
4855 always 0. This can be used to adjust some routing behavior on border routers
4856 based on some information from the request.
4857
4858 See RFC 2474, 2597, 3260 and 4594 for more information.
4859
4860http-response set-var(<var-name>) <expr> [ { if | unless } <condition> ]
4861
4862 This is used to set the contents of a variable. The variable is declared
4863 inline.
4864
4865 Arguments:
4866 <var-name> The name of the variable starts with an indication about its
4867 scope. The scopes allowed are:
4868 "proc" : the variable is shared with the whole process
4869 "sess" : the variable is shared with the whole session
4870 "txn" : the variable is shared with the transaction
4871 (request and response)
4872 "req" : the variable is shared only during request
4873 processing
4874 "res" : the variable is shared only during response
4875 processing
4876 This prefix is followed by a name. The separator is a '.'.
4877 The name may only contain characters 'a-z', 'A-Z', '0-9', '.'
4878 and '_'.
4879
4880 <expr> Is a standard HAProxy expression formed by a sample-fetch
4881 followed by some converters.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004882
4883 Example:
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004884 http-response set-var(sess.last_redir) res.hdr(location)
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004885
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004886http-response silent-drop [ { if | unless } <condition> ]
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004887
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004888 This stops the evaluation of the rules and makes the client-facing connection
4889 suddenly disappear using a system-dependent way that tries to prevent the
4890 client from being notified. The effect it then that the client still sees an
4891 established connection while there's none on HAProxy. The purpose is to
4892 achieve a comparable effect to "tarpit" except that it doesn't use any local
4893 resource at all on the machine running HAProxy. It can resist much higher
4894 loads than "tarpit", and slow down stronger attackers. It is important to
4895 understand the impact of using this mechanism. All stateful equipment placed
4896 between the client and HAProxy (firewalls, proxies, load balancers) will also
4897 keep the established connection for a long time and may suffer from this
4898 action.
4899 On modern Linux systems running with enough privileges, the TCP_REPAIR socket
4900 option is used to block the emission of a TCP reset. On other systems, the
4901 socket's TTL is reduced to 1 so that the TCP reset doesn't pass the first
4902 router, though it's still delivered to local networks. Do not use it unless
4903 you fully understand how it works.
Baptiste Assmannfabcbe02014-04-24 22:16:59 +02004904
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004905http-response track-sc0 <key> [table <table>] [ { if | unless } <condition> ]
4906http-response track-sc1 <key> [table <table>] [ { if | unless } <condition> ]
4907http-response track-sc2 <key> [table <table>] [ { if | unless } <condition> ]
Willy Tarreaue365c0b2013-06-11 16:06:12 +02004908
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02004909 This enables tracking of sticky counters from current response. Please refer
4910 to "http-request track-sc" for a complete description. The only difference
4911 from "http-request track-sc" is the <key> sample expression can only make use
4912 of samples in response (e.g. res.*, status etc.) and samples below Layer 6
4913 (e.g. SSL-related samples, see section 7.3.4). If the sample is not
4914 supported, haproxy will fail and warn while parsing the config.
4915
4916http-response unset-var(<var-name>) [ { if | unless } <condition> ]
4917
4918 This is used to unset a variable. See "http-response set-var" for details
4919 about <var-name>.
4920
4921 Example:
4922 http-response unset-var(sess.last_redir)
4923
Baptiste Assmann5ecb77f2013-10-06 23:24:13 +02004924
Willy Tarreau30631952015-08-06 15:05:24 +02004925http-reuse { never | safe | aggressive | always }
4926 Declare how idle HTTP connections may be shared between requests
4927
4928 May be used in sections: defaults | frontend | listen | backend
4929 yes | no | yes | yes
4930
4931 By default, a connection established between haproxy and the backend server
Olivier Houchard86006a52018-12-14 19:37:49 +01004932 which is considered safe for reuse is moved back to the server's idle
4933 connections pool so that any other request can make use of it. This is the
4934 "safe" strategy below.
Willy Tarreau30631952015-08-06 15:05:24 +02004935
4936 The argument indicates the desired connection reuse strategy :
4937
Olivier Houchard86006a52018-12-14 19:37:49 +01004938 - "never" : idle connections are never shared between sessions. This mode
4939 may be enforced to cancel a different strategy inherited from
4940 a defaults section or for troubleshooting. For example, if an
4941 old bogus application considers that multiple requests over
4942 the same connection come from the same client and it is not
4943 possible to fix the application, it may be desirable to
4944 disable connection sharing in a single backend. An example of
4945 such an application could be an old haproxy using cookie
4946 insertion in tunnel mode and not checking any request past the
4947 first one.
Willy Tarreau30631952015-08-06 15:05:24 +02004948
Olivier Houchard86006a52018-12-14 19:37:49 +01004949 - "safe" : this is the default and the recommended strategy. The first
4950 request of a session is always sent over its own connection,
4951 and only subsequent requests may be dispatched over other
4952 existing connections. This ensures that in case the server
4953 closes the connection when the request is being sent, the
4954 browser can decide to silently retry it. Since it is exactly
4955 equivalent to regular keep-alive, there should be no side
4956 effects.
Willy Tarreau30631952015-08-06 15:05:24 +02004957
4958 - "aggressive" : this mode may be useful in webservices environments where
4959 all servers are not necessarily known and where it would be
4960 appreciable to deliver most first requests over existing
4961 connections. In this case, first requests are only delivered
4962 over existing connections that have been reused at least once,
4963 proving that the server correctly supports connection reuse.
4964 It should only be used when it's sure that the client can
4965 retry a failed request once in a while and where the benefit
4966 of aggressive connection reuse significantly outweights the
4967 downsides of rare connection failures.
4968
4969 - "always" : this mode is only recommended when the path to the server is
4970 known for never breaking existing connections quickly after
4971 releasing them. It allows the first request of a session to be
4972 sent to an existing connection. This can provide a significant
4973 performance increase over the "safe" strategy when the backend
4974 is a cache farm, since such components tend to show a
Davor Ocelice9ed2812017-12-25 17:49:28 +01004975 consistent behavior and will benefit from the connection
Willy Tarreau30631952015-08-06 15:05:24 +02004976 sharing. It is recommended that the "http-keep-alive" timeout
4977 remains low in this mode so that no dead connections remain
4978 usable. In most cases, this will lead to the same performance
4979 gains as "aggressive" but with more risks. It should only be
4980 used when it improves the situation over "aggressive".
4981
4982 When http connection sharing is enabled, a great care is taken to respect the
Davor Ocelice9ed2812017-12-25 17:49:28 +01004983 connection properties and compatibility. Specifically :
4984 - connections made with "usesrc" followed by a client-dependent value
4985 ("client", "clientip", "hdr_ip") are marked private and never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004986
4987 - connections sent to a server with a TLS SNI extension are marked private
Davor Ocelice9ed2812017-12-25 17:49:28 +01004988 and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004989
Lukas Tribusfd9b68c2018-10-27 20:06:59 +02004990 - connections with certain bogus authentication schemes (relying on the
4991 connection) like NTLM are detected, marked private and are never shared;
Willy Tarreau30631952015-08-06 15:05:24 +02004992
4993 No connection pool is involved, once a session dies, the last idle connection
4994 it was attached to is deleted at the same time. This ensures that connections
4995 may not last after all sessions are closed.
4996
4997 Note: connection reuse improves the accuracy of the "server maxconn" setting,
4998 because almost no new connection will be established while idle connections
4999 remain available. This is particularly true with the "always" strategy.
5000
5001 See also : "option http-keep-alive", "server maxconn"
5002
5003
Mark Lamourinec2247f02012-01-04 13:02:01 -05005004http-send-name-header [<header>]
5005 Add the server name to a request. Use the header string given by <header>
5006
5007 May be used in sections: defaults | frontend | listen | backend
5008 yes | no | yes | yes
5009
5010 Arguments :
5011
5012 <header> The header string to use to send the server name
5013
5014 The "http-send-name-header" statement causes the name of the target
Davor Ocelice9ed2812017-12-25 17:49:28 +01005015 server to be added to the headers of an HTTP request. The name
Mark Lamourinec2247f02012-01-04 13:02:01 -05005016 is added with the header string proved.
5017
5018 See also : "server"
5019
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005020id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +02005021 Set a persistent ID to a proxy.
5022 May be used in sections : defaults | frontend | listen | backend
5023 no | yes | yes | yes
5024 Arguments : none
5025
5026 Set a persistent ID for the proxy. This ID must be unique and positive.
5027 An unused ID will automatically be assigned if unset. The first assigned
5028 value will be 1. This ID is currently only returned in statistics.
Krzysztof Piotr Oledzkif58a9622008-02-23 01:19:10 +01005029
5030
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005031ignore-persist { if | unless } <condition>
5032 Declare a condition to ignore persistence
5033 May be used in sections: defaults | frontend | listen | backend
Cyril Bonté4288c5a2018-03-12 22:02:59 +01005034 no | no | yes | yes
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005035
5036 By default, when cookie persistence is enabled, every requests containing
5037 the cookie are unconditionally persistent (assuming the target server is up
5038 and running).
5039
5040 The "ignore-persist" statement allows one to declare various ACL-based
5041 conditions which, when met, will cause a request to ignore persistence.
5042 This is sometimes useful to load balance requests for static files, which
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03005043 often don't require persistence. This can also be used to fully disable
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005044 persistence for a specific User-Agent (for example, some web crawler bots).
5045
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005046 The persistence is ignored when an "if" condition is met, or unless an
5047 "unless" condition is met.
5048
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03005049 Example:
5050 acl url_static path_beg /static /images /img /css
5051 acl url_static path_end .gif .png .jpg .css .js
5052 ignore-persist if url_static
5053
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005054 See also : "force-persist", "cookie", and section 7 about ACL usage.
5055
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005056load-server-state-from-file { global | local | none }
5057 Allow seamless reload of HAProxy
5058 May be used in sections: defaults | frontend | listen | backend
5059 yes | no | yes | yes
5060
5061 This directive points HAProxy to a file where server state from previous
5062 running process has been saved. That way, when starting up, before handling
5063 traffic, the new process can apply old states to servers exactly has if no
Davor Ocelice9ed2812017-12-25 17:49:28 +01005064 reload occurred. The purpose of the "load-server-state-from-file" directive is
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005065 to tell haproxy which file to use. For now, only 2 arguments to either prevent
5066 loading state or load states from a file containing all backends and servers.
5067 The state file can be generated by running the command "show servers state"
5068 over the stats socket and redirect output.
5069
Davor Ocelice9ed2812017-12-25 17:49:28 +01005070 The format of the file is versioned and is very specific. To understand it,
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005071 please read the documentation of the "show servers state" command (chapter
Willy Tarreau1af20c72017-06-23 16:01:14 +02005072 9.3 of Management Guide).
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005073
5074 Arguments:
5075 global load the content of the file pointed by the global directive
5076 named "server-state-file".
5077
5078 local load the content of the file pointed by the directive
5079 "server-state-file-name" if set. If not set, then the backend
5080 name is used as a file name.
5081
5082 none don't load any stat for this backend
5083
5084 Notes:
Willy Tarreaue5a60682016-11-09 14:54:53 +01005085 - server's IP address is preserved across reloads by default, but the
5086 order can be changed thanks to the server's "init-addr" setting. This
5087 means that an IP address change performed on the CLI at run time will
Davor Ocelice9ed2812017-12-25 17:49:28 +01005088 be preserved, and that any change to the local resolver (e.g. /etc/hosts)
Willy Tarreaue5a60682016-11-09 14:54:53 +01005089 will possibly not have any effect if the state file is in use.
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005090
5091 - server's weight is applied from previous running process unless it has
5092 has changed between previous and new configuration files.
5093
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005094 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005095
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005096 global
5097 stats socket /tmp/socket
5098 server-state-file /tmp/server_state
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005099
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005100 defaults
5101 load-server-state-from-file global
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005102
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005103 backend bk
5104 server s1 127.0.0.1:22 check weight 11
5105 server s2 127.0.0.1:22 check weight 12
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005106
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005107
5108 Then one can run :
5109
5110 socat /tmp/socket - <<< "show servers state" > /tmp/server_state
5111
5112 Content of the file /tmp/server_state would be like this:
5113
5114 1
5115 # <field names skipped for the doc example>
5116 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5117 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5118
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005119 Example: Minimal configuration
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005120
5121 global
5122 stats socket /tmp/socket
5123 server-state-base /etc/haproxy/states
5124
5125 defaults
5126 load-server-state-from-file local
5127
5128 backend bk
5129 server s1 127.0.0.1:22 check weight 11
5130 server s2 127.0.0.1:22 check weight 12
5131
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005132
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02005133 Then one can run :
5134
5135 socat /tmp/socket - <<< "show servers state bk" > /etc/haproxy/states/bk
5136
5137 Content of the file /etc/haproxy/states/bk would be like this:
5138
5139 1
5140 # <field names skipped for the doc example>
5141 1 bk 1 s1 127.0.0.1 2 0 11 11 4 6 3 4 6 0 0
5142 1 bk 2 s2 127.0.0.1 2 0 12 12 4 6 3 4 6 0 0
5143
5144 See also: "server-state-file", "server-state-file-name", and
5145 "show servers state"
5146
Cyril Bonté0d4bf012010-04-25 23:21:46 +02005147
Willy Tarreau2769aa02007-12-27 18:26:09 +01005148log global
Willy Tarreauadb345d2018-11-12 07:56:13 +01005149log <address> [len <length>] [format <format>] <facility> [<level> [<minlevel>]]
William Lallemand0f99e342011-10-12 17:50:54 +02005150no log
Willy Tarreau2769aa02007-12-27 18:26:09 +01005151 Enable per-instance logging of events and traffic.
5152 May be used in sections : defaults | frontend | listen | backend
5153 yes | yes | yes | yes
William Lallemand0f99e342011-10-12 17:50:54 +02005154
5155 Prefix :
5156 no should be used when the logger list must be flushed. For example,
5157 if you don't want to inherit from the default logger list. This
5158 prefix does not allow arguments.
5159
Willy Tarreau2769aa02007-12-27 18:26:09 +01005160 Arguments :
5161 global should be used when the instance's logging parameters are the
5162 same as the global ones. This is the most common usage. "global"
5163 replaces <address>, <facility> and <level> with those of the log
5164 entries found in the "global" section. Only one "log global"
5165 statement may be used per instance, and this form takes no other
5166 parameter.
5167
5168 <address> indicates where to send the logs. It takes the same format as
5169 for the "global" section's logs, and can be one of :
5170
5171 - An IPv4 address optionally followed by a colon (':') and a UDP
5172 port. If no port is specified, 514 is used by default (the
5173 standard syslog port).
5174
David du Colombier24bb5f52011-03-17 10:40:23 +01005175 - An IPv6 address followed by a colon (':') and optionally a UDP
5176 port. If no port is specified, 514 is used by default (the
5177 standard syslog port).
5178
Willy Tarreau2769aa02007-12-27 18:26:09 +01005179 - A filesystem path to a UNIX domain socket, keeping in mind
5180 considerations for chroot (be sure the path is accessible
5181 inside the chroot) and uid/gid (be sure the path is
Davor Ocelice9ed2812017-12-25 17:49:28 +01005182 appropriately writable).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005183
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005184 - A file descriptor number in the form "fd@<number>", which may
5185 point to a pipe, terminal, or socket. In this case unbuffered
5186 logs are used and one writev() call per log is performed. This
5187 is a bit expensive but acceptable for most workloads. Messages
5188 sent this way will not be truncated but may be dropped, in
5189 which case the DroppedLogs counter will be incremented. The
5190 writev() call is atomic even on pipes for messages up to
5191 PIPE_BUF size, which POSIX recommends to be at least 512 and
5192 which is 4096 bytes on most modern operating systems. Any
5193 larger message may be interleaved with messages from other
5194 processes. Exceptionally for debugging purposes the file
5195 descriptor may also be directed to a file, but doing so will
5196 significantly slow haproxy down as non-blocking calls will be
5197 ignored. Also there will be no way to purge nor rotate this
5198 file without restarting the process. Note that the configured
5199 syslog format is preserved, so the output is suitable for use
Willy Tarreauc1b06452018-11-12 11:57:56 +01005200 with a TCP syslog server. See also the "short" and "raw"
5201 formats below.
Willy Tarreau5a32ecc2018-11-12 07:34:59 +01005202
5203 - "stdout" / "stderr", which are respectively aliases for "fd@1"
5204 and "fd@2", see above.
5205
5206 You may want to reference some environment variables in the
5207 address parameter, see section 2.3 about environment variables.
Willy Tarreaudad36a32013-03-11 01:20:04 +01005208
Willy Tarreau18324f52014-06-27 18:10:07 +02005209 <length> is an optional maximum line length. Log lines larger than this
5210 value will be truncated before being sent. The reason is that
5211 syslog servers act differently on log line length. All servers
5212 support the default value of 1024, but some servers simply drop
5213 larger lines while others do log them. If a server supports long
5214 lines, it may make sense to set this value here in order to avoid
5215 truncating long lines. Similarly, if a server drops long lines,
5216 it is preferable to truncate them before sending them. Accepted
5217 values are 80 to 65535 inclusive. The default value of 1024 is
5218 generally fine for all standard usages. Some specific cases of
Davor Ocelice9ed2812017-12-25 17:49:28 +01005219 long captures or JSON-formatted logs may require larger values.
Willy Tarreau18324f52014-06-27 18:10:07 +02005220
Willy Tarreauadb345d2018-11-12 07:56:13 +01005221 <format> is the log format used when generating syslog messages. It may be
5222 one of the following :
5223
5224 rfc3164 The RFC3164 syslog message format. This is the default.
5225 (https://tools.ietf.org/html/rfc3164)
5226
5227 rfc5424 The RFC5424 syslog message format.
5228 (https://tools.ietf.org/html/rfc5424)
5229
Willy Tarreaue8746a02018-11-12 08:45:00 +01005230 short A message containing only a level between angle brackets such as
5231 '<3>', followed by the text. The PID, date, time, process name
5232 and system name are omitted. This is designed to be used with a
5233 local log server. This format is compatible with what the
5234 systemd logger consumes.
5235
Willy Tarreauc1b06452018-11-12 11:57:56 +01005236 raw A message containing only the text. The level, PID, date, time,
5237 process name and system name are omitted. This is designed to
5238 be used in containers or during development, where the severity
5239 only depends on the file descriptor used (stdout/stderr).
5240
Willy Tarreau2769aa02007-12-27 18:26:09 +01005241 <facility> must be one of the 24 standard syslog facilities :
5242
Willy Tarreaue8746a02018-11-12 08:45:00 +01005243 kern user mail daemon auth syslog lpr news
5244 uucp cron auth2 ftp ntp audit alert cron2
5245 local0 local1 local2 local3 local4 local5 local6 local7
5246
Willy Tarreauc1b06452018-11-12 11:57:56 +01005247 Note that the facility is ignored for the "short" and "raw"
5248 formats, but still required as a positional field. It is
5249 recommended to use "daemon" in this case to make it clear that
5250 it's only supposed to be used locally.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005251
5252 <level> is optional and can be specified to filter outgoing messages. By
5253 default, all messages are sent. If a level is specified, only
5254 messages with a severity at least as important as this level
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005255 will be sent. An optional minimum level can be specified. If it
5256 is set, logs emitted with a more severe level than this one will
5257 be capped to this level. This is used to avoid sending "emerg"
5258 messages on all terminals on some default syslog configurations.
5259 Eight levels are known :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005260
5261 emerg alert crit err warning notice info debug
5262
William Lallemand0f99e342011-10-12 17:50:54 +02005263 It is important to keep in mind that it is the frontend which decides what to
5264 log from a connection, and that in case of content switching, the log entries
5265 from the backend will be ignored. Connections are logged at level "info".
Willy Tarreaucc6c8912009-02-22 10:53:55 +01005266
5267 However, backend log declaration define how and where servers status changes
5268 will be logged. Level "notice" will be used to indicate a server going up,
5269 "warning" will be used for termination signals and definitive service
5270 termination, and "alert" will be used for when a server goes down.
5271
5272 Note : According to RFC3164, messages are truncated to 1024 bytes before
5273 being emitted.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005274
5275 Example :
5276 log global
Willy Tarreauc1b06452018-11-12 11:57:56 +01005277 log stdout format short daemon # send log to systemd
5278 log stdout format raw daemon # send everything to stdout
5279 log stderr format raw daemon notice # send important events to stderr
Willy Tarreauf7edefa2009-05-10 17:20:05 +02005280 log 127.0.0.1:514 local0 notice # only send important events
5281 log 127.0.0.1:514 local0 notice notice # same but limit output level
William Lallemandb2f07452015-05-12 14:27:13 +02005282 log "${LOCAL_SYSLOG}:514" local0 notice # send to local server
Willy Tarreaudad36a32013-03-11 01:20:04 +01005283
Willy Tarreau2769aa02007-12-27 18:26:09 +01005284
William Lallemand48940402012-01-30 16:47:22 +01005285log-format <string>
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005286 Specifies the log format string to use for traffic logs
5287 May be used in sections: defaults | frontend | listen | backend
5288 yes | yes | yes | no
William Lallemand48940402012-01-30 16:47:22 +01005289
Willy Tarreaufb4e7ea2015-01-07 14:55:17 +01005290 This directive specifies the log format string that will be used for all logs
5291 resulting from traffic passing through the frontend using this line. If the
5292 directive is used in a defaults section, all subsequent frontends will use
5293 the same log format. Please see section 8.2.4 which covers the log format
5294 string in depth.
William Lallemand48940402012-01-30 16:47:22 +01005295
Guillaume de Lafond29f45602017-03-31 19:52:15 +02005296 "log-format" directive overrides previous "option tcplog", "log-format" and
5297 "option httplog" directives.
5298
Dragan Dosen7ad31542015-09-28 17:16:47 +02005299log-format-sd <string>
5300 Specifies the RFC5424 structured-data log format string
5301 May be used in sections: defaults | frontend | listen | backend
5302 yes | yes | yes | no
5303
5304 This directive specifies the RFC5424 structured-data log format string that
5305 will be used for all logs resulting from traffic passing through the frontend
5306 using this line. If the directive is used in a defaults section, all
5307 subsequent frontends will use the same log format. Please see section 8.2.4
5308 which covers the log format string in depth.
5309
5310 See https://tools.ietf.org/html/rfc5424#section-6.3 for more information
5311 about the RFC5424 structured-data part.
5312
5313 Note : This log format string will be used only for loggers that have set
5314 log format to "rfc5424".
5315
5316 Example :
5317 log-format-sd [exampleSDID@1234\ bytes=\"%B\"\ status=\"%ST\"]
5318
5319
Willy Tarreau094af4e2015-01-07 15:03:42 +01005320log-tag <string>
5321 Specifies the log tag to use for all outgoing logs
5322 May be used in sections: defaults | frontend | listen | backend
5323 yes | yes | yes | yes
5324
5325 Sets the tag field in the syslog header to this string. It defaults to the
5326 log-tag set in the global section, otherwise the program name as launched
5327 from the command line, which usually is "haproxy". Sometimes it can be useful
5328 to differentiate between multiple processes running on the same host, or to
5329 differentiate customer instances running in the same process. In the backend,
5330 logs about servers up/down will use this tag. As a hint, it can be convenient
5331 to set a log-tag related to a hosted customer in a defaults section then put
5332 all the frontends and backends for that customer, then start another customer
5333 in a new defaults section. See also the global "log-tag" directive.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005334
Willy Tarreauc35362a2014-04-25 13:58:37 +02005335max-keep-alive-queue <value>
5336 Set the maximum server queue size for maintaining keep-alive connections
5337 May be used in sections: defaults | frontend | listen | backend
5338 yes | no | yes | yes
5339
5340 HTTP keep-alive tries to reuse the same server connection whenever possible,
5341 but sometimes it can be counter-productive, for example if a server has a lot
5342 of connections while other ones are idle. This is especially true for static
5343 servers.
5344
5345 The purpose of this setting is to set a threshold on the number of queued
5346 connections at which haproxy stops trying to reuse the same server and prefers
5347 to find another one. The default value, -1, means there is no limit. A value
5348 of zero means that keep-alive requests will never be queued. For very close
5349 servers which can be reached with a low latency and which are not sensible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01005350 breaking keep-alive, a low value is recommended (e.g. local static server can
Willy Tarreauc35362a2014-04-25 13:58:37 +02005351 use a value of 10 or less). For remote servers suffering from a high latency,
5352 higher values might be needed to cover for the latency and/or the cost of
5353 picking a different server.
5354
5355 Note that this has no impact on responses which are maintained to the same
5356 server consecutively to a 401 response. They will still go to the same server
5357 even if they have to be queued.
5358
5359 See also : "option http-server-close", "option prefer-last-server", server
5360 "maxconn" and cookie persistence.
5361
Olivier Houcharda4d4fdf2018-12-14 19:27:06 +01005362max-session-srv-conns <nb>
5363 Set the maximum number of outgoing connections we can keep idling for a given
5364 client session. The default is 5 (it precisely equals MAX_SRV_LIST which is
5365 defined at build time).
Willy Tarreauc35362a2014-04-25 13:58:37 +02005366
Willy Tarreau2769aa02007-12-27 18:26:09 +01005367maxconn <conns>
5368 Fix the maximum number of concurrent connections on a frontend
5369 May be used in sections : defaults | frontend | listen | backend
5370 yes | yes | yes | no
5371 Arguments :
5372 <conns> is the maximum number of concurrent connections the frontend will
5373 accept to serve. Excess connections will be queued by the system
5374 in the socket's listen queue and will be served once a connection
5375 closes.
5376
5377 If the system supports it, it can be useful on big sites to raise this limit
5378 very high so that haproxy manages connection queues, instead of leaving the
5379 clients with unanswered connection attempts. This value should not exceed the
5380 global maxconn. Also, keep in mind that a connection contains two buffers
Baptiste Assmann79fb45d2016-03-06 23:34:31 +01005381 of tune.bufsize (16kB by default) each, as well as some other data resulting
5382 in about 33 kB of RAM being consumed per established connection. That means
5383 that a medium system equipped with 1GB of RAM can withstand around
5384 20000-25000 concurrent connections if properly tuned.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005385
5386 Also, when <conns> is set to large values, it is possible that the servers
5387 are not sized to accept such loads, and for this reason it is generally wise
5388 to assign them some reasonable connection limits.
5389
Willy Tarreauc8d5b952019-02-27 17:25:52 +01005390 When this value is set to zero, which is the default, the global "maxconn"
5391 value is used.
Vincent Bernat6341be52012-06-27 17:18:30 +02005392
Willy Tarreau2769aa02007-12-27 18:26:09 +01005393 See also : "server", global section's "maxconn", "fullconn"
5394
5395
5396mode { tcp|http|health }
5397 Set the running mode or protocol of the instance
5398 May be used in sections : defaults | frontend | listen | backend
5399 yes | yes | yes | yes
5400 Arguments :
5401 tcp The instance will work in pure TCP mode. A full-duplex connection
5402 will be established between clients and servers, and no layer 7
5403 examination will be performed. This is the default mode. It
5404 should be used for SSL, SSH, SMTP, ...
5405
5406 http The instance will work in HTTP mode. The client request will be
5407 analyzed in depth before connecting to any server. Any request
5408 which is not RFC-compliant will be rejected. Layer 7 filtering,
5409 processing and switching will be possible. This is the mode which
5410 brings HAProxy most of its value.
5411
5412 health The instance will work in "health" mode. It will just reply "OK"
Willy Tarreau82569f92012-09-27 23:48:56 +02005413 to incoming connections and close the connection. Alternatively,
5414 If the "httpchk" option is set, "HTTP/1.0 200 OK" will be sent
5415 instead. Nothing will be logged in either case. This mode is used
5416 to reply to external components health checks. This mode is
5417 deprecated and should not be used anymore as it is possible to do
5418 the same and even better by combining TCP or HTTP modes with the
5419 "monitor" keyword.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005420
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005421 When doing content switching, it is mandatory that the frontend and the
5422 backend are in the same mode (generally HTTP), otherwise the configuration
5423 will be refused.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005424
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005425 Example :
Willy Tarreau2769aa02007-12-27 18:26:09 +01005426 defaults http_instances
5427 mode http
5428
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005429 See also : "monitor", "monitor-net"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005430
Willy Tarreau0ba27502007-12-24 16:55:16 +01005431
Cyril Bontéf0c60612010-02-06 14:44:47 +01005432monitor fail { if | unless } <condition>
Willy Tarreau2769aa02007-12-27 18:26:09 +01005433 Add a condition to report a failure to a monitor HTTP request.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005434 May be used in sections : defaults | frontend | listen | backend
5435 no | yes | yes | no
Willy Tarreau0ba27502007-12-24 16:55:16 +01005436 Arguments :
5437 if <cond> the monitor request will fail if the condition is satisfied,
5438 and will succeed otherwise. The condition should describe a
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005439 combined test which must induce a failure if all conditions
Willy Tarreau0ba27502007-12-24 16:55:16 +01005440 are met, for instance a low number of servers both in a
5441 backend and its backup.
5442
5443 unless <cond> the monitor request will succeed only if the condition is
5444 satisfied, and will fail otherwise. Such a condition may be
5445 based on a test on the presence of a minimum number of active
5446 servers in a list of backends.
5447
5448 This statement adds a condition which can force the response to a monitor
5449 request to report a failure. By default, when an external component queries
5450 the URI dedicated to monitoring, a 200 response is returned. When one of the
5451 conditions above is met, haproxy will return 503 instead of 200. This is
5452 very useful to report a site failure to an external component which may base
5453 routing advertisements between multiple sites on the availability reported by
5454 haproxy. In this case, one would rely on an ACL involving the "nbsrv"
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005455 criterion. Note that "monitor fail" only works in HTTP mode. Both status
5456 messages may be tweaked using "errorfile" or "errorloc" if needed.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005457
5458 Example:
5459 frontend www
Willy Tarreau2769aa02007-12-27 18:26:09 +01005460 mode http
Willy Tarreau0ba27502007-12-24 16:55:16 +01005461 acl site_dead nbsrv(dynamic) lt 2
5462 acl site_dead nbsrv(static) lt 2
5463 monitor-uri /site_alive
5464 monitor fail if site_dead
5465
Willy Tarreauae94d4d2011-05-11 16:28:49 +02005466 See also : "monitor-net", "monitor-uri", "errorfile", "errorloc"
Willy Tarreau2769aa02007-12-27 18:26:09 +01005467
5468
5469monitor-net <source>
5470 Declare a source network which is limited to monitor requests
5471 May be used in sections : defaults | frontend | listen | backend
5472 yes | yes | yes | no
5473 Arguments :
5474 <source> is the source IPv4 address or network which will only be able to
5475 get monitor responses to any request. It can be either an IPv4
5476 address, a host name, or an address followed by a slash ('/')
5477 followed by a mask.
5478
5479 In TCP mode, any connection coming from a source matching <source> will cause
5480 the connection to be immediately closed without any log. This allows another
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005481 equipment to probe the port and verify that it is still listening, without
Willy Tarreau2769aa02007-12-27 18:26:09 +01005482 forwarding the connection to a remote server.
5483
5484 In HTTP mode, a connection coming from a source matching <source> will be
5485 accepted, the following response will be sent without waiting for a request,
5486 then the connection will be closed : "HTTP/1.0 200 OK". This is normally
5487 enough for any front-end HTTP probe to detect that the service is UP and
Willy Tarreau82569f92012-09-27 23:48:56 +02005488 running without forwarding the request to a backend server. Note that this
5489 response is sent in raw format, without any transformation. This is important
5490 as it means that it will not be SSL-encrypted on SSL listeners.
Willy Tarreau2769aa02007-12-27 18:26:09 +01005491
Willy Tarreau82569f92012-09-27 23:48:56 +02005492 Monitor requests are processed very early, just after tcp-request connection
5493 ACLs which are the only ones able to block them. These connections are short
5494 lived and never wait for any data from the client. They cannot be logged, and
5495 it is the intended purpose. They are only used to report HAProxy's health to
5496 an upper component, nothing more. Please note that "monitor fail" rules do
5497 not apply to connections intercepted by "monitor-net".
Willy Tarreau2769aa02007-12-27 18:26:09 +01005498
Willy Tarreau95cd2832010-03-04 23:36:33 +01005499 Last, please note that only one "monitor-net" statement can be specified in
5500 a frontend. If more than one is found, only the last one will be considered.
Cyril Bonté108cf6e2012-04-21 23:30:29 +02005501
Willy Tarreau2769aa02007-12-27 18:26:09 +01005502 Example :
5503 # addresses .252 and .253 are just probing us.
5504 frontend www
5505 monitor-net 192.168.0.252/31
5506
5507 See also : "monitor fail", "monitor-uri"
5508
5509
5510monitor-uri <uri>
5511 Intercept a URI used by external components' monitor requests
5512 May be used in sections : defaults | frontend | listen | backend
5513 yes | yes | yes | no
5514 Arguments :
5515 <uri> is the exact URI which we want to intercept to return HAProxy's
5516 health status instead of forwarding the request.
5517
5518 When an HTTP request referencing <uri> will be received on a frontend,
5519 HAProxy will not forward it nor log it, but instead will return either
5520 "HTTP/1.0 200 OK" or "HTTP/1.0 503 Service unavailable", depending on failure
5521 conditions defined with "monitor fail". This is normally enough for any
5522 front-end HTTP probe to detect that the service is UP and running without
5523 forwarding the request to a backend server. Note that the HTTP method, the
5524 version and all headers are ignored, but the request must at least be valid
5525 at the HTTP level. This keyword may only be used with an HTTP-mode frontend.
5526
Willy Tarreau721d8e02017-12-01 18:25:08 +01005527 Monitor requests are processed very early, just after the request is parsed
5528 and even before any "http-request" or "block" rulesets. The only rulesets
5529 applied before are the tcp-request ones. They cannot be logged either, and it
5530 is the intended purpose. They are only used to report HAProxy's health to an
5531 upper component, nothing more. However, it is possible to add any number of
5532 conditions using "monitor fail" and ACLs so that the result can be adjusted
5533 to whatever check can be imagined (most often the number of available servers
5534 in a backend).
Willy Tarreau2769aa02007-12-27 18:26:09 +01005535
5536 Example :
5537 # Use /haproxy_test to report haproxy's status
5538 frontend www
5539 mode http
5540 monitor-uri /haproxy_test
5541
5542 See also : "monitor fail", "monitor-net"
5543
Willy Tarreau0ba27502007-12-24 16:55:16 +01005544
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005545option abortonclose
5546no option abortonclose
5547 Enable or disable early dropping of aborted requests pending in queues.
5548 May be used in sections : defaults | frontend | listen | backend
5549 yes | no | yes | yes
5550 Arguments : none
5551
5552 In presence of very high loads, the servers will take some time to respond.
5553 The per-instance connection queue will inflate, and the response time will
5554 increase respective to the size of the queue times the average per-session
5555 response time. When clients will wait for more than a few seconds, they will
Willy Tarreau198a7442008-01-17 12:05:32 +01005556 often hit the "STOP" button on their browser, leaving a useless request in
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005557 the queue, and slowing down other users, and the servers as well, because the
5558 request will eventually be served, then aborted at the first error
5559 encountered while delivering the response.
5560
5561 As there is no way to distinguish between a full STOP and a simple output
5562 close on the client side, HTTP agents should be conservative and consider
5563 that the client might only have closed its output channel while waiting for
5564 the response. However, this introduces risks of congestion when lots of users
5565 do the same, and is completely useless nowadays because probably no client at
5566 all will close the session while waiting for the response. Some HTTP agents
Davor Ocelice9ed2812017-12-25 17:49:28 +01005567 support this behavior (Squid, Apache, HAProxy), and others do not (TUX, most
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005568 hardware-based load balancers). So the probability for a closed input channel
Willy Tarreau198a7442008-01-17 12:05:32 +01005569 to represent a user hitting the "STOP" button is close to 100%, and the risk
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005570 of being the single component to break rare but valid traffic is extremely
5571 low, which adds to the temptation to be able to abort a session early while
5572 still not served and not pollute the servers.
5573
Davor Ocelice9ed2812017-12-25 17:49:28 +01005574 In HAProxy, the user can choose the desired behavior using the option
5575 "abortonclose". By default (without the option) the behavior is HTTP
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005576 compliant and aborted requests will be served. But when the option is
5577 specified, a session with an incoming channel closed will be aborted while
5578 it is still possible, either pending in the queue for a connection slot, or
5579 during the connection establishment if the server has not yet acknowledged
5580 the connection request. This considerably reduces the queue size and the load
5581 on saturated servers when users are tempted to click on STOP, which in turn
Willy Tarreaud72758d2010-01-12 10:42:19 +01005582 reduces the response time for other users.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005583
5584 If this option has been enabled in a "defaults" section, it can be disabled
5585 in a specific instance by prepending the "no" keyword before it.
5586
5587 See also : "timeout queue" and server's "maxconn" and "maxqueue" parameters
5588
5589
Willy Tarreau4076a152009-04-02 15:18:36 +02005590option accept-invalid-http-request
5591no option accept-invalid-http-request
5592 Enable or disable relaxing of HTTP request parsing
5593 May be used in sections : defaults | frontend | listen | backend
5594 yes | yes | yes | no
5595 Arguments : none
5596
Willy Tarreau91852eb2015-05-01 13:26:00 +02005597 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005598 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005599 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005600 forbidden characters are essentially used to build attacks exploiting server
5601 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5602 server will emit invalid header names for whatever reason (configuration,
5603 implementation) and the issue will not be immediately fixed. In such a case,
5604 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau422246e2012-01-07 23:54:13 +01005605 even if that does not make sense, by specifying this option. Similarly, the
5606 list of characters allowed to appear in a URI is well defined by RFC3986, and
5607 chars 0-31, 32 (space), 34 ('"'), 60 ('<'), 62 ('>'), 92 ('\'), 94 ('^'), 96
5608 ('`'), 123 ('{'), 124 ('|'), 125 ('}'), 127 (delete) and anything above are
Davor Ocelice9ed2812017-12-25 17:49:28 +01005609 not allowed at all. HAProxy always blocks a number of them (0..32, 127). The
Willy Tarreau91852eb2015-05-01 13:26:00 +02005610 remaining ones are blocked by default unless this option is enabled. This
Willy Tarreau13317662015-05-01 13:47:08 +02005611 option also relaxes the test on the HTTP version, it allows HTTP/0.9 requests
5612 to pass through (no version specified) and multiple digits for both the major
5613 and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005614
5615 This option should never be enabled by default as it hides application bugs
5616 and open security breaches. It should only be deployed after a problem has
5617 been confirmed.
5618
5619 When this option is enabled, erroneous header names will still be accepted in
5620 requests, but the complete request will be captured in order to permit later
Willy Tarreau422246e2012-01-07 23:54:13 +01005621 analysis using the "show errors" request on the UNIX stats socket. Similarly,
5622 requests containing invalid chars in the URI part will be logged. Doing this
Willy Tarreau4076a152009-04-02 15:18:36 +02005623 also helps confirming that the issue has been solved.
5624
5625 If this option has been enabled in a "defaults" section, it can be disabled
5626 in a specific instance by prepending the "no" keyword before it.
5627
5628 See also : "option accept-invalid-http-response" and "show errors" on the
5629 stats socket.
5630
5631
5632option accept-invalid-http-response
5633no option accept-invalid-http-response
5634 Enable or disable relaxing of HTTP response parsing
5635 May be used in sections : defaults | frontend | listen | backend
5636 yes | no | yes | yes
5637 Arguments : none
5638
Willy Tarreau91852eb2015-05-01 13:26:00 +02005639 By default, HAProxy complies with RFC7230 in terms of message parsing. This
Willy Tarreau4076a152009-04-02 15:18:36 +02005640 means that invalid characters in header names are not permitted and cause an
Davor Ocelice9ed2812017-12-25 17:49:28 +01005641 error to be returned to the client. This is the desired behavior as such
Willy Tarreau4076a152009-04-02 15:18:36 +02005642 forbidden characters are essentially used to build attacks exploiting server
5643 weaknesses, and bypass security filtering. Sometimes, a buggy browser or
5644 server will emit invalid header names for whatever reason (configuration,
5645 implementation) and the issue will not be immediately fixed. In such a case,
5646 it is possible to relax HAProxy's header name parser to accept any character
Willy Tarreau91852eb2015-05-01 13:26:00 +02005647 even if that does not make sense, by specifying this option. This option also
5648 relaxes the test on the HTTP version format, it allows multiple digits for
5649 both the major and the minor version.
Willy Tarreau4076a152009-04-02 15:18:36 +02005650
5651 This option should never be enabled by default as it hides application bugs
5652 and open security breaches. It should only be deployed after a problem has
5653 been confirmed.
5654
5655 When this option is enabled, erroneous header names will still be accepted in
5656 responses, but the complete response will be captured in order to permit
5657 later analysis using the "show errors" request on the UNIX stats socket.
5658 Doing this also helps confirming that the issue has been solved.
5659
5660 If this option has been enabled in a "defaults" section, it can be disabled
5661 in a specific instance by prepending the "no" keyword before it.
5662
5663 See also : "option accept-invalid-http-request" and "show errors" on the
5664 stats socket.
5665
5666
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005667option allbackups
5668no option allbackups
5669 Use either all backup servers at a time or only the first one
5670 May be used in sections : defaults | frontend | listen | backend
5671 yes | no | yes | yes
5672 Arguments : none
5673
5674 By default, the first operational backup server gets all traffic when normal
5675 servers are all down. Sometimes, it may be preferred to use multiple backups
5676 at once, because one will not be enough. When "option allbackups" is enabled,
5677 the load balancing will be performed among all backup servers when all normal
5678 ones are unavailable. The same load balancing algorithm will be used and the
5679 servers' weights will be respected. Thus, there will not be any priority
5680 order between the backup servers anymore.
5681
5682 This option is mostly used with static server farms dedicated to return a
5683 "sorry" page when an application is completely offline.
5684
5685 If this option has been enabled in a "defaults" section, it can be disabled
5686 in a specific instance by prepending the "no" keyword before it.
5687
5688
5689option checkcache
5690no option checkcache
Godbach7056a352013-12-11 20:01:07 +08005691 Analyze all server responses and block responses with cacheable cookies
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005692 May be used in sections : defaults | frontend | listen | backend
5693 yes | no | yes | yes
5694 Arguments : none
5695
5696 Some high-level frameworks set application cookies everywhere and do not
5697 always let enough control to the developer to manage how the responses should
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005698 be cached. When a session cookie is returned on a cacheable object, there is a
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005699 high risk of session crossing or stealing between users traversing the same
5700 caches. In some situations, it is better to block the response than to let
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02005701 some sensitive session information go in the wild.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005702
5703 The option "checkcache" enables deep inspection of all server responses for
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005704 strict compliance with HTTP specification in terms of cacheability. It
Willy Tarreau198a7442008-01-17 12:05:32 +01005705 carefully checks "Cache-control", "Pragma" and "Set-cookie" headers in server
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005706 response to check if there's a risk of caching a cookie on a client-side
5707 proxy. When this option is enabled, the only responses which can be delivered
Willy Tarreau198a7442008-01-17 12:05:32 +01005708 to the client are :
Davor Ocelice9ed2812017-12-25 17:49:28 +01005709 - all those without "Set-Cookie" header;
Willy Tarreauc55ddce2017-12-21 11:41:38 +01005710 - all those with a return code other than 200, 203, 204, 206, 300, 301,
5711 404, 405, 410, 414, 501, provided that the server has not set a
Davor Ocelice9ed2812017-12-25 17:49:28 +01005712 "Cache-control: public" header field;
Willy Tarreau24ea0bc2017-12-21 11:32:55 +01005713 - all those that result from a request using a method other than GET, HEAD,
5714 OPTIONS, TRACE, provided that the server has not set a 'Cache-Control:
Davor Ocelice9ed2812017-12-25 17:49:28 +01005715 public' header field;
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005716 - those with a 'Pragma: no-cache' header
5717 - those with a 'Cache-control: private' header
5718 - those with a 'Cache-control: no-store' header
5719 - those with a 'Cache-control: max-age=0' header
5720 - those with a 'Cache-control: s-maxage=0' header
5721 - those with a 'Cache-control: no-cache' header
5722 - those with a 'Cache-control: no-cache="set-cookie"' header
5723 - those with a 'Cache-control: no-cache="set-cookie,' header
5724 (allowing other fields after set-cookie)
5725
5726 If a response doesn't respect these requirements, then it will be blocked
Willy Tarreau198a7442008-01-17 12:05:32 +01005727 just as if it was from an "rspdeny" filter, with an "HTTP 502 bad gateway".
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005728 The session state shows "PH--" meaning that the proxy blocked the response
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01005729 during headers processing. Additionally, an alert will be sent in the logs so
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005730 that admins are informed that there's something to be fixed.
5731
5732 Due to the high impact on the application, the application should be tested
5733 in depth with the option enabled before going to production. It is also a
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01005734 good practice to always activate it during tests, even if it is not used in
Davor Ocelice9ed2812017-12-25 17:49:28 +01005735 production, as it will report potentially dangerous application behaviors.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005736
5737 If this option has been enabled in a "defaults" section, it can be disabled
5738 in a specific instance by prepending the "no" keyword before it.
5739
5740
5741option clitcpka
5742no option clitcpka
5743 Enable or disable the sending of TCP keepalive packets on the client side
5744 May be used in sections : defaults | frontend | listen | backend
5745 yes | yes | yes | no
5746 Arguments : none
5747
5748 When there is a firewall or any session-aware component between a client and
5749 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01005750 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005751 components decides to expire a session which has remained idle for too long.
5752
5753 Enabling socket-level TCP keep-alives makes the system regularly send packets
5754 to the other end of the connection, leaving it active. The delay between
5755 keep-alive probes is controlled by the system only and depends both on the
5756 operating system and its tuning parameters.
5757
5758 It is important to understand that keep-alive packets are neither emitted nor
5759 received at the application level. It is only the network stacks which sees
5760 them. For this reason, even if one side of the proxy already uses keep-alives
5761 to maintain its connection alive, those keep-alive packets will not be
5762 forwarded to the other side of the proxy.
5763
5764 Please note that this has nothing to do with HTTP keep-alive.
5765
5766 Using option "clitcpka" enables the emission of TCP keep-alive probes on the
5767 client side of a connection, which should help when session expirations are
5768 noticed between HAProxy and a client.
5769
5770 If this option has been enabled in a "defaults" section, it can be disabled
5771 in a specific instance by prepending the "no" keyword before it.
5772
5773 See also : "option srvtcpka", "option tcpka"
5774
5775
Willy Tarreau0ba27502007-12-24 16:55:16 +01005776option contstats
5777 Enable continuous traffic statistics updates
5778 May be used in sections : defaults | frontend | listen | backend
5779 yes | yes | yes | no
5780 Arguments : none
5781
5782 By default, counters used for statistics calculation are incremented
5783 only when a session finishes. It works quite well when serving small
5784 objects, but with big ones (for example large images or archives) or
5785 with A/V streaming, a graph generated from haproxy counters looks like
Willy Tarreaudef0d222016-11-08 22:03:00 +01005786 a hedgehog. With this option enabled counters get incremented frequently
5787 along the session, typically every 5 seconds, which is often enough to
5788 produce clean graphs. Recounting touches a hotpath directly so it is not
5789 not enabled by default, as it can cause a lot of wakeups for very large
5790 session counts and cause a small performance drop.
Willy Tarreau0ba27502007-12-24 16:55:16 +01005791
5792
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005793option dontlog-normal
5794no option dontlog-normal
5795 Enable or disable logging of normal, successful connections
5796 May be used in sections : defaults | frontend | listen | backend
5797 yes | yes | yes | no
5798 Arguments : none
5799
5800 There are large sites dealing with several thousand connections per second
5801 and for which logging is a major pain. Some of them are even forced to turn
5802 logs off and cannot debug production issues. Setting this option ensures that
5803 normal connections, those which experience no error, no timeout, no retry nor
5804 redispatch, will not be logged. This leaves disk space for anomalies. In HTTP
5805 mode, the response status code is checked and return codes 5xx will still be
5806 logged.
5807
5808 It is strongly discouraged to use this option as most of the time, the key to
5809 complex issues is in the normal logs which will not be logged here. If you
5810 need to separate logs, see the "log-separate-errors" option instead.
5811
Willy Tarreauc57f0e22009-05-10 13:12:33 +02005812 See also : "log", "dontlognull", "log-separate-errors" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02005813 logging.
5814
5815
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005816option dontlognull
5817no option dontlognull
5818 Enable or disable logging of null connections
5819 May be used in sections : defaults | frontend | listen | backend
5820 yes | yes | yes | no
5821 Arguments : none
5822
5823 In certain environments, there are components which will regularly connect to
5824 various systems to ensure that they are still alive. It can be the case from
5825 another load balancer as well as from monitoring systems. By default, even a
5826 simple port probe or scan will produce a log. If those connections pollute
5827 the logs too much, it is possible to enable option "dontlognull" to indicate
5828 that a connection on which no data has been transferred will not be logged,
Willy Tarreau0f228a02015-05-01 15:37:53 +02005829 which typically corresponds to those probes. Note that errors will still be
5830 returned to the client and accounted for in the stats. If this is not what is
5831 desired, option http-ignore-probes can be used instead.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005832
5833 It is generally recommended not to use this option in uncontrolled
Davor Ocelice9ed2812017-12-25 17:49:28 +01005834 environments (e.g. internet), otherwise scans and other malicious activities
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005835 would not be logged.
5836
5837 If this option has been enabled in a "defaults" section, it can be disabled
5838 in a specific instance by prepending the "no" keyword before it.
5839
Willy Tarreau0f228a02015-05-01 15:37:53 +02005840 See also : "log", "http-ignore-probes", "monitor-net", "monitor-uri", and
5841 section 8 about logging.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005842
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005843
Christopher Faulet315b39c2018-09-21 16:26:19 +02005844option forceclose (deprecated)
5845no option forceclose (deprecated)
5846 This is an alias for "option httpclose". Thus this option is deprecated.
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005847
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005848 See also : "option httpclose" and "option http-pretend-keepalive"
Willy Tarreaubf1f8162007-12-28 17:42:56 +01005849
5850
Willy Tarreau87cf5142011-08-19 22:57:24 +02005851option forwardfor [ except <network> ] [ header <name> ] [ if-none ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01005852 Enable insertion of the X-Forwarded-For header to requests sent to servers
5853 May be used in sections : defaults | frontend | listen | backend
5854 yes | yes | yes | yes
5855 Arguments :
5856 <network> is an optional argument used to disable this option for sources
5857 matching <network>
Ross Westaf72a1d2008-08-03 10:51:45 +02005858 <name> an optional argument to specify a different "X-Forwarded-For"
Willy Tarreaud72758d2010-01-12 10:42:19 +01005859 header name.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005860
5861 Since HAProxy works in reverse-proxy mode, the servers see its IP address as
5862 their client address. This is sometimes annoying when the client's IP address
5863 is expected in server logs. To solve this problem, the well-known HTTP header
5864 "X-Forwarded-For" may be added by HAProxy to all requests sent to the server.
5865 This header contains a value representing the client's IP address. Since this
5866 header is always appended at the end of the existing header list, the server
5867 must be configured to always use the last occurrence of this header only. See
Ross Westaf72a1d2008-08-03 10:51:45 +02005868 the server's manual to find how to enable use of this standard header. Note
5869 that only the last occurrence of the header must be used, since it is really
5870 possible that the client has already brought one.
5871
Willy Tarreaud72758d2010-01-12 10:42:19 +01005872 The keyword "header" may be used to supply a different header name to replace
Ross Westaf72a1d2008-08-03 10:51:45 +02005873 the default "X-Forwarded-For". This can be useful where you might already
Davor Ocelice9ed2812017-12-25 17:49:28 +01005874 have a "X-Forwarded-For" header from a different application (e.g. stunnel),
Willy Tarreaud72758d2010-01-12 10:42:19 +01005875 and you need preserve it. Also if your backend server doesn't use the
Davor Ocelice9ed2812017-12-25 17:49:28 +01005876 "X-Forwarded-For" header and requires different one (e.g. Zeus Web Servers
Ross Westaf72a1d2008-08-03 10:51:45 +02005877 require "X-Cluster-Client-IP").
Willy Tarreauc27debf2008-01-06 08:57:02 +01005878
5879 Sometimes, a same HAProxy instance may be shared between a direct client
5880 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
5881 used to decrypt HTTPS traffic). It is possible to disable the addition of the
5882 header for a known source address or network by adding the "except" keyword
5883 followed by the network address. In this case, any source IP matching the
5884 network will not cause an addition of this header. Most common uses are with
5885 private networks or 127.0.0.1.
5886
Willy Tarreau87cf5142011-08-19 22:57:24 +02005887 Alternatively, the keyword "if-none" states that the header will only be
5888 added if it is not present. This should only be used in perfectly trusted
5889 environment, as this might cause a security issue if headers reaching haproxy
5890 are under the control of the end-user.
5891
Willy Tarreauc27debf2008-01-06 08:57:02 +01005892 This option may be specified either in the frontend or in the backend. If at
Ross Westaf72a1d2008-08-03 10:51:45 +02005893 least one of them uses it, the header will be added. Note that the backend's
5894 setting of the header subargument takes precedence over the frontend's if
Willy Tarreau87cf5142011-08-19 22:57:24 +02005895 both are defined. In the case of the "if-none" argument, if at least one of
5896 the frontend or the backend does not specify it, it wants the addition to be
5897 mandatory, so it wins.
Willy Tarreauc27debf2008-01-06 08:57:02 +01005898
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +02005899 Example :
Willy Tarreauc27debf2008-01-06 08:57:02 +01005900 # Public HTTP address also used by stunnel on the same machine
5901 frontend www
5902 mode http
5903 option forwardfor except 127.0.0.1 # stunnel already adds the header
5904
Ross Westaf72a1d2008-08-03 10:51:45 +02005905 # Those servers want the IP Address in X-Client
5906 backend www
5907 mode http
5908 option forwardfor header X-Client
5909
Willy Tarreau87cf5142011-08-19 22:57:24 +02005910 See also : "option httpclose", "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02005911 "option http-keep-alive"
Willy Tarreauc27debf2008-01-06 08:57:02 +01005912
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02005913
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005914option http-buffer-request
5915no option http-buffer-request
5916 Enable or disable waiting for whole HTTP request body before proceeding
5917 May be used in sections : defaults | frontend | listen | backend
5918 yes | yes | yes | yes
5919 Arguments : none
5920
5921 It is sometimes desirable to wait for the body of an HTTP request before
5922 taking a decision. This is what is being done by "balance url_param" for
5923 example. The first use case is to buffer requests from slow clients before
5924 connecting to the server. Another use case consists in taking the routing
5925 decision based on the request body's contents. This option placed in a
5926 frontend or backend forces the HTTP processing to wait until either the whole
5927 body is received, or the request buffer is full, or the first chunk is
5928 complete in case of chunked encoding. It can have undesired side effects with
Tim Düsterhus4896c442016-11-29 02:15:19 +01005929 some applications abusing HTTP by expecting unbuffered transmissions between
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005930 the frontend and the backend, so this should definitely not be used by
5931 default.
5932
Baptiste Assmanneccdf432015-10-28 13:49:01 +01005933 See also : "option http-no-delay", "timeout http-request"
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02005934
5935
Willy Tarreau0f228a02015-05-01 15:37:53 +02005936option http-ignore-probes
5937no option http-ignore-probes
5938 Enable or disable logging of null connections and request timeouts
5939 May be used in sections : defaults | frontend | listen | backend
5940 yes | yes | yes | no
5941 Arguments : none
5942
5943 Recently some browsers started to implement a "pre-connect" feature
5944 consisting in speculatively connecting to some recently visited web sites
5945 just in case the user would like to visit them. This results in many
5946 connections being established to web sites, which end up in 408 Request
5947 Timeout if the timeout strikes first, or 400 Bad Request when the browser
5948 decides to close them first. These ones pollute the log and feed the error
5949 counters. There was already "option dontlognull" but it's insufficient in
5950 this case. Instead, this option does the following things :
5951 - prevent any 400/408 message from being sent to the client if nothing
Davor Ocelice9ed2812017-12-25 17:49:28 +01005952 was received over a connection before it was closed;
5953 - prevent any log from being emitted in this situation;
Willy Tarreau0f228a02015-05-01 15:37:53 +02005954 - prevent any error counter from being incremented
5955
5956 That way the empty connection is silently ignored. Note that it is better
5957 not to use this unless it is clear that it is needed, because it will hide
5958 real problems. The most common reason for not receiving a request and seeing
5959 a 408 is due to an MTU inconsistency between the client and an intermediary
5960 element such as a VPN, which blocks too large packets. These issues are
5961 generally seen with POST requests as well as GET with large cookies. The logs
5962 are often the only way to detect them.
5963
5964 If this option has been enabled in a "defaults" section, it can be disabled
5965 in a specific instance by prepending the "no" keyword before it.
5966
5967 See also : "log", "dontlognull", "errorfile", and section 8 about logging.
5968
5969
Willy Tarreau16bfb022010-01-16 19:48:41 +01005970option http-keep-alive
5971no option http-keep-alive
5972 Enable or disable HTTP keep-alive from client to server
5973 May be used in sections : defaults | frontend | listen | backend
5974 yes | yes | yes | yes
5975 Arguments : none
5976
Willy Tarreau70dffda2014-01-30 03:07:23 +01005977 By default HAProxy operates in keep-alive mode with regards to persistent
5978 connections: for each connection it processes each request and response, and
Christopher Faulet315b39c2018-09-21 16:26:19 +02005979 leaves the connection idle on both sides between the end of a response and
5980 the start of a new request. This mode may be changed by several options such
5981 as "option http-server-close", "option httpclose" or "option http-tunnel".
5982 This option allows to set back the keep-alive mode, which can be useful when
5983 another mode was used in a defaults section.
Willy Tarreau70dffda2014-01-30 03:07:23 +01005984
5985 Setting "option http-keep-alive" enables HTTP keep-alive mode on the client-
5986 and server- sides. This provides the lowest latency on the client side (slow
Willy Tarreau16bfb022010-01-16 19:48:41 +01005987 network) and the fastest session reuse on the server side at the expense
5988 of maintaining idle connections to the servers. In general, it is possible
5989 with this option to achieve approximately twice the request rate that the
5990 "http-server-close" option achieves on small objects. There are mainly two
5991 situations where this option may be useful :
5992
5993 - when the server is non-HTTP compliant and authenticates the connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01005994 instead of requests (e.g. NTLM authentication)
Willy Tarreau16bfb022010-01-16 19:48:41 +01005995
5996 - when the cost of establishing the connection to the server is significant
5997 compared to the cost of retrieving the associated object from the server.
5998
5999 This last case can happen when the server is a fast static server of cache.
6000 In this case, the server will need to be properly tuned to support high enough
6001 connection counts because connections will last until the client sends another
6002 request.
6003
6004 If the client request has to go to another backend or another server due to
6005 content switching or the load balancing algorithm, the idle connection will
Willy Tarreau9420b122013-12-15 18:58:25 +01006006 immediately be closed and a new one re-opened. Option "prefer-last-server" is
6007 available to try optimize server selection so that if the server currently
6008 attached to an idle connection is usable, it will be used.
Willy Tarreau16bfb022010-01-16 19:48:41 +01006009
Willy Tarreau16bfb022010-01-16 19:48:41 +01006010 At the moment, logs will not indicate whether requests came from the same
6011 session or not. The accept date reported in the logs corresponds to the end
6012 of the previous request, and the request time corresponds to the time spent
6013 waiting for a new request. The keep-alive request time is still bound to the
6014 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6015 not set.
6016
Cyril Bonté653dcd62014-02-20 00:13:15 +01006017 This option disables and replaces any previous "option httpclose", "option
Christopher Faulet315b39c2018-09-21 16:26:19 +02006018 http-server-close" or "option http-tunnel". When backend and frontend options
6019 differ, all of these 4 options have precedence over "option http-keep-alive".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006020
Christopher Faulet315b39c2018-09-21 16:26:19 +02006021 See also : "option httpclose",, "option http-server-close",
Willy Tarreau9420b122013-12-15 18:58:25 +01006022 "option prefer-last-server", "option http-pretend-keepalive",
Frédéric Lécaille93d33162019-03-06 09:35:59 +01006023 and "1.1. The HTTP transaction model".
Willy Tarreau16bfb022010-01-16 19:48:41 +01006024
6025
Willy Tarreau96e31212011-05-30 18:10:30 +02006026option http-no-delay
6027no option http-no-delay
6028 Instruct the system to favor low interactive delays over performance in HTTP
6029 May be used in sections : defaults | frontend | listen | backend
6030 yes | yes | yes | yes
6031 Arguments : none
6032
6033 In HTTP, each payload is unidirectional and has no notion of interactivity.
6034 Any agent is expected to queue data somewhat for a reasonably low delay.
6035 There are some very rare server-to-server applications that abuse the HTTP
6036 protocol and expect the payload phase to be highly interactive, with many
6037 interleaved data chunks in both directions within a single request. This is
6038 absolutely not supported by the HTTP specification and will not work across
6039 most proxies or servers. When such applications attempt to do this through
6040 haproxy, it works but they will experience high delays due to the network
6041 optimizations which favor performance by instructing the system to wait for
6042 enough data to be available in order to only send full packets. Typical
6043 delays are around 200 ms per round trip. Note that this only happens with
6044 abnormal uses. Normal uses such as CONNECT requests nor WebSockets are not
6045 affected.
6046
6047 When "option http-no-delay" is present in either the frontend or the backend
6048 used by a connection, all such optimizations will be disabled in order to
6049 make the exchanges as fast as possible. Of course this offers no guarantee on
6050 the functionality, as it may break at any other place. But if it works via
6051 HAProxy, it will work as fast as possible. This option should never be used
6052 by default, and should never be used at all unless such a buggy application
6053 is discovered. The impact of using this option is an increase of bandwidth
6054 usage and CPU usage, which may significantly lower performance in high
6055 latency environments.
6056
Willy Tarreau9fbe18e2015-05-01 22:42:08 +02006057 See also : "option http-buffer-request"
6058
Willy Tarreau96e31212011-05-30 18:10:30 +02006059
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006060option http-pretend-keepalive
6061no option http-pretend-keepalive
6062 Define whether haproxy will announce keepalive to the server or not
6063 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet98db9762018-09-21 10:25:19 +02006064 yes | no | yes | yes
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006065 Arguments : none
6066
Christopher Faulet315b39c2018-09-21 16:26:19 +02006067 When running with "option http-server-close" or "option httpclose", haproxy
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006068 adds a "Connection: close" header to the request forwarded to the server.
6069 Unfortunately, when some servers see this header, they automatically refrain
6070 from using the chunked encoding for responses of unknown length, while this
6071 is totally unrelated. The immediate effect is that this prevents haproxy from
6072 maintaining the client connection alive. A second effect is that a client or
6073 a cache could receive an incomplete response without being aware of it, and
6074 consider the response complete.
6075
6076 By setting "option http-pretend-keepalive", haproxy will make the server
6077 believe it will keep the connection alive. The server will then not fall back
6078 to the abnormal undesired above. When haproxy gets the whole response, it
6079 will close the connection with the server just as it would do with the
Christopher Faulet315b39c2018-09-21 16:26:19 +02006080 "option httpclose". That way the client gets a normal response and the
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006081 connection is correctly closed on the server side.
6082
6083 It is recommended not to enable this option by default, because most servers
6084 will more efficiently close the connection themselves after the last packet,
6085 and release its buffers slightly earlier. Also, the added packet on the
6086 network could slightly reduce the overall peak performance. However it is
6087 worth noting that when this option is enabled, haproxy will have slightly
6088 less work to do. So if haproxy is the bottleneck on the whole architecture,
6089 enabling this option might save a few CPU cycles.
6090
Christopher Faulet98db9762018-09-21 10:25:19 +02006091 This option may be set in backend and listen sections. Using it in a frontend
6092 section will be ignored and a warning will be reported during startup. It is
6093 a backend related option, so there is no real reason to set it on a
6094 frontend. This option may be combined with "option httpclose", which will
6095 cause keepalive to be announced to the server and close to be announced to
6096 the client. This practice is discouraged though.
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006097
6098 If this option has been enabled in a "defaults" section, it can be disabled
6099 in a specific instance by prepending the "no" keyword before it.
6100
Christopher Faulet315b39c2018-09-21 16:26:19 +02006101 See also : "option httpclose", "option http-server-close", and
Willy Tarreau16bfb022010-01-16 19:48:41 +01006102 "option http-keep-alive"
Willy Tarreau8a8e1d92010-04-05 16:15:16 +02006103
Willy Tarreauc27debf2008-01-06 08:57:02 +01006104
Willy Tarreaub608feb2010-01-02 22:47:18 +01006105option http-server-close
6106no option http-server-close
6107 Enable or disable HTTP connection closing on the server side
6108 May be used in sections : defaults | frontend | listen | backend
6109 yes | yes | yes | yes
6110 Arguments : none
6111
Willy Tarreau70dffda2014-01-30 03:07:23 +01006112 By default HAProxy operates in keep-alive mode with regards to persistent
6113 connections: for each connection it processes each request and response, and
6114 leaves the connection idle on both sides between the end of a response and
6115 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006116 as "option http-server-close", "option httpclose" or "option http-tunnel".
6117 Setting "option http-server-close" enables HTTP connection-close mode on the
6118 server side while keeping the ability to support HTTP keep-alive and
6119 pipelining on the client side. This provides the lowest latency on the client
6120 side (slow network) and the fastest session reuse on the server side to save
6121 server resources, similarly to "option httpclose". It also permits
6122 non-keepalive capable servers to be served in keep-alive mode to the clients
6123 if they conform to the requirements of RFC7230. Please note that some servers
6124 do not always conform to those requirements when they see "Connection: close"
6125 in the request. The effect will be that keep-alive will never be used. A
6126 workaround consists in enabling "option http-pretend-keepalive".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006127
6128 At the moment, logs will not indicate whether requests came from the same
6129 session or not. The accept date reported in the logs corresponds to the end
6130 of the previous request, and the request time corresponds to the time spent
6131 waiting for a new request. The keep-alive request time is still bound to the
Willy Tarreaub16a5742010-01-10 14:46:16 +01006132 timeout defined by "timeout http-keep-alive" or "timeout http-request" if
6133 not set.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006134
6135 This option may be set both in a frontend and in a backend. It is enabled if
6136 at least one of the frontend or backend holding a connection has it enabled.
Christopher Faulet315b39c2018-09-21 16:26:19 +02006137 It disables and replaces any previous "option httpclose", "option http-tunnel"
6138 or "option http-keep-alive". Please check section 4 ("Proxies") to see how
6139 this option combines with others when frontend and backend options differ.
Willy Tarreaub608feb2010-01-02 22:47:18 +01006140
6141 If this option has been enabled in a "defaults" section, it can be disabled
6142 in a specific instance by prepending the "no" keyword before it.
6143
Christopher Faulet315b39c2018-09-21 16:26:19 +02006144 See also : "option httpclose", "option http-pretend-keepalive",
6145 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreaub608feb2010-01-02 22:47:18 +01006146
6147
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006148option http-tunnel
6149no option http-tunnel
6150 Disable or enable HTTP connection processing after first transaction
6151 May be used in sections : defaults | frontend | listen | backend
Christopher Faulet4212a302018-09-21 10:42:19 +02006152 yes | yes | yes | no
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006153 Arguments : none
6154
Willy Tarreau70dffda2014-01-30 03:07:23 +01006155 By default HAProxy operates in keep-alive mode with regards to persistent
6156 connections: for each connection it processes each request and response, and
6157 leaves the connection idle on both sides between the end of a response and
6158 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006159 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006160
6161 Option "http-tunnel" disables any HTTP processing past the first request and
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03006162 the first response. This is the mode which was used by default in versions
Willy Tarreau70dffda2014-01-30 03:07:23 +01006163 1.0 to 1.5-dev21. It is the mode with the lowest processing overhead, which
6164 is normally not needed anymore unless in very specific cases such as when
6165 using an in-house protocol that looks like HTTP but is not compatible, or
6166 just to log one request per client in order to reduce log size. Note that
6167 everything which works at the HTTP level, including header parsing/addition,
6168 cookie processing or content switching will only work for the first request
6169 and will be ignored after the first response.
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006170
Christopher Faulet4212a302018-09-21 10:42:19 +02006171 This option may be set on frontend and listen sections. Using it on a backend
6172 section will be ignored and a warning will be reported during the startup. It
6173 is a frontend related option, so there is no real reason to set it on a
6174 backend.
6175
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006176 If this option has been enabled in a "defaults" section, it can be disabled
6177 in a specific instance by prepending the "no" keyword before it.
6178
Christopher Faulet315b39c2018-09-21 16:26:19 +02006179 See also : "option httpclose", "option http-server-close",
6180 "option http-keep-alive", and "1.1. The HTTP transaction model".
Willy Tarreau02bce8b2014-01-30 00:15:28 +01006181
6182
Willy Tarreau88d349d2010-01-25 12:15:43 +01006183option http-use-proxy-header
Cyril Bontéf0c60612010-02-06 14:44:47 +01006184no option http-use-proxy-header
Willy Tarreau88d349d2010-01-25 12:15:43 +01006185 Make use of non-standard Proxy-Connection header instead of Connection
6186 May be used in sections : defaults | frontend | listen | backend
6187 yes | yes | yes | no
6188 Arguments : none
6189
Lukas Tribus23953682017-04-28 13:24:30 +00006190 While RFC7230 explicitly states that HTTP/1.1 agents must use the
Willy Tarreau88d349d2010-01-25 12:15:43 +01006191 Connection header to indicate their wish of persistent or non-persistent
6192 connections, both browsers and proxies ignore this header for proxied
6193 connections and make use of the undocumented, non-standard Proxy-Connection
6194 header instead. The issue begins when trying to put a load balancer between
6195 browsers and such proxies, because there will be a difference between what
6196 haproxy understands and what the client and the proxy agree on.
6197
6198 By setting this option in a frontend, haproxy can automatically switch to use
6199 that non-standard header if it sees proxied requests. A proxied request is
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006200 defined here as one where the URI begins with neither a '/' nor a '*'. This
6201 is incompatible with the HTTP tunnel mode. Note that this option can only be
6202 specified in a frontend and will affect the request along its whole life.
Willy Tarreau88d349d2010-01-25 12:15:43 +01006203
Willy Tarreau844a7e72010-01-31 21:46:18 +01006204 Also, when this option is set, a request which requires authentication will
6205 automatically switch to use proxy authentication headers if it is itself a
6206 proxied request. That makes it possible to check or enforce authentication in
6207 front of an existing proxy.
6208
Willy Tarreau88d349d2010-01-25 12:15:43 +01006209 This option should normally never be used, except in front of a proxy.
6210
Christopher Faulet315b39c2018-09-21 16:26:19 +02006211 See also : "option httpclose", and "option http-server-close".
Willy Tarreau88d349d2010-01-25 12:15:43 +01006212
6213
Willy Tarreau68ad3a42018-10-22 11:49:15 +02006214option http-use-htx
6215no option http-use-htx
6216 Switch to the new HTX internal representation for HTTP protocol elements
6217 May be used in sections : defaults | frontend | listen | backend
6218 yes | yes | yes | yes
6219 Arguments : none
6220
6221 By default, the HTTP protocol is processed as-is. Inserting, deleting, or
6222 modifying a header field requires to rewrite the affected part in the buffer
6223 and to move the buffer's tail accordingly. Since this principle has deep
6224 roots in haproxy, the HTTP/2 protocol is converted to HTTP/1.1 before being
6225 processed this way. It also results in the inability to establish HTTP/2
6226 connections to servers because of the loss of HTTP/2 semantics in the HTTP/1
6227 representation.
6228
6229 HTX is the name of a totally new native internal representation for the HTTP
6230 protocol, that is agnostic to the version and aims at preserving semantics
6231 all along the chain. It relies on a fast parsing, tokenizing and indexing of
6232 the protocol elements so that no more memory moves are necessary and that
6233 most elements are directly accessed. This mechanism is still limited to the
6234 most basic operations (no compression, filters, Lua, applets, cache, etc).
6235 But it supports using either HTTP/1 or HTTP/2 on any side regardless of the
6236 other side's version.
6237
6238 This option indicates that HTX needs to be used. It will cause errors to be
6239 emitted if incompatible features are used, but will allow H2 to be selected
6240 as a server protocol. It is recommended to use this option on new reasonably
6241 simple configurations, but since the feature still has incomplete functional
6242 coverage, it is not enabled by default.
6243
6244 See also : "mode http"
6245
6246
Willy Tarreaud63335a2010-02-26 12:56:52 +01006247option httpchk
6248option httpchk <uri>
6249option httpchk <method> <uri>
6250option httpchk <method> <uri> <version>
6251 Enable HTTP protocol to check on the servers health
6252 May be used in sections : defaults | frontend | listen | backend
6253 yes | no | yes | yes
6254 Arguments :
6255 <method> is the optional HTTP method used with the requests. When not set,
6256 the "OPTIONS" method is used, as it generally requires low server
6257 processing and is easy to filter out from the logs. Any method
6258 may be used, though it is not recommended to invent non-standard
6259 ones.
6260
6261 <uri> is the URI referenced in the HTTP requests. It defaults to " / "
6262 which is accessible by default on almost any server, but may be
6263 changed to any other URI. Query strings are permitted.
6264
6265 <version> is the optional HTTP version string. It defaults to "HTTP/1.0"
6266 but some servers might behave incorrectly in HTTP 1.0, so turning
6267 it to HTTP/1.1 may sometimes help. Note that the Host field is
6268 mandatory in HTTP/1.1, and as a trick, it is possible to pass it
6269 after "\r\n" following the version string.
6270
6271 By default, server health checks only consist in trying to establish a TCP
6272 connection. When "option httpchk" is specified, a complete HTTP request is
6273 sent once the TCP connection is established, and responses 2xx and 3xx are
6274 considered valid, while all other ones indicate a server failure, including
6275 the lack of any response.
6276
6277 The port and interval are specified in the server configuration.
6278
6279 This option does not necessarily require an HTTP backend, it also works with
6280 plain TCP backends. This is particularly useful to check simple scripts bound
6281 to some dedicated ports using the inetd daemon.
6282
6283 Examples :
6284 # Relay HTTPS traffic to Apache instance and check service availability
6285 # using HTTP request "OPTIONS * HTTP/1.1" on port 80.
6286 backend https_relay
6287 mode tcp
6288 option httpchk OPTIONS * HTTP/1.1\r\nHost:\ www
6289 server apache1 192.168.1.1:443 check port 80
6290
Simon Hormanafc47ee2013-11-25 10:46:35 +09006291 See also : "option ssl-hello-chk", "option smtpchk", "option mysql-check",
6292 "option pgsql-check", "http-check" and the "check", "port" and
6293 "inter" server options.
Willy Tarreaud63335a2010-02-26 12:56:52 +01006294
6295
Willy Tarreauc27debf2008-01-06 08:57:02 +01006296option httpclose
6297no option httpclose
Christopher Faulet315b39c2018-09-21 16:26:19 +02006298 Enable or disable HTTP connection closing
Willy Tarreauc27debf2008-01-06 08:57:02 +01006299 May be used in sections : defaults | frontend | listen | backend
6300 yes | yes | yes | yes
6301 Arguments : none
6302
Willy Tarreau70dffda2014-01-30 03:07:23 +01006303 By default HAProxy operates in keep-alive mode with regards to persistent
6304 connections: for each connection it processes each request and response, and
6305 leaves the connection idle on both sides between the end of a response and
6306 the start of a new request. This mode may be changed by several options such
Christopher Faulet315b39c2018-09-21 16:26:19 +02006307 as "option http-server-close", "option httpclose" or "option http-tunnel".
Willy Tarreau70dffda2014-01-30 03:07:23 +01006308
Christopher Faulet315b39c2018-09-21 16:26:19 +02006309 If "option httpclose" is set, HAProxy will close connections with the server
6310 and the client as soon as the request and the response are received. It will
6311 alos check if a "Connection: close" header is already set in each direction,
6312 and will add one if missing. Any "Connection" header different from "close"
6313 will also be removed.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006314
Christopher Faulet315b39c2018-09-21 16:26:19 +02006315 This option may also be combined with "option http-pretend-keepalive", which
6316 will disable sending of the "Connection: close" header, but will still cause
6317 the connection to be closed once the whole response is received.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006318
6319 This option may be set both in a frontend and in a backend. It is enabled if
6320 at least one of the frontend or backend holding a connection has it enabled.
Cyril Bonté653dcd62014-02-20 00:13:15 +01006321 It disables and replaces any previous "option http-server-close",
Christopher Faulet315b39c2018-09-21 16:26:19 +02006322 "option http-keep-alive" or "option http-tunnel". Please check section 4
6323 ("Proxies") to see how this option combines with others when frontend and
6324 backend options differ.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006325
6326 If this option has been enabled in a "defaults" section, it can be disabled
6327 in a specific instance by prepending the "no" keyword before it.
6328
Christopher Faulet315b39c2018-09-21 16:26:19 +02006329 See also : "option http-server-close" and "1.1. The HTTP transaction model".
Willy Tarreauc27debf2008-01-06 08:57:02 +01006330
6331
Emeric Brun3a058f32009-06-30 18:26:00 +02006332option httplog [ clf ]
Willy Tarreauc27debf2008-01-06 08:57:02 +01006333 Enable logging of HTTP request, session state and timers
6334 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01006335 yes | yes | yes | no
Emeric Brun3a058f32009-06-30 18:26:00 +02006336 Arguments :
6337 clf if the "clf" argument is added, then the output format will be
6338 the CLF format instead of HAProxy's default HTTP format. You can
6339 use this when you need to feed HAProxy's logs through a specific
Davor Ocelice9ed2812017-12-25 17:49:28 +01006340 log analyzer which only support the CLF format and which is not
Emeric Brun3a058f32009-06-30 18:26:00 +02006341 extensible.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006342
6343 By default, the log output format is very poor, as it only contains the
6344 source and destination addresses, and the instance name. By specifying
6345 "option httplog", each log line turns into a much richer format including,
6346 but not limited to, the HTTP request, the connection timers, the session
6347 status, the connections numbers, the captured headers and cookies, the
6348 frontend, backend and server name, and of course the source address and
6349 ports.
6350
PiBa-NLbd556bf2014-12-11 21:31:54 +01006351 Specifying only "option httplog" will automatically clear the 'clf' mode
6352 if it was set by default.
Emeric Brun3a058f32009-06-30 18:26:00 +02006353
Guillaume de Lafond29f45602017-03-31 19:52:15 +02006354 "option httplog" overrides any previous "log-format" directive.
6355
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006356 See also : section 8 about logging.
Willy Tarreauc27debf2008-01-06 08:57:02 +01006357
Willy Tarreau55165fe2009-05-10 12:02:55 +02006358
6359option http_proxy
6360no option http_proxy
6361 Enable or disable plain HTTP proxy mode
6362 May be used in sections : defaults | frontend | listen | backend
6363 yes | yes | yes | yes
6364 Arguments : none
6365
6366 It sometimes happens that people need a pure HTTP proxy which understands
6367 basic proxy requests without caching nor any fancy feature. In this case,
6368 it may be worth setting up an HAProxy instance with the "option http_proxy"
6369 set. In this mode, no server is declared, and the connection is forwarded to
6370 the IP address and port found in the URL after the "http://" scheme.
6371
6372 No host address resolution is performed, so this only works when pure IP
6373 addresses are passed. Since this option's usage perimeter is rather limited,
Lukas Tribusf01a9cd2016-02-03 18:09:37 +01006374 it will probably be used only by experts who know they need exactly it. This
6375 is incompatible with the HTTP tunnel mode.
Willy Tarreau55165fe2009-05-10 12:02:55 +02006376
6377 If this option has been enabled in a "defaults" section, it can be disabled
6378 in a specific instance by prepending the "no" keyword before it.
6379
6380 Example :
6381 # this backend understands HTTP proxy requests and forwards them directly.
6382 backend direct_forward
6383 option httpclose
6384 option http_proxy
6385
6386 See also : "option httpclose"
6387
Willy Tarreau211ad242009-10-03 21:45:07 +02006388
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006389option independent-streams
6390no option independent-streams
6391 Enable or disable independent timeout processing for both directions
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006392 May be used in sections : defaults | frontend | listen | backend
6393 yes | yes | yes | yes
6394 Arguments : none
6395
6396 By default, when data is sent over a socket, both the write timeout and the
6397 read timeout for that socket are refreshed, because we consider that there is
6398 activity on that socket, and we have no other means of guessing if we should
6399 receive data or not.
6400
Davor Ocelice9ed2812017-12-25 17:49:28 +01006401 While this default behavior is desirable for almost all applications, there
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006402 exists a situation where it is desirable to disable it, and only refresh the
6403 read timeout if there are incoming data. This happens on sessions with large
6404 timeouts and low amounts of exchanged data such as telnet session. If the
6405 server suddenly disappears, the output data accumulates in the system's
6406 socket buffers, both timeouts are correctly refreshed, and there is no way
6407 to know the server does not receive them, so we don't timeout. However, when
6408 the underlying protocol always echoes sent data, it would be enough by itself
6409 to detect the issue using the read timeout. Note that this problem does not
6410 happen with more verbose protocols because data won't accumulate long in the
6411 socket buffers.
6412
6413 When this option is set on the frontend, it will disable read timeout updates
6414 on data sent to the client. There probably is little use of this case. When
6415 the option is set on the backend, it will disable read timeout updates on
6416 data sent to the server. Doing so will typically break large HTTP posts from
6417 slow lines, so use it with caution.
6418
Lukas Tribus745f15e2018-11-08 12:41:42 +01006419 Note: older versions used to call this setting "option independant-streams"
Jamie Gloudon801a0a32012-08-25 00:18:33 -04006420 with a spelling mistake. This spelling is still supported but
6421 deprecated.
6422
Willy Tarreauce887fd2012-05-12 12:50:00 +02006423 See also : "timeout client", "timeout server" and "timeout tunnel"
Willy Tarreauf27b5ea2009-10-03 22:01:18 +02006424
6425
Gabor Lekenyb4c81e42010-09-29 18:17:05 +02006426option ldap-check
6427 Use LDAPv3 health checks for server testing
6428 May be used in sections : defaults | frontend | listen | backend
6429 yes | no | yes | yes
6430 Arguments : none
6431
6432 It is possible to test that the server correctly talks LDAPv3 instead of just
6433 testing that it accepts the TCP connection. When this option is set, an
6434 LDAPv3 anonymous simple bind message is sent to the server, and the response
6435 is analyzed to find an LDAPv3 bind response message.
6436
6437 The server is considered valid only when the LDAP response contains success
6438 resultCode (http://tools.ietf.org/html/rfc4511#section-4.1.9).
6439
6440 Logging of bind requests is server dependent see your documentation how to
6441 configure it.
6442
6443 Example :
6444 option ldap-check
6445
6446 See also : "option httpchk"
6447
6448
Simon Horman98637e52014-06-20 12:30:16 +09006449option external-check
6450 Use external processes for server health checks
6451 May be used in sections : defaults | frontend | listen | backend
6452 yes | no | yes | yes
6453
6454 It is possible to test the health of a server using an external command.
6455 This is achieved by running the executable set using "external-check
6456 command".
6457
6458 Requires the "external-check" global to be set.
6459
6460 See also : "external-check", "external-check command", "external-check path"
6461
6462
Willy Tarreau211ad242009-10-03 21:45:07 +02006463option log-health-checks
6464no option log-health-checks
Willy Tarreaubef1b322014-05-13 21:01:39 +02006465 Enable or disable logging of health checks status updates
Willy Tarreau211ad242009-10-03 21:45:07 +02006466 May be used in sections : defaults | frontend | listen | backend
6467 yes | no | yes | yes
6468 Arguments : none
6469
Willy Tarreaubef1b322014-05-13 21:01:39 +02006470 By default, failed health check are logged if server is UP and successful
6471 health checks are logged if server is DOWN, so the amount of additional
6472 information is limited.
Willy Tarreau211ad242009-10-03 21:45:07 +02006473
Willy Tarreaubef1b322014-05-13 21:01:39 +02006474 When this option is enabled, any change of the health check status or to
6475 the server's health will be logged, so that it becomes possible to know
6476 that a server was failing occasional checks before crashing, or exactly when
6477 it failed to respond a valid HTTP status, then when the port started to
6478 reject connections, then when the server stopped responding at all.
6479
Davor Ocelice9ed2812017-12-25 17:49:28 +01006480 Note that status changes not caused by health checks (e.g. enable/disable on
Willy Tarreaubef1b322014-05-13 21:01:39 +02006481 the CLI) are intentionally not logged by this option.
Willy Tarreau211ad242009-10-03 21:45:07 +02006482
Willy Tarreaubef1b322014-05-13 21:01:39 +02006483 See also: "option httpchk", "option ldap-check", "option mysql-check",
6484 "option pgsql-check", "option redis-check", "option smtpchk",
6485 "option tcp-check", "log" and section 8 about logging.
Willy Tarreau211ad242009-10-03 21:45:07 +02006486
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006487
6488option log-separate-errors
6489no option log-separate-errors
6490 Change log level for non-completely successful connections
6491 May be used in sections : defaults | frontend | listen | backend
6492 yes | yes | yes | no
6493 Arguments : none
6494
6495 Sometimes looking for errors in logs is not easy. This option makes haproxy
6496 raise the level of logs containing potentially interesting information such
6497 as errors, timeouts, retries, redispatches, or HTTP status codes 5xx. The
6498 level changes from "info" to "err". This makes it possible to log them
6499 separately to a different file with most syslog daemons. Be careful not to
6500 remove them from the original file, otherwise you would lose ordering which
6501 provides very important information.
6502
6503 Using this option, large sites dealing with several thousand connections per
6504 second may log normal traffic to a rotating buffer and only archive smaller
6505 error logs.
6506
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006507 See also : "log", "dontlognull", "dontlog-normal" and section 8 about
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +02006508 logging.
6509
Willy Tarreauc27debf2008-01-06 08:57:02 +01006510
6511option logasap
6512no option logasap
6513 Enable or disable early logging of HTTP requests
6514 May be used in sections : defaults | frontend | listen | backend
6515 yes | yes | yes | no
6516 Arguments : none
6517
6518 By default, HTTP requests are logged upon termination so that the total
6519 transfer time and the number of bytes appear in the logs. When large objects
6520 are being transferred, it may take a while before the request appears in the
6521 logs. Using "option logasap", the request gets logged as soon as the server
6522 sends the complete headers. The only missing information in the logs will be
6523 the total number of bytes which will indicate everything except the amount
6524 of data transferred, and the total time which will not take the transfer
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01006525 time into account. In such a situation, it's a good practice to capture the
Willy Tarreauc27debf2008-01-06 08:57:02 +01006526 "Content-Length" response header so that the logs at least indicate how many
6527 bytes are expected to be transferred.
6528
Willy Tarreaucc6c8912009-02-22 10:53:55 +01006529 Examples :
6530 listen http_proxy 0.0.0.0:80
6531 mode http
6532 option httplog
6533 option logasap
6534 log 192.168.2.200 local3
6535
6536 >>> Feb 6 12:14:14 localhost \
6537 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
6538 static/srv1 9/10/7/14/+30 200 +243 - - ---- 3/1/1/1/0 1/0 \
6539 "GET /image.iso HTTP/1.0"
6540
Willy Tarreauc57f0e22009-05-10 13:12:33 +02006541 See also : "option httplog", "capture response header", and section 8 about
Willy Tarreauc27debf2008-01-06 08:57:02 +01006542 logging.
6543
6544
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006545option mysql-check [ user <username> [ post-41 ] ]
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006546 Use MySQL health checks for server testing
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006547 May be used in sections : defaults | frontend | listen | backend
6548 yes | no | yes | yes
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006549 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +02006550 <username> This is the username which will be used when connecting to MySQL
6551 server.
Nenad Merdanovic6639a7c2014-05-30 14:26:32 +02006552 post-41 Send post v4.1 client compatible checks
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006553
6554 If you specify a username, the check consists of sending two MySQL packet,
6555 one Client Authentication packet, and one QUIT packet, to correctly close
Davor Ocelice9ed2812017-12-25 17:49:28 +01006556 MySQL session. We then parse the MySQL Handshake Initialization packet and/or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006557 Error packet. It is a basic but useful test which does not produce error nor
6558 aborted connect on the server. However, it requires adding an authorization
6559 in the MySQL table, like this :
6560
6561 USE mysql;
6562 INSERT INTO user (Host,User) values ('<ip_of_haproxy>','<username>');
6563 FLUSH PRIVILEGES;
6564
6565 If you don't specify a username (it is deprecated and not recommended), the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006566 check only consists in parsing the Mysql Handshake Initialization packet or
Hervé COMMOWICK8776f1b2010-10-18 15:58:36 +02006567 Error packet, we don't send anything in this mode. It was reported that it
6568 can generate lockout if check is too frequent and/or if there is not enough
6569 traffic. In fact, you need in this case to check MySQL "max_connect_errors"
6570 value as if a connection is established successfully within fewer than MySQL
6571 "max_connect_errors" attempts after a previous connection was interrupted,
6572 the error count for the host is cleared to zero. If HAProxy's server get
6573 blocked, the "FLUSH HOSTS" statement is the only way to unblock it.
6574
6575 Remember that this does not check database presence nor database consistency.
6576 To do this, you can use an external check with xinetd for example.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006577
Hervé COMMOWICK212f7782011-06-10 14:05:59 +02006578 The check requires MySQL >=3.22, for older version, please use TCP check.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006579
6580 Most often, an incoming MySQL server needs to see the client's IP address for
6581 various purposes, including IP privilege matching and connection logging.
6582 When possible, it is often wise to masquerade the client's IP address when
6583 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006584 which requires the transparent proxy feature to be compiled in, and the MySQL
6585 server to route the client via the machine hosting haproxy.
Hervé COMMOWICK698ae002010-01-12 09:25:13 +01006586
6587 See also: "option httpchk"
6588
6589
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006590option nolinger
6591no option nolinger
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006592 Enable or disable immediate session resource cleaning after close
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006593 May be used in sections: defaults | frontend | listen | backend
6594 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006595 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006596
Davor Ocelice9ed2812017-12-25 17:49:28 +01006597 When clients or servers abort connections in a dirty way (e.g. they are
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006598 physically disconnected), the session timeouts triggers and the session is
6599 closed. But it will remain in FIN_WAIT1 state for some time in the system,
6600 using some resources and possibly limiting the ability to establish newer
6601 connections.
6602
6603 When this happens, it is possible to activate "option nolinger" which forces
6604 the system to immediately remove any socket's pending data on close. Thus,
6605 the session is instantly purged from the system's tables. This usually has
6606 side effects such as increased number of TCP resets due to old retransmits
6607 getting immediately rejected. Some firewalls may sometimes complain about
6608 this too.
6609
6610 For this reason, it is not recommended to use this option when not absolutely
6611 needed. You know that you need it when you have thousands of FIN_WAIT1
6612 sessions on your system (TIME_WAIT ones do not count).
6613
6614 This option may be used both on frontends and backends, depending on the side
6615 where it is required. Use it on the frontend for clients, and on the backend
6616 for servers.
6617
6618 If this option has been enabled in a "defaults" section, it can be disabled
6619 in a specific instance by prepending the "no" keyword before it.
6620
6621
Willy Tarreau55165fe2009-05-10 12:02:55 +02006622option originalto [ except <network> ] [ header <name> ]
6623 Enable insertion of the X-Original-To header to requests sent to servers
6624 May be used in sections : defaults | frontend | listen | backend
6625 yes | yes | yes | yes
6626 Arguments :
6627 <network> is an optional argument used to disable this option for sources
6628 matching <network>
6629 <name> an optional argument to specify a different "X-Original-To"
6630 header name.
6631
6632 Since HAProxy can work in transparent mode, every request from a client can
6633 be redirected to the proxy and HAProxy itself can proxy every request to a
6634 complex SQUID environment and the destination host from SO_ORIGINAL_DST will
6635 be lost. This is annoying when you want access rules based on destination ip
6636 addresses. To solve this problem, a new HTTP header "X-Original-To" may be
6637 added by HAProxy to all requests sent to the server. This header contains a
6638 value representing the original destination IP address. Since this must be
6639 configured to always use the last occurrence of this header only. Note that
6640 only the last occurrence of the header must be used, since it is really
6641 possible that the client has already brought one.
6642
6643 The keyword "header" may be used to supply a different header name to replace
6644 the default "X-Original-To". This can be useful where you might already
6645 have a "X-Original-To" header from a different application, and you need
6646 preserve it. Also if your backend server doesn't use the "X-Original-To"
6647 header and requires different one.
6648
6649 Sometimes, a same HAProxy instance may be shared between a direct client
6650 access and a reverse-proxy access (for instance when an SSL reverse-proxy is
6651 used to decrypt HTTPS traffic). It is possible to disable the addition of the
6652 header for a known source address or network by adding the "except" keyword
6653 followed by the network address. In this case, any source IP matching the
6654 network will not cause an addition of this header. Most common uses are with
6655 private networks or 127.0.0.1.
6656
6657 This option may be specified either in the frontend or in the backend. If at
6658 least one of them uses it, the header will be added. Note that the backend's
6659 setting of the header subargument takes precedence over the frontend's if
6660 both are defined.
6661
Willy Tarreau55165fe2009-05-10 12:02:55 +02006662 Examples :
6663 # Original Destination address
6664 frontend www
6665 mode http
6666 option originalto except 127.0.0.1
6667
6668 # Those servers want the IP Address in X-Client-Dst
6669 backend www
6670 mode http
6671 option originalto header X-Client-Dst
6672
Christopher Faulet315b39c2018-09-21 16:26:19 +02006673 See also : "option httpclose", "option http-server-close".
Willy Tarreau55165fe2009-05-10 12:02:55 +02006674
6675
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006676option persist
6677no option persist
6678 Enable or disable forced persistence on down servers
6679 May be used in sections: defaults | frontend | listen | backend
6680 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01006681 Arguments : none
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006682
6683 When an HTTP request reaches a backend with a cookie which references a dead
6684 server, by default it is redispatched to another server. It is possible to
6685 force the request to be sent to the dead server first using "option persist"
6686 if absolutely needed. A common use case is when servers are under extreme
6687 load and spend their time flapping. In this case, the users would still be
6688 directed to the server they opened the session on, in the hope they would be
6689 correctly served. It is recommended to use "option redispatch" in conjunction
6690 with this option so that in the event it would not be possible to connect to
6691 the server at all (server definitely dead), the client would finally be
6692 redirected to another valid server.
6693
6694 If this option has been enabled in a "defaults" section, it can be disabled
6695 in a specific instance by prepending the "no" keyword before it.
6696
Willy Tarreau4de91492010-01-22 19:10:05 +01006697 See also : "option redispatch", "retries", "force-persist"
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006698
6699
Willy Tarreau0c122822013-12-15 18:49:01 +01006700option pgsql-check [ user <username> ]
6701 Use PostgreSQL health checks for server testing
6702 May be used in sections : defaults | frontend | listen | backend
6703 yes | no | yes | yes
6704 Arguments :
6705 <username> This is the username which will be used when connecting to
6706 PostgreSQL server.
6707
6708 The check sends a PostgreSQL StartupMessage and waits for either
6709 Authentication request or ErrorResponse message. It is a basic but useful
6710 test which does not produce error nor aborted connect on the server.
6711 This check is identical with the "mysql-check".
6712
6713 See also: "option httpchk"
6714
6715
Willy Tarreau9420b122013-12-15 18:58:25 +01006716option prefer-last-server
6717no option prefer-last-server
6718 Allow multiple load balanced requests to remain on the same server
6719 May be used in sections: defaults | frontend | listen | backend
6720 yes | no | yes | yes
6721 Arguments : none
6722
6723 When the load balancing algorithm in use is not deterministic, and a previous
6724 request was sent to a server to which haproxy still holds a connection, it is
6725 sometimes desirable that subsequent requests on a same session go to the same
6726 server as much as possible. Note that this is different from persistence, as
6727 we only indicate a preference which haproxy tries to apply without any form
6728 of warranty. The real use is for keep-alive connections sent to servers. When
6729 this option is used, haproxy will try to reuse the same connection that is
6730 attached to the server instead of rebalancing to another server, causing a
6731 close of the connection. This can make sense for static file servers. It does
Willy Tarreau068621e2013-12-23 15:11:25 +01006732 not make much sense to use this in combination with hashing algorithms. Note,
6733 haproxy already automatically tries to stick to a server which sends a 401 or
Lukas Tribus80512b12018-10-27 20:07:40 +02006734 to a proxy which sends a 407 (authentication required), when the load
6735 balancing algorithm is not deterministic. This is mandatory for use with the
6736 broken NTLM authentication challenge, and significantly helps in
Willy Tarreau068621e2013-12-23 15:11:25 +01006737 troubleshooting some faulty applications. Option prefer-last-server might be
6738 desirable in these environments as well, to avoid redistributing the traffic
6739 after every other response.
Willy Tarreau9420b122013-12-15 18:58:25 +01006740
6741 If this option has been enabled in a "defaults" section, it can be disabled
6742 in a specific instance by prepending the "no" keyword before it.
6743
6744 See also: "option http-keep-alive"
6745
6746
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006747option redispatch
Joseph Lynch726ab712015-05-11 23:25:34 -07006748option redispatch <interval>
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006749no option redispatch
6750 Enable or disable session redistribution in case of connection failure
6751 May be used in sections: defaults | frontend | listen | backend
6752 yes | no | yes | yes
Joseph Lynch726ab712015-05-11 23:25:34 -07006753 Arguments :
6754 <interval> The optional integer value that controls how often redispatches
6755 occur when retrying connections. Positive value P indicates a
6756 redispatch is desired on every Pth retry, and negative value
Davor Ocelice9ed2812017-12-25 17:49:28 +01006757 N indicate a redispatch is desired on the Nth retry prior to the
Joseph Lynch726ab712015-05-11 23:25:34 -07006758 last retry. For example, the default of -1 preserves the
Davor Ocelice9ed2812017-12-25 17:49:28 +01006759 historical behavior of redispatching on the last retry, a
Joseph Lynch726ab712015-05-11 23:25:34 -07006760 positive value of 1 would indicate a redispatch on every retry,
6761 and a positive value of 3 would indicate a redispatch on every
6762 third retry. You can disable redispatches with a value of 0.
6763
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006764
6765 In HTTP mode, if a server designated by a cookie is down, clients may
6766 definitely stick to it because they cannot flush the cookie, so they will not
6767 be able to access the service anymore.
6768
Willy Tarreau59884a62019-01-02 14:48:31 +01006769 Specifying "option redispatch" will allow the proxy to break cookie or
6770 consistent hash based persistence and redistribute them to a working server.
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006771
Joseph Lynch726ab712015-05-11 23:25:34 -07006772 It also allows to retry connections to another server in case of multiple
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006773 connection failures. Of course, it requires having "retries" set to a nonzero
6774 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01006775
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006776 This form is the preferred form, which replaces both the "redispatch" and
6777 "redisp" keywords.
6778
6779 If this option has been enabled in a "defaults" section, it can be disabled
6780 in a specific instance by prepending the "no" keyword before it.
6781
Willy Tarreau4de91492010-01-22 19:10:05 +01006782 See also : "redispatch", "retries", "force-persist"
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006783
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006784
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006785option redis-check
6786 Use redis health checks for server testing
6787 May be used in sections : defaults | frontend | listen | backend
6788 yes | no | yes | yes
6789 Arguments : none
6790
6791 It is possible to test that the server correctly talks REDIS protocol instead
6792 of just testing that it accepts the TCP connection. When this option is set,
6793 a PING redis command is sent to the server, and the response is analyzed to
6794 find the "+PONG" response message.
6795
6796 Example :
6797 option redis-check
6798
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03006799 See also : "option httpchk", "option tcp-check", "tcp-check expect"
Hervé COMMOWICKec032d62011-08-05 16:23:48 +02006800
6801
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006802option smtpchk
6803option smtpchk <hello> <domain>
6804 Use SMTP health checks for server testing
6805 May be used in sections : defaults | frontend | listen | backend
6806 yes | no | yes | yes
Willy Tarreaud72758d2010-01-12 10:42:19 +01006807 Arguments :
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006808 <hello> is an optional argument. It is the "hello" command to use. It can
Lukas Tribus27935782018-10-01 02:00:16 +02006809 be either "HELO" (for SMTP) or "EHLO" (for ESMTP). All other
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006810 values will be turned into the default command ("HELO").
6811
6812 <domain> is the domain name to present to the server. It may only be
6813 specified (and is mandatory) if the hello command has been
6814 specified. By default, "localhost" is used.
6815
6816 When "option smtpchk" is set, the health checks will consist in TCP
6817 connections followed by an SMTP command. By default, this command is
6818 "HELO localhost". The server's return code is analyzed and only return codes
6819 starting with a "2" will be considered as valid. All other responses,
6820 including a lack of response will constitute an error and will indicate a
6821 dead server.
6822
6823 This test is meant to be used with SMTP servers or relays. Depending on the
6824 request, it is possible that some servers do not log each connection attempt,
Davor Ocelice9ed2812017-12-25 17:49:28 +01006825 so you may want to experiment to improve the behavior. Using telnet on port
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006826 25 is often easier than adjusting the configuration.
6827
6828 Most often, an incoming SMTP server needs to see the client's IP address for
6829 various purposes, including spam filtering, anti-spoofing and logging. When
6830 possible, it is often wise to masquerade the client's IP address when
6831 connecting to the server using the "usesrc" argument of the "source" keyword,
Willy Tarreau29fbe512015-08-20 19:35:14 +02006832 which requires the transparent proxy feature to be compiled in.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006833
6834 Example :
6835 option smtpchk HELO mydomain.org
6836
6837 See also : "option httpchk", "source"
6838
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01006839
Krzysztof Piotr Oledzkiaeebf9b2009-10-04 15:43:17 +02006840option socket-stats
6841no option socket-stats
6842
6843 Enable or disable collecting & providing separate statistics for each socket.
6844 May be used in sections : defaults | frontend | listen | backend
6845 yes | yes | yes | no
6846
6847 Arguments : none
6848
6849
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006850option splice-auto
6851no option splice-auto
6852 Enable or disable automatic kernel acceleration on sockets in both directions
6853 May be used in sections : defaults | frontend | listen | backend
6854 yes | yes | yes | yes
6855 Arguments : none
6856
6857 When this option is enabled either on a frontend or on a backend, haproxy
6858 will automatically evaluate the opportunity to use kernel tcp splicing to
Davor Ocelice9ed2812017-12-25 17:49:28 +01006859 forward data between the client and the server, in either direction. HAProxy
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006860 uses heuristics to estimate if kernel splicing might improve performance or
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01006861 not. Both directions are handled independently. Note that the heuristics used
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006862 are not much aggressive in order to limit excessive use of splicing. This
6863 option requires splicing to be enabled at compile time, and may be globally
6864 disabled with the global option "nosplice". Since splice uses pipes, using it
6865 requires that there are enough spare pipes.
6866
6867 Important note: kernel-based TCP splicing is a Linux-specific feature which
6868 first appeared in kernel 2.6.25. It offers kernel-based acceleration to
6869 transfer data between sockets without copying these data to user-space, thus
6870 providing noticeable performance gains and CPU cycles savings. Since many
6871 early implementations are buggy, corrupt data and/or are inefficient, this
6872 feature is not enabled by default, and it should be used with extreme care.
6873 While it is not possible to detect the correctness of an implementation,
6874 2.6.29 is the first version offering a properly working implementation. In
6875 case of doubt, splicing may be globally disabled using the global "nosplice"
6876 keyword.
6877
6878 Example :
6879 option splice-auto
6880
6881 If this option has been enabled in a "defaults" section, it can be disabled
6882 in a specific instance by prepending the "no" keyword before it.
6883
6884 See also : "option splice-request", "option splice-response", and global
6885 options "nosplice" and "maxpipes"
6886
6887
6888option splice-request
6889no option splice-request
6890 Enable or disable automatic kernel acceleration on sockets for requests
6891 May be used in sections : defaults | frontend | listen | backend
6892 yes | yes | yes | yes
6893 Arguments : none
6894
6895 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006896 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006897 the client to the server. It might still use the recv/send scheme if there
6898 are no spare pipes left. This option requires splicing to be enabled at
6899 compile time, and may be globally disabled with the global option "nosplice".
6900 Since splice uses pipes, using it requires that there are enough spare pipes.
6901
6902 Important note: see "option splice-auto" for usage limitations.
6903
6904 Example :
6905 option splice-request
6906
6907 If this option has been enabled in a "defaults" section, it can be disabled
6908 in a specific instance by prepending the "no" keyword before it.
6909
6910 See also : "option splice-auto", "option splice-response", and global options
6911 "nosplice" and "maxpipes"
6912
6913
6914option splice-response
6915no option splice-response
6916 Enable or disable automatic kernel acceleration on sockets for responses
6917 May be used in sections : defaults | frontend | listen | backend
6918 yes | yes | yes | yes
6919 Arguments : none
6920
6921 When this option is enabled either on a frontend or on a backend, haproxy
Jamie Gloudonaaa21002012-08-25 00:18:33 -04006922 will use kernel tcp splicing whenever possible to forward data going from
Willy Tarreauff4f82d2009-02-06 11:28:13 +01006923 the server to the client. It might still use the recv/send scheme if there
6924 are no spare pipes left. This option requires splicing to be enabled at
6925 compile time, and may be globally disabled with the global option "nosplice".
6926 Since splice uses pipes, using it requires that there are enough spare pipes.
6927
6928 Important note: see "option splice-auto" for usage limitations.
6929
6930 Example :
6931 option splice-response
6932
6933 If this option has been enabled in a "defaults" section, it can be disabled
6934 in a specific instance by prepending the "no" keyword before it.
6935
6936 See also : "option splice-auto", "option splice-request", and global options
6937 "nosplice" and "maxpipes"
6938
6939
Christopher Fauletba7bc162016-11-07 21:07:38 +01006940option spop-check
6941 Use SPOP health checks for server testing
6942 May be used in sections : defaults | frontend | listen | backend
6943 no | no | no | yes
6944 Arguments : none
6945
6946 It is possible to test that the server correctly talks SPOP protocol instead
6947 of just testing that it accepts the TCP connection. When this option is set,
6948 a HELLO handshake is performed between HAProxy and the server, and the
6949 response is analyzed to check no error is reported.
6950
6951 Example :
6952 option spop-check
6953
6954 See also : "option httpchk"
6955
6956
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006957option srvtcpka
6958no option srvtcpka
6959 Enable or disable the sending of TCP keepalive packets on the server side
6960 May be used in sections : defaults | frontend | listen | backend
6961 yes | no | yes | yes
6962 Arguments : none
6963
6964 When there is a firewall or any session-aware component between a client and
6965 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01006966 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01006967 components decides to expire a session which has remained idle for too long.
6968
6969 Enabling socket-level TCP keep-alives makes the system regularly send packets
6970 to the other end of the connection, leaving it active. The delay between
6971 keep-alive probes is controlled by the system only and depends both on the
6972 operating system and its tuning parameters.
6973
6974 It is important to understand that keep-alive packets are neither emitted nor
6975 received at the application level. It is only the network stacks which sees
6976 them. For this reason, even if one side of the proxy already uses keep-alives
6977 to maintain its connection alive, those keep-alive packets will not be
6978 forwarded to the other side of the proxy.
6979
6980 Please note that this has nothing to do with HTTP keep-alive.
6981
6982 Using option "srvtcpka" enables the emission of TCP keep-alive probes on the
6983 server side of a connection, which should help when session expirations are
6984 noticed between HAProxy and a server.
6985
6986 If this option has been enabled in a "defaults" section, it can be disabled
6987 in a specific instance by prepending the "no" keyword before it.
6988
6989 See also : "option clitcpka", "option tcpka"
6990
6991
Willy Tarreaua453bdd2008-01-08 19:50:52 +01006992option ssl-hello-chk
6993 Use SSLv3 client hello health checks for server testing
6994 May be used in sections : defaults | frontend | listen | backend
6995 yes | no | yes | yes
6996 Arguments : none
6997
6998 When some SSL-based protocols are relayed in TCP mode through HAProxy, it is
6999 possible to test that the server correctly talks SSL instead of just testing
7000 that it accepts the TCP connection. When "option ssl-hello-chk" is set, pure
7001 SSLv3 client hello messages are sent once the connection is established to
7002 the server, and the response is analyzed to find an SSL server hello message.
7003 The server is considered valid only when the response contains this server
7004 hello message.
7005
7006 All servers tested till there correctly reply to SSLv3 client hello messages,
7007 and most servers tested do not even log the requests containing only hello
7008 messages, which is appreciable.
7009
Willy Tarreau763a95b2012-10-04 23:15:39 +02007010 Note that this check works even when SSL support was not built into haproxy
7011 because it forges the SSL message. When SSL support is available, it is best
7012 to use native SSL health checks instead of this one.
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007013
Willy Tarreau763a95b2012-10-04 23:15:39 +02007014 See also: "option httpchk", "check-ssl"
7015
Willy Tarreaua453bdd2008-01-08 19:50:52 +01007016
Willy Tarreaued179852013-12-16 01:07:00 +01007017option tcp-check
7018 Perform health checks using tcp-check send/expect sequences
7019 May be used in sections: defaults | frontend | listen | backend
7020 yes | no | yes | yes
7021
7022 This health check method is intended to be combined with "tcp-check" command
7023 lists in order to support send/expect types of health check sequences.
7024
7025 TCP checks currently support 4 modes of operations :
7026 - no "tcp-check" directive : the health check only consists in a connection
7027 attempt, which remains the default mode.
7028
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007029 - "tcp-check send" or "tcp-check send-binary" only is mentioned : this is
Willy Tarreaued179852013-12-16 01:07:00 +01007030 used to send a string along with a connection opening. With some
7031 protocols, it helps sending a "QUIT" message for example that prevents
7032 the server from logging a connection error for each health check. The
7033 check result will still be based on the ability to open the connection
7034 only.
7035
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007036 - "tcp-check expect" only is mentioned : this is used to test a banner.
Willy Tarreaued179852013-12-16 01:07:00 +01007037 The connection is opened and haproxy waits for the server to present some
7038 contents which must validate some rules. The check result will be based
7039 on the matching between the contents and the rules. This is suited for
7040 POP, IMAP, SMTP, FTP, SSH, TELNET.
7041
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007042 - both "tcp-check send" and "tcp-check expect" are mentioned : this is
Davor Ocelice9ed2812017-12-25 17:49:28 +01007043 used to test a hello-type protocol. HAProxy sends a message, the server
7044 responds and its response is analyzed. the check result will be based on
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007045 the matching between the response contents and the rules. This is often
Willy Tarreaued179852013-12-16 01:07:00 +01007046 suited for protocols which require a binding or a request/response model.
7047 LDAP, MySQL, Redis and SSL are example of such protocols, though they
7048 already all have their dedicated checks with a deeper understanding of
7049 the respective protocols.
7050 In this mode, many questions may be sent and many answers may be
Davor Ocelice9ed2812017-12-25 17:49:28 +01007051 analyzed.
Willy Tarreaued179852013-12-16 01:07:00 +01007052
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007053 A fifth mode can be used to insert comments in different steps of the
7054 script.
7055
7056 For each tcp-check rule you create, you can add a "comment" directive,
7057 followed by a string. This string will be reported in the log and stderr
7058 in debug mode. It is useful to make user-friendly error reporting.
7059 The "comment" is of course optional.
7060
7061
Willy Tarreaued179852013-12-16 01:07:00 +01007062 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007063 # perform a POP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007064 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007065 tcp-check expect string +OK\ POP3\ ready comment POP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007066
Davor Ocelice9ed2812017-12-25 17:49:28 +01007067 # perform an IMAP check (analyze only server's banner)
Willy Tarreaued179852013-12-16 01:07:00 +01007068 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007069 tcp-check expect string *\ OK\ IMAP4\ ready comment IMAP\ protocol
Willy Tarreaued179852013-12-16 01:07:00 +01007070
7071 # look for the redis master server after ensuring it speaks well
7072 # redis protocol, then it exits properly.
Davor Ocelice9ed2812017-12-25 17:49:28 +01007073 # (send a command then analyze the response 3 times)
Willy Tarreaued179852013-12-16 01:07:00 +01007074 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007075 tcp-check comment PING\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007076 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02007077 tcp-check expect string +PONG
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007078 tcp-check comment role\ check
Willy Tarreaued179852013-12-16 01:07:00 +01007079 tcp-check send info\ replication\r\n
7080 tcp-check expect string role:master
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007081 tcp-check comment QUIT\ phase
Willy Tarreaued179852013-12-16 01:07:00 +01007082 tcp-check send QUIT\r\n
7083 tcp-check expect string +OK
7084
Davor Ocelice9ed2812017-12-25 17:49:28 +01007085 forge a HTTP request, then analyze the response
Willy Tarreaued179852013-12-16 01:07:00 +01007086 (send many headers before analyzing)
7087 option tcp-check
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007088 tcp-check comment forge\ and\ send\ HTTP\ request
Willy Tarreaued179852013-12-16 01:07:00 +01007089 tcp-check send HEAD\ /\ HTTP/1.1\r\n
7090 tcp-check send Host:\ www.mydomain.com\r\n
7091 tcp-check send User-Agent:\ HAProxy\ tcpcheck\r\n
7092 tcp-check send \r\n
Baptiste Assmannd60a9e52015-04-25 16:27:23 +02007093 tcp-check expect rstring HTTP/1\..\ (2..|3..) comment check\ HTTP\ response
Willy Tarreaued179852013-12-16 01:07:00 +01007094
7095
7096 See also : "tcp-check expect", "tcp-check send"
7097
7098
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007099option tcp-smart-accept
7100no option tcp-smart-accept
7101 Enable or disable the saving of one ACK packet during the accept sequence
7102 May be used in sections : defaults | frontend | listen | backend
7103 yes | yes | yes | no
7104 Arguments : none
7105
7106 When an HTTP connection request comes in, the system acknowledges it on
7107 behalf of HAProxy, then the client immediately sends its request, and the
7108 system acknowledges it too while it is notifying HAProxy about the new
7109 connection. HAProxy then reads the request and responds. This means that we
7110 have one TCP ACK sent by the system for nothing, because the request could
7111 very well be acknowledged by HAProxy when it sends its response.
7112
7113 For this reason, in HTTP mode, HAProxy automatically asks the system to avoid
7114 sending this useless ACK on platforms which support it (currently at least
7115 Linux). It must not cause any problem, because the system will send it anyway
7116 after 40 ms if the response takes more time than expected to come.
7117
7118 During complex network debugging sessions, it may be desirable to disable
7119 this optimization because delayed ACKs can make troubleshooting more complex
7120 when trying to identify where packets are delayed. It is then possible to
Davor Ocelice9ed2812017-12-25 17:49:28 +01007121 fall back to normal behavior by specifying "no option tcp-smart-accept".
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007122
7123 It is also possible to force it for non-HTTP proxies by simply specifying
7124 "option tcp-smart-accept". For instance, it can make sense with some services
7125 such as SMTP where the server speaks first.
7126
7127 It is recommended to avoid forcing this option in a defaults section. In case
7128 of doubt, consider setting it back to automatic values by prepending the
7129 "default" keyword before it, or disabling it using the "no" keyword.
7130
Willy Tarreaud88edf22009-06-14 15:48:17 +02007131 See also : "option tcp-smart-connect"
7132
7133
7134option tcp-smart-connect
7135no option tcp-smart-connect
7136 Enable or disable the saving of one ACK packet during the connect sequence
7137 May be used in sections : defaults | frontend | listen | backend
7138 yes | no | yes | yes
7139 Arguments : none
7140
7141 On certain systems (at least Linux), HAProxy can ask the kernel not to
7142 immediately send an empty ACK upon a connection request, but to directly
7143 send the buffer request instead. This saves one packet on the network and
7144 thus boosts performance. It can also be useful for some servers, because they
7145 immediately get the request along with the incoming connection.
7146
7147 This feature is enabled when "option tcp-smart-connect" is set in a backend.
7148 It is not enabled by default because it makes network troubleshooting more
7149 complex.
7150
7151 It only makes sense to enable it with protocols where the client speaks first
7152 such as HTTP. In other situations, if there is no data to send in place of
7153 the ACK, a normal ACK is sent.
7154
7155 If this option has been enabled in a "defaults" section, it can be disabled
7156 in a specific instance by prepending the "no" keyword before it.
7157
7158 See also : "option tcp-smart-accept"
7159
Willy Tarreau9ea05a72009-06-14 12:07:01 +02007160
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007161option tcpka
7162 Enable or disable the sending of TCP keepalive packets on both sides
7163 May be used in sections : defaults | frontend | listen | backend
7164 yes | yes | yes | yes
7165 Arguments : none
7166
7167 When there is a firewall or any session-aware component between a client and
7168 a server, and when the protocol involves very long sessions with long idle
Davor Ocelice9ed2812017-12-25 17:49:28 +01007169 periods (e.g. remote desktops), there is a risk that one of the intermediate
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007170 components decides to expire a session which has remained idle for too long.
7171
7172 Enabling socket-level TCP keep-alives makes the system regularly send packets
7173 to the other end of the connection, leaving it active. The delay between
7174 keep-alive probes is controlled by the system only and depends both on the
7175 operating system and its tuning parameters.
7176
7177 It is important to understand that keep-alive packets are neither emitted nor
7178 received at the application level. It is only the network stacks which sees
7179 them. For this reason, even if one side of the proxy already uses keep-alives
7180 to maintain its connection alive, those keep-alive packets will not be
7181 forwarded to the other side of the proxy.
7182
7183 Please note that this has nothing to do with HTTP keep-alive.
7184
7185 Using option "tcpka" enables the emission of TCP keep-alive probes on both
7186 the client and server sides of a connection. Note that this is meaningful
7187 only in "defaults" or "listen" sections. If this option is used in a
7188 frontend, only the client side will get keep-alives, and if this option is
7189 used in a backend, only the server side will get keep-alives. For this
7190 reason, it is strongly recommended to explicitly use "option clitcpka" and
7191 "option srvtcpka" when the configuration is split between frontends and
7192 backends.
7193
7194 See also : "option clitcpka", "option srvtcpka"
7195
Willy Tarreau844e3c52008-01-11 16:28:18 +01007196
7197option tcplog
7198 Enable advanced logging of TCP connections with session state and timers
7199 May be used in sections : defaults | frontend | listen | backend
Tim Duesterhus9ad9f352018-02-05 20:52:27 +01007200 yes | yes | yes | no
Willy Tarreau844e3c52008-01-11 16:28:18 +01007201 Arguments : none
7202
7203 By default, the log output format is very poor, as it only contains the
7204 source and destination addresses, and the instance name. By specifying
7205 "option tcplog", each log line turns into a much richer format including, but
7206 not limited to, the connection timers, the session status, the connections
7207 numbers, the frontend, backend and server name, and of course the source
7208 address and ports. This option is useful for pure TCP proxies in order to
7209 find which of the client or server disconnects or times out. For normal HTTP
7210 proxies, it's better to use "option httplog" which is even more complete.
7211
Guillaume de Lafond29f45602017-03-31 19:52:15 +02007212 "option tcplog" overrides any previous "log-format" directive.
7213
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007214 See also : "option httplog", and section 8 about logging.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007215
7216
Willy Tarreau844e3c52008-01-11 16:28:18 +01007217option transparent
7218no option transparent
7219 Enable client-side transparent proxying
7220 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +01007221 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +01007222 Arguments : none
7223
7224 This option was introduced in order to provide layer 7 persistence to layer 3
7225 load balancers. The idea is to use the OS's ability to redirect an incoming
7226 connection for a remote address to a local process (here HAProxy), and let
7227 this process know what address was initially requested. When this option is
7228 used, sessions without cookies will be forwarded to the original destination
7229 IP address of the incoming request (which should match that of another
7230 equipment), while requests with cookies will still be forwarded to the
7231 appropriate server.
7232
7233 Note that contrary to a common belief, this option does NOT make HAProxy
7234 present the client's IP to the server when establishing the connection.
7235
Willy Tarreaua1146052011-03-01 09:51:54 +01007236 See also: the "usesrc" argument of the "source" keyword, and the
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007237 "transparent" option of the "bind" keyword.
Willy Tarreau844e3c52008-01-11 16:28:18 +01007238
Willy Tarreaubf1f8162007-12-28 17:42:56 +01007239
Simon Horman98637e52014-06-20 12:30:16 +09007240external-check command <command>
7241 Executable to run when performing an external-check
7242 May be used in sections : defaults | frontend | listen | backend
7243 yes | no | yes | yes
7244
7245 Arguments :
7246 <command> is the external command to run
7247
Simon Horman98637e52014-06-20 12:30:16 +09007248 The arguments passed to the to the command are:
7249
Cyril Bonté777be862014-12-02 21:21:35 +01007250 <proxy_address> <proxy_port> <server_address> <server_port>
Simon Horman98637e52014-06-20 12:30:16 +09007251
Cyril Bonté777be862014-12-02 21:21:35 +01007252 The <proxy_address> and <proxy_port> are derived from the first listener
7253 that is either IPv4, IPv6 or a UNIX socket. In the case of a UNIX socket
7254 listener the proxy_address will be the path of the socket and the
7255 <proxy_port> will be the string "NOT_USED". In a backend section, it's not
7256 possible to determine a listener, and both <proxy_address> and <proxy_port>
7257 will have the string value "NOT_USED".
Simon Horman98637e52014-06-20 12:30:16 +09007258
Cyril Bonté72cda2a2014-12-27 22:28:39 +01007259 Some values are also provided through environment variables.
7260
7261 Environment variables :
7262 HAPROXY_PROXY_ADDR The first bind address if available (or empty if not
7263 applicable, for example in a "backend" section).
7264
7265 HAPROXY_PROXY_ID The backend id.
7266
7267 HAPROXY_PROXY_NAME The backend name.
7268
7269 HAPROXY_PROXY_PORT The first bind port if available (or empty if not
7270 applicable, for example in a "backend" section or
7271 for a UNIX socket).
7272
7273 HAPROXY_SERVER_ADDR The server address.
7274
7275 HAPROXY_SERVER_CURCONN The current number of connections on the server.
7276
7277 HAPROXY_SERVER_ID The server id.
7278
7279 HAPROXY_SERVER_MAXCONN The server max connections.
7280
7281 HAPROXY_SERVER_NAME The server name.
7282
7283 HAPROXY_SERVER_PORT The server port if available (or empty for a UNIX
7284 socket).
7285
7286 PATH The PATH environment variable used when executing
7287 the command may be set using "external-check path".
7288
Simon Horman98637e52014-06-20 12:30:16 +09007289 If the command executed and exits with a zero status then the check is
7290 considered to have passed, otherwise the check is considered to have
7291 failed.
7292
7293 Example :
7294 external-check command /bin/true
7295
7296 See also : "external-check", "option external-check", "external-check path"
7297
7298
7299external-check path <path>
7300 The value of the PATH environment variable used when running an external-check
7301 May be used in sections : defaults | frontend | listen | backend
7302 yes | no | yes | yes
7303
7304 Arguments :
7305 <path> is the path used when executing external command to run
7306
7307 The default path is "".
7308
7309 Example :
7310 external-check path "/usr/bin:/bin"
7311
7312 See also : "external-check", "option external-check",
7313 "external-check command"
7314
7315
Emeric Brun647caf12009-06-30 17:57:00 +02007316persist rdp-cookie
Hervé COMMOWICKa3eb39c2011-08-05 18:48:51 +02007317persist rdp-cookie(<name>)
Emeric Brun647caf12009-06-30 17:57:00 +02007318 Enable RDP cookie-based persistence
7319 May be used in sections : defaults | frontend | listen | backend
7320 yes | no | yes | yes
7321 Arguments :
7322 <name> is the optional name of the RDP cookie to check. If omitted, the
Willy Tarreau61e28f22010-05-16 22:31:05 +02007323 default cookie name "msts" will be used. There currently is no
7324 valid reason to change this name.
Emeric Brun647caf12009-06-30 17:57:00 +02007325
7326 This statement enables persistence based on an RDP cookie. The RDP cookie
7327 contains all information required to find the server in the list of known
Davor Ocelice9ed2812017-12-25 17:49:28 +01007328 servers. So when this option is set in the backend, the request is analyzed
Emeric Brun647caf12009-06-30 17:57:00 +02007329 and if an RDP cookie is found, it is decoded. If it matches a known server
7330 which is still UP (or if "option persist" is set), then the connection is
7331 forwarded to this server.
7332
7333 Note that this only makes sense in a TCP backend, but for this to work, the
7334 frontend must have waited long enough to ensure that an RDP cookie is present
7335 in the request buffer. This is the same requirement as with the "rdp-cookie"
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01007336 load-balancing method. Thus it is highly recommended to put all statements in
Emeric Brun647caf12009-06-30 17:57:00 +02007337 a single "listen" section.
7338
Willy Tarreau61e28f22010-05-16 22:31:05 +02007339 Also, it is important to understand that the terminal server will emit this
7340 RDP cookie only if it is configured for "token redirection mode", which means
7341 that the "IP address redirection" option is disabled.
7342
Emeric Brun647caf12009-06-30 17:57:00 +02007343 Example :
7344 listen tse-farm
7345 bind :3389
7346 # wait up to 5s for an RDP cookie in the request
7347 tcp-request inspect-delay 5s
7348 tcp-request content accept if RDP_COOKIE
7349 # apply RDP cookie persistence
7350 persist rdp-cookie
7351 # if server is unknown, let's balance on the same cookie.
Cyril Bontédc4d9032012-04-08 21:57:39 +02007352 # alternatively, "balance leastconn" may be useful too.
Emeric Brun647caf12009-06-30 17:57:00 +02007353 balance rdp-cookie
7354 server srv1 1.1.1.1:3389
7355 server srv2 1.1.1.2:3389
7356
Simon Hormanab814e02011-06-24 14:50:20 +09007357 See also : "balance rdp-cookie", "tcp-request", the "req_rdp_cookie" ACL and
7358 the rdp_cookie pattern fetch function.
Emeric Brun647caf12009-06-30 17:57:00 +02007359
7360
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007361rate-limit sessions <rate>
7362 Set a limit on the number of new sessions accepted per second on a frontend
7363 May be used in sections : defaults | frontend | listen | backend
7364 yes | yes | yes | no
7365 Arguments :
7366 <rate> The <rate> parameter is an integer designating the maximum number
7367 of new sessions per second to accept on the frontend.
7368
7369 When the frontend reaches the specified number of new sessions per second, it
7370 stops accepting new connections until the rate drops below the limit again.
7371 During this time, the pending sessions will be kept in the socket's backlog
7372 (in system buffers) and haproxy will not even be aware that sessions are
7373 pending. When applying very low limit on a highly loaded service, it may make
7374 sense to increase the socket's backlog using the "backlog" keyword.
7375
7376 This feature is particularly efficient at blocking connection-based attacks
7377 or service abuse on fragile servers. Since the session rate is measured every
7378 millisecond, it is extremely accurate. Also, the limit applies immediately,
7379 no delay is needed at all to detect the threshold.
7380
7381 Example : limit the connection rate on SMTP to 10 per second max
7382 listen smtp
7383 mode tcp
7384 bind :25
7385 rate-limit sessions 10
Panagiotis Panagiotopoulos7282d8e2016-02-11 16:37:15 +02007386 server smtp1 127.0.0.1:1025
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007387
Willy Tarreaua17c2d92011-07-25 08:16:20 +02007388 Note : when the maximum rate is reached, the frontend's status is not changed
7389 but its sockets appear as "WAITING" in the statistics if the
7390 "socket-stats" option is enabled.
Willy Tarreau3a7d2072009-03-05 23:48:25 +01007391
7392 See also : the "backlog" keyword and the "fe_sess_rate" ACL criterion.
7393
7394
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007395redirect location <loc> [code <code>] <option> [{if | unless} <condition>]
7396redirect prefix <pfx> [code <code>] <option> [{if | unless} <condition>]
7397redirect scheme <sch> [code <code>] <option> [{if | unless} <condition>]
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007398 Return an HTTP redirection if/unless a condition is matched
7399 May be used in sections : defaults | frontend | listen | backend
7400 no | yes | yes | yes
7401
7402 If/unless the condition is matched, the HTTP request will lead to a redirect
Willy Tarreauf285f542010-01-03 20:03:03 +01007403 response. If no condition is specified, the redirect applies unconditionally.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007404
Willy Tarreau0140f252008-11-19 21:07:09 +01007405 Arguments :
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007406 <loc> With "redirect location", the exact value in <loc> is placed into
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007407 the HTTP "Location" header. When used in an "http-request" rule,
7408 <loc> value follows the log-format rules and can include some
7409 dynamic values (see Custom Log Format in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007410
7411 <pfx> With "redirect prefix", the "Location" header is built from the
7412 concatenation of <pfx> and the complete URI path, including the
7413 query string, unless the "drop-query" option is specified (see
7414 below). As a special case, if <pfx> equals exactly "/", then
7415 nothing is inserted before the original URI. It allows one to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007416 redirect to the same URL (for instance, to insert a cookie). When
7417 used in an "http-request" rule, <pfx> value follows the log-format
7418 rules and can include some dynamic values (see Custom Log Format
7419 in section 8.2.4).
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007420
7421 <sch> With "redirect scheme", then the "Location" header is built by
7422 concatenating <sch> with "://" then the first occurrence of the
7423 "Host" header, and then the URI path, including the query string
7424 unless the "drop-query" option is specified (see below). If no
7425 path is found or if the path is "*", then "/" is used instead. If
7426 no "Host" header is found, then an empty host component will be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03007427 returned, which most recent browsers interpret as redirecting to
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007428 the same host. This directive is mostly used to redirect HTTP to
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007429 HTTPS. When used in an "http-request" rule, <sch> value follows
7430 the log-format rules and can include some dynamic values (see
7431 Custom Log Format in section 8.2.4).
Willy Tarreau0140f252008-11-19 21:07:09 +01007432
7433 <code> The code is optional. It indicates which type of HTTP redirection
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007434 is desired. Only codes 301, 302, 303, 307 and 308 are supported,
7435 with 302 used by default if no code is specified. 301 means
7436 "Moved permanently", and a browser may cache the Location. 302
Baptiste Assmannea849c02015-08-03 11:42:50 +02007437 means "Moved temporarily" and means that the browser should not
Willy Tarreaub67fdc42013-03-29 19:28:11 +01007438 cache the redirection. 303 is equivalent to 302 except that the
7439 browser will fetch the location with a GET method. 307 is just
7440 like 302 but makes it clear that the same method must be reused.
7441 Likewise, 308 replaces 301 if the same method must be used.
Willy Tarreau0140f252008-11-19 21:07:09 +01007442
7443 <option> There are several options which can be specified to adjust the
Davor Ocelice9ed2812017-12-25 17:49:28 +01007444 expected behavior of a redirection :
Willy Tarreau0140f252008-11-19 21:07:09 +01007445
7446 - "drop-query"
7447 When this keyword is used in a prefix-based redirection, then the
7448 location will be set without any possible query-string, which is useful
7449 for directing users to a non-secure page for instance. It has no effect
7450 with a location-type redirect.
7451
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007452 - "append-slash"
7453 This keyword may be used in conjunction with "drop-query" to redirect
7454 users who use a URL not ending with a '/' to the same one with the '/'.
7455 It can be useful to ensure that search engines will only see one URL.
7456 For this, a return code 301 is preferred.
7457
Willy Tarreau0140f252008-11-19 21:07:09 +01007458 - "set-cookie NAME[=value]"
7459 A "Set-Cookie" header will be added with NAME (and optionally "=value")
7460 to the response. This is sometimes used to indicate that a user has
7461 been seen, for instance to protect against some types of DoS. No other
7462 cookie option is added, so the cookie will be a session cookie. Note
7463 that for a browser, a sole cookie name without an equal sign is
7464 different from a cookie with an equal sign.
7465
7466 - "clear-cookie NAME[=]"
7467 A "Set-Cookie" header will be added with NAME (and optionally "="), but
7468 with the "Max-Age" attribute set to zero. This will tell the browser to
7469 delete this cookie. It is useful for instance on logout pages. It is
7470 important to note that clearing the cookie "NAME" will not remove a
7471 cookie set with "NAME=value". You have to clear the cookie "NAME=" for
7472 that, because the browser makes the difference.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007473
7474 Example: move the login URL only to HTTPS.
7475 acl clear dst_port 80
7476 acl secure dst_port 8080
7477 acl login_page url_beg /login
Willy Tarreau0140f252008-11-19 21:07:09 +01007478 acl logout url_beg /logout
Willy Tarreau79da4692008-11-19 20:03:04 +01007479 acl uid_given url_reg /login?userid=[^&]+
Willy Tarreau0140f252008-11-19 21:07:09 +01007480 acl cookie_set hdr_sub(cookie) SEEN=1
7481
7482 redirect prefix https://mysite.com set-cookie SEEN=1 if !cookie_set
Willy Tarreau79da4692008-11-19 20:03:04 +01007483 redirect prefix https://mysite.com if login_page !secure
7484 redirect prefix http://mysite.com drop-query if login_page !uid_given
7485 redirect location http://mysite.com/ if !login_page secure
Willy Tarreau0140f252008-11-19 21:07:09 +01007486 redirect location / clear-cookie USERID= if logout
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007487
Willy Tarreau81e3b4f2010-01-10 00:42:19 +01007488 Example: send redirects for request for articles without a '/'.
7489 acl missing_slash path_reg ^/article/[^/]*$
7490 redirect code 301 prefix / drop-query append-slash if missing_slash
7491
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007492 Example: redirect all HTTP traffic to HTTPS when SSL is handled by haproxy.
David BERARDe7153042012-11-03 00:11:31 +01007493 redirect scheme https if !{ ssl_fc }
Willy Tarreau2e1dca82012-09-12 08:43:15 +02007494
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007495 Example: append 'www.' prefix in front of all hosts not having it
Coen Rosdorff596659b2016-04-11 11:33:49 +02007496 http-request redirect code 301 location \
7497 http://www.%[hdr(host)]%[capture.req.uri] \
7498 unless { hdr_beg(host) -i www }
Thierry FOURNIERd18cd0f2013-11-29 12:15:45 +01007499
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007500 See section 7 about ACL usage.
Willy Tarreaub463dfb2008-06-07 23:08:56 +02007501
7502
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007503redisp (deprecated)
7504redispatch (deprecated)
7505 Enable or disable session redistribution in case of connection failure
7506 May be used in sections: defaults | frontend | listen | backend
7507 yes | no | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007508 Arguments : none
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007509
7510 In HTTP mode, if a server designated by a cookie is down, clients may
7511 definitely stick to it because they cannot flush the cookie, so they will not
7512 be able to access the service anymore.
7513
7514 Specifying "redispatch" will allow the proxy to break their persistence and
7515 redistribute them to a working server.
7516
7517 It also allows to retry last connection to another server in case of multiple
7518 connection failures. Of course, it requires having "retries" set to a nonzero
7519 value.
Willy Tarreaud72758d2010-01-12 10:42:19 +01007520
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01007521 This form is deprecated, do not use it in any new configuration, use the new
7522 "option redispatch" instead.
7523
7524 See also : "option redispatch"
7525
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007526
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007527reqadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007528 Add a header at the end of the HTTP request
7529 May be used in sections : defaults | frontend | listen | backend
7530 no | yes | yes | yes
7531 Arguments :
7532 <string> is the complete line to be added. Any space or known delimiter
7533 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007534 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007535
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007536 <cond> is an optional matching condition built from ACLs. It makes it
7537 possible to ignore this rule when other conditions are not met.
7538
Willy Tarreau303c0352008-01-17 19:01:39 +01007539 A new line consisting in <string> followed by a line feed will be added after
7540 the last header of an HTTP request.
7541
7542 Header transformations only apply to traffic which passes through HAProxy,
7543 and not to traffic generated by HAProxy, such as health-checks or error
7544 responses.
7545
Willy Tarreau8abd4cd2010-01-31 14:30:44 +01007546 Example : add "X-Proto: SSL" to requests coming via port 81
7547 acl is-ssl dst_port 81
7548 reqadd X-Proto:\ SSL if is-ssl
7549
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007550 See also: "rspadd", "http-request", section 6 about HTTP header manipulation,
7551 and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007552
7553
Willy Tarreau5321c422010-01-28 20:35:13 +01007554reqallow <search> [{if | unless} <cond>]
7555reqiallow <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007556 Definitely allow an HTTP request if a line matches a regular expression
7557 May be used in sections : defaults | frontend | listen | backend
7558 no | yes | yes | yes
7559 Arguments :
7560 <search> is the regular expression applied to HTTP headers and to the
7561 request line. This is an extended regular expression. Parenthesis
7562 grouping is supported and no preliminary backslash is required.
7563 Any space or known delimiter must be escaped using a backslash
7564 ('\'). The pattern applies to a full line at a time. The
7565 "reqallow" keyword strictly matches case while "reqiallow"
7566 ignores case.
7567
Willy Tarreau5321c422010-01-28 20:35:13 +01007568 <cond> is an optional matching condition built from ACLs. It makes it
7569 possible to ignore this rule when other conditions are not met.
7570
Willy Tarreau303c0352008-01-17 19:01:39 +01007571 A request containing any line which matches extended regular expression
7572 <search> will mark the request as allowed, even if any later test would
7573 result in a deny. The test applies both to the request line and to request
7574 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007575 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007576
7577 It is easier, faster and more powerful to use ACLs to write access policies.
7578 Reqdeny, reqallow and reqpass should be avoided in new designs.
7579
7580 Example :
7581 # allow www.* but refuse *.local
7582 reqiallow ^Host:\ www\.
7583 reqideny ^Host:\ .*\.local
7584
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007585 See also: "reqdeny", "block", "http-request", section 6 about HTTP header
7586 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007587
7588
Willy Tarreau5321c422010-01-28 20:35:13 +01007589reqdel <search> [{if | unless} <cond>]
7590reqidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007591 Delete all headers matching a regular expression in an HTTP request
7592 May be used in sections : defaults | frontend | listen | backend
7593 no | yes | yes | yes
7594 Arguments :
7595 <search> is the regular expression applied to HTTP headers and to the
7596 request line. This is an extended regular expression. Parenthesis
7597 grouping is supported and no preliminary backslash is required.
7598 Any space or known delimiter must be escaped using a backslash
7599 ('\'). The pattern applies to a full line at a time. The "reqdel"
7600 keyword strictly matches case while "reqidel" ignores case.
7601
Willy Tarreau5321c422010-01-28 20:35:13 +01007602 <cond> is an optional matching condition built from ACLs. It makes it
7603 possible to ignore this rule when other conditions are not met.
7604
Willy Tarreau303c0352008-01-17 19:01:39 +01007605 Any header line matching extended regular expression <search> in the request
7606 will be completely deleted. Most common use of this is to remove unwanted
7607 and/or dangerous headers or cookies from a request before passing it to the
7608 next servers.
7609
7610 Header transformations only apply to traffic which passes through HAProxy,
7611 and not to traffic generated by HAProxy, such as health-checks or error
7612 responses. Keep in mind that header names are not case-sensitive.
7613
7614 Example :
7615 # remove X-Forwarded-For header and SERVER cookie
7616 reqidel ^X-Forwarded-For:.*
7617 reqidel ^Cookie:.*SERVER=
7618
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007619 See also: "reqadd", "reqrep", "rspdel", "http-request", section 6 about
7620 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007621
7622
Willy Tarreau5321c422010-01-28 20:35:13 +01007623reqdeny <search> [{if | unless} <cond>]
7624reqideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007625 Deny an HTTP request if a line matches a regular expression
7626 May be used in sections : defaults | frontend | listen | backend
7627 no | yes | yes | yes
7628 Arguments :
7629 <search> is the regular expression applied to HTTP headers and to the
7630 request line. This is an extended regular expression. Parenthesis
7631 grouping is supported and no preliminary backslash is required.
7632 Any space or known delimiter must be escaped using a backslash
7633 ('\'). The pattern applies to a full line at a time. The
7634 "reqdeny" keyword strictly matches case while "reqideny" ignores
7635 case.
7636
Willy Tarreau5321c422010-01-28 20:35:13 +01007637 <cond> is an optional matching condition built from ACLs. It makes it
7638 possible to ignore this rule when other conditions are not met.
7639
Willy Tarreau303c0352008-01-17 19:01:39 +01007640 A request containing any line which matches extended regular expression
7641 <search> will mark the request as denied, even if any later test would
7642 result in an allow. The test applies both to the request line and to request
7643 headers. Keep in mind that URLs in request line are case-sensitive while
Willy Tarreaud72758d2010-01-12 10:42:19 +01007644 header names are not.
Willy Tarreau303c0352008-01-17 19:01:39 +01007645
Willy Tarreauced27012008-01-17 20:35:34 +01007646 A denied request will generate an "HTTP 403 forbidden" response once the
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01007647 complete request has been parsed. This is consistent with what is practiced
Willy Tarreaud72758d2010-01-12 10:42:19 +01007648 using ACLs.
Willy Tarreauced27012008-01-17 20:35:34 +01007649
Willy Tarreau303c0352008-01-17 19:01:39 +01007650 It is easier, faster and more powerful to use ACLs to write access policies.
7651 Reqdeny, reqallow and reqpass should be avoided in new designs.
7652
7653 Example :
7654 # refuse *.local, then allow www.*
7655 reqideny ^Host:\ .*\.local
7656 reqiallow ^Host:\ www\.
7657
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007658 See also: "reqallow", "rspdeny", "block", "http-request", section 6 about
7659 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007660
7661
Willy Tarreau5321c422010-01-28 20:35:13 +01007662reqpass <search> [{if | unless} <cond>]
7663reqipass <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007664 Ignore any HTTP request line matching a regular expression in next rules
7665 May be used in sections : defaults | frontend | listen | backend
7666 no | yes | yes | yes
7667 Arguments :
7668 <search> is the regular expression applied to HTTP headers and to the
7669 request line. This is an extended regular expression. Parenthesis
7670 grouping is supported and no preliminary backslash is required.
7671 Any space or known delimiter must be escaped using a backslash
7672 ('\'). The pattern applies to a full line at a time. The
7673 "reqpass" keyword strictly matches case while "reqipass" ignores
7674 case.
7675
Willy Tarreau5321c422010-01-28 20:35:13 +01007676 <cond> is an optional matching condition built from ACLs. It makes it
7677 possible to ignore this rule when other conditions are not met.
7678
Willy Tarreau303c0352008-01-17 19:01:39 +01007679 A request containing any line which matches extended regular expression
7680 <search> will skip next rules, without assigning any deny or allow verdict.
7681 The test applies both to the request line and to request headers. Keep in
7682 mind that URLs in request line are case-sensitive while header names are not.
7683
7684 It is easier, faster and more powerful to use ACLs to write access policies.
7685 Reqdeny, reqallow and reqpass should be avoided in new designs.
7686
7687 Example :
7688 # refuse *.local, then allow www.*, but ignore "www.private.local"
7689 reqipass ^Host:\ www.private\.local
7690 reqideny ^Host:\ .*\.local
7691 reqiallow ^Host:\ www\.
7692
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007693 See also: "reqallow", "reqdeny", "block", "http-request", section 6 about
7694 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007695
7696
Willy Tarreau5321c422010-01-28 20:35:13 +01007697reqrep <search> <string> [{if | unless} <cond>]
7698reqirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007699 Replace a regular expression with a string in an HTTP request line
7700 May be used in sections : defaults | frontend | listen | backend
7701 no | yes | yes | yes
7702 Arguments :
7703 <search> is the regular expression applied to HTTP headers and to the
7704 request line. This is an extended regular expression. Parenthesis
7705 grouping is supported and no preliminary backslash is required.
7706 Any space or known delimiter must be escaped using a backslash
7707 ('\'). The pattern applies to a full line at a time. The "reqrep"
7708 keyword strictly matches case while "reqirep" ignores case.
7709
7710 <string> is the complete line to be added. Any space or known delimiter
7711 must be escaped using a backslash ('\'). References to matched
7712 pattern groups are possible using the common \N form, with N
7713 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007714 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007715
Willy Tarreau5321c422010-01-28 20:35:13 +01007716 <cond> is an optional matching condition built from ACLs. It makes it
7717 possible to ignore this rule when other conditions are not met.
7718
Willy Tarreau303c0352008-01-17 19:01:39 +01007719 Any line matching extended regular expression <search> in the request (both
7720 the request line and header lines) will be completely replaced with <string>.
7721 Most common use of this is to rewrite URLs or domain names in "Host" headers.
7722
7723 Header transformations only apply to traffic which passes through HAProxy,
7724 and not to traffic generated by HAProxy, such as health-checks or error
7725 responses. Note that for increased readability, it is suggested to add enough
7726 spaces between the request and the response. Keep in mind that URLs in
7727 request line are case-sensitive while header names are not.
7728
7729 Example :
7730 # replace "/static/" with "/" at the beginning of any request path.
Dmitry Sivachenko7823de32012-05-16 14:00:26 +04007731 reqrep ^([^\ :]*)\ /static/(.*) \1\ /\2
Willy Tarreau303c0352008-01-17 19:01:39 +01007732 # replace "www.mydomain.com" with "www" in the host name.
7733 reqirep ^Host:\ www.mydomain.com Host:\ www
7734
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007735 See also: "reqadd", "reqdel", "rsprep", "tune.bufsize", "http-request",
7736 section 6 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007737
7738
Willy Tarreau5321c422010-01-28 20:35:13 +01007739reqtarpit <search> [{if | unless} <cond>]
7740reqitarpit <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007741 Tarpit an HTTP request containing a line matching a regular expression
7742 May be used in sections : defaults | frontend | listen | backend
7743 no | yes | yes | yes
7744 Arguments :
7745 <search> is the regular expression applied to HTTP headers and to the
7746 request line. This is an extended regular expression. Parenthesis
7747 grouping is supported and no preliminary backslash is required.
7748 Any space or known delimiter must be escaped using a backslash
7749 ('\'). The pattern applies to a full line at a time. The
7750 "reqtarpit" keyword strictly matches case while "reqitarpit"
7751 ignores case.
7752
Willy Tarreau5321c422010-01-28 20:35:13 +01007753 <cond> is an optional matching condition built from ACLs. It makes it
7754 possible to ignore this rule when other conditions are not met.
7755
Willy Tarreau303c0352008-01-17 19:01:39 +01007756 A request containing any line which matches extended regular expression
7757 <search> will be tarpitted, which means that it will connect to nowhere, will
Willy Tarreauced27012008-01-17 20:35:34 +01007758 be kept open for a pre-defined time, then will return an HTTP error 500 so
7759 that the attacker does not suspect it has been tarpitted. The status 500 will
7760 be reported in the logs, but the completion flags will indicate "PT". The
Willy Tarreau303c0352008-01-17 19:01:39 +01007761 delay is defined by "timeout tarpit", or "timeout connect" if the former is
7762 not set.
7763
7764 The goal of the tarpit is to slow down robots attacking servers with
7765 identifiable requests. Many robots limit their outgoing number of connections
7766 and stay connected waiting for a reply which can take several minutes to
7767 come. Depending on the environment and attack, it may be particularly
7768 efficient at reducing the load on the network and firewalls.
7769
Willy Tarreau5321c422010-01-28 20:35:13 +01007770 Examples :
Davor Ocelice9ed2812017-12-25 17:49:28 +01007771 # ignore user-agents reporting any flavor of "Mozilla" or "MSIE", but
Willy Tarreau303c0352008-01-17 19:01:39 +01007772 # block all others.
7773 reqipass ^User-Agent:\.*(Mozilla|MSIE)
7774 reqitarpit ^User-Agent:
7775
Willy Tarreau5321c422010-01-28 20:35:13 +01007776 # block bad guys
7777 acl badguys src 10.1.0.3 172.16.13.20/28
7778 reqitarpit . if badguys
7779
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007780 See also: "reqallow", "reqdeny", "reqpass", "http-request", section 6
7781 about HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007782
7783
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007784retries <value>
7785 Set the number of retries to perform on a server after a connection failure
7786 May be used in sections: defaults | frontend | listen | backend
7787 yes | no | yes | yes
7788 Arguments :
7789 <value> is the number of times a connection attempt should be retried on
7790 a server when a connection either is refused or times out. The
7791 default value is 3.
7792
7793 It is important to understand that this value applies to the number of
7794 connection attempts, not full requests. When a connection has effectively
7795 been established to a server, there will be no more retry.
7796
7797 In order to avoid immediate reconnections to a server which is restarting,
Joseph Lynch726ab712015-05-11 23:25:34 -07007798 a turn-around timer of min("timeout connect", one second) is applied before
7799 a retry occurs.
Willy Tarreaue5c5ce92008-06-20 17:27:19 +02007800
7801 When "option redispatch" is set, the last retry may be performed on another
7802 server even if a cookie references a different server.
7803
7804 See also : "option redispatch"
7805
7806
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007807rspadd <string> [{if | unless} <cond>]
Willy Tarreau303c0352008-01-17 19:01:39 +01007808 Add a header at the end of the HTTP response
7809 May be used in sections : defaults | frontend | listen | backend
7810 no | yes | yes | yes
7811 Arguments :
7812 <string> is the complete line to be added. Any space or known delimiter
7813 must be escaped using a backslash ('\'). Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007814 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007815
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007816 <cond> is an optional matching condition built from ACLs. It makes it
7817 possible to ignore this rule when other conditions are not met.
7818
Willy Tarreau303c0352008-01-17 19:01:39 +01007819 A new line consisting in <string> followed by a line feed will be added after
7820 the last header of an HTTP response.
7821
7822 Header transformations only apply to traffic which passes through HAProxy,
7823 and not to traffic generated by HAProxy, such as health-checks or error
7824 responses.
7825
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007826 See also: "rspdel" "reqadd", "http-response", section 6 about HTTP header
7827 manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007828
7829
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007830rspdel <search> [{if | unless} <cond>]
7831rspidel <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007832 Delete all headers matching a regular expression in an HTTP response
7833 May be used in sections : defaults | frontend | listen | backend
7834 no | yes | yes | yes
7835 Arguments :
7836 <search> is the regular expression applied to HTTP headers and to the
7837 response line. This is an extended regular expression, so
7838 parenthesis grouping is supported and no preliminary backslash
7839 is required. Any space or known delimiter must be escaped using
7840 a backslash ('\'). The pattern applies to a full line at a time.
7841 The "rspdel" keyword strictly matches case while "rspidel"
7842 ignores case.
7843
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007844 <cond> is an optional matching condition built from ACLs. It makes it
7845 possible to ignore this rule when other conditions are not met.
7846
Willy Tarreau303c0352008-01-17 19:01:39 +01007847 Any header line matching extended regular expression <search> in the response
7848 will be completely deleted. Most common use of this is to remove unwanted
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02007849 and/or sensitive headers or cookies from a response before passing it to the
Willy Tarreau303c0352008-01-17 19:01:39 +01007850 client.
7851
7852 Header transformations only apply to traffic which passes through HAProxy,
7853 and not to traffic generated by HAProxy, such as health-checks or error
7854 responses. Keep in mind that header names are not case-sensitive.
7855
7856 Example :
7857 # remove the Server header from responses
Willy Tarreau5e80e022013-05-25 08:31:25 +02007858 rspidel ^Server:.*
Willy Tarreau303c0352008-01-17 19:01:39 +01007859
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007860 See also: "rspadd", "rsprep", "reqdel", "http-response", section 6 about
7861 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007862
7863
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007864rspdeny <search> [{if | unless} <cond>]
7865rspideny <search> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007866 Block an HTTP response if a line matches a regular expression
7867 May be used in sections : defaults | frontend | listen | backend
7868 no | yes | yes | yes
7869 Arguments :
7870 <search> is the regular expression applied to HTTP headers and to the
7871 response line. This is an extended regular expression, so
7872 parenthesis grouping is supported and no preliminary backslash
7873 is required. Any space or known delimiter must be escaped using
7874 a backslash ('\'). The pattern applies to a full line at a time.
7875 The "rspdeny" keyword strictly matches case while "rspideny"
7876 ignores case.
7877
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007878 <cond> is an optional matching condition built from ACLs. It makes it
7879 possible to ignore this rule when other conditions are not met.
7880
Willy Tarreau303c0352008-01-17 19:01:39 +01007881 A response containing any line which matches extended regular expression
7882 <search> will mark the request as denied. The test applies both to the
7883 response line and to response headers. Keep in mind that header names are not
7884 case-sensitive.
7885
7886 Main use of this keyword is to prevent sensitive information leak and to
Willy Tarreauced27012008-01-17 20:35:34 +01007887 block the response before it reaches the client. If a response is denied, it
7888 will be replaced with an HTTP 502 error so that the client never retrieves
7889 any sensitive data.
Willy Tarreau303c0352008-01-17 19:01:39 +01007890
7891 It is easier, faster and more powerful to use ACLs to write access policies.
7892 Rspdeny should be avoided in new designs.
7893
7894 Example :
7895 # Ensure that no content type matching ms-word will leak
7896 rspideny ^Content-type:\.*/ms-word
7897
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007898 See also: "reqdeny", "acl", "block", "http-response", section 6 about
7899 HTTP header manipulation and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007900
7901
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007902rsprep <search> <string> [{if | unless} <cond>]
7903rspirep <search> <string> [{if | unless} <cond>] (ignore case)
Willy Tarreau303c0352008-01-17 19:01:39 +01007904 Replace a regular expression with a string in an HTTP response line
7905 May be used in sections : defaults | frontend | listen | backend
7906 no | yes | yes | yes
7907 Arguments :
7908 <search> is the regular expression applied to HTTP headers and to the
7909 response line. This is an extended regular expression, so
7910 parenthesis grouping is supported and no preliminary backslash
7911 is required. Any space or known delimiter must be escaped using
7912 a backslash ('\'). The pattern applies to a full line at a time.
7913 The "rsprep" keyword strictly matches case while "rspirep"
7914 ignores case.
7915
7916 <string> is the complete line to be added. Any space or known delimiter
7917 must be escaped using a backslash ('\'). References to matched
7918 pattern groups are possible using the common \N form, with N
7919 being a single digit between 0 and 9. Please refer to section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007920 6 about HTTP header manipulation for more information.
Willy Tarreau303c0352008-01-17 19:01:39 +01007921
Willy Tarreaufdb563c2010-01-31 15:43:27 +01007922 <cond> is an optional matching condition built from ACLs. It makes it
7923 possible to ignore this rule when other conditions are not met.
7924
Willy Tarreau303c0352008-01-17 19:01:39 +01007925 Any line matching extended regular expression <search> in the response (both
7926 the response line and header lines) will be completely replaced with
7927 <string>. Most common use of this is to rewrite Location headers.
7928
7929 Header transformations only apply to traffic which passes through HAProxy,
7930 and not to traffic generated by HAProxy, such as health-checks or error
7931 responses. Note that for increased readability, it is suggested to add enough
7932 spaces between the request and the response. Keep in mind that header names
7933 are not case-sensitive.
7934
7935 Example :
7936 # replace "Location: 127.0.0.1:8080" with "Location: www.mydomain.com"
7937 rspirep ^Location:\ 127.0.0.1:8080 Location:\ www.mydomain.com
7938
Ruoshan Huangeb5a3632015-12-08 21:00:23 +08007939 See also: "rspadd", "rspdel", "reqrep", "http-response", section 6 about
7940 HTTP header manipulation, and section 7 about ACLs.
Willy Tarreau303c0352008-01-17 19:01:39 +01007941
7942
David du Colombier486df472011-03-17 10:40:26 +01007943server <name> <address>[:[port]] [param*]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007944 Declare a server in a backend
7945 May be used in sections : defaults | frontend | listen | backend
7946 no | no | yes | yes
7947 Arguments :
7948 <name> is the internal name assigned to this server. This name will
Davor Ocelice9ed2812017-12-25 17:49:28 +01007949 appear in logs and alerts. If "http-send-name-header" is
Mark Lamourinec2247f02012-01-04 13:02:01 -05007950 set, it will be added to the request header sent to the server.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007951
David du Colombier486df472011-03-17 10:40:26 +01007952 <address> is the IPv4 or IPv6 address of the server. Alternatively, a
7953 resolvable hostname is supported, but this name will be resolved
7954 during start-up. Address "0.0.0.0" or "*" has a special meaning.
7955 It indicates that the connection will be forwarded to the same IP
Willy Tarreaud669a4f2010-07-13 14:49:50 +02007956 address as the one from the client connection. This is useful in
7957 transparent proxy architectures where the client's connection is
7958 intercepted and haproxy must forward to the original destination
7959 address. This is more or less what the "transparent" keyword does
7960 except that with a server it's possible to limit concurrency and
Willy Tarreau24709282013-03-10 21:32:12 +01007961 to report statistics. Optionally, an address family prefix may be
7962 used before the address to force the family regardless of the
7963 address format, which can be useful to specify a path to a unix
7964 socket with no slash ('/'). Currently supported prefixes are :
7965 - 'ipv4@' -> address is always IPv4
7966 - 'ipv6@' -> address is always IPv6
7967 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02007968 - 'abns@' -> address is in abstract namespace (Linux only)
William Lallemand2fe7dd02018-09-11 16:51:29 +02007969 - 'sockpair@' -> address is the FD of a connected unix
7970 socket or of a socketpair. During a connection, the
7971 backend creates a pair of connected sockets, and passes
7972 one of them over the FD. The bind part will use the
7973 received socket as the client FD. Should be used
7974 carefully.
William Lallemandb2f07452015-05-12 14:27:13 +02007975 You may want to reference some environment variables in the
7976 address parameter, see section 2.3 about environment
Willy Tarreau6a031d12016-11-07 19:42:35 +01007977 variables. The "init-addr" setting can be used to modify the way
7978 IP addresses should be resolved upon startup.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007979
Willy Tarreaub6205fd2012-09-24 12:27:33 +02007980 <port> is an optional port specification. If set, all connections will
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007981 be sent to this port. If unset, the same port the client
7982 connected to will be used. The port may also be prefixed by a "+"
7983 or a "-". In this case, the server's port will be determined by
7984 adding this value to the client's port.
7985
7986 <param*> is a list of parameters for this server. The "server" keywords
7987 accepts an important number of options and has a complete section
Willy Tarreauc57f0e22009-05-10 13:12:33 +02007988 dedicated to it. Please refer to section 5 for more details.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007989
7990 Examples :
7991 server first 10.1.1.1:1080 cookie first check inter 1000
7992 server second 10.1.1.2:1080 cookie second check inter 1000
Willy Tarreau24709282013-03-10 21:32:12 +01007993 server transp ipv4@
William Lallemandb2f07452015-05-12 14:27:13 +02007994 server backup "${SRV_BACKUP}:1080" backup
7995 server www1_dc1 "${LAN_DC1}.101:80"
7996 server www1_dc2 "${LAN_DC2}.101:80"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01007997
Willy Tarreau55dcaf62015-09-27 15:03:15 +02007998 Note: regarding Linux's abstract namespace sockets, HAProxy uses the whole
7999 sun_path length is used for the address length. Some other programs
8000 such as socat use the string length only by default. Pass the option
8001 ",unix-tightsocklen=0" to any abstract socket definition in socat to
8002 make it compatible with HAProxy's.
8003
Mark Lamourinec2247f02012-01-04 13:02:01 -05008004 See also: "default-server", "http-send-name-header" and section 5 about
8005 server options
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008006
Baptiste Assmann01c6cc32015-08-23 11:45:29 +02008007server-state-file-name [<file>]
8008 Set the server state file to read, load and apply to servers available in
8009 this backend. It only applies when the directive "load-server-state-from-file"
8010 is set to "local". When <file> is not provided or if this directive is not
8011 set, then backend name is used. If <file> starts with a slash '/', then it is
8012 considered as an absolute path. Otherwise, <file> is concatenated to the
8013 global directive "server-state-file-base".
8014
8015 Example: the minimal configuration below would make HAProxy look for the
8016 state server file '/etc/haproxy/states/bk':
8017
8018 global
8019 server-state-file-base /etc/haproxy/states
8020
8021 backend bk
8022 load-server-state-from-file
8023
8024 See also: "server-state-file-base", "load-server-state-from-file", and
8025 "show servers state"
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008026
Frédéric Lécaillecb4502e2017-04-20 13:36:25 +02008027server-template <prefix> <num | range> <fqdn>[:<port>] [params*]
8028 Set a template to initialize servers with shared parameters.
8029 The names of these servers are built from <prefix> and <num | range> parameters.
8030 May be used in sections : defaults | frontend | listen | backend
8031 no | no | yes | yes
8032
8033 Arguments:
8034 <prefix> A prefix for the server names to be built.
8035
8036 <num | range>
8037 If <num> is provided, this template initializes <num> servers
8038 with 1 up to <num> as server name suffixes. A range of numbers
8039 <num_low>-<num_high> may also be used to use <num_low> up to
8040 <num_high> as server name suffixes.
8041
8042 <fqdn> A FQDN for all the servers this template initializes.
8043
8044 <port> Same meaning as "server" <port> argument (see "server" keyword).
8045
8046 <params*>
8047 Remaining server parameters among all those supported by "server"
8048 keyword.
8049
8050 Examples:
8051 # Initializes 3 servers with srv1, srv2 and srv3 as names,
8052 # google.com as FQDN, and health-check enabled.
8053 server-template srv 1-3 google.com:80 check
8054
8055 # or
8056 server-template srv 3 google.com:80 check
8057
8058 # would be equivalent to:
8059 server srv1 google.com:80 check
8060 server srv2 google.com:80 check
8061 server srv3 google.com:80 check
8062
8063
8064
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008065source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +02008066source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008067source <addr>[:<port>] [interface <name>]
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008068 Set the source address for outgoing connections
8069 May be used in sections : defaults | frontend | listen | backend
8070 yes | no | yes | yes
8071 Arguments :
8072 <addr> is the IPv4 address HAProxy will bind to before connecting to a
8073 server. This address is also used as a source for health checks.
Willy Tarreau24709282013-03-10 21:32:12 +01008074
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008075 The default value of 0.0.0.0 means that the system will select
Willy Tarreau24709282013-03-10 21:32:12 +01008076 the most appropriate address to reach its destination. Optionally
8077 an address family prefix may be used before the address to force
8078 the family regardless of the address format, which can be useful
8079 to specify a path to a unix socket with no slash ('/'). Currently
8080 supported prefixes are :
8081 - 'ipv4@' -> address is always IPv4
8082 - 'ipv6@' -> address is always IPv6
8083 - 'unix@' -> address is a path to a local unix socket
Willy Tarreauccfccef2014-05-10 01:49:15 +02008084 - 'abns@' -> address is in abstract namespace (Linux only)
Cyril Bonté307ee1e2015-09-28 23:16:06 +02008085 You may want to reference some environment variables in the
8086 address parameter, see section 2.3 about environment variables.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008087
8088 <port> is an optional port. It is normally not needed but may be useful
8089 in some very specific contexts. The default value of zero means
Willy Tarreauc6f4ce82009-06-10 11:09:37 +02008090 the system will select a free port. Note that port ranges are not
8091 supported in the backend. If you want to force port ranges, you
8092 have to specify them on each "server" line.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008093
8094 <addr2> is the IP address to present to the server when connections are
8095 forwarded in full transparent proxy mode. This is currently only
8096 supported on some patched Linux kernels. When this address is
8097 specified, clients connecting to the server will be presented
8098 with this address, while health checks will still use the address
8099 <addr>.
8100
8101 <port2> is the optional port to present to the server when connections
8102 are forwarded in full transparent proxy mode (see <addr2> above).
8103 The default value of zero means the system will select a free
8104 port.
8105
Willy Tarreaubce70882009-09-07 11:51:47 +02008106 <hdr> is the name of a HTTP header in which to fetch the IP to bind to.
8107 This is the name of a comma-separated header list which can
8108 contain multiple IP addresses. By default, the last occurrence is
8109 used. This is designed to work with the X-Forwarded-For header
Baptiste Assmannea3e73b2013-02-02 23:47:49 +01008110 and to automatically bind to the client's IP address as seen
Willy Tarreaubce70882009-09-07 11:51:47 +02008111 by previous proxy, typically Stunnel. In order to use another
8112 occurrence from the last one, please see the <occ> parameter
8113 below. When the header (or occurrence) is not found, no binding
8114 is performed so that the proxy's default IP address is used. Also
8115 keep in mind that the header name is case insensitive, as for any
8116 HTTP header.
8117
8118 <occ> is the occurrence number of a value to be used in a multi-value
8119 header. This is to be used in conjunction with "hdr_ip(<hdr>)",
Jamie Gloudonaaa21002012-08-25 00:18:33 -04008120 in order to specify which occurrence to use for the source IP
Willy Tarreaubce70882009-09-07 11:51:47 +02008121 address. Positive values indicate a position from the first
8122 occurrence, 1 being the first one. Negative values indicate
8123 positions relative to the last one, -1 being the last one. This
8124 is helpful for situations where an X-Forwarded-For header is set
8125 at the entry point of an infrastructure and must be used several
8126 proxy layers away. When this value is not specified, -1 is
8127 assumed. Passing a zero here disables the feature.
8128
Willy Tarreaud53f96b2009-02-04 18:46:54 +01008129 <name> is an optional interface name to which to bind to for outgoing
8130 traffic. On systems supporting this features (currently, only
8131 Linux), this allows one to bind all traffic to the server to
8132 this interface even if it is not the one the system would select
8133 based on routing tables. This should be used with extreme care.
8134 Note that using this option requires root privileges.
8135
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008136 The "source" keyword is useful in complex environments where a specific
8137 address only is allowed to connect to the servers. It may be needed when a
8138 private address must be used through a public gateway for instance, and it is
8139 known that the system cannot determine the adequate source address by itself.
8140
8141 An extension which is available on certain patched Linux kernels may be used
8142 through the "usesrc" optional keyword. It makes it possible to connect to the
8143 servers with an IP address which does not belong to the system itself. This
8144 is called "full transparent proxy mode". For this to work, the destination
8145 servers have to route their traffic back to this address through the machine
8146 running HAProxy, and IP forwarding must generally be enabled on this machine.
8147
8148 In this "full transparent proxy" mode, it is possible to force a specific IP
8149 address to be presented to the servers. This is not much used in fact. A more
8150 common use is to tell HAProxy to present the client's IP address. For this,
8151 there are two methods :
8152
8153 - present the client's IP and port addresses. This is the most transparent
8154 mode, but it can cause problems when IP connection tracking is enabled on
8155 the machine, because a same connection may be seen twice with different
8156 states. However, this solution presents the huge advantage of not
8157 limiting the system to the 64k outgoing address+port couples, because all
8158 of the client ranges may be used.
8159
8160 - present only the client's IP address and select a spare port. This
8161 solution is still quite elegant but slightly less transparent (downstream
8162 firewalls logs will not match upstream's). It also presents the downside
8163 of limiting the number of concurrent connections to the usual 64k ports.
8164 However, since the upstream and downstream ports are different, local IP
8165 connection tracking on the machine will not be upset by the reuse of the
8166 same session.
8167
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008168 This option sets the default source for all servers in the backend. It may
8169 also be specified in a "defaults" section. Finer source address specification
8170 is possible at the server level using the "source" server option. Refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008171 section 5 for more information.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008172
Baptiste Assmann91bd3372015-07-17 21:59:42 +02008173 In order to work, "usesrc" requires root privileges.
8174
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008175 Examples :
8176 backend private
8177 # Connect to the servers using our 192.168.1.200 source address
8178 source 192.168.1.200
8179
8180 backend transparent_ssl1
8181 # Connect to the SSL farm from the client's source address
8182 source 192.168.1.200 usesrc clientip
8183
8184 backend transparent_ssl2
8185 # Connect to the SSL farm from the client's source address and port
8186 # not recommended if IP conntrack is present on the local machine.
8187 source 192.168.1.200 usesrc client
8188
8189 backend transparent_ssl3
8190 # Connect to the SSL farm from the client's source address. It
8191 # is more conntrack-friendly.
8192 source 192.168.1.200 usesrc clientip
8193
8194 backend transparent_smtp
8195 # Connect to the SMTP farm from the client's source address/port
8196 # with Tproxy version 4.
8197 source 0.0.0.0 usesrc clientip
8198
Willy Tarreaubce70882009-09-07 11:51:47 +02008199 backend transparent_http
8200 # Connect to the servers using the client's IP as seen by previous
8201 # proxy.
8202 source 0.0.0.0 usesrc hdr_ip(x-forwarded-for,-1)
8203
Willy Tarreauc57f0e22009-05-10 13:12:33 +02008204 See also : the "source" server option in section 5, the Tproxy patches for
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008205 the Linux kernel on www.balabit.com, the "bind" keyword.
8206
Krzysztof Piotr Oledzki25b501a2008-01-06 16:36:16 +01008207
Willy Tarreau844e3c52008-01-11 16:28:18 +01008208srvtimeout <timeout> (deprecated)
8209 Set the maximum inactivity time on the server side.
8210 May be used in sections : defaults | frontend | listen | backend
8211 yes | no | yes | yes
8212 Arguments :
8213 <timeout> is the timeout value specified in milliseconds by default, but
8214 can be in any other unit if the number is suffixed by the unit,
8215 as explained at the top of this document.
8216
8217 The inactivity timeout applies when the server is expected to acknowledge or
8218 send data. In HTTP mode, this timeout is particularly important to consider
8219 during the first phase of the server's response, when it has to send the
8220 headers, as it directly represents the server's processing time for the
8221 request. To find out what value to put there, it's often good to start with
8222 what would be considered as unacceptable response times, then check the logs
8223 to observe the response time distribution, and adjust the value accordingly.
8224
8225 The value is specified in milliseconds by default, but can be in any other
8226 unit if the number is suffixed by the unit, as specified at the top of this
8227 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
8228 recommended that the client timeout remains equal to the server timeout in
8229 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +01008230 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +01008231 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +01008232 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreau844e3c52008-01-11 16:28:18 +01008233
8234 This parameter is specific to backends, but can be specified once for all in
8235 "defaults" sections. This is in fact one of the easiest solutions not to
8236 forget about it. An unspecified timeout results in an infinite timeout, which
8237 is not recommended. Such a usage is accepted and works but reports a warning
8238 during startup because it may results in accumulation of expired sessions in
8239 the system if the system's timeouts are not configured either.
8240
8241 This parameter is provided for compatibility but is currently deprecated.
8242 Please use "timeout server" instead.
8243
Willy Tarreauce887fd2012-05-12 12:50:00 +02008244 See also : "timeout server", "timeout tunnel", "timeout client" and
8245 "clitimeout".
Willy Tarreau844e3c52008-01-11 16:28:18 +01008246
8247
Cyril Bonté66c327d2010-10-12 00:14:37 +02008248stats admin { if | unless } <cond>
8249 Enable statistics admin level if/unless a condition is matched
8250 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008251 no | yes | yes | yes
Cyril Bonté66c327d2010-10-12 00:14:37 +02008252
8253 This statement enables the statistics admin level if/unless a condition is
8254 matched.
8255
8256 The admin level allows to enable/disable servers from the web interface. By
8257 default, statistics page is read-only for security reasons.
8258
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008259 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8260 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008261 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008262
Cyril Bonté23b39d92011-02-10 22:54:44 +01008263 Currently, the POST request is limited to the buffer size minus the reserved
8264 buffer space, which means that if the list of servers is too long, the
8265 request won't be processed. It is recommended to alter few servers at a
8266 time.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008267
8268 Example :
8269 # statistics admin level only for localhost
8270 backend stats_localhost
8271 stats enable
8272 stats admin if LOCALHOST
8273
8274 Example :
8275 # statistics admin level always enabled because of the authentication
8276 backend stats_auth
8277 stats enable
8278 stats auth admin:AdMiN123
8279 stats admin if TRUE
8280
8281 Example :
8282 # statistics admin level depends on the authenticated user
8283 userlist stats-auth
8284 group admin users admin
8285 user admin insecure-password AdMiN123
8286 group readonly users haproxy
8287 user haproxy insecure-password haproxy
8288
8289 backend stats_auth
8290 stats enable
8291 acl AUTH http_auth(stats-auth)
8292 acl AUTH_ADMIN http_auth_group(stats-auth) admin
8293 stats http-request auth unless AUTH
8294 stats admin if AUTH_ADMIN
8295
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008296 See also : "stats enable", "stats auth", "stats http-request", "nbproc",
8297 "bind-process", section 3.4 about userlists and section 7 about
8298 ACL usage.
Cyril Bonté66c327d2010-10-12 00:14:37 +02008299
8300
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008301stats auth <user>:<passwd>
8302 Enable statistics with authentication and grant access to an account
8303 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008304 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008305 Arguments :
8306 <user> is a user name to grant access to
8307
8308 <passwd> is the cleartext password associated to this user
8309
8310 This statement enables statistics with default settings, and restricts access
8311 to declared users only. It may be repeated as many times as necessary to
8312 allow as many users as desired. When a user tries to access the statistics
8313 without a valid account, a "401 Forbidden" response will be returned so that
8314 the browser asks the user to provide a valid user and password. The real
8315 which will be returned to the browser is configurable using "stats realm".
8316
8317 Since the authentication method is HTTP Basic Authentication, the passwords
8318 circulate in cleartext on the network. Thus, it was decided that the
8319 configuration file would also use cleartext passwords to remind the users
Willy Tarreau3c92c5f2011-08-28 09:45:47 +02008320 that those ones should not be sensitive and not shared with any other account.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008321
8322 It is also possible to reduce the scope of the proxies which appear in the
8323 report using "stats scope".
8324
8325 Though this statement alone is enough to enable statistics reporting, it is
8326 recommended to set all other settings in order to avoid relying on default
8327 unobvious parameters.
8328
8329 Example :
8330 # public access (limited to this backend only)
8331 backend public_www
8332 server srv1 192.168.0.1:80
8333 stats enable
8334 stats hide-version
8335 stats scope .
8336 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008337 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008338 stats auth admin1:AdMiN123
8339 stats auth admin2:AdMiN321
8340
8341 # internal monitoring access (unlimited)
8342 backend private_monitoring
8343 stats enable
8344 stats uri /admin?stats
8345 stats refresh 5s
8346
8347 See also : "stats enable", "stats realm", "stats scope", "stats uri"
8348
8349
8350stats enable
8351 Enable statistics reporting with default settings
8352 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008353 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008354 Arguments : none
8355
8356 This statement enables statistics reporting with default settings defined
8357 at build time. Unless stated otherwise, these settings are used :
8358 - stats uri : /haproxy?stats
8359 - stats realm : "HAProxy Statistics"
8360 - stats auth : no authentication
8361 - stats scope : no restriction
8362
8363 Though this statement alone is enough to enable statistics reporting, it is
8364 recommended to set all other settings in order to avoid relying on default
8365 unobvious parameters.
8366
8367 Example :
8368 # public access (limited to this backend only)
8369 backend public_www
8370 server srv1 192.168.0.1:80
8371 stats enable
8372 stats hide-version
8373 stats scope .
8374 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008375 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008376 stats auth admin1:AdMiN123
8377 stats auth admin2:AdMiN321
8378
8379 # internal monitoring access (unlimited)
8380 backend private_monitoring
8381 stats enable
8382 stats uri /admin?stats
8383 stats refresh 5s
8384
8385 See also : "stats auth", "stats realm", "stats uri"
8386
8387
Willy Tarreaud63335a2010-02-26 12:56:52 +01008388stats hide-version
8389 Enable statistics and hide HAProxy version reporting
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008390 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008391 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008392 Arguments : none
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008393
Willy Tarreaud63335a2010-02-26 12:56:52 +01008394 By default, the stats page reports some useful status information along with
8395 the statistics. Among them is HAProxy's version. However, it is generally
8396 considered dangerous to report precise version to anyone, as it can help them
8397 target known weaknesses with specific attacks. The "stats hide-version"
8398 statement removes the version from the statistics report. This is recommended
8399 for public sites or any site with a weak login/password.
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008400
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008401 Though this statement alone is enough to enable statistics reporting, it is
8402 recommended to set all other settings in order to avoid relying on default
8403 unobvious parameters.
8404
Willy Tarreaud63335a2010-02-26 12:56:52 +01008405 Example :
8406 # public access (limited to this backend only)
8407 backend public_www
8408 server srv1 192.168.0.1:80
Krzysztof Piotr Oledzki48cb2ae2009-10-02 22:51:14 +02008409 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008410 stats hide-version
8411 stats scope .
8412 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008413 stats realm HAProxy\ Statistics
Willy Tarreaud63335a2010-02-26 12:56:52 +01008414 stats auth admin1:AdMiN123
8415 stats auth admin2:AdMiN321
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008416
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008417 # internal monitoring access (unlimited)
8418 backend private_monitoring
8419 stats enable
Willy Tarreaud63335a2010-02-26 12:56:52 +01008420 stats uri /admin?stats
8421 stats refresh 5s
Krzysztof Piotr Oledzki15514c22010-01-04 16:03:09 +01008422
Willy Tarreaud63335a2010-02-26 12:56:52 +01008423 See also : "stats auth", "stats enable", "stats realm", "stats uri"
Willy Tarreau1d45b7c2009-08-16 10:29:18 +02008424
Willy Tarreau983e01e2010-01-11 18:42:06 +01008425
Cyril Bonté2be1b3f2010-09-30 23:46:30 +02008426stats http-request { allow | deny | auth [realm <realm>] }
8427 [ { if | unless } <condition> ]
8428 Access control for statistics
8429
8430 May be used in sections: defaults | frontend | listen | backend
8431 no | no | yes | yes
8432
8433 As "http-request", these set of options allow to fine control access to
8434 statistics. Each option may be followed by if/unless and acl.
8435 First option with matched condition (or option without condition) is final.
8436 For "deny" a 403 error will be returned, for "allow" normal processing is
8437 performed, for "auth" a 401/407 error code is returned so the client
8438 should be asked to enter a username and password.
8439
8440 There is no fixed limit to the number of http-request statements per
8441 instance.
8442
8443 See also : "http-request", section 3.4 about userlists and section 7
8444 about ACL usage.
8445
8446
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008447stats realm <realm>
8448 Enable statistics and set authentication realm
8449 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008450 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008451 Arguments :
8452 <realm> is the name of the HTTP Basic Authentication realm reported to
8453 the browser. The browser uses it to display it in the pop-up
8454 inviting the user to enter a valid username and password.
8455
8456 The realm is read as a single word, so any spaces in it should be escaped
8457 using a backslash ('\').
8458
8459 This statement is useful only in conjunction with "stats auth" since it is
8460 only related to authentication.
8461
8462 Though this statement alone is enough to enable statistics reporting, it is
8463 recommended to set all other settings in order to avoid relying on default
8464 unobvious parameters.
8465
8466 Example :
8467 # public access (limited to this backend only)
8468 backend public_www
8469 server srv1 192.168.0.1:80
8470 stats enable
8471 stats hide-version
8472 stats scope .
8473 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008474 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008475 stats auth admin1:AdMiN123
8476 stats auth admin2:AdMiN321
8477
8478 # internal monitoring access (unlimited)
8479 backend private_monitoring
8480 stats enable
8481 stats uri /admin?stats
8482 stats refresh 5s
8483
8484 See also : "stats auth", "stats enable", "stats uri"
8485
8486
8487stats refresh <delay>
8488 Enable statistics with automatic refresh
8489 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008490 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008491 Arguments :
8492 <delay> is the suggested refresh delay, specified in seconds, which will
8493 be returned to the browser consulting the report page. While the
8494 browser is free to apply any delay, it will generally respect it
8495 and refresh the page this every seconds. The refresh interval may
8496 be specified in any other non-default time unit, by suffixing the
8497 unit after the value, as explained at the top of this document.
8498
8499 This statement is useful on monitoring displays with a permanent page
8500 reporting the load balancer's activity. When set, the HTML report page will
8501 include a link "refresh"/"stop refresh" so that the user can select whether
8502 he wants automatic refresh of the page or not.
8503
8504 Though this statement alone is enough to enable statistics reporting, it is
8505 recommended to set all other settings in order to avoid relying on default
8506 unobvious parameters.
8507
8508 Example :
8509 # public access (limited to this backend only)
8510 backend public_www
8511 server srv1 192.168.0.1:80
8512 stats enable
8513 stats hide-version
8514 stats scope .
8515 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008516 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008517 stats auth admin1:AdMiN123
8518 stats auth admin2:AdMiN321
8519
8520 # internal monitoring access (unlimited)
8521 backend private_monitoring
8522 stats enable
8523 stats uri /admin?stats
8524 stats refresh 5s
8525
8526 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8527
8528
8529stats scope { <name> | "." }
8530 Enable statistics and limit access scope
8531 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008532 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008533 Arguments :
8534 <name> is the name of a listen, frontend or backend section to be
8535 reported. The special name "." (a single dot) designates the
8536 section in which the statement appears.
8537
8538 When this statement is specified, only the sections enumerated with this
8539 statement will appear in the report. All other ones will be hidden. This
8540 statement may appear as many times as needed if multiple sections need to be
8541 reported. Please note that the name checking is performed as simple string
8542 comparisons, and that it is never checked that a give section name really
8543 exists.
8544
8545 Though this statement alone is enough to enable statistics reporting, it is
8546 recommended to set all other settings in order to avoid relying on default
8547 unobvious parameters.
8548
8549 Example :
8550 # public access (limited to this backend only)
8551 backend public_www
8552 server srv1 192.168.0.1:80
8553 stats enable
8554 stats hide-version
8555 stats scope .
8556 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008557 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008558 stats auth admin1:AdMiN123
8559 stats auth admin2:AdMiN321
8560
8561 # internal monitoring access (unlimited)
8562 backend private_monitoring
8563 stats enable
8564 stats uri /admin?stats
8565 stats refresh 5s
8566
8567 See also : "stats auth", "stats enable", "stats realm", "stats uri"
8568
Willy Tarreaud63335a2010-02-26 12:56:52 +01008569
Willy Tarreauc9705a12010-07-27 20:05:50 +02008570stats show-desc [ <desc> ]
Willy Tarreaud63335a2010-02-26 12:56:52 +01008571 Enable reporting of a description on the statistics page.
8572 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008573 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008574
Willy Tarreauc9705a12010-07-27 20:05:50 +02008575 <desc> is an optional description to be reported. If unspecified, the
Willy Tarreaud63335a2010-02-26 12:56:52 +01008576 description from global section is automatically used instead.
8577
8578 This statement is useful for users that offer shared services to their
8579 customers, where node or description should be different for each customer.
8580
8581 Though this statement alone is enough to enable statistics reporting, it is
8582 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008583 unobvious parameters. By default description is not shown.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008584
8585 Example :
8586 # internal monitoring access (unlimited)
8587 backend private_monitoring
8588 stats enable
8589 stats show-desc Master node for Europe, Asia, Africa
8590 stats uri /admin?stats
8591 stats refresh 5s
8592
8593 See also: "show-node", "stats enable", "stats uri" and "description" in
8594 global section.
8595
8596
8597stats show-legends
Willy Tarreaued2119c2014-04-24 22:10:39 +02008598 Enable reporting additional information on the statistics page
8599 May be used in sections : defaults | frontend | listen | backend
8600 yes | yes | yes | yes
8601 Arguments : none
8602
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03008603 Enable reporting additional information on the statistics page :
Willy Tarreaud63335a2010-02-26 12:56:52 +01008604 - cap: capabilities (proxy)
8605 - mode: one of tcp, http or health (proxy)
8606 - id: SNMP ID (proxy, socket, server)
8607 - IP (socket, server)
8608 - cookie (backend, server)
8609
8610 Though this statement alone is enough to enable statistics reporting, it is
8611 recommended to set all other settings in order to avoid relying on default
Davor Ocelice9ed2812017-12-25 17:49:28 +01008612 unobvious parameters. Default behavior is not to show this information.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008613
8614 See also: "stats enable", "stats uri".
8615
8616
8617stats show-node [ <name> ]
8618 Enable reporting of a host name on the statistics page.
8619 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008620 yes | yes | yes | yes
Willy Tarreaud63335a2010-02-26 12:56:52 +01008621 Arguments:
8622 <name> is an optional name to be reported. If unspecified, the
8623 node name from global section is automatically used instead.
8624
8625 This statement is useful for users that offer shared services to their
8626 customers, where node or description might be different on a stats page
Davor Ocelice9ed2812017-12-25 17:49:28 +01008627 provided for each customer. Default behavior is not to show host name.
Willy Tarreaud63335a2010-02-26 12:56:52 +01008628
8629 Though this statement alone is enough to enable statistics reporting, it is
8630 recommended to set all other settings in order to avoid relying on default
8631 unobvious parameters.
8632
8633 Example:
8634 # internal monitoring access (unlimited)
8635 backend private_monitoring
8636 stats enable
8637 stats show-node Europe-1
8638 stats uri /admin?stats
8639 stats refresh 5s
8640
8641 See also: "show-desc", "stats enable", "stats uri", and "node" in global
8642 section.
8643
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008644
8645stats uri <prefix>
8646 Enable statistics and define the URI prefix to access them
8647 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaued2119c2014-04-24 22:10:39 +02008648 yes | yes | yes | yes
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008649 Arguments :
8650 <prefix> is the prefix of any URI which will be redirected to stats. This
8651 prefix may contain a question mark ('?') to indicate part of a
8652 query string.
8653
8654 The statistics URI is intercepted on the relayed traffic, so it appears as a
8655 page within the normal application. It is strongly advised to ensure that the
8656 selected URI will never appear in the application, otherwise it will never be
8657 possible to reach it in the application.
8658
8659 The default URI compiled in haproxy is "/haproxy?stats", but this may be
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01008660 changed at build time, so it's better to always explicitly specify it here.
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008661 It is generally a good idea to include a question mark in the URI so that
8662 intermediate proxies refrain from caching the results. Also, since any string
8663 beginning with the prefix will be accepted as a stats request, the question
8664 mark helps ensuring that no valid URI will begin with the same words.
8665
8666 It is sometimes very convenient to use "/" as the URI prefix, and put that
8667 statement in a "listen" instance of its own. That makes it easy to dedicate
8668 an address or a port to statistics only.
8669
8670 Though this statement alone is enough to enable statistics reporting, it is
8671 recommended to set all other settings in order to avoid relying on default
8672 unobvious parameters.
8673
8674 Example :
8675 # public access (limited to this backend only)
8676 backend public_www
8677 server srv1 192.168.0.1:80
8678 stats enable
8679 stats hide-version
8680 stats scope .
8681 stats uri /admin?stats
Davor Ocelice9ed2812017-12-25 17:49:28 +01008682 stats realm HAProxy\ Statistics
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008683 stats auth admin1:AdMiN123
8684 stats auth admin2:AdMiN321
8685
8686 # internal monitoring access (unlimited)
8687 backend private_monitoring
8688 stats enable
8689 stats uri /admin?stats
8690 stats refresh 5s
8691
8692 See also : "stats auth", "stats enable", "stats realm"
8693
8694
Willy Tarreaud63335a2010-02-26 12:56:52 +01008695stick match <pattern> [table <table>] [{if | unless} <cond>]
8696 Define a request pattern matching condition to stick a user to a server
Willy Tarreaueabeafa2008-01-16 16:17:06 +01008697 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaud63335a2010-02-26 12:56:52 +01008698 no | no | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008699
8700 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008701 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008702 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008703 will be analyzed in the hope to find a matching entry in a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008704 stickiness table. This rule is mandatory.
8705
8706 <table> is an optional stickiness table name. If unspecified, the same
8707 backend's table is used. A stickiness table is declared using
8708 the "stick-table" statement.
8709
8710 <cond> is an optional matching condition. It makes it possible to match
8711 on a certain criterion only when other conditions are met (or
8712 not met). For instance, it could be used to match on a source IP
8713 address except when a request passes through a known proxy, in
8714 which case we'd match on a header containing that IP address.
8715
8716 Some protocols or applications require complex stickiness rules and cannot
8717 always simply rely on cookies nor hashing. The "stick match" statement
8718 describes a rule to extract the stickiness criterion from an incoming request
8719 or connection. See section 7 for a complete list of possible patterns and
8720 transformation rules.
8721
8722 The table has to be declared using the "stick-table" statement. It must be of
8723 a type compatible with the pattern. By default it is the one which is present
8724 in the same backend. It is possible to share a table with other backends by
8725 referencing it using the "table" keyword. If another table is referenced,
8726 the server's ID inside the backends are used. By default, all server IDs
8727 start at 1 in each backend, so the server ordering is enough. But in case of
8728 doubt, it is highly recommended to force server IDs using their "id" setting.
8729
8730 It is possible to restrict the conditions where a "stick match" statement
8731 will apply, using "if" or "unless" followed by a condition. See section 7 for
8732 ACL based conditions.
8733
8734 There is no limit on the number of "stick match" statements. The first that
8735 applies and matches will cause the request to be directed to the same server
8736 as was used for the request which created the entry. That way, multiple
8737 matches can be used as fallbacks.
8738
8739 The stick rules are checked after the persistence cookies, so they will not
8740 affect stickiness if a cookie has already been used to select a server. That
8741 way, it becomes very easy to insert cookies and match on IP addresses in
8742 order to maintain stickiness between HTTP and HTTPS.
8743
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008744 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8745 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008746 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008747
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008748 Example :
8749 # forward SMTP users to the same server they just used for POP in the
8750 # last 30 minutes
8751 backend pop
8752 mode tcp
8753 balance roundrobin
8754 stick store-request src
8755 stick-table type ip size 200k expire 30m
8756 server s1 192.168.1.1:110
8757 server s2 192.168.1.1:110
8758
8759 backend smtp
8760 mode tcp
8761 balance roundrobin
8762 stick match src table pop
8763 server s1 192.168.1.1:25
8764 server s2 192.168.1.1:25
8765
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008766 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008767 about ACLs and samples fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008768
8769
8770stick on <pattern> [table <table>] [{if | unless} <condition>]
8771 Define a request pattern to associate a user to a server
8772 May be used in sections : defaults | frontend | listen | backend
8773 no | no | yes | yes
8774
8775 Note : This form is exactly equivalent to "stick match" followed by
8776 "stick store-request", all with the same arguments. Please refer
8777 to both keywords for details. It is only provided as a convenience
8778 for writing more maintainable configurations.
8779
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008780 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8781 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008782 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008783
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008784 Examples :
8785 # The following form ...
Willy Tarreauec579d82010-02-26 19:15:04 +01008786 stick on src table pop if !localhost
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008787
8788 # ...is strictly equivalent to this one :
8789 stick match src table pop if !localhost
8790 stick store-request src table pop if !localhost
8791
8792
8793 # Use cookie persistence for HTTP, and stick on source address for HTTPS as
8794 # well as HTTP without cookie. Share the same table between both accesses.
8795 backend http
8796 mode http
8797 balance roundrobin
8798 stick on src table https
8799 cookie SRV insert indirect nocache
8800 server s1 192.168.1.1:80 cookie s1
8801 server s2 192.168.1.1:80 cookie s2
8802
8803 backend https
8804 mode tcp
8805 balance roundrobin
8806 stick-table type ip size 200k expire 30m
8807 stick on src
8808 server s1 192.168.1.1:443
8809 server s2 192.168.1.1:443
8810
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008811 See also : "stick match", "stick store-request", "nbproc" and "bind-process".
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008812
8813
8814stick store-request <pattern> [table <table>] [{if | unless} <condition>]
8815 Define a request pattern used to create an entry in a stickiness table
8816 May be used in sections : defaults | frontend | listen | backend
8817 no | no | yes | yes
8818
8819 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02008820 <pattern> is a sample expression rule as described in section 7.3. It
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008821 describes what elements of the incoming request or connection
Davor Ocelice9ed2812017-12-25 17:49:28 +01008822 will be analyzed, extracted and stored in the table once a
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008823 server is selected.
8824
8825 <table> is an optional stickiness table name. If unspecified, the same
8826 backend's table is used. A stickiness table is declared using
8827 the "stick-table" statement.
8828
8829 <cond> is an optional storage condition. It makes it possible to store
8830 certain criteria only when some conditions are met (or not met).
8831 For instance, it could be used to store the source IP address
8832 except when the request passes through a known proxy, in which
8833 case we'd store a converted form of a header containing that IP
8834 address.
8835
8836 Some protocols or applications require complex stickiness rules and cannot
8837 always simply rely on cookies nor hashing. The "stick store-request" statement
8838 describes a rule to decide what to extract from the request and when to do
8839 it, in order to store it into a stickiness table for further requests to
8840 match it using the "stick match" statement. Obviously the extracted part must
8841 make sense and have a chance to be matched in a further request. Storing a
8842 client's IP address for instance often makes sense. Storing an ID found in a
8843 URL parameter also makes sense. Storing a source port will almost never make
8844 any sense because it will be randomly matched. See section 7 for a complete
8845 list of possible patterns and transformation rules.
8846
8847 The table has to be declared using the "stick-table" statement. It must be of
8848 a type compatible with the pattern. By default it is the one which is present
8849 in the same backend. It is possible to share a table with other backends by
8850 referencing it using the "table" keyword. If another table is referenced,
8851 the server's ID inside the backends are used. By default, all server IDs
8852 start at 1 in each backend, so the server ordering is enough. But in case of
8853 doubt, it is highly recommended to force server IDs using their "id" setting.
8854
8855 It is possible to restrict the conditions where a "stick store-request"
8856 statement will apply, using "if" or "unless" followed by a condition. This
8857 condition will be evaluated while parsing the request, so any criteria can be
8858 used. See section 7 for ACL based conditions.
8859
8860 There is no limit on the number of "stick store-request" statements, but
8861 there is a limit of 8 simultaneous stores per request or response. This
8862 makes it possible to store up to 8 criteria, all extracted from either the
8863 request or the response, regardless of the number of rules. Only the 8 first
8864 ones which match will be kept. Using this, it is possible to feed multiple
8865 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01008866 another protocol or access method. Using multiple store-request rules with
8867 the same table is possible and may be used to find the best criterion to rely
8868 on, by arranging the rules by decreasing preference order. Only the first
8869 extracted criterion for a given table will be stored. All subsequent store-
8870 request rules referencing the same table will be skipped and their ACLs will
8871 not be evaluated.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008872
8873 The "store-request" rules are evaluated once the server connection has been
8874 established, so that the table will contain the real server that processed
8875 the request.
8876
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008877 Note : Consider not using this feature in multi-process mode (nbproc > 1)
8878 unless you know what you do : memory is not shared between the
Davor Ocelice9ed2812017-12-25 17:49:28 +01008879 processes, which can result in random behaviors.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008880
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008881 Example :
8882 # forward SMTP users to the same server they just used for POP in the
8883 # last 30 minutes
8884 backend pop
8885 mode tcp
8886 balance roundrobin
8887 stick store-request src
8888 stick-table type ip size 200k expire 30m
8889 server s1 192.168.1.1:110
8890 server s2 192.168.1.1:110
8891
8892 backend smtp
8893 mode tcp
8894 balance roundrobin
8895 stick match src table pop
8896 server s1 192.168.1.1:25
8897 server s2 192.168.1.1:25
8898
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008899 See also : "stick-table", "stick on", "nbproc", "bind-process" and section 7
Willy Tarreaube722a22014-06-13 16:31:59 +02008900 about ACLs and sample fetching.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008901
8902
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008903stick-table type {ip | integer | string [len <length>] | binary [len <length>]}
Emeric Brunf099e792010-09-27 12:05:28 +02008904 size <size> [expire <expire>] [nopurge] [peers <peersect>]
8905 [store <data_type>]*
Godbach64cef792013-12-04 16:08:22 +08008906 Configure the stickiness table for the current section
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008907 May be used in sections : defaults | frontend | listen | backend
Willy Tarreauc00cdc22010-06-06 16:48:26 +02008908 no | yes | yes | yes
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008909
8910 Arguments :
8911 ip a table declared with "type ip" will only store IPv4 addresses.
8912 This form is very compact (about 50 bytes per entry) and allows
8913 very fast entry lookup and stores with almost no overhead. This
8914 is mainly used to store client source IP addresses.
8915
David du Colombier9a6d3c92011-03-17 10:40:24 +01008916 ipv6 a table declared with "type ipv6" will only store IPv6 addresses.
8917 This form is very compact (about 60 bytes per entry) and allows
8918 very fast entry lookup and stores with almost no overhead. This
8919 is mainly used to store client source IP addresses.
8920
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008921 integer a table declared with "type integer" will store 32bit integers
8922 which can represent a client identifier found in a request for
8923 instance.
8924
8925 string a table declared with "type string" will store substrings of up
8926 to <len> characters. If the string provided by the pattern
8927 extractor is larger than <len>, it will be truncated before
8928 being stored. During matching, at most <len> characters will be
8929 compared between the string in the table and the extracted
8930 pattern. When not specified, the string is automatically limited
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008931 to 32 characters.
8932
8933 binary a table declared with "type binary" will store binary blocks
8934 of <len> bytes. If the block provided by the pattern
8935 extractor is larger than <len>, it will be truncated before
Willy Tarreaube722a22014-06-13 16:31:59 +02008936 being stored. If the block provided by the sample expression
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008937 is shorter than <len>, it will be padded by 0. When not
8938 specified, the block is automatically limited to 32 bytes.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008939
8940 <length> is the maximum number of characters that will be stored in a
Emeric Brun7c6b82e2010-09-24 16:34:28 +02008941 "string" type table (See type "string" above). Or the number
8942 of bytes of the block in "binary" type table. Be careful when
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008943 changing this parameter as memory usage will proportionally
8944 increase.
8945
8946 <size> is the maximum number of entries that can fit in the table. This
Cyril Bonté78caf842010-03-10 22:41:43 +01008947 value directly impacts memory usage. Count approximately
8948 50 bytes per entry, plus the size of a string if any. The size
8949 supports suffixes "k", "m", "g" for 2^10, 2^20 and 2^30 factors.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008950
8951 [nopurge] indicates that we refuse to purge older entries when the table
8952 is full. When not specified and the table is full when haproxy
8953 wants to store an entry in it, it will flush a few of the oldest
8954 entries in order to release some space for the new ones. This is
Davor Ocelice9ed2812017-12-25 17:49:28 +01008955 most often the desired behavior. In some specific cases, it
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008956 be desirable to refuse new entries instead of purging the older
8957 ones. That may be the case when the amount of data to store is
8958 far above the hardware limits and we prefer not to offer access
8959 to new clients than to reject the ones already connected. When
8960 using this parameter, be sure to properly set the "expire"
8961 parameter (see below).
8962
Emeric Brunf099e792010-09-27 12:05:28 +02008963 <peersect> is the name of the peers section to use for replication. Entries
8964 which associate keys to server IDs are kept synchronized with
8965 the remote peers declared in this section. All entries are also
8966 automatically learned from the local peer (old process) during a
8967 soft restart.
8968
Willy Tarreau1abc6732015-05-01 19:21:02 +02008969 NOTE : each peers section may be referenced only by tables
8970 belonging to the same unique process.
Cyril Bonté02ff8ef2010-12-14 22:48:49 +01008971
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008972 <expire> defines the maximum duration of an entry in the table since it
8973 was last created, refreshed or matched. The expiration delay is
8974 defined using the standard time format, similarly as the various
8975 timeouts. The maximum duration is slightly above 24 days. See
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03008976 section 2.4 for more information. If this delay is not specified,
Cyril Bontédc4d9032012-04-08 21:57:39 +02008977 the session won't automatically expire, but older entries will
Willy Tarreaub937b7e2010-01-12 15:27:54 +01008978 be removed once full. Be sure not to use the "nopurge" parameter
8979 if not expiration delay is specified.
8980
Willy Tarreau08d5f982010-06-06 13:34:54 +02008981 <data_type> is used to store additional information in the stick-table. This
8982 may be used by ACLs in order to control various criteria related
8983 to the activity of the client matching the stick-table. For each
8984 item specified here, the size of each entry will be inflated so
Willy Tarreauc9705a12010-07-27 20:05:50 +02008985 that the additional data can fit. Several data types may be
8986 stored with an entry. Multiple data types may be specified after
8987 the "store" keyword, as a comma-separated list. Alternatively,
8988 it is possible to repeat the "store" keyword followed by one or
8989 several data types. Except for the "server_id" type which is
8990 automatically detected and enabled, all data types must be
8991 explicitly declared to be stored. If an ACL references a data
8992 type which is not stored, the ACL will simply not match. Some
8993 data types require an argument which must be passed just after
8994 the type between parenthesis. See below for the supported data
8995 types and their arguments.
8996
8997 The data types that can be stored with an entry are the following :
8998 - server_id : this is an integer which holds the numeric ID of the server a
8999 request was assigned to. It is used by the "stick match", "stick store",
9000 and "stick on" rules. It is automatically enabled when referenced.
9001
9002 - gpc0 : first General Purpose Counter. It is a positive 32-bit integer
9003 integer which may be used for anything. Most of the time it will be used
9004 to put a special tag on some entries, for instance to note that a
Davor Ocelice9ed2812017-12-25 17:49:28 +01009005 specific behavior was detected and must be known for future matches.
Willy Tarreauc9705a12010-07-27 20:05:50 +02009006
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009007 - gpc0_rate(<period>) : increment rate of the first General Purpose Counter
9008 over a period. It is a positive 32-bit integer integer which may be used
9009 for anything. Just like <gpc0>, it counts events, but instead of keeping
Davor Ocelice9ed2812017-12-25 17:49:28 +01009010 a cumulative number, it maintains the rate at which the counter is
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009011 incremented. Most of the time it will be used to measure the frequency of
Davor Ocelice9ed2812017-12-25 17:49:28 +01009012 occurrence of certain events (e.g. requests to a specific URL).
Willy Tarreauba2ffd12013-05-29 15:54:14 +02009013
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009014 - gpc1 : second General Purpose Counter. It is a positive 32-bit integer
9015 integer which may be used for anything. Most of the time it will be used
9016 to put a special tag on some entries, for instance to note that a
9017 specific behavior was detected and must be known for future matches.
9018
9019 - gpc1_rate(<period>) : increment rate of the second General Purpose Counter
9020 over a period. It is a positive 32-bit integer integer which may be used
9021 for anything. Just like <gpc1>, it counts events, but instead of keeping
9022 a cumulative number, it maintains the rate at which the counter is
9023 incremented. Most of the time it will be used to measure the frequency of
9024 occurrence of certain events (e.g. requests to a specific URL).
9025
Willy Tarreauc9705a12010-07-27 20:05:50 +02009026 - conn_cnt : Connection Count. It is a positive 32-bit integer which counts
9027 the absolute number of connections received from clients which matched
9028 this entry. It does not mean the connections were accepted, just that
9029 they were received.
9030
9031 - conn_cur : Current Connections. It is a positive 32-bit integer which
9032 stores the concurrent connection counts for the entry. It is incremented
9033 once an incoming connection matches the entry, and decremented once the
9034 connection leaves. That way it is possible to know at any time the exact
9035 number of concurrent connections for an entry.
9036
9037 - conn_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9038 integer parameter <period> which indicates in milliseconds the length
9039 of the period over which the average is measured. It reports the average
9040 incoming connection rate over that period, in connections per period. The
9041 result is an integer which can be matched using ACLs.
9042
9043 - sess_cnt : Session Count. It is a positive 32-bit integer which counts
9044 the absolute number of sessions received from clients which matched this
9045 entry. A session is a connection that was accepted by the layer 4 rules.
9046
9047 - sess_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9048 integer parameter <period> which indicates in milliseconds the length
9049 of the period over which the average is measured. It reports the average
9050 incoming session rate over that period, in sessions per period. The
9051 result is an integer which can be matched using ACLs.
9052
9053 - http_req_cnt : HTTP request Count. It is a positive 32-bit integer which
9054 counts the absolute number of HTTP requests received from clients which
9055 matched this entry. It does not matter whether they are valid requests or
9056 not. Note that this is different from sessions when keep-alive is used on
9057 the client side.
9058
9059 - http_req_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9060 integer parameter <period> which indicates in milliseconds the length
9061 of the period over which the average is measured. It reports the average
9062 HTTP request rate over that period, in requests per period. The result is
9063 an integer which can be matched using ACLs. It does not matter whether
9064 they are valid requests or not. Note that this is different from sessions
9065 when keep-alive is used on the client side.
9066
9067 - http_err_cnt : HTTP Error Count. It is a positive 32-bit integer which
9068 counts the absolute number of HTTP requests errors induced by clients
9069 which matched this entry. Errors are counted on invalid and truncated
9070 requests, as well as on denied or tarpitted requests, and on failed
9071 authentications. If the server responds with 4xx, then the request is
9072 also counted as an error since it's an error triggered by the client
Davor Ocelice9ed2812017-12-25 17:49:28 +01009073 (e.g. vulnerability scan).
Willy Tarreauc9705a12010-07-27 20:05:50 +02009074
9075 - http_err_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9076 integer parameter <period> which indicates in milliseconds the length
9077 of the period over which the average is measured. It reports the average
9078 HTTP request error rate over that period, in requests per period (see
9079 http_err_cnt above for what is accounted as an error). The result is an
9080 integer which can be matched using ACLs.
9081
9082 - bytes_in_cnt : client to server byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009083 integer which counts the cumulative number of bytes received from clients
Willy Tarreauc9705a12010-07-27 20:05:50 +02009084 which matched this entry. Headers are included in the count. This may be
9085 used to limit abuse of upload features on photo or video servers.
9086
9087 - bytes_in_rate(<period>) : frequency counter (takes 12 bytes). It takes an
9088 integer parameter <period> which indicates in milliseconds the length
9089 of the period over which the average is measured. It reports the average
9090 incoming bytes rate over that period, in bytes per period. It may be used
9091 to detect users which upload too much and too fast. Warning: with large
9092 uploads, it is possible that the amount of uploaded data will be counted
9093 once upon termination, thus causing spikes in the average transfer speed
9094 instead of having a smooth one. This may partially be smoothed with
9095 "option contstats" though this is not perfect yet. Use of byte_in_cnt is
9096 recommended for better fairness.
9097
9098 - bytes_out_cnt : server to client byte count. It is a positive 64-bit
Davor Ocelice9ed2812017-12-25 17:49:28 +01009099 integer which counts the cumulative number of bytes sent to clients which
Willy Tarreauc9705a12010-07-27 20:05:50 +02009100 matched this entry. Headers are included in the count. This may be used
9101 to limit abuse of bots sucking the whole site.
9102
9103 - bytes_out_rate(<period>) : frequency counter (takes 12 bytes). It takes
9104 an integer parameter <period> which indicates in milliseconds the length
9105 of the period over which the average is measured. It reports the average
9106 outgoing bytes rate over that period, in bytes per period. It may be used
9107 to detect users which download too much and too fast. Warning: with large
9108 transfers, it is possible that the amount of transferred data will be
9109 counted once upon termination, thus causing spikes in the average
9110 transfer speed instead of having a smooth one. This may partially be
9111 smoothed with "option contstats" though this is not perfect yet. Use of
9112 byte_out_cnt is recommended for better fairness.
Willy Tarreau08d5f982010-06-06 13:34:54 +02009113
Willy Tarreauc00cdc22010-06-06 16:48:26 +02009114 There is only one stick-table per proxy. At the moment of writing this doc,
9115 it does not seem useful to have multiple tables per proxy. If this happens
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009116 to be required, simply create a dummy backend with a stick-table in it and
9117 reference it.
9118
9119 It is important to understand that stickiness based on learning information
9120 has some limitations, including the fact that all learned associations are
Baptiste Assmann123ff042016-03-06 23:29:28 +01009121 lost upon restart unless peers are properly configured to transfer such
9122 information upon restart (recommended). In general it can be good as a
9123 complement but not always as an exclusive stickiness.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009124
Willy Tarreauc9705a12010-07-27 20:05:50 +02009125 Last, memory requirements may be important when storing many data types.
9126 Indeed, storing all indicators above at once in each entry requires 116 bytes
9127 per entry, or 116 MB for a 1-million entries table. This is definitely not
9128 something that can be ignored.
9129
9130 Example:
9131 # Keep track of counters of up to 1 million IP addresses over 5 minutes
9132 # and store a general purpose counter and the average connection rate
9133 # computed over a sliding window of 30 seconds.
9134 stick-table type ip size 1m expire 5m store gpc0,conn_rate(30s)
9135
Jarno Huuskonene0ee0be2017-07-04 10:35:12 +03009136 See also : "stick match", "stick on", "stick store-request", section 2.4
David du Colombiera13d1b92011-03-17 10:40:22 +01009137 about time format and section 7 about ACLs.
Willy Tarreaub937b7e2010-01-12 15:27:54 +01009138
9139
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009140stick store-response <pattern> [table <table>] [{if | unless} <condition>]
Baptiste Assmann2f2d2ec2016-03-06 23:27:24 +01009141 Define a response pattern used to create an entry in a stickiness table
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009142 May be used in sections : defaults | frontend | listen | backend
9143 no | no | yes | yes
9144
9145 Arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009146 <pattern> is a sample expression rule as described in section 7.3. It
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009147 describes what elements of the response or connection will
Davor Ocelice9ed2812017-12-25 17:49:28 +01009148 be analyzed, extracted and stored in the table once a
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009149 server is selected.
9150
9151 <table> is an optional stickiness table name. If unspecified, the same
9152 backend's table is used. A stickiness table is declared using
9153 the "stick-table" statement.
9154
9155 <cond> is an optional storage condition. It makes it possible to store
9156 certain criteria only when some conditions are met (or not met).
9157 For instance, it could be used to store the SSL session ID only
9158 when the response is a SSL server hello.
9159
9160 Some protocols or applications require complex stickiness rules and cannot
9161 always simply rely on cookies nor hashing. The "stick store-response"
9162 statement describes a rule to decide what to extract from the response and
9163 when to do it, in order to store it into a stickiness table for further
9164 requests to match it using the "stick match" statement. Obviously the
9165 extracted part must make sense and have a chance to be matched in a further
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009166 request. Storing an ID found in a header of a response makes sense.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009167 See section 7 for a complete list of possible patterns and transformation
9168 rules.
9169
9170 The table has to be declared using the "stick-table" statement. It must be of
9171 a type compatible with the pattern. By default it is the one which is present
9172 in the same backend. It is possible to share a table with other backends by
9173 referencing it using the "table" keyword. If another table is referenced,
9174 the server's ID inside the backends are used. By default, all server IDs
9175 start at 1 in each backend, so the server ordering is enough. But in case of
9176 doubt, it is highly recommended to force server IDs using their "id" setting.
9177
9178 It is possible to restrict the conditions where a "stick store-response"
9179 statement will apply, using "if" or "unless" followed by a condition. This
9180 condition will be evaluated while parsing the response, so any criteria can
9181 be used. See section 7 for ACL based conditions.
9182
9183 There is no limit on the number of "stick store-response" statements, but
9184 there is a limit of 8 simultaneous stores per request or response. This
9185 makes it possible to store up to 8 criteria, all extracted from either the
9186 request or the response, regardless of the number of rules. Only the 8 first
9187 ones which match will be kept. Using this, it is possible to feed multiple
9188 tables at once in the hope to increase the chance to recognize a user on
Willy Tarreau9667a802013-12-09 12:52:13 +01009189 another protocol or access method. Using multiple store-response rules with
9190 the same table is possible and may be used to find the best criterion to rely
9191 on, by arranging the rules by decreasing preference order. Only the first
9192 extracted criterion for a given table will be stored. All subsequent store-
9193 response rules referencing the same table will be skipped and their ACLs will
9194 not be evaluated. However, even if a store-request rule references a table, a
9195 store-response rule may also use the same table. This means that each table
9196 may learn exactly one element from the request and one element from the
9197 response at once.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009198
9199 The table will contain the real server that processed the request.
9200
9201 Example :
9202 # Learn SSL session ID from both request and response and create affinity.
9203 backend https
9204 mode tcp
9205 balance roundrobin
Cyril Bontédc4d9032012-04-08 21:57:39 +02009206 # maximum SSL session ID length is 32 bytes.
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009207 stick-table type binary len 32 size 30k expire 30m
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009208
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009209 acl clienthello req_ssl_hello_type 1
9210 acl serverhello rep_ssl_hello_type 2
9211
9212 # use tcp content accepts to detects ssl client and server hello.
9213 tcp-request inspect-delay 5s
9214 tcp-request content accept if clienthello
9215
9216 # no timeout on response inspect delay by default.
9217 tcp-response content accept if serverhello
Cyril Bonté108cf6e2012-04-21 23:30:29 +02009218
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009219 # SSL session ID (SSLID) may be present on a client or server hello.
9220 # Its length is coded on 1 byte at offset 43 and its value starts
9221 # at offset 44.
9222
9223 # Match and learn on request if client hello.
9224 stick on payload_lv(43,1) if clienthello
9225
9226 # Learn on response if server hello.
9227 stick store-response payload_lv(43,1) if serverhello
Cyril Bontédc4d9032012-04-08 21:57:39 +02009228
Emeric Brun6a1cefa2010-09-24 18:15:17 +02009229 server s1 192.168.1.1:443
9230 server s2 192.168.1.1:443
9231
9232 See also : "stick-table", "stick on", and section 7 about ACLs and pattern
9233 extraction.
9234
9235
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009236tcp-check connect [params*]
9237 Opens a new connection
9238 May be used in sections: defaults | frontend | listen | backend
9239 no | no | yes | yes
9240
9241 When an application lies on more than a single TCP port or when HAProxy
9242 load-balance many services in a single backend, it makes sense to probe all
9243 the services individually before considering a server as operational.
9244
9245 When there are no TCP port configured on the server line neither server port
9246 directive, then the 'tcp-check connect port <port>' must be the first step
9247 of the sequence.
9248
9249 In a tcp-check ruleset a 'connect' is required, it is also mandatory to start
9250 the ruleset with a 'connect' rule. Purpose is to ensure admin know what they
9251 do.
9252
9253 Parameters :
9254 They are optional and can be used to describe how HAProxy should open and
9255 use the TCP connection.
9256
9257 port if not set, check port or server port is used.
9258 It tells HAProxy where to open the connection to.
9259 <port> must be a valid TCP port source integer, from 1 to 65535.
9260
9261 send-proxy send a PROXY protocol string
9262
9263 ssl opens a ciphered connection
9264
9265 Examples:
9266 # check HTTP and HTTPs services on a server.
9267 # first open port 80 thanks to server line port directive, then
9268 # tcp-check opens port 443, ciphered and run a request on it:
9269 option tcp-check
9270 tcp-check connect
9271 tcp-check send GET\ /\ HTTP/1.0\r\n
9272 tcp-check send Host:\ haproxy.1wt.eu\r\n
9273 tcp-check send \r\n
9274 tcp-check expect rstring (2..|3..)
9275 tcp-check connect port 443 ssl
9276 tcp-check send GET\ /\ HTTP/1.0\r\n
9277 tcp-check send Host:\ haproxy.1wt.eu\r\n
9278 tcp-check send \r\n
9279 tcp-check expect rstring (2..|3..)
9280 server www 10.0.0.1 check port 80
9281
9282 # check both POP and IMAP from a single server:
9283 option tcp-check
9284 tcp-check connect port 110
9285 tcp-check expect string +OK\ POP3\ ready
9286 tcp-check connect port 143
9287 tcp-check expect string *\ OK\ IMAP4\ ready
9288 server mail 10.0.0.1 check
9289
9290 See also : "option tcp-check", "tcp-check send", "tcp-check expect"
9291
9292
9293tcp-check expect [!] <match> <pattern>
Davor Ocelice9ed2812017-12-25 17:49:28 +01009294 Specify data to be collected and analyzed during a generic health check
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009295 May be used in sections: defaults | frontend | listen | backend
9296 no | no | yes | yes
9297
9298 Arguments :
9299 <match> is a keyword indicating how to look for a specific pattern in the
9300 response. The keyword may be one of "string", "rstring" or
9301 binary.
9302 The keyword may be preceded by an exclamation mark ("!") to negate
9303 the match. Spaces are allowed between the exclamation mark and the
9304 keyword. See below for more details on the supported keywords.
9305
9306 <pattern> is the pattern to look for. It may be a string or a regular
9307 expression. If the pattern contains spaces, they must be escaped
9308 with the usual backslash ('\').
9309 If the match is set to binary, then the pattern must be passed as
Davor Ocelice9ed2812017-12-25 17:49:28 +01009310 a series of hexadecimal digits in an even number. Each sequence of
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009311 two digits will represent a byte. The hexadecimal digits may be
9312 used upper or lower case.
9313
9314
9315 The available matches are intentionally similar to their http-check cousins :
9316
9317 string <string> : test the exact string matches in the response buffer.
9318 A health check response will be considered valid if the
9319 response's buffer contains this exact string. If the
9320 "string" keyword is prefixed with "!", then the response
9321 will be considered invalid if the body contains this
9322 string. This can be used to look for a mandatory pattern
9323 in a protocol response, or to detect a failure when a
9324 specific error appears in a protocol banner.
9325
9326 rstring <regex> : test a regular expression on the response buffer.
9327 A health check response will be considered valid if the
9328 response's buffer matches this expression. If the
9329 "rstring" keyword is prefixed with "!", then the response
9330 will be considered invalid if the body matches the
9331 expression.
9332
9333 binary <hexstring> : test the exact string in its hexadecimal form matches
9334 in the response buffer. A health check response will
9335 be considered valid if the response's buffer contains
9336 this exact hexadecimal string.
9337 Purpose is to match data on binary protocols.
9338
9339 It is important to note that the responses will be limited to a certain size
9340 defined by the global "tune.chksize" option, which defaults to 16384 bytes.
9341 Thus, too large responses may not contain the mandatory pattern when using
9342 "string", "rstring" or binary. If a large response is absolutely required, it
9343 is possible to change the default max size by setting the global variable.
9344 However, it is worth keeping in mind that parsing very large responses can
9345 waste some CPU cycles, especially when regular expressions are used, and that
9346 it is always better to focus the checks on smaller resources. Also, in its
9347 current state, the check will not find any string nor regex past a null
9348 character in the response. Similarly it is not possible to request matching
9349 the null character.
9350
9351 Examples :
9352 # perform a POP check
9353 option tcp-check
9354 tcp-check expect string +OK\ POP3\ ready
9355
9356 # perform an IMAP check
9357 option tcp-check
9358 tcp-check expect string *\ OK\ IMAP4\ ready
9359
9360 # look for the redis master server
9361 option tcp-check
9362 tcp-check send PING\r\n
Baptiste Assmanna3322992015-08-04 10:12:18 +02009363 tcp-check expect string +PONG
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009364 tcp-check send info\ replication\r\n
9365 tcp-check expect string role:master
9366 tcp-check send QUIT\r\n
9367 tcp-check expect string +OK
9368
9369
9370 See also : "option tcp-check", "tcp-check connect", "tcp-check send",
9371 "tcp-check send-binary", "http-check expect", tune.chksize
9372
9373
9374tcp-check send <data>
9375 Specify a string to be sent as a question during a generic health check
9376 May be used in sections: defaults | frontend | listen | backend
9377 no | no | yes | yes
9378
9379 <data> : the data to be sent as a question during a generic health check
9380 session. For now, <data> must be a string.
9381
9382 Examples :
9383 # look for the redis master server
9384 option tcp-check
9385 tcp-check send info\ replication\r\n
9386 tcp-check expect string role:master
9387
9388 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9389 "tcp-check send-binary", tune.chksize
9390
9391
Davor Ocelice9ed2812017-12-25 17:49:28 +01009392tcp-check send-binary <hexstring>
9393 Specify a hex digits string to be sent as a binary question during a raw
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009394 tcp health check
9395 May be used in sections: defaults | frontend | listen | backend
9396 no | no | yes | yes
9397
9398 <data> : the data to be sent as a question during a generic health check
9399 session. For now, <data> must be a string.
Davor Ocelice9ed2812017-12-25 17:49:28 +01009400 <hexstring> : test the exact string in its hexadecimal form matches in the
Willy Tarreau938c7fe2014-04-25 14:21:39 +02009401 response buffer. A health check response will be considered
9402 valid if the response's buffer contains this exact
9403 hexadecimal string.
9404 Purpose is to send binary data to ask on binary protocols.
9405
9406 Examples :
9407 # redis check in binary
9408 option tcp-check
9409 tcp-check send-binary 50494e470d0a # PING\r\n
9410 tcp-check expect binary 2b504F4e47 # +PONG
9411
9412
9413 See also : "option tcp-check", "tcp-check connect", "tcp-check expect",
9414 "tcp-check send", tune.chksize
9415
9416
Willy Tarreaue9656522010-08-17 15:40:09 +02009417tcp-request connection <action> [{if | unless} <condition>]
9418 Perform an action on an incoming connection depending on a layer 4 condition
Willy Tarreau1a687942010-05-23 22:40:30 +02009419 May be used in sections : defaults | frontend | listen | backend
9420 no | yes | yes | no
Willy Tarreaue9656522010-08-17 15:40:09 +02009421 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009422 <action> defines the action to perform if the condition applies. See
9423 below.
Willy Tarreau1a687942010-05-23 22:40:30 +02009424
Willy Tarreaue9656522010-08-17 15:40:09 +02009425 <condition> is a standard layer4-only ACL-based condition (see section 7).
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009426
9427 Immediately after acceptance of a new incoming connection, it is possible to
9428 evaluate some conditions to decide whether this connection must be accepted
Willy Tarreaue9656522010-08-17 15:40:09 +02009429 or dropped or have its counters tracked. Those conditions cannot make use of
9430 any data contents because the connection has not been read from yet, and the
9431 buffers are not yet allocated. This is used to selectively and very quickly
9432 accept or drop connections from various sources with a very low overhead. If
9433 some contents need to be inspected in order to take the decision, the
9434 "tcp-request content" statements must be used instead.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009435
Willy Tarreaue9656522010-08-17 15:40:09 +02009436 The "tcp-request connection" rules are evaluated in their exact declaration
9437 order. If no rule matches or if there is no rule, the default action is to
9438 accept the incoming connection. There is no specific limit to the number of
9439 rules which may be inserted.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009440
Willy Tarreaua9083d02015-05-08 15:27:59 +02009441 Four types of actions are supported :
Willy Tarreaue9656522010-08-17 15:40:09 +02009442 - accept :
9443 accepts the connection if the condition is true (when used with "if")
9444 or false (when used with "unless"). The first such rule executed ends
9445 the rules evaluation.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009446
Willy Tarreaue9656522010-08-17 15:40:09 +02009447 - reject :
9448 rejects the connection if the condition is true (when used with "if")
9449 or false (when used with "unless"). The first such rule executed ends
9450 the rules evaluation. Rejected connections do not even become a
9451 session, which is why they are accounted separately for in the stats,
9452 as "denied connections". They are not considered for the session
9453 rate-limit and are not logged either. The reason is that these rules
9454 should only be used to filter extremely high connection rates such as
9455 the ones encountered during a massive DDoS attack. Under these extreme
9456 conditions, the simple action of logging each event would make the
9457 system collapse and would considerably lower the filtering capacity. If
9458 logging is absolutely desired, then "tcp-request content" rules should
Willy Tarreau4f614292016-10-21 17:49:36 +02009459 be used instead, as "tcp-request session" rules will not log either.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009460
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009461 - expect-proxy layer4 :
9462 configures the client-facing connection to receive a PROXY protocol
9463 header before any byte is read from the socket. This is equivalent to
9464 having the "accept-proxy" keyword on the "bind" line, except that using
9465 the TCP rule allows the PROXY protocol to be accepted only for certain
9466 IP address ranges using an ACL. This is convenient when multiple layers
9467 of load balancers are passed through by traffic coming from public
9468 hosts.
9469
Bertrand Jacquin90759682016-06-06 15:35:39 +01009470 - expect-netscaler-cip layer4 :
9471 configures the client-facing connection to receive a NetScaler Client
9472 IP insertion protocol header before any byte is read from the socket.
9473 This is equivalent to having the "accept-netscaler-cip" keyword on the
9474 "bind" line, except that using the TCP rule allows the PROXY protocol
9475 to be accepted only for certain IP address ranges using an ACL. This
9476 is convenient when multiple layers of load balancers are passed
9477 through by traffic coming from public hosts.
9478
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009479 - capture <sample> len <length> :
9480 This only applies to "tcp-request content" rules. It captures sample
9481 expression <sample> from the request buffer, and converts it to a
9482 string of at most <len> characters. The resulting string is stored into
9483 the next request "capture" slot, so it will possibly appear next to
9484 some captured HTTP headers. It will then automatically appear in the
9485 logs, and it will be possible to extract it using sample fetch rules to
9486 feed it into headers or anything. The length should be limited given
9487 that this size will be allocated for each capture during the whole
Willy Tarreaua9083d02015-05-08 15:27:59 +02009488 session life. Please check section 7.3 (Fetching samples) and "capture
9489 request header" for more information.
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009490
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009491 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>] :
Willy Tarreaue9656522010-08-17 15:40:09 +02009492 enables tracking of sticky counters from current connection. These
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +02009493 rules do not stop evaluation and do not change default action. The
9494 number of counters that may be simultaneously tracked by the same
9495 connection is set in MAX_SESS_STKCTR at build time (reported in
9496 haproxy -vv) whichs defaults to 3, so the track-sc number is between 0
9497 and (MAX_SESS_STCKTR-1). The first "track-sc0" rule executed enables
9498 tracking of the counters of the specified table as the first set. The
9499 first "track-sc1" rule executed enables tracking of the counters of the
9500 specified table as the second set. The first "track-sc2" rule executed
9501 enables tracking of the counters of the specified table as the third
9502 set. It is a recommended practice to use the first set of counters for
9503 the per-frontend counters and the second set for the per-backend ones.
9504 But this is just a guideline, all may be used everywhere.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009505
Willy Tarreaue9656522010-08-17 15:40:09 +02009506 These actions take one or two arguments :
Willy Tarreaube722a22014-06-13 16:31:59 +02009507 <key> is mandatory, and is a sample expression rule as described
Willy Tarreau74ca5042013-06-11 23:12:07 +02009508 in section 7.3. It describes what elements of the incoming
Davor Ocelice9ed2812017-12-25 17:49:28 +01009509 request or connection will be analyzed, extracted, combined,
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009510 and used to select which table entry to update the counters.
9511 Note that "tcp-request connection" cannot use content-based
9512 fetches.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009513
Willy Tarreaue9656522010-08-17 15:40:09 +02009514 <table> is an optional table to be used instead of the default one,
9515 which is the stick-table declared in the current proxy. All
9516 the counters for the matches and updates for the key will
9517 then be performed in that table until the session ends.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009518
Willy Tarreaue9656522010-08-17 15:40:09 +02009519 Once a "track-sc*" rule is executed, the key is looked up in the table
9520 and if it is not found, an entry is allocated for it. Then a pointer to
9521 that entry is kept during all the session's life, and this entry's
9522 counters are updated as often as possible, every time the session's
9523 counters are updated, and also systematically when the session ends.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009524 Counters are only updated for events that happen after the tracking has
9525 been started. For example, connection counters will not be updated when
9526 tracking layer 7 information, since the connection event happens before
9527 layer7 information is extracted.
9528
Willy Tarreaue9656522010-08-17 15:40:09 +02009529 If the entry tracks concurrent connection counters, one connection is
9530 counted for as long as the entry is tracked, and the entry will not
9531 expire during that time. Tracking counters also provides a performance
9532 advantage over just checking the keys, because only one table lookup is
9533 performed for all ACL checks that make use of it.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009534
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009535 - sc-inc-gpc0(<sc-id>):
9536 The "sc-inc-gpc0" increments the GPC0 counter according to the sticky
9537 counter designated by <sc-id>. If an error occurs, this action silently
9538 fails and the actions evaluation continues.
9539
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009540 - sc-inc-gpc1(<sc-id>):
9541 The "sc-inc-gpc1" increments the GPC1 counter according to the sticky
9542 counter designated by <sc-id>. If an error occurs, this action silently
9543 fails and the actions evaluation continues.
9544
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009545 - sc-set-gpt0(<sc-id>) <int>:
9546 This action sets the GPT0 tag according to the sticky counter designated
9547 by <sc-id> and the value of <int>. The expected result is a boolean. If
9548 an error occurs, this action silently fails and the actions evaluation
9549 continues.
9550
William Lallemand2e785f22016-05-25 01:48:42 +02009551 - set-src <expr> :
9552 Is used to set the source IP address to the value of specified
9553 expression. Useful if you want to mask source IP for privacy.
9554 If you want to provide an IP from a HTTP header use "http-request
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009555 set-src".
William Lallemand2e785f22016-05-25 01:48:42 +02009556
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009557 Arguments:
9558 <expr> Is a standard HAProxy expression formed by a sample-fetch
9559 followed by some converters.
William Lallemand2e785f22016-05-25 01:48:42 +02009560
9561 Example:
William Lallemand2e785f22016-05-25 01:48:42 +02009562 tcp-request connection set-src src,ipmask(24)
9563
Willy Tarreau0c630532016-10-21 17:52:58 +02009564 When possible, set-src preserves the original source port as long as the
9565 address family allows it, otherwise the source port is set to 0.
William Lallemand2e785f22016-05-25 01:48:42 +02009566
William Lallemand44be6402016-05-25 01:51:35 +02009567 - set-src-port <expr> :
9568 Is used to set the source port address to the value of specified
9569 expression.
9570
Cyril Bonté6c81d5f2018-10-17 00:14:51 +02009571 Arguments:
9572 <expr> Is a standard HAProxy expression formed by a sample-fetch
9573 followed by some converters.
William Lallemand44be6402016-05-25 01:51:35 +02009574
9575 Example:
William Lallemand44be6402016-05-25 01:51:35 +02009576 tcp-request connection set-src-port int(4000)
9577
Willy Tarreau0c630532016-10-21 17:52:58 +02009578 When possible, set-src-port preserves the original source address as long
9579 as the address family supports a port, otherwise it forces the source
9580 address to IPv4 "0.0.0.0" before rewriting the port.
William Lallemand44be6402016-05-25 01:51:35 +02009581
William Lallemand13e9b0c2016-05-25 02:34:07 +02009582 - set-dst <expr> :
9583 Is used to set the destination IP address to the value of specified
9584 expression. Useful if you want to mask IP for privacy in log.
9585 If you want to provide an IP from a HTTP header use "http-request
9586 set-dst". If you want to connect to the new address/port, use
9587 '0.0.0.0:0' as a server address in the backend.
9588
9589 <expr> Is a standard HAProxy expression formed by a sample-fetch
9590 followed by some converters.
9591
9592 Example:
9593
9594 tcp-request connection set-dst dst,ipmask(24)
9595 tcp-request connection set-dst ipv4(10.0.0.1)
9596
Willy Tarreau0c630532016-10-21 17:52:58 +02009597 When possible, set-dst preserves the original destination port as long as
9598 the address family allows it, otherwise the destination port is set to 0.
9599
William Lallemand13e9b0c2016-05-25 02:34:07 +02009600 - set-dst-port <expr> :
9601 Is used to set the destination port address to the value of specified
9602 expression. If you want to connect to the new address/port, use
9603 '0.0.0.0:0' as a server address in the backend.
9604
9605
9606 <expr> Is a standard HAProxy expression formed by a sample-fetch
9607 followed by some converters.
9608
9609 Example:
9610
9611 tcp-request connection set-dst-port int(4000)
9612
Willy Tarreau0c630532016-10-21 17:52:58 +02009613 When possible, set-dst-port preserves the original destination address as
9614 long as the address family supports a port, otherwise it forces the
9615 destination address to IPv4 "0.0.0.0" before rewriting the port.
9616
Willy Tarreau2d392c22015-08-24 01:43:45 +02009617 - "silent-drop" :
9618 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009619 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009620 to prevent the client from being notified. The effect it then that the
9621 client still sees an established connection while there's none on
9622 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9623 except that it doesn't use any local resource at all on the machine
9624 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009625 slow down stronger attackers. It is important to understand the impact
9626 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009627 client and HAProxy (firewalls, proxies, load balancers) will also keep
9628 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009629 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009630 TCP_REPAIR socket option is used to block the emission of a TCP
9631 reset. On other systems, the socket's TTL is reduced to 1 so that the
9632 TCP reset doesn't pass the first router, though it's still delivered to
9633 local networks. Do not use it unless you fully understand how it works.
9634
Willy Tarreaue9656522010-08-17 15:40:09 +02009635 Note that the "if/unless" condition is optional. If no condition is set on
9636 the action, it is simply performed unconditionally. That can be useful for
9637 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009638
Willy Tarreaue9656522010-08-17 15:40:09 +02009639 Example: accept all connections from white-listed hosts, reject too fast
9640 connection without counting them, and track accepted connections.
9641 This results in connection rate being capped from abusive sources.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009642
Willy Tarreaue9656522010-08-17 15:40:09 +02009643 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009644 tcp-request connection reject if { src_conn_rate gt 10 }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009645 tcp-request connection track-sc0 src
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009646
Willy Tarreaue9656522010-08-17 15:40:09 +02009647 Example: accept all connections from white-listed hosts, count all other
9648 connections and reject too fast ones. This results in abusive ones
9649 being blocked as long as they don't slow down.
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009650
Willy Tarreaue9656522010-08-17 15:40:09 +02009651 tcp-request connection accept if { src -f /etc/haproxy/whitelist.lst }
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009652 tcp-request connection track-sc0 src
9653 tcp-request connection reject if { sc0_conn_rate gt 10 }
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009654
Willy Tarreau4f0d9192013-06-11 20:40:55 +02009655 Example: enable the PROXY protocol for traffic coming from all known proxies.
9656
9657 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
9658
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009659 See section 7 about ACL usage.
9660
Willy Tarreau4f614292016-10-21 17:49:36 +02009661 See also : "tcp-request session", "tcp-request content", "stick-table"
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009662
9663
Willy Tarreaue9656522010-08-17 15:40:09 +02009664tcp-request content <action> [{if | unless} <condition>]
9665 Perform an action on a new session depending on a layer 4-7 condition
Willy Tarreau62644772008-07-16 18:36:06 +02009666 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009667 no | yes | yes | yes
Willy Tarreaue9656522010-08-17 15:40:09 +02009668 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009669 <action> defines the action to perform if the condition applies. See
9670 below.
Willy Tarreau62644772008-07-16 18:36:06 +02009671
Willy Tarreaue9656522010-08-17 15:40:09 +02009672 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
Willy Tarreau62644772008-07-16 18:36:06 +02009673
Davor Ocelice9ed2812017-12-25 17:49:28 +01009674 A request's contents can be analyzed at an early stage of request processing
Willy Tarreaue9656522010-08-17 15:40:09 +02009675 called "TCP content inspection". During this stage, ACL-based rules are
9676 evaluated every time the request contents are updated, until either an
9677 "accept" or a "reject" rule matches, or the TCP request inspection delay
9678 expires with no matching rule.
Willy Tarreau62644772008-07-16 18:36:06 +02009679
Willy Tarreaue9656522010-08-17 15:40:09 +02009680 The first difference between these rules and "tcp-request connection" rules
9681 is that "tcp-request content" rules can make use of contents to take a
9682 decision. Most often, these decisions will consider a protocol recognition or
9683 validity. The second difference is that content-based rules can be used in
Willy Tarreauf3338342014-01-28 21:40:28 +01009684 both frontends and backends. In case of HTTP keep-alive with the client, all
9685 tcp-request content rules are evaluated again, so haproxy keeps a record of
9686 what sticky counters were assigned by a "tcp-request connection" versus a
9687 "tcp-request content" rule, and flushes all the content-related ones after
9688 processing an HTTP request, so that they may be evaluated again by the rules
9689 being evaluated again for the next request. This is of particular importance
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009690 when the rule tracks some L7 information or when it is conditioned by an
Willy Tarreauf3338342014-01-28 21:40:28 +01009691 L7-based ACL, since tracking may change between requests.
Willy Tarreau62644772008-07-16 18:36:06 +02009692
Willy Tarreaue9656522010-08-17 15:40:09 +02009693 Content-based rules are evaluated in their exact declaration order. If no
9694 rule matches or if there is no rule, the default action is to accept the
9695 contents. There is no specific limit to the number of rules which may be
9696 inserted.
Willy Tarreau62644772008-07-16 18:36:06 +02009697
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009698 Several types of actions are supported :
Willy Tarreau18bf01e2014-06-13 16:18:52 +02009699 - accept : the request is accepted
9700 - reject : the request is rejected and the connection is closed
9701 - capture : the specified sample expression is captured
Patrick Hemmer268a7072018-05-11 12:52:31 -04009702 - set-priority-class <expr> | set-priority-offset <expr>
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009703 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009704 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009705 - sc-inc-gpc1(<sc-id>)
Thierry Fournierb9125672016-03-29 19:34:37 +02009706 - sc-set-gpt0(<sc-id>) <int>
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009707 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +01009708 - unset-var(<var-name>)
Willy Tarreau2d392c22015-08-24 01:43:45 +02009709 - silent-drop
Davor Ocelice9ed2812017-12-25 17:49:28 +01009710 - send-spoe-group <engine-name> <group-name>
Willy Tarreau62644772008-07-16 18:36:06 +02009711
Willy Tarreaue9656522010-08-17 15:40:09 +02009712 They have the same meaning as their counter-parts in "tcp-request connection"
9713 so please refer to that section for a complete description.
Willy Tarreau62644772008-07-16 18:36:06 +02009714
Willy Tarreauf3338342014-01-28 21:40:28 +01009715 While there is nothing mandatory about it, it is recommended to use the
9716 track-sc0 in "tcp-request connection" rules, track-sc1 for "tcp-request
9717 content" rules in the frontend, and track-sc2 for "tcp-request content"
9718 rules in the backend, because that makes the configuration more readable
9719 and easier to troubleshoot, but this is just a guideline and all counters
9720 may be used everywhere.
Willy Tarreau62644772008-07-16 18:36:06 +02009721
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009722 Note that the "if/unless" condition is optional. If no condition is set on
Willy Tarreaue9656522010-08-17 15:40:09 +02009723 the action, it is simply performed unconditionally. That can be useful for
9724 "track-sc*" actions as well as for changing the default action to a reject.
Willy Tarreau62644772008-07-16 18:36:06 +02009725
Willy Tarreaue9656522010-08-17 15:40:09 +02009726 It is perfectly possible to match layer 7 contents with "tcp-request content"
Willy Tarreauc0239e02012-04-16 14:42:55 +02009727 rules, since HTTP-specific ACL matches are able to preliminarily parse the
9728 contents of a buffer before extracting the required data. If the buffered
9729 contents do not parse as a valid HTTP message, then the ACL does not match.
9730 The parser which is involved there is exactly the same as for all other HTTP
Willy Tarreauf3338342014-01-28 21:40:28 +01009731 processing, so there is no risk of parsing something differently. In an HTTP
9732 backend connected to from an HTTP frontend, it is guaranteed that HTTP
9733 contents will always be immediately present when the rule is evaluated first.
Willy Tarreau62644772008-07-16 18:36:06 +02009734
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009735 Tracking layer7 information is also possible provided that the information
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009736 are present when the rule is processed. The rule processing engine is able to
9737 wait until the inspect delay expires when the data to be tracked is not yet
9738 available.
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009739
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009740 The "set-var" is used to set the content of a variable. The variable is
Willy Tarreau4f614292016-10-21 17:49:36 +02009741 declared inline. For "tcp-request session" rules, only session-level
9742 variables can be used, without any layer7 contents.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009743
Daniel Schneller0b547052016-03-21 20:46:57 +01009744 <var-name> The name of the variable starts with an indication about
9745 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +01009746 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +01009747 "sess" : the variable is shared with the whole session
9748 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009749 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +01009750 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009751 processing
Daniel Schneller0b547052016-03-21 20:46:57 +01009752 "res" : the variable is shared only during response
9753 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009754 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +01009755 The name may only contain characters 'a-z', 'A-Z', '0-9',
9756 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009757
9758 <expr> Is a standard HAProxy expression formed by a sample-fetch
9759 followed by some converters.
9760
Christopher Faulet85d79c92016-11-09 16:54:56 +01009761 The "unset-var" is used to unset a variable. See above for details about
9762 <var-name>.
9763
Patrick Hemmer268a7072018-05-11 12:52:31 -04009764 The "set-priority-class" is used to set the queue priority class of the
9765 current request. The value must be a sample expression which converts to an
9766 integer in the range -2047..2047. Results outside this range will be
9767 truncated. The priority class determines the order in which queued requests
9768 are processed. Lower values have higher priority.
9769
9770 The "set-priority-offset" is used to set the queue priority timestamp offset
9771 of the current request. The value must be a sample expression which converts
9772 to an integer in the range -524287..524287. Results outside this range will be
9773 truncated. When a request is queued, it is ordered first by the priority
9774 class, then by the current timestamp adjusted by the given offset in
9775 milliseconds. Lower values have higher priority.
9776 Note that the resulting timestamp is is only tracked with enough precision for
9777 524,287ms (8m44s287ms). If the request is queued long enough to where the
9778 adjusted timestamp exceeds this value, it will be misidentified as highest
9779 priority. Thus it is important to set "timeout queue" to a value, where when
9780 combined with the offset, does not exceed this limit.
9781
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009782 The "send-spoe-group" is used to trigger sending of a group of SPOE
9783 messages. To do so, the SPOE engine used to send messages must be defined, as
9784 well as the SPOE group to send. Of course, the SPOE engine must refer to an
9785 existing SPOE filter. If not engine name is provided on the SPOE filter line,
9786 the SPOE agent name must be used.
9787
9788 <engine-name> The SPOE engine name.
9789
9790 <group-name> The SPOE group name as specified in the engine configuration.
9791
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009792 Example:
9793
9794 tcp-request content set-var(sess.my_var) src
Christopher Faulet85d79c92016-11-09 16:54:56 +01009795 tcp-request content unset-var(sess.my_var2)
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009796
Willy Tarreau62644772008-07-16 18:36:06 +02009797 Example:
Willy Tarreaue9656522010-08-17 15:40:09 +02009798 # Accept HTTP requests containing a Host header saying "example.com"
9799 # and reject everything else.
9800 acl is_host_com hdr(Host) -i example.com
9801 tcp-request inspect-delay 30s
Willy Tarreauc0239e02012-04-16 14:42:55 +02009802 tcp-request content accept if is_host_com
Willy Tarreaue9656522010-08-17 15:40:09 +02009803 tcp-request content reject
9804
9805 Example:
Willy Tarreau62644772008-07-16 18:36:06 +02009806 # reject SMTP connection if client speaks first
9807 tcp-request inspect-delay 30s
9808 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009809 tcp-request content reject if content_present
Willy Tarreau62644772008-07-16 18:36:06 +02009810
9811 # Forward HTTPS connection only if client speaks
9812 tcp-request inspect-delay 30s
9813 acl content_present req_len gt 0
Willy Tarreau68c03ab2010-08-06 15:08:45 +02009814 tcp-request content accept if content_present
Willy Tarreaue9656522010-08-17 15:40:09 +02009815 tcp-request content reject
9816
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009817 Example:
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009818 # Track the last IP(stick-table type string) from X-Forwarded-For
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009819 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009820 tcp-request content track-sc0 hdr(x-forwarded-for,-1)
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009821 # Or track the last IP(stick-table type ip|ipv6) from X-Forwarded-For
9822 tcp-request content track-sc0 req.hdr_ip(x-forwarded-for,-1)
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009823
9824 Example:
9825 # track request counts per "base" (concatenation of Host+URL)
9826 tcp-request inspect-delay 10s
Willy Tarreau4d54c7c2014-09-16 15:48:15 +02009827 tcp-request content track-sc0 base table req-rate
Willy Tarreau5d5b5d82012-12-09 12:00:04 +01009828
Willy Tarreaue9656522010-08-17 15:40:09 +02009829 Example: track per-frontend and per-backend counters, block abusers at the
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009830 frontend when the backend detects abuse(and marks gpc0).
Willy Tarreaue9656522010-08-17 15:40:09 +02009831
9832 frontend http
Davor Ocelice9ed2812017-12-25 17:49:28 +01009833 # Use General Purpose Counter 0 in SC0 as a global abuse counter
Willy Tarreaue9656522010-08-17 15:40:09 +02009834 # protecting all our sites
9835 stick-table type ip size 1m expire 5m store gpc0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009836 tcp-request connection track-sc0 src
9837 tcp-request connection reject if { sc0_get_gpc0 gt 0 }
Willy Tarreaue9656522010-08-17 15:40:09 +02009838 ...
9839 use_backend http_dynamic if { path_end .php }
9840
9841 backend http_dynamic
9842 # if a source makes too fast requests to this dynamic site (tracked
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009843 # by SC1), block it globally in the frontend.
Willy Tarreaue9656522010-08-17 15:40:09 +02009844 stick-table type ip size 1m expire 5m store http_req_rate(10s)
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009845 acl click_too_fast sc1_http_req_rate gt 10
Jarno Huuskonene5ae7022017-04-03 14:36:21 +03009846 acl mark_as_abuser sc0_inc_gpc0(http) gt 0
Willy Tarreaube4a3ef2013-06-17 15:04:07 +02009847 tcp-request content track-sc1 src
Willy Tarreaue9656522010-08-17 15:40:09 +02009848 tcp-request content reject if click_too_fast mark_as_abuser
Willy Tarreau62644772008-07-16 18:36:06 +02009849
Willy Tarreauc57f0e22009-05-10 13:12:33 +02009850 See section 7 about ACL usage.
Willy Tarreau62644772008-07-16 18:36:06 +02009851
Jarno Huuskonen95b012b2017-04-06 13:59:14 +03009852 See also : "tcp-request connection", "tcp-request session",
9853 "tcp-request inspect-delay", and "http-request".
Willy Tarreau62644772008-07-16 18:36:06 +02009854
9855
9856tcp-request inspect-delay <timeout>
9857 Set the maximum allowed time to wait for data during content inspection
9858 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaufb356202010-08-03 14:02:05 +02009859 no | yes | yes | yes
Willy Tarreau62644772008-07-16 18:36:06 +02009860 Arguments :
9861 <timeout> is the timeout value specified in milliseconds by default, but
9862 can be in any other unit if the number is suffixed by the unit,
9863 as explained at the top of this document.
9864
9865 People using haproxy primarily as a TCP relay are often worried about the
9866 risk of passing any type of protocol to a server without any analysis. In
9867 order to be able to analyze the request contents, we must first withhold
9868 the data then analyze them. This statement simply enables withholding of
9869 data for at most the specified amount of time.
9870
Willy Tarreaufb356202010-08-03 14:02:05 +02009871 TCP content inspection applies very early when a connection reaches a
9872 frontend, then very early when the connection is forwarded to a backend. This
9873 means that a connection may experience a first delay in the frontend and a
9874 second delay in the backend if both have tcp-request rules.
9875
Willy Tarreau62644772008-07-16 18:36:06 +02009876 Note that when performing content inspection, haproxy will evaluate the whole
9877 rules for every new chunk which gets in, taking into account the fact that
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +01009878 those data are partial. If no rule matches before the aforementioned delay,
Willy Tarreau62644772008-07-16 18:36:06 +02009879 a last check is performed upon expiration, this time considering that the
Willy Tarreaud869b242009-03-15 14:43:58 +01009880 contents are definitive. If no delay is set, haproxy will not wait at all
9881 and will immediately apply a verdict based on the available information.
9882 Obviously this is unlikely to be very useful and might even be racy, so such
9883 setups are not recommended.
Willy Tarreau62644772008-07-16 18:36:06 +02009884
9885 As soon as a rule matches, the request is released and continues as usual. If
9886 the timeout is reached and no rule matches, the default policy will be to let
9887 it pass through unaffected.
9888
9889 For most protocols, it is enough to set it to a few seconds, as most clients
9890 send the full request immediately upon connection. Add 3 or more seconds to
9891 cover TCP retransmits but that's all. For some protocols, it may make sense
Willy Tarreaud72758d2010-01-12 10:42:19 +01009892 to use large values, for instance to ensure that the client never talks
Davor Ocelice9ed2812017-12-25 17:49:28 +01009893 before the server (e.g. SMTP), or to wait for a client to talk before passing
9894 data to the server (e.g. SSL). Note that the client timeout must cover at
Willy Tarreaub824b002010-09-29 16:36:16 +02009895 least the inspection delay, otherwise it will expire first. If the client
9896 closes the connection or if the buffer is full, the delay immediately expires
9897 since the contents will not be able to change anymore.
Willy Tarreau62644772008-07-16 18:36:06 +02009898
Willy Tarreau55165fe2009-05-10 12:02:55 +02009899 See also : "tcp-request content accept", "tcp-request content reject",
Willy Tarreau62644772008-07-16 18:36:06 +02009900 "timeout client".
9901
9902
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009903tcp-response content <action> [{if | unless} <condition>]
9904 Perform an action on a session response depending on a layer 4-7 condition
9905 May be used in sections : defaults | frontend | listen | backend
9906 no | no | yes | yes
9907 Arguments :
Willy Tarreauc870bfd2015-09-28 18:47:38 +02009908 <action> defines the action to perform if the condition applies. See
9909 below.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009910
9911 <condition> is a standard layer 4-7 ACL-based condition (see section 7).
9912
Davor Ocelice9ed2812017-12-25 17:49:28 +01009913 Response contents can be analyzed at an early stage of response processing
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009914 called "TCP content inspection". During this stage, ACL-based rules are
9915 evaluated every time the response contents are updated, until either an
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009916 "accept", "close" or a "reject" rule matches, or a TCP response inspection
9917 delay is set and expires with no matching rule.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009918
9919 Most often, these decisions will consider a protocol recognition or validity.
9920
9921 Content-based rules are evaluated in their exact declaration order. If no
9922 rule matches or if there is no rule, the default action is to accept the
9923 contents. There is no specific limit to the number of rules which may be
9924 inserted.
9925
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009926 Several types of actions are supported :
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009927 - accept :
9928 accepts the response if the condition is true (when used with "if")
9929 or false (when used with "unless"). The first such rule executed ends
9930 the rules evaluation.
9931
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009932 - close :
9933 immediately closes the connection with the server if the condition is
9934 true (when used with "if"), or false (when used with "unless"). The
9935 first such rule executed ends the rules evaluation. The main purpose of
9936 this action is to force a connection to be finished between a client
9937 and a server after an exchange when the application protocol expects
9938 some long time outs to elapse first. The goal is to eliminate idle
Jarno Huuskonen0e82b922014-04-12 18:22:19 +03009939 connections which take significant resources on servers with certain
Willy Tarreaucc1e04b2013-09-11 23:20:29 +02009940 protocols.
9941
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009942 - reject :
9943 rejects the response if the condition is true (when used with "if")
9944 or false (when used with "unless"). The first such rule executed ends
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009945 the rules evaluation. Rejected session are immediately closed.
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009946
Thierry FOURNIER4834bc72015-06-06 19:29:07 +02009947 - set-var(<var-name>) <expr>
9948 Sets a variable.
9949
Christopher Faulet85d79c92016-11-09 16:54:56 +01009950 - unset-var(<var-name>)
9951 Unsets a variable.
9952
Thierry FOURNIERe0627bd2015-08-04 08:20:33 +02009953 - sc-inc-gpc0(<sc-id>):
9954 This action increments the GPC0 counter according to the sticky
9955 counter designated by <sc-id>. If an error occurs, this action fails
9956 silently and the actions evaluation continues.
9957
Frédéric Lécaille6778b272018-01-29 15:22:53 +01009958 - sc-inc-gpc1(<sc-id>):
9959 This action increments the GPC1 counter according to the sticky
9960 counter designated by <sc-id>. If an error occurs, this action fails
9961 silently and the actions evaluation continues.
9962
Thierry FOURNIER236657b2015-08-19 08:25:14 +02009963 - sc-set-gpt0(<sc-id>) <int> :
9964 This action sets the GPT0 tag according to the sticky counter designated
9965 by <sc-id> and the value of <int>. The expected result is a boolean. If
9966 an error occurs, this action silently fails and the actions evaluation
9967 continues.
9968
Willy Tarreau2d392c22015-08-24 01:43:45 +02009969 - "silent-drop" :
9970 This stops the evaluation of the rules and makes the client-facing
Davor Ocelice9ed2812017-12-25 17:49:28 +01009971 connection suddenly disappear using a system-dependent way that tries
Willy Tarreau2d392c22015-08-24 01:43:45 +02009972 to prevent the client from being notified. The effect it then that the
9973 client still sees an established connection while there's none on
9974 HAProxy. The purpose is to achieve a comparable effect to "tarpit"
9975 except that it doesn't use any local resource at all on the machine
9976 running HAProxy. It can resist much higher loads than "tarpit", and
Davor Ocelice9ed2812017-12-25 17:49:28 +01009977 slow down stronger attackers. It is important to understand the impact
9978 of using this mechanism. All stateful equipment placed between the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009979 client and HAProxy (firewalls, proxies, load balancers) will also keep
9980 the established connection for a long time and may suffer from this
Davor Ocelice9ed2812017-12-25 17:49:28 +01009981 action. On modern Linux systems running with enough privileges, the
Willy Tarreau2d392c22015-08-24 01:43:45 +02009982 TCP_REPAIR socket option is used to block the emission of a TCP
9983 reset. On other systems, the socket's TTL is reduced to 1 so that the
9984 TCP reset doesn't pass the first router, though it's still delivered to
9985 local networks. Do not use it unless you fully understand how it works.
9986
Christopher Faulet76c09ef2017-09-21 11:03:52 +02009987 - send-spoe-group <engine-name> <group-name>
9988 Send a group of SPOE messages.
9989
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009990 Note that the "if/unless" condition is optional. If no condition is set on
9991 the action, it is simply performed unconditionally. That can be useful for
9992 for changing the default action to a reject.
9993
Jamie Gloudonaaa21002012-08-25 00:18:33 -04009994 It is perfectly possible to match layer 7 contents with "tcp-response
9995 content" rules, but then it is important to ensure that a full response has
9996 been buffered, otherwise no contents will match. In order to achieve this,
9997 the best solution involves detecting the HTTP protocol during the inspection
Emeric Brun0a3b67f2010-09-24 15:34:53 +02009998 period.
9999
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010000 The "set-var" is used to set the content of a variable. The variable is
10001 declared inline.
10002
Daniel Schneller0b547052016-03-21 20:46:57 +010010003 <var-name> The name of the variable starts with an indication about
10004 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010010005 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010010006 "sess" : the variable is shared with the whole session
10007 "txn" : the variable is shared with the transaction
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010008 (request and response)
Daniel Schneller0b547052016-03-21 20:46:57 +010010009 "req" : the variable is shared only during request
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010010 processing
Daniel Schneller0b547052016-03-21 20:46:57 +010010011 "res" : the variable is shared only during response
10012 processing
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010013 This prefix is followed by a name. The separator is a '.'.
Christopher Fauletb71557a2016-10-31 10:49:03 +010010014 The name may only contain characters 'a-z', 'A-Z', '0-9',
10015 '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020010016
10017 <expr> Is a standard HAProxy expression formed by a sample-fetch
10018 followed by some converters.
10019
10020 Example:
10021
10022 tcp-request content set-var(sess.my_var) src
10023
Christopher Faulet85d79c92016-11-09 16:54:56 +010010024 The "unset-var" is used to unset a variable. See above for details about
10025 <var-name>.
10026
10027 Example:
10028
10029 tcp-request content unset-var(sess.my_var)
10030
Christopher Faulet76c09ef2017-09-21 11:03:52 +020010031 The "send-spoe-group" is used to trigger sending of a group of SPOE
10032 messages. To do so, the SPOE engine used to send messages must be defined, as
10033 well as the SPOE group to send. Of course, the SPOE engine must refer to an
10034 existing SPOE filter. If not engine name is provided on the SPOE filter line,
10035 the SPOE agent name must be used.
10036
10037 <engine-name> The SPOE engine name.
10038
10039 <group-name> The SPOE group name as specified in the engine configuration.
10040
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010041 See section 7 about ACL usage.
10042
10043 See also : "tcp-request content", "tcp-response inspect-delay"
10044
10045
Willy Tarreau4f614292016-10-21 17:49:36 +020010046tcp-request session <action> [{if | unless} <condition>]
10047 Perform an action on a validated session depending on a layer 5 condition
10048 May be used in sections : defaults | frontend | listen | backend
10049 no | yes | yes | no
10050 Arguments :
10051 <action> defines the action to perform if the condition applies. See
10052 below.
10053
10054 <condition> is a standard layer5-only ACL-based condition (see section 7).
10055
Davor Ocelice9ed2812017-12-25 17:49:28 +010010056 Once a session is validated, (i.e. after all handshakes have been completed),
Willy Tarreau4f614292016-10-21 17:49:36 +020010057 it is possible to evaluate some conditions to decide whether this session
10058 must be accepted or dropped or have its counters tracked. Those conditions
10059 cannot make use of any data contents because no buffers are allocated yet and
10060 the processing cannot wait at this stage. The main use case it to copy some
10061 early information into variables (since variables are accessible in the
10062 session), or to keep track of some information collected after the handshake,
10063 such as SSL-level elements (SNI, ciphers, client cert's CN) or information
Davor Ocelice9ed2812017-12-25 17:49:28 +010010064 from the PROXY protocol header (e.g. track a source forwarded this way). The
Willy Tarreau4f614292016-10-21 17:49:36 +020010065 extracted information can thus be copied to a variable or tracked using
10066 "track-sc" rules. Of course it is also possible to decide to accept/reject as
10067 with other rulesets. Most operations performed here could also be performed
10068 in "tcp-request content" rules, except that in HTTP these rules are evaluated
10069 for each new request, and that might not always be acceptable. For example a
10070 rule might increment a counter on each evaluation. It would also be possible
10071 that a country is resolved by geolocation from the source IP address,
10072 assigned to a session-wide variable, then the source address rewritten from
10073 an HTTP header for all requests. If some contents need to be inspected in
10074 order to take the decision, the "tcp-request content" statements must be used
10075 instead.
10076
10077 The "tcp-request session" rules are evaluated in their exact declaration
10078 order. If no rule matches or if there is no rule, the default action is to
10079 accept the incoming session. There is no specific limit to the number of
10080 rules which may be inserted.
10081
10082 Several types of actions are supported :
10083 - accept : the request is accepted
10084 - reject : the request is rejected and the connection is closed
10085 - { track-sc0 | track-sc1 | track-sc2 } <key> [table <table>]
10086 - sc-inc-gpc0(<sc-id>)
Frédéric Lécaille6778b272018-01-29 15:22:53 +010010087 - sc-inc-gpc1(<sc-id>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010088 - sc-set-gpt0(<sc-id>) <int>
10089 - set-var(<var-name>) <expr>
Christopher Faulet85d79c92016-11-09 16:54:56 +010010090 - unset-var(<var-name>)
Willy Tarreau4f614292016-10-21 17:49:36 +020010091 - silent-drop
10092
10093 These actions have the same meaning as their respective counter-parts in
10094 "tcp-request connection" and "tcp-request content", so please refer to these
10095 sections for a complete description.
10096
10097 Note that the "if/unless" condition is optional. If no condition is set on
10098 the action, it is simply performed unconditionally. That can be useful for
10099 "track-sc*" actions as well as for changing the default action to a reject.
10100
10101 Example: track the original source address by default, or the one advertised
10102 in the PROXY protocol header for connection coming from the local
10103 proxies. The first connection-level rule enables receipt of the
10104 PROXY protocol for these ones, the second rule tracks whatever
10105 address we decide to keep after optional decoding.
10106
10107 tcp-request connection expect-proxy layer4 if { src -f proxies.lst }
10108 tcp-request session track-sc0 src
10109
10110 Example: accept all sessions from white-listed hosts, reject too fast
10111 sessions without counting them, and track accepted sessions.
10112 This results in session rate being capped from abusive sources.
10113
10114 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10115 tcp-request session reject if { src_sess_rate gt 10 }
10116 tcp-request session track-sc0 src
10117
10118 Example: accept all sessions from white-listed hosts, count all other
10119 sessions and reject too fast ones. This results in abusive ones
10120 being blocked as long as they don't slow down.
10121
10122 tcp-request session accept if { src -f /etc/haproxy/whitelist.lst }
10123 tcp-request session track-sc0 src
10124 tcp-request session reject if { sc0_sess_rate gt 10 }
10125
10126 See section 7 about ACL usage.
10127
10128 See also : "tcp-request connection", "tcp-request content", "stick-table"
10129
10130
Emeric Brun0a3b67f2010-09-24 15:34:53 +020010131tcp-response inspect-delay <timeout>
10132 Set the maximum allowed time to wait for a response during content inspection
10133 May be used in sections : defaults | frontend | listen | backend
10134 no | no | yes | yes
10135 Arguments :
10136 <timeout> is the timeout value specified in milliseconds by default, but
10137 can be in any other unit if the number is suffixed by the unit,
10138 as explained at the top of this document.
10139
10140 See also : "tcp-response content", "tcp-request inspect-delay".
10141
10142
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010143timeout check <timeout>
10144 Set additional check timeout, but only after a connection has been already
10145 established.
10146
10147 May be used in sections: defaults | frontend | listen | backend
10148 yes | no | yes | yes
10149 Arguments:
10150 <timeout> is the timeout value specified in milliseconds by default, but
10151 can be in any other unit if the number is suffixed by the unit,
10152 as explained at the top of this document.
10153
10154 If set, haproxy uses min("timeout connect", "inter") as a connect timeout
10155 for check and "timeout check" as an additional read timeout. The "min" is
Davor Ocelice9ed2812017-12-25 17:49:28 +010010156 used so that people running with *very* long "timeout connect" (e.g. those
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010157 who needed this due to the queue or tarpit) do not slow down their checks.
Willy Tarreaud7550a22010-02-10 05:10:19 +010010158 (Please also note that there is no valid reason to have such long connect
10159 timeouts, because "timeout queue" and "timeout tarpit" can always be used to
10160 avoid that).
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010161
10162 If "timeout check" is not set haproxy uses "inter" for complete check
10163 timeout (connect + read) exactly like all <1.3.15 version.
10164
10165 In most cases check request is much simpler and faster to handle than normal
10166 requests and people may want to kick out laggy servers so this timeout should
Willy Tarreau41a340d2008-01-22 12:25:31 +010010167 be smaller than "timeout server".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010168
10169 This parameter is specific to backends, but can be specified once for all in
10170 "defaults" sections. This is in fact one of the easiest solutions not to
10171 forget about it.
10172
Willy Tarreau41a340d2008-01-22 12:25:31 +010010173 See also: "timeout connect", "timeout queue", "timeout server",
10174 "timeout tarpit".
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010175
10176
Willy Tarreau0ba27502007-12-24 16:55:16 +010010177timeout client <timeout>
10178timeout clitimeout <timeout> (deprecated)
10179 Set the maximum inactivity time on the client side.
10180 May be used in sections : defaults | frontend | listen | backend
10181 yes | yes | yes | no
10182 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010183 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010184 can be in any other unit if the number is suffixed by the unit,
10185 as explained at the top of this document.
10186
10187 The inactivity timeout applies when the client is expected to acknowledge or
10188 send data. In HTTP mode, this timeout is particularly important to consider
10189 during the first phase, when the client sends the request, and during the
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010190 response while it is reading data sent by the server. That said, for the
10191 first phase, it is preferable to set the "timeout http-request" to better
10192 protect HAProxy from Slowloris like attacks. The value is specified in
10193 milliseconds by default, but can be in any other unit if the number is
Willy Tarreau0ba27502007-12-24 16:55:16 +010010194 suffixed by the unit, as specified at the top of this document. In TCP mode
10195 (and to a lesser extent, in HTTP mode), it is highly recommended that the
10196 client timeout remains equal to the server timeout in order to avoid complex
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010197 situations to debug. It is a good practice to cover one or several TCP packet
Willy Tarreau0ba27502007-12-24 16:55:16 +010010198 losses by specifying timeouts that are slightly above multiples of 3 seconds
Davor Ocelice9ed2812017-12-25 17:49:28 +010010199 (e.g. 4 or 5 seconds). If some long-lived sessions are mixed with short-lived
10200 sessions (e.g. WebSocket and HTTP), it's worth considering "timeout tunnel",
Willy Tarreau05cdd962014-05-10 14:30:07 +020010201 which overrides "timeout client" and "timeout server" for tunnels, as well as
10202 "timeout client-fin" for half-closed connections.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010203
10204 This parameter is specific to frontends, but can be specified once for all in
10205 "defaults" sections. This is in fact one of the easiest solutions not to
10206 forget about it. An unspecified timeout results in an infinite timeout, which
10207 is not recommended. Such a usage is accepted and works but reports a warning
10208 during startup because it may results in accumulation of expired sessions in
10209 the system if the system's timeouts are not configured either.
10210
Willy Tarreau95c4e142017-11-26 12:18:55 +010010211 This also applies to HTTP/2 connections, which will be closed with GOAWAY.
Lukas Tribus75df9d72017-11-24 19:05:12 +010010212
Willy Tarreau0ba27502007-12-24 16:55:16 +010010213 This parameter replaces the old, deprecated "clitimeout". It is recommended
10214 to use it to write new configurations. The form "timeout clitimeout" is
10215 provided only by backwards compatibility but its use is strongly discouraged.
10216
Baptiste Assmann2e1941e2016-03-06 23:24:12 +010010217 See also : "clitimeout", "timeout server", "timeout tunnel",
10218 "timeout http-request".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010219
10220
Willy Tarreau05cdd962014-05-10 14:30:07 +020010221timeout client-fin <timeout>
10222 Set the inactivity timeout on the client side for half-closed connections.
10223 May be used in sections : defaults | frontend | listen | backend
10224 yes | yes | yes | no
10225 Arguments :
10226 <timeout> is the timeout value specified in milliseconds by default, but
10227 can be in any other unit if the number is suffixed by the unit,
10228 as explained at the top of this document.
10229
10230 The inactivity timeout applies when the client is expected to acknowledge or
10231 send data while one direction is already shut down. This timeout is different
10232 from "timeout client" in that it only applies to connections which are closed
10233 in one direction. This is particularly useful to avoid keeping connections in
10234 FIN_WAIT state for too long when clients do not disconnect cleanly. This
10235 problem is particularly common long connections such as RDP or WebSocket.
10236 Note that this timeout can override "timeout tunnel" when a connection shuts
Willy Tarreau599391a2017-11-24 10:16:00 +010010237 down in one direction. It is applied to idle HTTP/2 connections once a GOAWAY
10238 frame was sent, often indicating an expectation that the connection quickly
10239 ends.
Willy Tarreau05cdd962014-05-10 14:30:07 +020010240
10241 This parameter is specific to frontends, but can be specified once for all in
10242 "defaults" sections. By default it is not set, so half-closed connections
10243 will use the other timeouts (timeout.client or timeout.tunnel).
10244
10245 See also : "timeout client", "timeout server-fin", and "timeout tunnel".
10246
10247
Willy Tarreau0ba27502007-12-24 16:55:16 +010010248timeout connect <timeout>
10249timeout contimeout <timeout> (deprecated)
10250 Set the maximum time to wait for a connection attempt to a server to succeed.
10251 May be used in sections : defaults | frontend | listen | backend
10252 yes | no | yes | yes
10253 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010254 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau0ba27502007-12-24 16:55:16 +010010255 can be in any other unit if the number is suffixed by the unit,
10256 as explained at the top of this document.
10257
10258 If the server is located on the same LAN as haproxy, the connection should be
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010259 immediate (less than a few milliseconds). Anyway, it is a good practice to
Willy Tarreaud72758d2010-01-12 10:42:19 +010010260 cover one or several TCP packet losses by specifying timeouts that are
Davor Ocelice9ed2812017-12-25 17:49:28 +010010261 slightly above multiples of 3 seconds (e.g. 4 or 5 seconds). By default, the
Krzysztof Piotr Oledzki5259dfe2008-01-21 01:54:06 +010010262 connect timeout also presets both queue and tarpit timeouts to the same value
10263 if these have not been specified.
Willy Tarreau0ba27502007-12-24 16:55:16 +010010264
10265 This parameter is specific to backends, but can be specified once for all in
10266 "defaults" sections. This is in fact one of the easiest solutions not to
10267 forget about it. An unspecified timeout results in an infinite timeout, which
10268 is not recommended. Such a usage is accepted and works but reports a warning
10269 during startup because it may results in accumulation of failed sessions in
10270 the system if the system's timeouts are not configured either.
10271
10272 This parameter replaces the old, deprecated "contimeout". It is recommended
10273 to use it to write new configurations. The form "timeout contimeout" is
10274 provided only by backwards compatibility but its use is strongly discouraged.
10275
Willy Tarreau41a340d2008-01-22 12:25:31 +010010276 See also: "timeout check", "timeout queue", "timeout server", "contimeout",
10277 "timeout tarpit".
Willy Tarreau0ba27502007-12-24 16:55:16 +010010278
10279
Willy Tarreaub16a5742010-01-10 14:46:16 +010010280timeout http-keep-alive <timeout>
10281 Set the maximum allowed time to wait for a new HTTP request to appear
10282 May be used in sections : defaults | frontend | listen | backend
10283 yes | yes | yes | yes
10284 Arguments :
10285 <timeout> is the timeout value specified in milliseconds by default, but
10286 can be in any other unit if the number is suffixed by the unit,
10287 as explained at the top of this document.
10288
10289 By default, the time to wait for a new request in case of keep-alive is set
10290 by "timeout http-request". However this is not always convenient because some
10291 people want very short keep-alive timeouts in order to release connections
10292 faster, and others prefer to have larger ones but still have short timeouts
10293 once the request has started to present itself.
10294
10295 The "http-keep-alive" timeout covers these needs. It will define how long to
10296 wait for a new HTTP request to start coming after a response was sent. Once
10297 the first byte of request has been seen, the "http-request" timeout is used
10298 to wait for the complete request to come. Note that empty lines prior to a
10299 new request do not refresh the timeout and are not counted as a new request.
10300
10301 There is also another difference between the two timeouts : when a connection
10302 expires during timeout http-keep-alive, no error is returned, the connection
10303 just closes. If the connection expires in "http-request" while waiting for a
10304 connection to complete, a HTTP 408 error is returned.
10305
10306 In general it is optimal to set this value to a few tens to hundreds of
10307 milliseconds, to allow users to fetch all objects of a page at once but
Davor Ocelice9ed2812017-12-25 17:49:28 +010010308 without waiting for further clicks. Also, if set to a very small value (e.g.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010309 1 millisecond) it will probably only accept pipelined requests but not the
10310 non-pipelined ones. It may be a nice trade-off for very large sites running
Patrick Mézard2382ad62010-05-09 10:43:32 +020010311 with tens to hundreds of thousands of clients.
Willy Tarreaub16a5742010-01-10 14:46:16 +010010312
10313 If this parameter is not set, the "http-request" timeout applies, and if both
10314 are not set, "timeout client" still applies at the lower level. It should be
10315 set in the frontend to take effect, unless the frontend is in TCP mode, in
10316 which case the HTTP backend's timeout will be used.
10317
Willy Tarreau95c4e142017-11-26 12:18:55 +010010318 When using HTTP/2 "timeout client" is applied instead. This is so we can keep
10319 using short keep-alive timeouts in HTTP/1.1 while using longer ones in HTTP/2
Lukas Tribus75df9d72017-11-24 19:05:12 +010010320 (where we only have one connection per client and a connection setup).
10321
Willy Tarreaub16a5742010-01-10 14:46:16 +010010322 See also : "timeout http-request", "timeout client".
10323
10324
Willy Tarreau036fae02008-01-06 13:24:40 +010010325timeout http-request <timeout>
10326 Set the maximum allowed time to wait for a complete HTTP request
10327 May be used in sections : defaults | frontend | listen | backend
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010328 yes | yes | yes | yes
Willy Tarreau036fae02008-01-06 13:24:40 +010010329 Arguments :
Willy Tarreau844e3c52008-01-11 16:28:18 +010010330 <timeout> is the timeout value specified in milliseconds by default, but
Willy Tarreau036fae02008-01-06 13:24:40 +010010331 can be in any other unit if the number is suffixed by the unit,
10332 as explained at the top of this document.
10333
10334 In order to offer DoS protection, it may be required to lower the maximum
10335 accepted time to receive a complete HTTP request without affecting the client
10336 timeout. This helps protecting against established connections on which
10337 nothing is sent. The client timeout cannot offer a good protection against
10338 this abuse because it is an inactivity timeout, which means that if the
10339 attacker sends one character every now and then, the timeout will not
10340 trigger. With the HTTP request timeout, no matter what speed the client
Willy Tarreau2705a612014-05-23 17:38:34 +020010341 types, the request will be aborted if it does not complete in time. When the
10342 timeout expires, an HTTP 408 response is sent to the client to inform it
10343 about the problem, and the connection is closed. The logs will report
10344 termination codes "cR". Some recent browsers are having problems with this
Davor Ocelice9ed2812017-12-25 17:49:28 +010010345 standard, well-documented behavior, so it might be needed to hide the 408
Willy Tarreau0f228a02015-05-01 15:37:53 +020010346 code using "option http-ignore-probes" or "errorfile 408 /dev/null". See
10347 more details in the explanations of the "cR" termination code in section 8.5.
Willy Tarreau036fae02008-01-06 13:24:40 +010010348
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010349 By default, this timeout only applies to the header part of the request,
10350 and not to any data. As soon as the empty line is received, this timeout is
10351 not used anymore. When combined with "option http-buffer-request", this
10352 timeout also applies to the body of the request..
10353 It is used again on keep-alive connections to wait for a second
Willy Tarreaub16a5742010-01-10 14:46:16 +010010354 request if "timeout http-keep-alive" is not set.
Willy Tarreau036fae02008-01-06 13:24:40 +010010355
10356 Generally it is enough to set it to a few seconds, as most clients send the
10357 full request immediately upon connection. Add 3 or more seconds to cover TCP
Davor Ocelice9ed2812017-12-25 17:49:28 +010010358 retransmits but that's all. Setting it to very low values (e.g. 50 ms) will
Willy Tarreau036fae02008-01-06 13:24:40 +010010359 generally work on local networks as long as there are no packet losses. This
10360 will prevent people from sending bare HTTP requests using telnet.
10361
10362 If this parameter is not set, the client timeout still applies between each
Willy Tarreaucd7afc02009-07-12 10:03:17 +020010363 chunk of the incoming request. It should be set in the frontend to take
10364 effect, unless the frontend is in TCP mode, in which case the HTTP backend's
10365 timeout will be used.
Willy Tarreau036fae02008-01-06 13:24:40 +010010366
Willy Tarreau0f228a02015-05-01 15:37:53 +020010367 See also : "errorfile", "http-ignore-probes", "timeout http-keep-alive", and
Baptiste Assmanneccdf432015-10-28 13:49:01 +010010368 "timeout client", "option http-buffer-request".
Willy Tarreau036fae02008-01-06 13:24:40 +010010369
Willy Tarreau844e3c52008-01-11 16:28:18 +010010370
10371timeout queue <timeout>
10372 Set the maximum time to wait in the queue for a connection slot to be free
10373 May be used in sections : defaults | frontend | listen | backend
10374 yes | no | yes | yes
10375 Arguments :
10376 <timeout> is the timeout value specified in milliseconds by default, but
10377 can be in any other unit if the number is suffixed by the unit,
10378 as explained at the top of this document.
10379
10380 When a server's maxconn is reached, connections are left pending in a queue
10381 which may be server-specific or global to the backend. In order not to wait
10382 indefinitely, a timeout is applied to requests pending in the queue. If the
10383 timeout is reached, it is considered that the request will almost never be
10384 served, so it is dropped and a 503 error is returned to the client.
10385
10386 The "timeout queue" statement allows to fix the maximum time for a request to
10387 be left pending in a queue. If unspecified, the same value as the backend's
10388 connection timeout ("timeout connect") is used, for backwards compatibility
10389 with older versions with no "timeout queue" parameter.
10390
10391 See also : "timeout connect", "contimeout".
10392
10393
10394timeout server <timeout>
10395timeout srvtimeout <timeout> (deprecated)
10396 Set the maximum inactivity time on the server side.
10397 May be used in sections : defaults | frontend | listen | backend
10398 yes | no | yes | yes
10399 Arguments :
10400 <timeout> is the timeout value specified in milliseconds by default, but
10401 can be in any other unit if the number is suffixed by the unit,
10402 as explained at the top of this document.
10403
10404 The inactivity timeout applies when the server is expected to acknowledge or
10405 send data. In HTTP mode, this timeout is particularly important to consider
10406 during the first phase of the server's response, when it has to send the
10407 headers, as it directly represents the server's processing time for the
10408 request. To find out what value to put there, it's often good to start with
10409 what would be considered as unacceptable response times, then check the logs
10410 to observe the response time distribution, and adjust the value accordingly.
10411
10412 The value is specified in milliseconds by default, but can be in any other
10413 unit if the number is suffixed by the unit, as specified at the top of this
10414 document. In TCP mode (and to a lesser extent, in HTTP mode), it is highly
10415 recommended that the client timeout remains equal to the server timeout in
10416 order to avoid complex situations to debug. Whatever the expected server
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010010417 response times, it is a good practice to cover at least one or several TCP
Willy Tarreau844e3c52008-01-11 16:28:18 +010010418 packet losses by specifying timeouts that are slightly above multiples of 3
Davor Ocelice9ed2812017-12-25 17:49:28 +010010419 seconds (e.g. 4 or 5 seconds minimum). If some long-lived sessions are mixed
10420 with short-lived sessions (e.g. WebSocket and HTTP), it's worth considering
Willy Tarreauce887fd2012-05-12 12:50:00 +020010421 "timeout tunnel", which overrides "timeout client" and "timeout server" for
10422 tunnels.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010423
10424 This parameter is specific to backends, but can be specified once for all in
10425 "defaults" sections. This is in fact one of the easiest solutions not to
10426 forget about it. An unspecified timeout results in an infinite timeout, which
10427 is not recommended. Such a usage is accepted and works but reports a warning
10428 during startup because it may results in accumulation of expired sessions in
10429 the system if the system's timeouts are not configured either.
10430
10431 This parameter replaces the old, deprecated "srvtimeout". It is recommended
10432 to use it to write new configurations. The form "timeout srvtimeout" is
10433 provided only by backwards compatibility but its use is strongly discouraged.
10434
Willy Tarreauce887fd2012-05-12 12:50:00 +020010435 See also : "srvtimeout", "timeout client" and "timeout tunnel".
Willy Tarreau844e3c52008-01-11 16:28:18 +010010436
Willy Tarreau05cdd962014-05-10 14:30:07 +020010437
10438timeout server-fin <timeout>
10439 Set the inactivity timeout on the server side for half-closed connections.
10440 May be used in sections : defaults | frontend | listen | backend
10441 yes | no | yes | yes
10442 Arguments :
10443 <timeout> is the timeout value specified in milliseconds by default, but
10444 can be in any other unit if the number is suffixed by the unit,
10445 as explained at the top of this document.
10446
10447 The inactivity timeout applies when the server is expected to acknowledge or
10448 send data while one direction is already shut down. This timeout is different
10449 from "timeout server" in that it only applies to connections which are closed
10450 in one direction. This is particularly useful to avoid keeping connections in
10451 FIN_WAIT state for too long when a remote server does not disconnect cleanly.
10452 This problem is particularly common long connections such as RDP or WebSocket.
10453 Note that this timeout can override "timeout tunnel" when a connection shuts
10454 down in one direction. This setting was provided for completeness, but in most
10455 situations, it should not be needed.
10456
10457 This parameter is specific to backends, but can be specified once for all in
10458 "defaults" sections. By default it is not set, so half-closed connections
10459 will use the other timeouts (timeout.server or timeout.tunnel).
10460
10461 See also : "timeout client-fin", "timeout server", and "timeout tunnel".
10462
Willy Tarreau844e3c52008-01-11 16:28:18 +010010463
10464timeout tarpit <timeout>
Cyril Bonté78caf842010-03-10 22:41:43 +010010465 Set the duration for which tarpitted connections will be maintained
Willy Tarreau844e3c52008-01-11 16:28:18 +010010466 May be used in sections : defaults | frontend | listen | backend
10467 yes | yes | yes | yes
10468 Arguments :
10469 <timeout> is the tarpit duration specified in milliseconds by default, but
10470 can be in any other unit if the number is suffixed by the unit,
10471 as explained at the top of this document.
10472
Jarno Huuskonene5ae7022017-04-03 14:36:21 +030010473 When a connection is tarpitted using "http-request tarpit" or
10474 "reqtarpit", it is maintained open with no activity for a certain
10475 amount of time, then closed. "timeout tarpit" defines how long it will
10476 be maintained open.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010477
10478 The value is specified in milliseconds by default, but can be in any other
10479 unit if the number is suffixed by the unit, as specified at the top of this
10480 document. If unspecified, the same value as the backend's connection timeout
10481 ("timeout connect") is used, for backwards compatibility with older versions
Cyril Bonté78caf842010-03-10 22:41:43 +010010482 with no "timeout tarpit" parameter.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010483
10484 See also : "timeout connect", "contimeout".
10485
10486
Willy Tarreauce887fd2012-05-12 12:50:00 +020010487timeout tunnel <timeout>
10488 Set the maximum inactivity time on the client and server side for tunnels.
10489 May be used in sections : defaults | frontend | listen | backend
10490 yes | no | yes | yes
10491 Arguments :
10492 <timeout> is the timeout value specified in milliseconds by default, but
10493 can be in any other unit if the number is suffixed by the unit,
10494 as explained at the top of this document.
10495
Jamie Gloudonaaa21002012-08-25 00:18:33 -040010496 The tunnel timeout applies when a bidirectional connection is established
Willy Tarreauce887fd2012-05-12 12:50:00 +020010497 between a client and a server, and the connection remains inactive in both
10498 directions. This timeout supersedes both the client and server timeouts once
10499 the connection becomes a tunnel. In TCP, this timeout is used as soon as no
Davor Ocelice9ed2812017-12-25 17:49:28 +010010500 analyzer remains attached to either connection (e.g. tcp content rules are
10501 accepted). In HTTP, this timeout is used when a connection is upgraded (e.g.
Willy Tarreauce887fd2012-05-12 12:50:00 +020010502 when switching to the WebSocket protocol, or forwarding a CONNECT request
10503 to a proxy), or after the first response when no keepalive/close option is
10504 specified.
10505
Willy Tarreau05cdd962014-05-10 14:30:07 +020010506 Since this timeout is usually used in conjunction with long-lived connections,
10507 it usually is a good idea to also set "timeout client-fin" to handle the
10508 situation where a client suddenly disappears from the net and does not
10509 acknowledge a close, or sends a shutdown and does not acknowledge pending
10510 data anymore. This can happen in lossy networks where firewalls are present,
10511 and is detected by the presence of large amounts of sessions in a FIN_WAIT
10512 state.
10513
Willy Tarreauce887fd2012-05-12 12:50:00 +020010514 The value is specified in milliseconds by default, but can be in any other
10515 unit if the number is suffixed by the unit, as specified at the top of this
10516 document. Whatever the expected normal idle time, it is a good practice to
10517 cover at least one or several TCP packet losses by specifying timeouts that
Davor Ocelice9ed2812017-12-25 17:49:28 +010010518 are slightly above multiples of 3 seconds (e.g. 4 or 5 seconds minimum).
Willy Tarreauce887fd2012-05-12 12:50:00 +020010519
10520 This parameter is specific to backends, but can be specified once for all in
10521 "defaults" sections. This is in fact one of the easiest solutions not to
10522 forget about it.
10523
10524 Example :
10525 defaults http
10526 option http-server-close
10527 timeout connect 5s
10528 timeout client 30s
Willy Tarreau05cdd962014-05-10 14:30:07 +020010529 timeout client-fin 30s
Willy Tarreauce887fd2012-05-12 12:50:00 +020010530 timeout server 30s
10531 timeout tunnel 1h # timeout to use with WebSocket and CONNECT
10532
Willy Tarreau05cdd962014-05-10 14:30:07 +020010533 See also : "timeout client", "timeout client-fin", "timeout server".
Willy Tarreauce887fd2012-05-12 12:50:00 +020010534
10535
Willy Tarreau844e3c52008-01-11 16:28:18 +010010536transparent (deprecated)
10537 Enable client-side transparent proxying
10538 May be used in sections : defaults | frontend | listen | backend
Willy Tarreau4b1f8592008-12-23 23:13:55 +010010539 yes | no | yes | yes
Willy Tarreau844e3c52008-01-11 16:28:18 +010010540 Arguments : none
10541
10542 This keyword was introduced in order to provide layer 7 persistence to layer
10543 3 load balancers. The idea is to use the OS's ability to redirect an incoming
10544 connection for a remote address to a local process (here HAProxy), and let
10545 this process know what address was initially requested. When this option is
10546 used, sessions without cookies will be forwarded to the original destination
10547 IP address of the incoming request (which should match that of another
10548 equipment), while requests with cookies will still be forwarded to the
10549 appropriate server.
10550
10551 The "transparent" keyword is deprecated, use "option transparent" instead.
10552
10553 Note that contrary to a common belief, this option does NOT make HAProxy
10554 present the client's IP to the server when establishing the connection.
10555
Willy Tarreau844e3c52008-01-11 16:28:18 +010010556 See also: "option transparent"
10557
William Lallemanda73203e2012-03-12 12:48:57 +010010558unique-id-format <string>
10559 Generate a unique ID for each request.
10560 May be used in sections : defaults | frontend | listen | backend
10561 yes | yes | yes | no
10562 Arguments :
10563 <string> is a log-format string.
10564
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010565 This keyword creates a ID for each request using the custom log format. A
10566 unique ID is useful to trace a request passing through many components of
10567 a complex infrastructure. The newly created ID may also be logged using the
10568 %ID tag the log-format string.
William Lallemanda73203e2012-03-12 12:48:57 +010010569
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010570 The format should be composed from elements that are guaranteed to be
10571 unique when combined together. For instance, if multiple haproxy instances
10572 are involved, it might be important to include the node name. It is often
10573 needed to log the incoming connection's source and destination addresses
10574 and ports. Note that since multiple requests may be performed over the same
10575 connection, including a request counter may help differentiate them.
10576 Similarly, a timestamp may protect against a rollover of the counter.
10577 Logging the process ID will avoid collisions after a service restart.
William Lallemanda73203e2012-03-12 12:48:57 +010010578
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010579 It is recommended to use hexadecimal notation for many fields since it
10580 makes them more compact and saves space in logs.
William Lallemanda73203e2012-03-12 12:48:57 +010010581
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010582 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010583
Julien Vehentf21be322014-03-07 08:27:34 -050010584 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010585
10586 will generate:
10587
10588 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10589
10590 See also: "unique-id-header"
10591
10592unique-id-header <name>
10593 Add a unique ID header in the HTTP request.
10594 May be used in sections : defaults | frontend | listen | backend
10595 yes | yes | yes | no
10596 Arguments :
10597 <name> is the name of the header.
10598
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010599 Add a unique-id header in the HTTP request sent to the server, using the
10600 unique-id-format. It can't work if the unique-id-format doesn't exist.
William Lallemanda73203e2012-03-12 12:48:57 +010010601
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010602 Example:
William Lallemanda73203e2012-03-12 12:48:57 +010010603
Julien Vehentf21be322014-03-07 08:27:34 -050010604 unique-id-format %{+X}o\ %ci:%cp_%fi:%fp_%Ts_%rt:%pid
William Lallemanda73203e2012-03-12 12:48:57 +010010605 unique-id-header X-Unique-ID
10606
10607 will generate:
10608
10609 X-Unique-ID: 7F000001:8296_7F00001E:1F90_4F7B0A69_0003:790A
10610
10611 See also: "unique-id-format"
Willy Tarreau844e3c52008-01-11 16:28:18 +010010612
Willy Tarreauf51658d2014-04-23 01:21:56 +020010613use_backend <backend> [{if | unless} <condition>]
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010614 Switch to a specific backend if/unless an ACL-based condition is matched.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010615 May be used in sections : defaults | frontend | listen | backend
10616 no | yes | yes | no
10617 Arguments :
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010618 <backend> is the name of a valid backend or "listen" section, or a
10619 "log-format" string resolving to a backend name.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010620
Willy Tarreauf51658d2014-04-23 01:21:56 +020010621 <condition> is a condition composed of ACLs, as described in section 7. If
10622 it is omitted, the rule is unconditionally applied.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010623
10624 When doing content-switching, connections arrive on a frontend and are then
10625 dispatched to various backends depending on a number of conditions. The
10626 relation between the conditions and the backends is described with the
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010627 "use_backend" keyword. While it is normally used with HTTP processing, it can
Davor Ocelice9ed2812017-12-25 17:49:28 +010010628 also be used in pure TCP, either without content using stateless ACLs (e.g.
Willy Tarreau1d0dfb12009-07-07 15:10:31 +020010629 source address validation) or combined with a "tcp-request" rule to wait for
10630 some payload.
Willy Tarreau844e3c52008-01-11 16:28:18 +010010631
10632 There may be as many "use_backend" rules as desired. All of these rules are
10633 evaluated in their declaration order, and the first one which matches will
10634 assign the backend.
10635
10636 In the first form, the backend will be used if the condition is met. In the
10637 second form, the backend will be used if the condition is not met. If no
10638 condition is valid, the backend defined with "default_backend" will be used.
10639 If no default backend is defined, either the servers in the same section are
10640 used (in case of a "listen" section) or, in case of a frontend, no server is
10641 used and a 503 service unavailable response is returned.
10642
Willy Tarreau51aecc72009-07-12 09:47:04 +020010643 Note that it is possible to switch from a TCP frontend to an HTTP backend. In
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010010644 this case, either the frontend has already checked that the protocol is HTTP,
Willy Tarreau51aecc72009-07-12 09:47:04 +020010645 and backend processing will immediately follow, or the backend will wait for
10646 a complete HTTP request to get in. This feature is useful when a frontend
10647 must decode several protocols on a unique port, one of them being HTTP.
10648
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010649 When <backend> is a simple name, it is resolved at configuration time, and an
10650 error is reported if the specified backend does not exist. If <backend> is
10651 a log-format string instead, no check may be done at configuration time, so
10652 the backend name is resolved dynamically at run time. If the resulting
10653 backend name does not correspond to any valid backend, no other rule is
10654 evaluated, and the default_backend directive is applied instead. Note that
10655 when using dynamic backend names, it is highly recommended to use a prefix
10656 that no other backend uses in order to ensure that an unauthorized backend
10657 cannot be forced from the request.
10658
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010659 It is worth mentioning that "use_backend" rules with an explicit name are
Bertrand Jacquin702d44f2013-11-19 11:43:06 +010010660 used to detect the association between frontends and backends to compute the
10661 backend's "fullconn" setting. This cannot be done for dynamic names.
10662
10663 See also: "default_backend", "tcp-request", "fullconn", "log-format", and
10664 section 7 about ACLs.
Willy Tarreaud72758d2010-01-12 10:42:19 +010010665
Willy Tarreau036fae02008-01-06 13:24:40 +010010666
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010667use-server <server> if <condition>
10668use-server <server> unless <condition>
10669 Only use a specific server if/unless an ACL-based condition is matched.
10670 May be used in sections : defaults | frontend | listen | backend
10671 no | no | yes | yes
10672 Arguments :
Cyril Bonté108cf6e2012-04-21 23:30:29 +020010673 <server> is the name of a valid server in the same backend section.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010674
10675 <condition> is a condition composed of ACLs, as described in section 7.
10676
10677 By default, connections which arrive to a backend are load-balanced across
10678 the available servers according to the configured algorithm, unless a
10679 persistence mechanism such as a cookie is used and found in the request.
10680
10681 Sometimes it is desirable to forward a particular request to a specific
10682 server without having to declare a dedicated backend for this server. This
10683 can be achieved using the "use-server" rules. These rules are evaluated after
10684 the "redirect" rules and before evaluating cookies, and they have precedence
10685 on them. There may be as many "use-server" rules as desired. All of these
10686 rules are evaluated in their declaration order, and the first one which
10687 matches will assign the server.
10688
10689 If a rule designates a server which is down, and "option persist" is not used
10690 and no force-persist rule was validated, it is ignored and evaluation goes on
10691 with the next rules until one matches.
10692
10693 In the first form, the server will be used if the condition is met. In the
10694 second form, the server will be used if the condition is not met. If no
10695 condition is valid, the processing continues and the server will be assigned
10696 according to other persistence mechanisms.
10697
10698 Note that even if a rule is matched, cookie processing is still performed but
10699 does not assign the server. This allows prefixed cookies to have their prefix
10700 stripped.
10701
10702 The "use-server" statement works both in HTTP and TCP mode. This makes it
10703 suitable for use with content-based inspection. For instance, a server could
10704 be selected in a farm according to the TLS SNI field. And if these servers
10705 have their weight set to zero, they will not be used for other traffic.
10706
10707 Example :
10708 # intercept incoming TLS requests based on the SNI field
10709 use-server www if { req_ssl_sni -i www.example.com }
10710 server www 192.168.0.1:443 weight 0
10711 use-server mail if { req_ssl_sni -i mail.example.com }
10712 server mail 192.168.0.1:587 weight 0
10713 use-server imap if { req_ssl_sni -i imap.example.com }
Lukas Tribus98a3e3f2017-03-26 12:55:35 +000010714 server imap 192.168.0.1:993 weight 0
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010715 # all the rest is forwarded to this server
10716 server default 192.168.0.2:443 check
10717
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010718 See also: "use_backend", section 5 about server and section 7 about ACLs.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020010719
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010720
Davor Ocelice9ed2812017-12-25 17:49:28 +0100107215. Bind and server options
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010722--------------------------
10723
10724The "bind", "server" and "default-server" keywords support a number of settings
10725depending on some build options and on the system HAProxy was built on. These
10726settings generally each consist in one word sometimes followed by a value,
10727written on the same line as the "bind" or "server" line. All these options are
10728described in this section.
10729
10730
107315.1. Bind options
10732-----------------
10733
10734The "bind" keyword supports a certain number of settings which are all passed
10735as arguments on the same line. The order in which those arguments appear makes
10736no importance, provided that they appear after the bind address. All of these
10737parameters are optional. Some of them consist in a single words (booleans),
10738while other ones expect a value after them. In this case, the value must be
10739provided immediately after the setting name.
10740
10741The currently supported settings are the following ones.
10742
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010743accept-netscaler-cip <magic number>
10744 Enforces the use of the NetScaler Client IP insertion protocol over any
10745 connection accepted by any of the TCP sockets declared on the same line. The
10746 NetScaler Client IP insertion protocol dictates the layer 3/4 addresses of
10747 the incoming connection to be used everywhere an address is used, with the
10748 only exception of "tcp-request connection" rules which will only see the
10749 real connection address. Logs will reflect the addresses indicated in the
10750 protocol, unless it is violated, in which case the real address will still
10751 be used. This keyword combined with support from external components can be
10752 used as an efficient and reliable alternative to the X-Forwarded-For
Bertrand Jacquin90759682016-06-06 15:35:39 +010010753 mechanism which is not always reliable and not even always usable. See also
10754 "tcp-request connection expect-netscaler-cip" for a finer-grained setting of
10755 which client is allowed to use the protocol.
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010010756
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010757accept-proxy
10758 Enforces the use of the PROXY protocol over any connection accepted by any of
Willy Tarreau77992672014-06-14 11:06:17 +020010759 the sockets declared on the same line. Versions 1 and 2 of the PROXY protocol
10760 are supported and correctly detected. The PROXY protocol dictates the layer
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010761 3/4 addresses of the incoming connection to be used everywhere an address is
10762 used, with the only exception of "tcp-request connection" rules which will
10763 only see the real connection address. Logs will reflect the addresses
10764 indicated in the protocol, unless it is violated, in which case the real
Davor Ocelice9ed2812017-12-25 17:49:28 +010010765 address will still be used. This keyword combined with support from external
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010766 components can be used as an efficient and reliable alternative to the
10767 X-Forwarded-For mechanism which is not always reliable and not even always
Willy Tarreau4f0d9192013-06-11 20:40:55 +020010768 usable. See also "tcp-request connection expect-proxy" for a finer-grained
10769 setting of which client is allowed to use the protocol.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010770
Olivier Houchardc2aae742017-09-22 18:26:28 +020010771allow-0rtt
Bertrand Jacquina25282b2018-08-14 00:56:13 +010010772 Allow receiving early data when using TLSv1.3. This is disabled by default,
Olivier Houchard69752962019-01-08 15:35:32 +010010773 due to security considerations. Because it is vulnerable to replay attacks,
10774 you should only allow if for requests that are safe to replay, ie requests
10775 that are idempotent. You can use the "wait-for-handshake" action for any
10776 request that wouldn't be safe with early data.
Olivier Houchardc2aae742017-09-22 18:26:28 +020010777
Willy Tarreauab861d32013-04-02 02:30:41 +020010778alpn <protocols>
10779 This enables the TLS ALPN extension and advertises the specified protocol
10780 list as supported on top of ALPN. The protocol list consists in a comma-
10781 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
10782 quotes). This requires that the SSL library is build with support for TLS
10783 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
Willy Tarreau95c4e142017-11-26 12:18:55 +010010784 initial NPN extension. ALPN is required to enable HTTP/2 on an HTTP frontend.
10785 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
10786 now obsolete NPN extension. At the time of writing this, most browsers still
10787 support both ALPN and NPN for HTTP/2 so a fallback to NPN may still work for
10788 a while. But ALPN must be used whenever possible. If both HTTP/2 and HTTP/1.1
10789 are expected to be supported, both versions can be advertised, in order of
10790 preference, like below :
10791
10792 bind :443 ssl crt pub.pem alpn h2,http/1.1
Willy Tarreauab861d32013-04-02 02:30:41 +020010793
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010794backlog <backlog>
Willy Tarreaue2711c72019-02-27 15:39:41 +010010795 Sets the socket's backlog to this value. If unspecified or 0, the frontend's
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010796 backlog is used instead, which generally defaults to the maxconn value.
10797
Emmanuel Hocdete7f2b732017-01-09 16:15:54 +010010798curves <curves>
10799 This setting is only available when support for OpenSSL was built in. It sets
10800 the string describing the list of elliptic curves algorithms ("curve suite")
10801 that are negotiated during the SSL/TLS handshake with ECDHE. The format of the
10802 string is a colon-delimited list of curve name.
10803 Example: "X25519:P-256" (without quote)
10804 When "curves" is set, "ecdhe" parameter is ignored.
10805
Emeric Brun7fb34422012-09-28 15:26:15 +020010806ecdhe <named curve>
10807 This setting is only available when support for OpenSSL was built in. It sets
Emeric Brun6924ef82013-03-06 14:08:53 +010010808 the named curve (RFC 4492) used to generate ECDH ephemeral keys. By default,
10809 used named curve is prime256v1.
Emeric Brun7fb34422012-09-28 15:26:15 +020010810
Emeric Brunfd33a262012-10-11 16:28:27 +020010811ca-file <cafile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010812 This setting is only available when support for OpenSSL was built in. It
10813 designates a PEM file from which to load CA certificates used to verify
10814 client's certificate.
10815
Emeric Brunb6dc9342012-09-28 17:55:37 +020010816ca-ignore-err [all|<errorID>,...]
10817 This setting is only available when support for OpenSSL was built in.
10818 Sets a comma separated list of errorIDs to ignore during verify at depth > 0.
10819 If set to 'all', all errors are ignored. SSL handshake is not aborted if an
10820 error is ignored.
10821
Christopher Faulet31af49d2015-06-09 17:29:50 +020010822ca-sign-file <cafile>
10823 This setting is only available when support for OpenSSL was built in. It
10824 designates a PEM file containing both the CA certificate and the CA private
10825 key used to create and sign server's certificates. This is a mandatory
10826 setting when the dynamic generation of certificates is enabled. See
10827 'generate-certificates' for details.
10828
Bertrand Jacquind4d0a232016-11-13 16:37:12 +000010829ca-sign-pass <passphrase>
Christopher Faulet31af49d2015-06-09 17:29:50 +020010830 This setting is only available when support for OpenSSL was built in. It is
10831 the CA private key passphrase. This setting is optional and used only when
10832 the dynamic generation of certificates is enabled. See
10833 'generate-certificates' for details.
10834
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010835ciphers <ciphers>
10836 This setting is only available when support for OpenSSL was built in. It sets
10837 the string describing the list of cipher algorithms ("cipher suite") that are
Bertrand Jacquin8cf7c1e2019-02-03 18:35:25 +000010838 negotiated during the SSL/TLS handshake up to TLSv1.2. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000010839 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
Dirkjan Bussink415150f2018-09-14 11:14:21 +020010840 information and recommendations see e.g.
10841 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
10842 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
10843 cipher configuration, please check the "ciphersuites" keyword.
10844
10845ciphersuites <ciphersuites>
10846 This setting is only available when support for OpenSSL was built in and
10847 OpenSSL 1.1.1 or later was used to build HAProxy. It sets the string describing
10848 the list of cipher algorithms ("cipher suite") that are negotiated during the
10849 TLSv1.3 handshake. The format of the string is defined in "man 1 ciphers" from
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000010850 OpenSSL man pages under the "ciphersuites" section. For cipher configuration
10851 for TLSv1.2 and earlier, please check the "ciphers" keyword.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010852
Emeric Brunfd33a262012-10-11 16:28:27 +020010853crl-file <crlfile>
Emeric Brun1a073b42012-09-28 17:07:34 +020010854 This setting is only available when support for OpenSSL was built in. It
10855 designates a PEM file from which to load certificate revocation list used
10856 to verify client's certificate.
10857
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010858crt <cert>
Alex Davies0fbf0162013-03-02 16:04:50 +000010859 This setting is only available when support for OpenSSL was built in. It
10860 designates a PEM file containing both the required certificates and any
10861 associated private keys. This file can be built by concatenating multiple
10862 PEM files into one (e.g. cat cert.pem key.pem > combined.pem). If your CA
10863 requires an intermediate certificate, this can also be concatenated into this
10864 file.
10865
10866 If the OpenSSL used supports Diffie-Hellman, parameters present in this file
10867 are loaded.
10868
10869 If a directory name is used instead of a PEM file, then all files found in
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010870 that directory will be loaded in alphabetic order unless their name ends with
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010871 '.issuer', '.ocsp' or '.sctl' (reserved extensions). This directive may be
10872 specified multiple times in order to load certificates from multiple files or
10873 directories. The certificates will be presented to clients who provide a
10874 valid TLS Server Name Indication field matching one of their CN or alt
Davor Ocelice9ed2812017-12-25 17:49:28 +010010875 subjects. Wildcards are supported, where a wildcard character '*' is used
10876 instead of the first hostname component (e.g. *.example.org matches
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010877 www.example.org but not www.sub.example.org).
Alex Davies0fbf0162013-03-02 16:04:50 +000010878
10879 If no SNI is provided by the client or if the SSL library does not support
10880 TLS extensions, or if the client provides an SNI hostname which does not
10881 match any certificate, then the first loaded certificate will be presented.
10882 This means that when loading certificates from a directory, it is highly
Cyril Bonté3180f7b2015-01-25 00:16:08 +010010883 recommended to load the default one first as a file or to ensure that it will
10884 always be the first one in the directory.
Alex Davies0fbf0162013-03-02 16:04:50 +000010885
Emeric Brune032bfa2012-09-28 13:01:45 +020010886 Note that the same cert may be loaded multiple times without side effects.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020010887
Davor Ocelice9ed2812017-12-25 17:49:28 +010010888 Some CAs (such as GoDaddy) offer a drop down list of server types that do not
Alex Davies0fbf0162013-03-02 16:04:50 +000010889 include HAProxy when obtaining a certificate. If this happens be sure to
Davor Ocelice9ed2812017-12-25 17:49:28 +010010890 choose a web server that the CA believes requires an intermediate CA (for
10891 GoDaddy, selection Apache Tomcat will get the correct bundle, but many
Alex Davies0fbf0162013-03-02 16:04:50 +000010892 others, e.g. nginx, result in a wrong bundle that will not work for some
10893 clients).
10894
Emeric Brun4147b2e2014-06-16 18:36:30 +020010895 For each PEM file, haproxy checks for the presence of file at the same path
10896 suffixed by ".ocsp". If such file is found, support for the TLS Certificate
10897 Status Request extension (also known as "OCSP stapling") is automatically
10898 enabled. The content of this file is optional. If not empty, it must contain
10899 a valid OCSP Response in DER format. In order to be valid an OCSP Response
10900 must comply with the following rules: it has to indicate a good status,
10901 it has to be a single response for the certificate of the PEM file, and it
10902 has to be valid at the moment of addition. If these rules are not respected
10903 the OCSP Response is ignored and a warning is emitted. In order to identify
10904 which certificate an OCSP Response applies to, the issuer's certificate is
10905 necessary. If the issuer's certificate is not found in the PEM file, it will
10906 be loaded from a file at the same path as the PEM file suffixed by ".issuer"
10907 if it exists otherwise it will fail with an error.
10908
Janusz Dziemidowicz2c701b52015-03-07 23:03:59 +010010909 For each PEM file, haproxy also checks for the presence of file at the same
10910 path suffixed by ".sctl". If such file is found, support for Certificate
10911 Transparency (RFC6962) TLS extension is enabled. The file must contain a
10912 valid Signed Certificate Timestamp List, as described in RFC. File is parsed
10913 to check basic syntax, but no signatures are verified.
10914
yanbzhu6c25e9e2016-01-05 12:52:02 -050010915 There are cases where it is desirable to support multiple key types, e.g. RSA
10916 and ECDSA in the cipher suites offered to the clients. This allows clients
10917 that support EC certificates to be able to use EC ciphers, while
10918 simultaneously supporting older, RSA only clients.
yanbzhud19630c2015-12-14 15:10:25 -050010919
10920 In order to provide this functionality, multiple PEM files, each with a
10921 different key type, are required. To associate these PEM files into a
10922 "cert bundle" that is recognized by haproxy, they must be named in the
10923 following way: All PEM files that are to be bundled must have the same base
10924 name, with a suffix indicating the key type. Currently, three suffixes are
10925 supported: rsa, dsa and ecdsa. For example, if www.example.com has two PEM
10926 files, an RSA file and an ECDSA file, they must be named: "example.pem.rsa"
10927 and "example.pem.ecdsa". The first part of the filename is arbitrary; only the
10928 suffix matters. To load this bundle into haproxy, specify the base name only:
10929
10930 Example : bind :8443 ssl crt example.pem
10931
yanbzhu6c25e9e2016-01-05 12:52:02 -050010932 Note that the suffix is not given to haproxy; this tells haproxy to look for
yanbzhud19630c2015-12-14 15:10:25 -050010933 a cert bundle.
10934
Davor Ocelice9ed2812017-12-25 17:49:28 +010010935 HAProxy will load all PEM files in the bundle at the same time to try to
yanbzhud19630c2015-12-14 15:10:25 -050010936 support multiple key types. PEM files are combined based on Common Name
10937 (CN) and Subject Alternative Name (SAN) to support SNI lookups. This means
10938 that even if you give haproxy a cert bundle, if there are no shared CN/SAN
10939 entries in the certificates in that bundle, haproxy will not be able to
10940 provide multi-cert support.
10941
10942 Assuming bundle in the example above contained the following:
10943
10944 Filename | CN | SAN
10945 -------------------+-----------------+-------------------
10946 example.pem.rsa | www.example.com | rsa.example.com
yanbzhu6c25e9e2016-01-05 12:52:02 -050010947 -------------------+-----------------+-------------------
yanbzhud19630c2015-12-14 15:10:25 -050010948 example.pem.ecdsa | www.example.com | ecdsa.example.com
10949 -------------------+-----------------+-------------------
10950
10951 Users connecting with an SNI of "www.example.com" will be able
10952 to use both RSA and ECDSA cipher suites. Users connecting with an SNI of
10953 "rsa.example.com" will only be able to use RSA cipher suites, and users
10954 connecting with "ecdsa.example.com" will only be able to use ECDSA cipher
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010955 suites. With BoringSSL and Openssl >= 1.1.1 multi-cert is natively supported,
10956 no need to bundle certificates. ECDSA certificate will be preferred if client
10957 support it.
yanbzhud19630c2015-12-14 15:10:25 -050010958
10959 If a directory name is given as the <cert> argument, haproxy will
10960 automatically search and load bundled files in that directory.
10961
10962 OSCP files (.ocsp) and issuer files (.issuer) are supported with multi-cert
10963 bundling. Each certificate can have its own .ocsp and .issuer file. At this
10964 time, sctl is not supported in multi-certificate bundling.
10965
Emeric Brunb6dc9342012-09-28 17:55:37 +020010966crt-ignore-err <errors>
Alex Davies0fbf0162013-03-02 16:04:50 +000010967 This setting is only available when support for OpenSSL was built in. Sets a
Davor Ocelice9ed2812017-12-25 17:49:28 +010010968 comma separated list of errorIDs to ignore during verify at depth == 0. If
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030010969 set to 'all', all errors are ignored. SSL handshake is not aborted if an error
Alex Davies0fbf0162013-03-02 16:04:50 +000010970 is ignored.
Emeric Brunb6dc9342012-09-28 17:55:37 +020010971
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010972crt-list <file>
10973 This setting is only available when support for OpenSSL was built in. It
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010974 designates a list of PEM file with an optional ssl configuration and a SNI
10975 filter per certificate, with the following format for each line :
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010976
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010977 <crtfile> [\[<sslbindconf> ...\]] [[!]<snifilter> ...]
10978
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020010979 sslbindconf support "npn", "alpn", "verify", "ca-file", "no-ca-names",
10980 crl-file", "ecdhe", "curves", "ciphers" configuration. With BoringSSL
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010981 and Openssl >= 1.1.1 "ssl-min-ver" and "ssl-max-ver" are also supported.
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010982 It override the configuration set in bind line for the certificate.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010983
Emmanuel Hocdet7c41a1b2013-05-07 20:20:06 +020010984 Wildcards are supported in the SNI filter. Negative filter are also supported,
10985 only useful in combination with a wildcard filter to exclude a particular SNI.
10986 The certificates will be presented to clients who provide a valid TLS Server
10987 Name Indication field matching one of the SNI filters. If no SNI filter is
10988 specified, the CN and alt subjects are used. This directive may be specified
10989 multiple times. See the "crt" option for more information. The default
10990 certificate is still needed to meet OpenSSL expectations. If it is not used,
10991 the 'strict-sni' option may be used.
Emmanuel Hocdetfe616562013-01-22 15:31:15 +010010992
yanbzhu6c25e9e2016-01-05 12:52:02 -050010993 Multi-cert bundling (see "crt") is supported with crt-list, as long as only
Emmanuel Hocdetd294aea2016-05-13 11:14:06 +020010994 the base name is given in the crt-list. SNI filter will do the same work on
Emmanuel Hocdet84e417d2017-08-16 11:33:17 +020010995 all bundled certificates. With BoringSSL and Openssl >= 1.1.1 multi-cert is
10996 natively supported, avoid multi-cert bundling. RSA and ECDSA certificates can
10997 be declared in a row, and set different ssl and filter parameter.
yanbzhud19630c2015-12-14 15:10:25 -050010998
Emmanuel Hocdet98263292016-12-29 18:26:15 +010010999 crt-list file example:
11000 cert1.pem
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011001 cert2.pem [alpn h2,http/1.1]
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011002 certW.pem *.domain.tld !secure.domain.tld
Emmanuel Hocdet05942112017-02-20 16:11:50 +010011003 certS.pem [curves X25519:P-256 ciphers ECDHE-ECDSA-AES256-GCM-SHA384] secure.domain.tld
Emmanuel Hocdet98263292016-12-29 18:26:15 +010011004
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011005defer-accept
11006 Is an optional keyword which is supported only on certain Linux kernels. It
11007 states that a connection will only be accepted once some data arrive on it,
11008 or at worst after the first retransmit. This should be used only on protocols
Davor Ocelice9ed2812017-12-25 17:49:28 +010011009 for which the client talks first (e.g. HTTP). It can slightly improve
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011010 performance by ensuring that most of the request is already available when
11011 the connection is accepted. On the other hand, it will not be able to detect
11012 connections which don't talk. It is important to note that this option is
11013 broken in all kernels up to 2.6.31, as the connection is never accepted until
11014 the client talks. This can cause issues with front firewalls which would see
11015 an established connection while the proxy will only see it in SYN_RECV. This
11016 option is only supported on TCPv4/TCPv6 sockets and ignored by other ones.
11017
William Lallemandf6975e92017-05-26 17:42:10 +020011018expose-fd listeners
11019 This option is only usable with the stats socket. It gives your stats socket
11020 the capability to pass listeners FD to another HAProxy process.
William Lallemande202b1e2017-06-01 17:38:56 +020011021 During a reload with the master-worker mode, the process is automatically
11022 reexecuted adding -x and one of the stats socket with this option.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011023 See also "-x" in the management guide.
William Lallemandf6975e92017-05-26 17:42:10 +020011024
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011025force-sslv3
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011026 This option enforces use of SSLv3 only on SSL connections instantiated from
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011027 this listener. SSLv3 is generally less expensive than the TLS counterparts
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011028 for high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011029 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011030
11031force-tlsv10
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011032 This option enforces use of TLSv1.0 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011033 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011034 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011035
11036force-tlsv11
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011037 This option enforces use of TLSv1.1 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011038 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011039 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011040
11041force-tlsv12
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011042 This option enforces use of TLSv1.2 only on SSL connections instantiated from
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011043 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011044 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011045
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011046force-tlsv13
11047 This option enforces use of TLSv1.3 only on SSL connections instantiated from
11048 this listener. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011049 "ssl-default-bind-options". See also "ssl-min-ver" and "ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011050
Christopher Faulet31af49d2015-06-09 17:29:50 +020011051generate-certificates
11052 This setting is only available when support for OpenSSL was built in. It
11053 enables the dynamic SSL certificates generation. A CA certificate and its
11054 private key are necessary (see 'ca-sign-file'). When HAProxy is configured as
11055 a transparent forward proxy, SSL requests generate errors because of a common
11056 name mismatch on the certificate presented to the client. With this option
11057 enabled, HAProxy will try to forge a certificate using the SNI hostname
11058 indicated by the client. This is done only if no certificate matches the SNI
11059 hostname (see 'crt-list'). If an error occurs, the default certificate is
11060 used, else the 'strict-sni' option is set.
11061 It can also be used when HAProxy is configured as a reverse proxy to ease the
11062 deployment of an architecture with many backends.
11063
11064 Creating a SSL certificate is an expensive operation, so a LRU cache is used
11065 to store forged certificates (see 'tune.ssl.ssl-ctx-cache-size'). It
Davor Ocelice9ed2812017-12-25 17:49:28 +010011066 increases the HAProxy's memory footprint to reduce latency when the same
Christopher Faulet31af49d2015-06-09 17:29:50 +020011067 certificate is used many times.
11068
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011069gid <gid>
11070 Sets the group of the UNIX sockets to the designated system gid. It can also
11071 be set by default in the global section's "unix-bind" statement. Note that
11072 some platforms simply ignore this. This setting is equivalent to the "group"
11073 setting except that the group ID is used instead of its name. This setting is
11074 ignored by non UNIX sockets.
11075
11076group <group>
11077 Sets the group of the UNIX sockets to the designated system group. It can
11078 also be set by default in the global section's "unix-bind" statement. Note
11079 that some platforms simply ignore this. This setting is equivalent to the
11080 "gid" setting except that the group name is used instead of its gid. This
11081 setting is ignored by non UNIX sockets.
11082
11083id <id>
11084 Fixes the socket ID. By default, socket IDs are automatically assigned, but
11085 sometimes it is more convenient to fix them to ease monitoring. This value
11086 must be strictly positive and unique within the listener/frontend. This
11087 option can only be used when defining only a single socket.
11088
11089interface <interface>
Lukas Tribusfce2e962013-02-12 22:13:19 +010011090 Restricts the socket to a specific interface. When specified, only packets
11091 received from that particular interface are processed by the socket. This is
11092 currently only supported on Linux. The interface must be a primary system
11093 interface, not an aliased interface. It is also possible to bind multiple
11094 frontends to the same address if they are bound to different interfaces. Note
11095 that binding to a network interface requires root privileges. This parameter
Jérôme Magnin61275192018-02-07 11:39:58 +010011096 is only compatible with TCPv4/TCPv6 sockets. When specified, return traffic
11097 uses the same interface as inbound traffic, and its associated routing table,
11098 even if there are explicit routes through different interfaces configured.
11099 This can prove useful to address asymmetric routing issues when the same
11100 client IP addresses need to be able to reach frontends hosted on different
11101 interfaces.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011102
Willy Tarreauabb175f2012-09-24 12:43:26 +020011103level <level>
11104 This setting is used with the stats sockets only to restrict the nature of
11105 the commands that can be issued on the socket. It is ignored by other
11106 sockets. <level> can be one of :
Davor Ocelice9ed2812017-12-25 17:49:28 +010011107 - "user" is the least privileged level; only non-sensitive stats can be
Willy Tarreauabb175f2012-09-24 12:43:26 +020011108 read, and no change is allowed. It would make sense on systems where it
11109 is not easy to restrict access to the socket.
11110 - "operator" is the default level and fits most common uses. All data can
Davor Ocelice9ed2812017-12-25 17:49:28 +010011111 be read, and only non-sensitive changes are permitted (e.g. clear max
Willy Tarreauabb175f2012-09-24 12:43:26 +020011112 counters).
Davor Ocelice9ed2812017-12-25 17:49:28 +010011113 - "admin" should be used with care, as everything is permitted (e.g. clear
Willy Tarreauabb175f2012-09-24 12:43:26 +020011114 all counters).
11115
Andjelko Iharosc4df59e2017-07-20 11:59:48 +020011116severity-output <format>
11117 This setting is used with the stats sockets only to configure severity
11118 level output prepended to informational feedback messages. Severity
11119 level of messages can range between 0 and 7, conforming to syslog
11120 rfc5424. Valid and successful socket commands requesting data
11121 (i.e. "show map", "get acl foo" etc.) will never have a severity level
11122 prepended. It is ignored by other sockets. <format> can be one of :
11123 - "none" (default) no severity level is prepended to feedback messages.
11124 - "number" severity level is prepended as a number.
11125 - "string" severity level is prepended as a string following the
11126 rfc5424 convention.
11127
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011128maxconn <maxconn>
11129 Limits the sockets to this number of concurrent connections. Extraneous
11130 connections will remain in the system's backlog until a connection is
11131 released. If unspecified, the limit will be the same as the frontend's
11132 maxconn. Note that in case of port ranges or multiple addresses, the same
11133 value will be applied to each socket. This setting enables different
11134 limitations on expensive sockets, for instance SSL entries which may easily
11135 eat all memory.
11136
11137mode <mode>
11138 Sets the octal mode used to define access permissions on the UNIX socket. It
11139 can also be set by default in the global section's "unix-bind" statement.
11140 Note that some platforms simply ignore this. This setting is ignored by non
11141 UNIX sockets.
11142
11143mss <maxseg>
11144 Sets the TCP Maximum Segment Size (MSS) value to be advertised on incoming
11145 connections. This can be used to force a lower MSS for certain specific
11146 ports, for instance for connections passing through a VPN. Note that this
11147 relies on a kernel feature which is theoretically supported under Linux but
11148 was buggy in all versions prior to 2.6.28. It may or may not work on other
11149 operating systems. It may also not change the advertised value but change the
11150 effective size of outgoing segments. The commonly advertised value for TCPv4
11151 over Ethernet networks is 1460 = 1500(MTU) - 40(IP+TCP). If this value is
11152 positive, it will be used as the advertised MSS. If it is negative, it will
11153 indicate by how much to reduce the incoming connection's advertised MSS for
11154 outgoing segments. This parameter is only compatible with TCP v4/v6 sockets.
11155
11156name <name>
11157 Sets an optional name for these sockets, which will be reported on the stats
11158 page.
11159
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011160namespace <name>
11161 On Linux, it is possible to specify which network namespace a socket will
11162 belong to. This directive makes it possible to explicitly bind a listener to
11163 a namespace different from the default one. Please refer to your operating
11164 system's documentation to find more details about network namespaces.
11165
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011166nice <nice>
11167 Sets the 'niceness' of connections initiated from the socket. Value must be
11168 in the range -1024..1024 inclusive, and defaults to zero. Positive values
11169 means that such connections are more friendly to others and easily offer
11170 their place in the scheduler. On the opposite, negative values mean that
11171 connections want to run with a higher priority than others. The difference
11172 only happens under high loads when the system is close to saturation.
11173 Negative values are appropriate for low-latency or administration services,
11174 and high values are generally recommended for CPU intensive tasks such as SSL
11175 processing or bulk transfers which are less sensible to latency. For example,
11176 it may make sense to use a positive value for an SMTP socket and a negative
11177 one for an RDP socket.
11178
Emmanuel Hocdet174dfe52017-07-28 15:01:05 +020011179no-ca-names
11180 This setting is only available when support for OpenSSL was built in. It
11181 prevents from send CA names in server hello message when ca-file is used.
11182
Emeric Brun9b3009b2012-10-05 11:55:06 +020011183no-sslv3
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011184 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011185 disables support for SSLv3 on any sockets instantiated from the listener when
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011186 SSL is supported. Note that SSLv2 is forced disabled in the code and cannot
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011187 be enabled using any configuration option. This option is also available on
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011188 global statement "ssl-default-bind-options". Use "ssl-min-ver" and
11189 "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011190
Emeric Brun90ad8722012-10-02 14:00:59 +020011191no-tls-tickets
11192 This setting is only available when support for OpenSSL was built in. It
11193 disables the stateless session resumption (RFC 5077 TLS Ticket
11194 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011195 session resumption is more expensive in CPU usage. This option is also
11196 available on global statement "ssl-default-bind-options".
Emeric Brun90ad8722012-10-02 14:00:59 +020011197
Emeric Brun9b3009b2012-10-05 11:55:06 +020011198no-tlsv10
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011199 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011200 disables support for TLSv1.0 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011201 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011202 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011203 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11204 and "ssl-max-ver" instead.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011205
Emeric Brun9b3009b2012-10-05 11:55:06 +020011206no-tlsv11
Emeric Brunf5da4932012-09-28 19:42:54 +020011207 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011208 disables support for TLSv1.1 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011209 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011210 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011211 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11212 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011213
Emeric Brun9b3009b2012-10-05 11:55:06 +020011214no-tlsv12
Emeric Brunf5da4932012-09-28 19:42:54 +020011215 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011216 disables support for TLSv1.2 on any sockets instantiated from the listener
Emeric Brun2cb7ae52012-10-05 14:14:21 +020011217 when SSL is supported. Note that SSLv2 is forced disabled in the code and
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011218 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011219 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11220 and "ssl-max-ver" instead.
Emeric Brunf5da4932012-09-28 19:42:54 +020011221
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011222no-tlsv13
11223 This setting is only available when support for OpenSSL was built in. It
11224 disables support for TLSv1.3 on any sockets instantiated from the listener
11225 when SSL is supported. Note that SSLv2 is forced disabled in the code and
11226 cannot be enabled using any configuration option. This option is also
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011227 available on global statement "ssl-default-bind-options". Use "ssl-min-ver"
11228 and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011229
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011230npn <protocols>
11231 This enables the NPN TLS extension and advertises the specified protocol list
11232 as supported on top of NPN. The protocol list consists in a comma-delimited
11233 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
11234 This requires that the SSL library is build with support for TLS extensions
Willy Tarreauab861d32013-04-02 02:30:41 +020011235 enabled (check with haproxy -vv). Note that the NPN extension has been
Willy Tarreau95c4e142017-11-26 12:18:55 +010011236 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11237 only available starting with OpenSSL 1.0.2. If HTTP/2 is desired on an older
11238 version of OpenSSL, NPN might still be used as most clients still support it
11239 at the time of writing this. It is possible to enable both NPN and ALPN
11240 though it probably doesn't make any sense out of testing.
Willy Tarreau6c9a3d52012-10-18 18:57:14 +020011241
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011242prefer-client-ciphers
11243 Use the client's preference when selecting the cipher suite, by default
11244 the server's preference is enforced. This option is also available on
11245 global statement "ssl-default-bind-options".
Lukas Tribus926594f2018-05-18 17:55:57 +020011246 Note that with OpenSSL >= 1.1.1 ChaCha20-Poly1305 is reprioritized anyway
11247 (without setting this option), if a ChaCha20-Poly1305 cipher is at the top of
11248 the client cipher list.
Lukas Tribus53ae85c2017-05-04 15:45:40 +000011249
Christopher Fauletc644fa92017-11-23 22:44:11 +010011250process <process-set>[/<thread-set>]
Willy Tarreaua36b3242019-02-02 13:14:34 +010011251 This restricts the list of processes or threads on which this listener is
Christopher Fauletc644fa92017-11-23 22:44:11 +010011252 allowed to run. It does not enforce any process but eliminates those which do
Davor Ocelice9ed2812017-12-25 17:49:28 +010011253 not match. If the frontend uses a "bind-process" setting, the intersection
Christopher Fauletc644fa92017-11-23 22:44:11 +010011254 between the two is applied. If in the end the listener is not allowed to run
11255 on any remaining process, a warning is emitted, and the listener will either
11256 run on the first process of the listener if a single process was specified,
11257 or on all of its processes if multiple processes were specified. If a thread
Davor Ocelice9ed2812017-12-25 17:49:28 +010011258 set is specified, it limits the threads allowed to process incoming
Willy Tarreaua36b3242019-02-02 13:14:34 +010011259 connections for this listener, for the the process set. If multiple processes
11260 and threads are configured, a warning is emitted, as it either results from a
11261 configuration error or a misunderstanding of these models. For the unlikely
11262 case where several ranges are needed, this directive may be repeated.
11263 <process-set> and <thread-set> must use the format
Christopher Fauletc644fa92017-11-23 22:44:11 +010011264
11265 all | odd | even | number[-[number]]
11266
11267 Ranges can be partially defined. The higher bound can be omitted. In such
11268 case, it is replaced by the corresponding maximum value. The main purpose of
11269 this directive is to be used with the stats sockets and have one different
11270 socket per process. The second purpose is to have multiple bind lines sharing
11271 the same IP:port but not the same process in a listener, so that the system
11272 can distribute the incoming connections into multiple queues and allow a
11273 smoother inter-process load balancing. Currently Linux 3.9 and above is known
11274 for supporting this. See also "bind-process" and "nbproc".
Willy Tarreau6ae1ba62014-05-07 19:01:58 +020011275
Christopher Fauleta717b992018-04-10 14:43:00 +020011276proto <name>
11277 Forces the multiplexer's protocol to use for the incoming connections. It
11278 must be compatible with the mode of the frontend (TCP or HTTP). It must also
11279 be usable on the frontend side. The list of available protocols is reported
11280 in haproxy -vv.
11281 Idea behind this optipon is to bypass the selection of the best multiplexer's
11282 protocol for all connections instantiated from this listening socket. For
Joseph Herlant71b4b152018-11-13 16:55:16 -080011283 instance, it is possible to force the http/2 on clear TCP by specifying "proto
Christopher Fauleta717b992018-04-10 14:43:00 +020011284 h2" on the bind line.
11285
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011286ssl
11287 This setting is only available when support for OpenSSL was built in. It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011288 enables SSL deciphering on connections instantiated from this listener. A
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011289 certificate is necessary (see "crt" above). All contents in the buffers will
11290 appear in clear text, so that ACLs and HTTP processing will only have access
Emmanuel Hocdetbd695fe2017-05-15 15:53:41 +020011291 to deciphered contents. SSLv3 is disabled per default, use "ssl-min-ver SSLv3"
11292 to enable it.
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011293
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011294ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11295 This option enforces use of <version> or lower on SSL connections instantiated
11296 from this listener. This option is also available on global statement
11297 "ssl-default-bind-options". See also "ssl-min-ver".
11298
11299ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
11300 This option enforces use of <version> or upper on SSL connections instantiated
11301 from this listener. This option is also available on global statement
11302 "ssl-default-bind-options". See also "ssl-max-ver".
11303
Emmanuel Hocdet65623372013-01-24 17:17:15 +010011304strict-sni
11305 This setting is only available when support for OpenSSL was built in. The
11306 SSL/TLS negotiation is allow only if the client provided an SNI which match
11307 a certificate. The default certificate is not used.
11308 See the "crt" option for more information.
11309
Willy Tarreau2af207a2015-02-04 00:45:58 +010011310tcp-ut <delay>
Tim Düsterhus4896c442016-11-29 02:15:19 +010011311 Sets the TCP User Timeout for all incoming connections instantiated from this
Willy Tarreau2af207a2015-02-04 00:45:58 +010011312 listening socket. This option is available on Linux since version 2.6.37. It
11313 allows haproxy to configure a timeout for sockets which contain data not
Davor Ocelice9ed2812017-12-25 17:49:28 +010011314 receiving an acknowledgment for the configured delay. This is especially
Willy Tarreau2af207a2015-02-04 00:45:58 +010011315 useful on long-lived connections experiencing long idle periods such as
11316 remote terminals or database connection pools, where the client and server
11317 timeouts must remain high to allow a long period of idle, but where it is
11318 important to detect that the client has disappeared in order to release all
11319 resources associated with its connection (and the server's session). The
11320 argument is a delay expressed in milliseconds by default. This only works
11321 for regular TCP connections, and is ignored for other protocols.
11322
Willy Tarreau1c862c52012-10-05 16:21:00 +020011323tfo
Lukas Tribus0defb902013-02-13 23:35:39 +010011324 Is an optional keyword which is supported only on Linux kernels >= 3.7. It
Willy Tarreau1c862c52012-10-05 16:21:00 +020011325 enables TCP Fast Open on the listening socket, which means that clients which
11326 support this feature will be able to send a request and receive a response
11327 during the 3-way handshake starting from second connection, thus saving one
11328 round-trip after the first connection. This only makes sense with protocols
11329 that use high connection rates and where each round trip matters. This can
11330 possibly cause issues with many firewalls which do not accept data on SYN
11331 packets, so this option should only be enabled once well tested. This option
Lukas Tribus0999f762013-04-02 16:43:24 +020011332 is only supported on TCPv4/TCPv6 sockets and ignored by other ones. You may
11333 need to build HAProxy with USE_TFO=1 if your libc doesn't define
11334 TCP_FASTOPEN.
Willy Tarreau1c862c52012-10-05 16:21:00 +020011335
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011336tls-ticket-keys <keyfile>
11337 Sets the TLS ticket keys file to load the keys from. The keys need to be 48
Emeric Brun9e754772019-01-10 17:51:55 +010011338 or 80 bytes long, depending if aes128 or aes256 is used, encoded with base64
11339 with one line per key (ex. openssl rand 80 | openssl base64 -A | xargs echo).
11340 The first key determines the key length used for next keys: you can't mix
11341 aes128 and aes256 keys. Number of keys is specified by the TLS_TICKETS_NO
11342 build option (default 3) and at least as many keys need to be present in
11343 the file. Last TLS_TICKETS_NO keys will be used for decryption and the
11344 penultimate one for encryption. This enables easy key rotation by just
11345 appending new key to the file and reloading the process. Keys must be
11346 periodically rotated (ex. every 12h) or Perfect Forward Secrecy is
11347 compromised. It is also a good idea to keep the keys off any permanent
Nenad Merdanovic188ad3e2015-02-27 19:56:50 +010011348 storage such as hard drives (hint: use tmpfs and don't swap those files).
11349 Lifetime hint can be changed using tune.ssl.timeout.
11350
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011351transparent
11352 Is an optional keyword which is supported only on certain Linux kernels. It
11353 indicates that the addresses will be bound even if they do not belong to the
11354 local machine, and that packets targeting any of these addresses will be
11355 intercepted just as if the addresses were locally configured. This normally
11356 requires that IP forwarding is enabled. Caution! do not use this with the
11357 default address '*', as it would redirect any traffic for the specified port.
11358 This keyword is available only when HAProxy is built with USE_LINUX_TPROXY=1.
11359 This parameter is only compatible with TCPv4 and TCPv6 sockets, depending on
11360 kernel version. Some distribution kernels include backports of the feature,
11361 so check for support with your vendor.
11362
Willy Tarreau77e3af92012-11-24 15:07:23 +010011363v4v6
11364 Is an optional keyword which is supported only on most recent systems
11365 including Linux kernels >= 2.4.21. It is used to bind a socket to both IPv4
11366 and IPv6 when it uses the default address. Doing so is sometimes necessary
11367 on systems which bind to IPv6 only by default. It has no effect on non-IPv6
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011368 sockets, and is overridden by the "v6only" option.
Willy Tarreau77e3af92012-11-24 15:07:23 +010011369
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011370v6only
11371 Is an optional keyword which is supported only on most recent systems
11372 including Linux kernels >= 2.4.21. It is used to bind a socket to IPv6 only
11373 when it uses the default address. Doing so is sometimes preferred to doing it
Willy Tarreau77e3af92012-11-24 15:07:23 +010011374 system-wide as it is per-listener. It has no effect on non-IPv6 sockets and
11375 has precedence over the "v4v6" option.
Willy Tarreau9b6700f2012-11-24 11:55:28 +010011376
Willy Tarreaub6205fd2012-09-24 12:27:33 +020011377uid <uid>
11378 Sets the owner of the UNIX sockets to the designated system uid. It can also
11379 be set by default in the global section's "unix-bind" statement. Note that
11380 some platforms simply ignore this. This setting is equivalent to the "user"
11381 setting except that the user numeric ID is used instead of its name. This
11382 setting is ignored by non UNIX sockets.
11383
11384user <user>
11385 Sets the owner of the UNIX sockets to the designated system user. It can also
11386 be set by default in the global section's "unix-bind" statement. Note that
11387 some platforms simply ignore this. This setting is equivalent to the "uid"
11388 setting except that the user name is used instead of its uid. This setting is
11389 ignored by non UNIX sockets.
11390
Emeric Brun1a073b42012-09-28 17:07:34 +020011391verify [none|optional|required]
11392 This setting is only available when support for OpenSSL was built in. If set
11393 to 'none', client certificate is not requested. This is the default. In other
11394 cases, a client certificate is requested. If the client does not provide a
11395 certificate after the request and if 'verify' is set to 'required', then the
11396 handshake is aborted, while it would have succeeded if set to 'optional'. The
Emeric Brunfd33a262012-10-11 16:28:27 +020011397 certificate provided by the client is always verified using CAs from
11398 'ca-file' and optional CRLs from 'crl-file'. On verify failure the handshake
11399 is aborted, regardless of the 'verify' option, unless the error code exactly
11400 matches one of those listed with 'ca-ignore-err' or 'crt-ignore-err'.
Willy Tarreau4a5cade2012-04-05 21:09:48 +020011401
Willy Tarreaub6205fd2012-09-24 12:27:33 +0200114025.2. Server and default-server options
Cyril Bontéf0c60612010-02-06 14:44:47 +010011403------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020011404
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011405The "server" and "default-server" keywords support a certain number of settings
11406which are all passed as arguments on the server line. The order in which those
11407arguments appear does not count, and they are all optional. Some of those
11408settings are single words (booleans) while others expect one or several values
11409after them. In this case, the values must immediately follow the setting name.
11410Except default-server, all those settings must be specified after the server's
11411address if they are used:
Willy Tarreau6a06a402007-07-15 20:15:28 +020011412
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011413 server <name> <address>[:port] [settings ...]
Krzysztof Piotr Oledzkic6df0662010-01-05 16:38:49 +010011414 default-server [settings ...]
Willy Tarreau6a06a402007-07-15 20:15:28 +020011415
Frédéric Lécailled2376272017-03-21 18:52:12 +010011416Note that all these settings are supported both by "server" and "default-server"
11417keywords, except "id" which is only supported by "server".
11418
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011419The currently supported settings are the following ones.
Willy Tarreau0ba27502007-12-24 16:55:16 +010011420
Willy Tarreauceb4ac92012-04-28 00:41:46 +020011421addr <ipv4|ipv6>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011422 Using the "addr" parameter, it becomes possible to use a different IP address
Baptiste Assmann13f83532016-03-06 23:14:36 +010011423 to send health-checks or to probe the agent-check. On some servers, it may be
11424 desirable to dedicate an IP address to specific component able to perform
11425 complex tests which are more suitable to health-checks than the application.
11426 This parameter is ignored if the "check" parameter is not set. See also the
11427 "port" parameter.
Willy Tarreau6a06a402007-07-15 20:15:28 +020011428
Simon Hormand60d6912013-11-25 10:46:36 +090011429agent-check
11430 Enable an auxiliary agent check which is run independently of a regular
Willy Tarreau81f5d942013-12-09 20:51:51 +010011431 health check. An agent health check is performed by making a TCP connection
Willy Tarreau7a0139e2018-12-16 08:42:56 +010011432 to the port set by the "agent-port" parameter and reading an ASCII string
11433 terminated by the first '\r' or '\n' met. The string is made of a series of
11434 words delimited by spaces, tabs or commas in any order, each consisting of :
Simon Hormand60d6912013-11-25 10:46:36 +090011435
Willy Tarreau81f5d942013-12-09 20:51:51 +010011436 - An ASCII representation of a positive integer percentage, e.g. "75%".
Simon Hormand60d6912013-11-25 10:46:36 +090011437 Values in this format will set the weight proportional to the initial
Willy Tarreauc5af3a62014-10-07 15:27:33 +020011438 weight of a server as configured when haproxy starts. Note that a zero
11439 weight is reported on the stats page as "DRAIN" since it has the same
11440 effect on the server (it's removed from the LB farm).
Simon Hormand60d6912013-11-25 10:46:36 +090011441
Davor Ocelice9ed2812017-12-25 17:49:28 +010011442 - The string "maxconn:" followed by an integer (no space between). Values
11443 in this format will set the maxconn of a server. The maximum number of
11444 connections advertised needs to be multiplied by the number of load
11445 balancers and different backends that use this health check to get the
11446 total number of connections the server might receive. Example: maxconn:30
Nenad Merdanovic174dd372016-04-24 23:10:06 +020011447
Willy Tarreau81f5d942013-12-09 20:51:51 +010011448 - The word "ready". This will turn the server's administrative state to the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011449 READY mode, thus canceling any DRAIN or MAINT state
Simon Hormand60d6912013-11-25 10:46:36 +090011450
Willy Tarreau81f5d942013-12-09 20:51:51 +010011451 - The word "drain". This will turn the server's administrative state to the
11452 DRAIN mode, thus it will not accept any new connections other than those
11453 that are accepted via persistence.
Simon Hormand60d6912013-11-25 10:46:36 +090011454
Willy Tarreau81f5d942013-12-09 20:51:51 +010011455 - The word "maint". This will turn the server's administrative state to the
11456 MAINT mode, thus it will not accept any new connections at all, and health
11457 checks will be stopped.
Simon Hormand60d6912013-11-25 10:46:36 +090011458
Willy Tarreau81f5d942013-12-09 20:51:51 +010011459 - The words "down", "failed", or "stopped", optionally followed by a
11460 description string after a sharp ('#'). All of these mark the server's
11461 operating state as DOWN, but since the word itself is reported on the stats
11462 page, the difference allows an administrator to know if the situation was
11463 expected or not : the service may intentionally be stopped, may appear up
Davor Ocelice9ed2812017-12-25 17:49:28 +010011464 but fail some validity tests, or may be seen as down (e.g. missing process,
Willy Tarreau81f5d942013-12-09 20:51:51 +010011465 or port not responding).
Simon Hormand60d6912013-11-25 10:46:36 +090011466
Willy Tarreau81f5d942013-12-09 20:51:51 +010011467 - The word "up" sets back the server's operating state as UP if health checks
11468 also report that the service is accessible.
Simon Hormand60d6912013-11-25 10:46:36 +090011469
Willy Tarreau81f5d942013-12-09 20:51:51 +010011470 Parameters which are not advertised by the agent are not changed. For
11471 example, an agent might be designed to monitor CPU usage and only report a
11472 relative weight and never interact with the operating status. Similarly, an
11473 agent could be designed as an end-user interface with 3 radio buttons
11474 allowing an administrator to change only the administrative state. However,
11475 it is important to consider that only the agent may revert its own actions,
11476 so if a server is set to DRAIN mode or to DOWN state using the agent, the
11477 agent must implement the other equivalent actions to bring the service into
11478 operations again.
Simon Hormand60d6912013-11-25 10:46:36 +090011479
Simon Horman2f1f9552013-11-25 10:46:37 +090011480 Failure to connect to the agent is not considered an error as connectivity
11481 is tested by the regular health check which is enabled by the "check"
Willy Tarreau81f5d942013-12-09 20:51:51 +010011482 parameter. Warning though, it is not a good idea to stop an agent after it
11483 reports "down", since only an agent reporting "up" will be able to turn the
11484 server up again. Note that the CLI on the Unix stats socket is also able to
Willy Tarreau989222a2016-01-15 10:26:26 +010011485 force an agent's result in order to work around a bogus agent if needed.
Simon Horman2f1f9552013-11-25 10:46:37 +090011486
Willy Tarreau81f5d942013-12-09 20:51:51 +010011487 Requires the "agent-port" parameter to be set. See also the "agent-inter"
Frédéric Lécailled2376272017-03-21 18:52:12 +010011488 and "no-agent-check" parameters.
Simon Hormand60d6912013-11-25 10:46:36 +090011489
James Brown55f9ff12015-10-21 18:19:05 -070011490agent-send <string>
11491 If this option is specified, haproxy will send the given string (verbatim)
11492 to the agent server upon connection. You could, for example, encode
11493 the backend name into this string, which would enable your agent to send
11494 different responses based on the backend. Make sure to include a '\n' if
11495 you want to terminate your request with a newline.
11496
Simon Hormand60d6912013-11-25 10:46:36 +090011497agent-inter <delay>
11498 The "agent-inter" parameter sets the interval between two agent checks
11499 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11500
11501 Just as with every other time-based parameter, it may be entered in any
11502 other explicit unit among { us, ms, s, m, h, d }. The "agent-inter"
11503 parameter also serves as a timeout for agent checks "timeout check" is
11504 not set. In order to reduce "resonance" effects when multiple servers are
11505 hosted on the same hardware, the agent and health checks of all servers
11506 are started with a small time offset between them. It is also possible to
11507 add some random noise in the agent and health checks interval using the
11508 global "spread-checks" keyword. This makes sense for instance when a lot
11509 of backends use the same servers.
11510
11511 See also the "agent-check" and "agent-port" parameters.
11512
Misiek768d8602017-01-09 09:52:43 +010011513agent-addr <addr>
11514 The "agent-addr" parameter sets address for agent check.
11515
11516 You can offload agent-check to another target, so you can make single place
11517 managing status and weights of servers defined in haproxy in case you can't
11518 make self-aware and self-managing services. You can specify both IP or
11519 hostname, it will be resolved.
11520
Simon Hormand60d6912013-11-25 10:46:36 +090011521agent-port <port>
11522 The "agent-port" parameter sets the TCP port used for agent checks.
11523
11524 See also the "agent-check" and "agent-inter" parameters.
11525
Olivier Houchardc7566002018-11-20 23:33:50 +010011526alpn <protocols>
11527 This enables the TLS ALPN extension and advertises the specified protocol
11528 list as supported on top of ALPN. The protocol list consists in a comma-
11529 delimited list of protocol names, for instance: "http/1.1,http/1.0" (without
11530 quotes). This requires that the SSL library is build with support for TLS
11531 extensions enabled (check with haproxy -vv). The ALPN extension replaces the
11532 initial NPN extension. ALPN is required to connect to HTTP/2 servers.
11533 Versions of OpenSSL prior to 1.0.2 didn't support ALPN and only supposed the
11534 now obsolete NPN extension.
11535 If both HTTP/2 and HTTP/1.1 are expected to be supported, both versions can
11536 be advertised, in order of preference, like below :
11537
11538 server 127.0.0.1:443 ssl crt pub.pem alpn h2,http/1.1
11539
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011540backup
11541 When "backup" is present on a server line, the server is only used in load
11542 balancing when all other non-backup servers are unavailable. Requests coming
11543 with a persistence cookie referencing the server will always be served
11544 though. By default, only the first operational backup server is used, unless
Frédéric Lécailled2376272017-03-21 18:52:12 +010011545 the "allbackups" option is set in the backend. See also the "no-backup" and
11546 "allbackups" options.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011547
Emeric Brunef42d922012-10-11 16:11:36 +020011548ca-file <cafile>
11549 This setting is only available when support for OpenSSL was built in. It
11550 designates a PEM file from which to load CA certificates used to verify
11551 server's certificate.
11552
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011553check
11554 This option enables health checks on the server. By default, a server is
Patrick Mézardb7aeec62012-01-22 16:01:22 +010011555 always considered available. If "check" is set, the server is available when
11556 accepting periodic TCP connections, to ensure that it is really able to serve
11557 requests. The default address and port to send the tests to are those of the
11558 server, and the default source is the same as the one defined in the
11559 backend. It is possible to change the address using the "addr" parameter, the
11560 port using the "port" parameter, the source address using the "source"
11561 address, and the interval and timers using the "inter", "rise" and "fall"
Simon Hormanafc47ee2013-11-25 10:46:35 +090011562 parameters. The request method is define in the backend using the "httpchk",
11563 "smtpchk", "mysql-check", "pgsql-check" and "ssl-hello-chk" options. Please
Frédéric Lécailled2376272017-03-21 18:52:12 +010011564 refer to those options and parameters for more information. See also
11565 "no-check" option.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011566
Willy Tarreau6c16adc2012-10-05 00:04:16 +020011567check-send-proxy
11568 This option forces emission of a PROXY protocol line with outgoing health
11569 checks, regardless of whether the server uses send-proxy or not for the
11570 normal traffic. By default, the PROXY protocol is enabled for health checks
11571 if it is already enabled for normal traffic and if no "port" nor "addr"
11572 directive is present. However, if such a directive is present, the
11573 "check-send-proxy" option needs to be used to force the use of the
11574 protocol. See also the "send-proxy" option for more information.
11575
Olivier Houchard92150142018-12-21 19:47:01 +010011576check-alpn <protocols>
11577 Defines which protocols to advertise with ALPN. The protocol list consists in
11578 a comma-delimited list of protocol names, for instance: "http/1.1,http/1.0"
11579 (without quotes). If it is not set, the server ALPN is used.
11580
Jérôme Magninae9bb762018-12-09 16:08:26 +010011581check-sni <sni>
Olivier Houchard9130a962017-10-17 17:33:43 +020011582 This option allows you to specify the SNI to be used when doing health checks
Jérôme Magninae9bb762018-12-09 16:08:26 +010011583 over SSL. It is only possible to use a string to set <sni>. If you want to
11584 set a SNI for proxied traffic, see "sni".
Olivier Houchard9130a962017-10-17 17:33:43 +020011585
Willy Tarreau763a95b2012-10-04 23:15:39 +020011586check-ssl
11587 This option forces encryption of all health checks over SSL, regardless of
11588 whether the server uses SSL or not for the normal traffic. This is generally
11589 used when an explicit "port" or "addr" directive is specified and SSL health
11590 checks are not inherited. It is important to understand that this option
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030011591 inserts an SSL transport layer below the checks, so that a simple TCP connect
Willy Tarreau763a95b2012-10-04 23:15:39 +020011592 check becomes an SSL connect, which replaces the old ssl-hello-chk. The most
11593 common use is to send HTTPS checks by combining "httpchk" with SSL checks.
Davor Ocelice9ed2812017-12-25 17:49:28 +010011594 All SSL settings are common to health checks and traffic (e.g. ciphers).
Frédéric Lécailled2376272017-03-21 18:52:12 +010011595 See the "ssl" option for more information and "no-check-ssl" to disable
11596 this option.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011597
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011598ciphers <ciphers>
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011599 This setting is only available when support for OpenSSL was built in. This
11600 option sets the string describing the list of cipher algorithms that is
11601 negotiated during the SSL/TLS handshake with the server. The format of the
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011602 string is defined in "man 1 ciphers" from OpenSSL man pages. For background
11603 information and recommendations see e.g.
11604 (https://wiki.mozilla.org/Security/Server_Side_TLS) and
11605 (https://mozilla.github.io/server-side-tls/ssl-config-generator/). For TLSv1.3
11606 cipher configuration, please check the "ciphersuites" keyword.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011607
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011608ciphersuites <ciphersuites>
11609 This setting is only available when support for OpenSSL was built in and
11610 OpenSSL 1.1.1 or later was used to build HAProxy. This option sets the string
11611 describing the list of cipher algorithms that is negotiated during the TLS
11612 1.3 handshake with the server. The format of the string is defined in
Bertrand Jacquin4f03ab02019-02-03 18:48:49 +000011613 "man 1 ciphers" from OpenSSL man pages under the "ciphersuites" section.
11614 For cipher configuration for TLSv1.2 and earlier, please check the "ciphers"
11615 keyword.
Dirkjan Bussink415150f2018-09-14 11:14:21 +020011616
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011617cookie <value>
11618 The "cookie" parameter sets the cookie value assigned to the server to
11619 <value>. This value will be checked in incoming requests, and the first
11620 operational server possessing the same value will be selected. In return, in
11621 cookie insertion or rewrite modes, this value will be assigned to the cookie
11622 sent to the client. There is nothing wrong in having several servers sharing
11623 the same cookie value, and it is in fact somewhat common between normal and
11624 backup servers. See also the "cookie" keyword in backend section.
11625
Emeric Brunef42d922012-10-11 16:11:36 +020011626crl-file <crlfile>
11627 This setting is only available when support for OpenSSL was built in. It
11628 designates a PEM file from which to load certificate revocation list used
11629 to verify server's certificate.
11630
Emeric Bruna7aa3092012-10-26 12:58:00 +020011631crt <cert>
11632 This setting is only available when support for OpenSSL was built in.
11633 It designates a PEM file from which to load both a certificate and the
11634 associated private key. This file can be built by concatenating both PEM
11635 files into one. This certificate will be sent if the server send a client
11636 certificate request.
11637
Willy Tarreau96839092010-03-29 10:02:24 +020011638disabled
11639 The "disabled" keyword starts the server in the "disabled" state. That means
11640 that it is marked down in maintenance mode, and no connection other than the
11641 ones allowed by persist mode will reach it. It is very well suited to setup
11642 new servers, because normal traffic will never reach them, while it is still
11643 possible to test the service by making use of the force-persist mechanism.
Frédéric Lécailled2376272017-03-21 18:52:12 +010011644 See also "enabled" setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011645
Frédéric Lécailled2376272017-03-21 18:52:12 +010011646enabled
11647 This option may be used as 'server' setting to reset any 'disabled'
11648 setting which would have been inherited from 'default-server' directive as
11649 default value.
11650 It may also be used as 'default-server' setting to reset any previous
11651 'default-server' 'disabled' setting.
Willy Tarreau96839092010-03-29 10:02:24 +020011652
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011653error-limit <count>
Willy Tarreau983e01e2010-01-11 18:42:06 +010011654 If health observing is enabled, the "error-limit" parameter specifies the
11655 number of consecutive errors that triggers event selected by the "on-error"
11656 option. By default it is set to 10 consecutive errors.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011657
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011658 See also the "check", "error-limit" and "on-error".
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011659
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011660fall <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011661 The "fall" parameter states that a server will be considered as dead after
11662 <count> consecutive unsuccessful health checks. This value defaults to 3 if
11663 unspecified. See also the "check", "inter" and "rise" parameters.
11664
Emeric Brun8694b9a2012-10-05 14:39:07 +020011665force-sslv3
11666 This option enforces use of SSLv3 only when SSL is used to communicate with
11667 the server. SSLv3 is generally less expensive than the TLS counterparts for
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011668 high connection rates. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011669 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011670
11671force-tlsv10
11672 This option enforces use of TLSv1.0 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011673 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011674 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011675
11676force-tlsv11
11677 This option enforces use of TLSv1.1 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011678 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011679 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011680
11681force-tlsv12
11682 This option enforces use of TLSv1.2 only when SSL is used to communicate with
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011683 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011684 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emeric Brun8694b9a2012-10-05 14:39:07 +020011685
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011686force-tlsv13
11687 This option enforces use of TLSv1.3 only when SSL is used to communicate with
11688 the server. This option is also available on global statement
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011689 "ssl-default-server-options". See also "ssl-min-ver" and ssl-max-ver".
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011690
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011691id <value>
Willy Tarreau53fb4ae2009-10-04 23:04:08 +020011692 Set a persistent ID for the server. This ID must be positive and unique for
11693 the proxy. An unused ID will automatically be assigned if unset. The first
11694 assigned value will be 1. This ID is currently only returned in statistics.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011695
Willy Tarreau6a031d12016-11-07 19:42:35 +010011696init-addr {last | libc | none | <ip>},[...]*
11697 Indicate in what order the server's address should be resolved upon startup
11698 if it uses an FQDN. Attempts are made to resolve the address by applying in
Davor Ocelice9ed2812017-12-25 17:49:28 +010011699 turn each of the methods mentioned in the comma-delimited list. The first
Willy Tarreau6a031d12016-11-07 19:42:35 +010011700 method which succeeds is used. If the end of the list is reached without
11701 finding a working method, an error is thrown. Method "last" suggests to pick
11702 the address which appears in the state file (see "server-state-file"). Method
11703 "libc" uses the libc's internal resolver (gethostbyname() or getaddrinfo()
11704 depending on the operating system and build options). Method "none"
11705 specifically indicates that the server should start without any valid IP
11706 address in a down state. It can be useful to ignore some DNS issues upon
11707 startup, waiting for the situation to get fixed later. Finally, an IP address
11708 (IPv4 or IPv6) may be provided. It can be the currently known address of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011709 server (e.g. filled by a configuration generator), or the address of a dummy
Willy Tarreau6a031d12016-11-07 19:42:35 +010011710 server used to catch old sessions and present them with a decent error
11711 message for example. When the "first" load balancing algorithm is used, this
11712 IP address could point to a fake server used to trigger the creation of new
11713 instances on the fly. This option defaults to "last,libc" indicating that the
11714 previous address found in the state file (if any) is used first, otherwise
11715 the libc's resolver is used. This ensures continued compatibility with the
Davor Ocelice9ed2812017-12-25 17:49:28 +010011716 historic behavior.
Willy Tarreau6a031d12016-11-07 19:42:35 +010011717
11718 Example:
11719 defaults
11720 # never fail on address resolution
11721 default-server init-addr last,libc,none
11722
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011723inter <delay>
11724fastinter <delay>
11725downinter <delay>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011726 The "inter" parameter sets the interval between two consecutive health checks
11727 to <delay> milliseconds. If left unspecified, the delay defaults to 2000 ms.
11728 It is also possible to use "fastinter" and "downinter" to optimize delays
11729 between checks depending on the server state :
11730
Pieter Baauw44fc9df2015-09-17 21:30:46 +020011731 Server state | Interval used
11732 ----------------------------------------+----------------------------------
11733 UP 100% (non-transitional) | "inter"
11734 ----------------------------------------+----------------------------------
11735 Transitionally UP (going down "fall"), | "fastinter" if set,
11736 Transitionally DOWN (going up "rise"), | "inter" otherwise.
11737 or yet unchecked. |
11738 ----------------------------------------+----------------------------------
11739 DOWN 100% (non-transitional) | "downinter" if set,
11740 | "inter" otherwise.
11741 ----------------------------------------+----------------------------------
Willy Tarreaud72758d2010-01-12 10:42:19 +010011742
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011743 Just as with every other time-based parameter, they can be entered in any
11744 other explicit unit among { us, ms, s, m, h, d }. The "inter" parameter also
11745 serves as a timeout for health checks sent to servers if "timeout check" is
11746 not set. In order to reduce "resonance" effects when multiple servers are
Simon Hormand60d6912013-11-25 10:46:36 +090011747 hosted on the same hardware, the agent and health checks of all servers
11748 are started with a small time offset between them. It is also possible to
11749 add some random noise in the agent and health checks interval using the
11750 global "spread-checks" keyword. This makes sense for instance when a lot
11751 of backends use the same servers.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011752
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011753maxconn <maxconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011754 The "maxconn" parameter specifies the maximal number of concurrent
11755 connections that will be sent to this server. If the number of incoming
11756 concurrent requests goes higher than this value, they will be queued, waiting
11757 for a connection to be released. This parameter is very important as it can
11758 save fragile servers from going down under extreme loads. If a "minconn"
11759 parameter is specified, the limit becomes dynamic. The default value is "0"
11760 which means unlimited. See also the "minconn" and "maxqueue" parameters, and
11761 the backend's "fullconn" keyword.
11762
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011763maxqueue <maxqueue>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011764 The "maxqueue" parameter specifies the maximal number of connections which
11765 will wait in the queue for this server. If this limit is reached, next
11766 requests will be redispatched to other servers instead of indefinitely
11767 waiting to be served. This will break persistence but may allow people to
11768 quickly re-log in when the server they try to connect to is dying. The
11769 default value is "0" which means the queue is unlimited. See also the
11770 "maxconn" and "minconn" parameters.
11771
Willy Tarreau9c538e02019-01-23 10:21:49 +010011772max-reuse <count>
11773 The "max-reuse" argument indicates the HTTP connection processors that they
11774 should not reuse a server connection more than this number of times to send
11775 new requests. Permitted values are -1 (the default), which disables this
11776 limit, or any positive value. Value zero will effectively disable keep-alive.
11777 This is only used to work around certain server bugs which cause them to leak
11778 resources over time. The argument is not necessarily respected by the lower
11779 layers as there might be technical limitations making it impossible to
11780 enforce. At least HTTP/2 connections to servers will respect it.
11781
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011782minconn <minconn>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011783 When the "minconn" parameter is set, the maxconn limit becomes a dynamic
11784 limit following the backend's load. The server will always accept at least
11785 <minconn> connections, never more than <maxconn>, and the limit will be on
11786 the ramp between both values when the backend has less than <fullconn>
11787 concurrent connections. This makes it possible to limit the load on the
11788 server during normal loads, but push it further for important loads without
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010011789 overloading the server during exceptional loads. See also the "maxconn"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020011790 and "maxqueue" parameters, as well as the "fullconn" backend keyword.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011791
Willy Tarreaud72f0f32015-10-13 14:50:22 +020011792namespace <name>
11793 On Linux, it is possible to specify which network namespace a socket will
11794 belong to. This directive makes it possible to explicitly bind a server to
11795 a namespace different from the default one. Please refer to your operating
11796 system's documentation to find more details about network namespaces.
11797
Frédéric Lécailled2376272017-03-21 18:52:12 +010011798no-agent-check
11799 This option may be used as "server" setting to reset any "agent-check"
11800 setting which would have been inherited from "default-server" directive as
11801 default value.
11802 It may also be used as "default-server" setting to reset any previous
11803 "default-server" "agent-check" setting.
11804
11805no-backup
11806 This option may be used as "server" setting to reset any "backup"
11807 setting which would have been inherited from "default-server" directive as
11808 default value.
11809 It may also be used as "default-server" setting to reset any previous
11810 "default-server" "backup" setting.
11811
11812no-check
11813 This option may be used as "server" setting to reset any "check"
11814 setting which would have been inherited from "default-server" directive as
11815 default value.
11816 It may also be used as "default-server" setting to reset any previous
11817 "default-server" "check" setting.
11818
11819no-check-ssl
11820 This option may be used as "server" setting to reset any "check-ssl"
11821 setting which would have been inherited from "default-server" directive as
11822 default value.
11823 It may also be used as "default-server" setting to reset any previous
11824 "default-server" "check-ssl" setting.
11825
Frédéric Lécailled2376272017-03-21 18:52:12 +010011826no-send-proxy
11827 This option may be used as "server" setting to reset any "send-proxy"
11828 setting which would have been inherited from "default-server" directive as
11829 default value.
11830 It may also be used as "default-server" setting to reset any previous
11831 "default-server" "send-proxy" setting.
11832
11833no-send-proxy-v2
11834 This option may be used as "server" setting to reset any "send-proxy-v2"
11835 setting which would have been inherited from "default-server" directive as
11836 default value.
11837 It may also be used as "default-server" setting to reset any previous
11838 "default-server" "send-proxy-v2" setting.
11839
11840no-send-proxy-v2-ssl
11841 This option may be used as "server" setting to reset any "send-proxy-v2-ssl"
11842 setting which would have been inherited from "default-server" directive as
11843 default value.
11844 It may also be used as "default-server" setting to reset any previous
11845 "default-server" "send-proxy-v2-ssl" setting.
11846
11847no-send-proxy-v2-ssl-cn
11848 This option may be used as "server" setting to reset any "send-proxy-v2-ssl-cn"
11849 setting which would have been inherited from "default-server" directive as
11850 default value.
11851 It may also be used as "default-server" setting to reset any previous
11852 "default-server" "send-proxy-v2-ssl-cn" setting.
11853
11854no-ssl
11855 This option may be used as "server" setting to reset any "ssl"
11856 setting which would have been inherited from "default-server" directive as
11857 default value.
11858 It may also be used as "default-server" setting to reset any previous
11859 "default-server" "ssl" setting.
11860
Willy Tarreau2a3fb1c2015-02-05 16:47:07 +010011861no-ssl-reuse
11862 This option disables SSL session reuse when SSL is used to communicate with
11863 the server. It will force the server to perform a full handshake for every
11864 new connection. It's probably only useful for benchmarking, troubleshooting,
11865 and for paranoid users.
11866
Emeric Brun9b3009b2012-10-05 11:55:06 +020011867no-sslv3
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011868 This option disables support for SSLv3 when SSL is used to communicate with
11869 the server. Note that SSLv2 is disabled in the code and cannot be enabled
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011870 using any configuration option. Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011871
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011872 Supported in default-server: No
11873
Emeric Brunf9c5c472012-10-11 15:28:34 +020011874no-tls-tickets
11875 This setting is only available when support for OpenSSL was built in. It
11876 disables the stateless session resumption (RFC 5077 TLS Ticket
11877 extension) and force to use stateful session resumption. Stateless
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011878 session resumption is more expensive in CPU usage for servers. This option
11879 is also available on global statement "ssl-default-server-options".
Frédéric Lécailled2376272017-03-21 18:52:12 +010011880 See also "tls-tickets".
Emeric Brunf9c5c472012-10-11 15:28:34 +020011881
Emeric Brun9b3009b2012-10-05 11:55:06 +020011882no-tlsv10
Emeric Brun8694b9a2012-10-05 14:39:07 +020011883 This option disables support for TLSv1.0 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011884 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11885 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011886 often makes sense to disable it when communicating with local servers. This
11887 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011888 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011889
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011890 Supported in default-server: No
11891
Emeric Brun9b3009b2012-10-05 11:55:06 +020011892no-tlsv11
Emeric Brun8694b9a2012-10-05 14:39:07 +020011893 This option disables support for TLSv1.1 when SSL is used to communicate with
Emeric Brunf5da4932012-09-28 19:42:54 +020011894 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11895 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011896 often makes sense to disable it when communicating with local servers. This
11897 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011898 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011899
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011900 Supported in default-server: No
11901
Emeric Brun9b3009b2012-10-05 11:55:06 +020011902no-tlsv12
Emeric Brun8694b9a2012-10-05 14:39:07 +020011903 This option disables support for TLSv1.2 when SSL is used to communicate with
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011904 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11905 using any configuration option. TLSv1 is more expensive than SSLv3 so it
Emeric Brun2c86cbf2014-10-30 15:56:50 +010011906 often makes sense to disable it when communicating with local servers. This
11907 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011908 Use "ssl-min-ver" and "ssl-max-ver" instead.
Emmanuel Hocdet42fb9802017-03-30 19:29:39 +020011909
11910 Supported in default-server: No
11911
11912no-tlsv13
11913 This option disables support for TLSv1.3 when SSL is used to communicate with
11914 the server. Note that SSLv2 is disabled in the code and cannot be enabled
11915 using any configuration option. TLSv1 is more expensive than SSLv3 so it
11916 often makes sense to disable it when communicating with local servers. This
11917 option is also available on global statement "ssl-default-server-options".
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020011918 Use "ssl-min-ver" and "ssl-max-ver" instead.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020011919
Emmanuel Hocdet6cb2d1e2017-03-30 14:43:31 +020011920 Supported in default-server: No
11921
Frédéric Lécailled2376272017-03-21 18:52:12 +010011922no-verifyhost
11923 This option may be used as "server" setting to reset any "verifyhost"
11924 setting which would have been inherited from "default-server" directive as
11925 default value.
11926 It may also be used as "default-server" setting to reset any previous
11927 "default-server" "verifyhost" setting.
Willy Tarreau763a95b2012-10-04 23:15:39 +020011928
Simon Hormanfa461682011-06-25 09:39:49 +090011929non-stick
11930 Never add connections allocated to this sever to a stick-table.
11931 This may be used in conjunction with backup to ensure that
11932 stick-table persistence is disabled for backup servers.
11933
Olivier Houchardc7566002018-11-20 23:33:50 +010011934npn <protocols>
11935 This enables the NPN TLS extension and advertises the specified protocol list
11936 as supported on top of NPN. The protocol list consists in a comma-delimited
11937 list of protocol names, for instance: "http/1.1,http/1.0" (without quotes).
11938 This requires that the SSL library is build with support for TLS extensions
11939 enabled (check with haproxy -vv). Note that the NPN extension has been
11940 replaced with the ALPN extension (see the "alpn" keyword), though this one is
11941 only available starting with OpenSSL 1.0.2.
11942
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011943observe <mode>
11944 This option enables health adjusting based on observing communication with
11945 the server. By default this functionality is disabled and enabling it also
11946 requires to enable health checks. There are two supported modes: "layer4" and
11947 "layer7". In layer4 mode, only successful/unsuccessful tcp connections are
11948 significant. In layer7, which is only allowed for http proxies, responses
11949 received from server are verified, like valid/wrong http code, unparsable
Willy Tarreau150d1462012-03-10 08:19:02 +010011950 headers, a timeout, etc. Valid status codes include 100 to 499, 501 and 505.
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011951
11952 See also the "check", "on-error" and "error-limit".
11953
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010011954on-error <mode>
Krzysztof Piotr Oledzki97f07b82009-12-15 22:31:24 +010011955 Select what should happen when enough consecutive errors are detected.
11956 Currently, four modes are available:
11957 - fastinter: force fastinter
11958 - fail-check: simulate a failed check, also forces fastinter (default)
11959 - sudden-death: simulate a pre-fatal failed health check, one more failed
11960 check will mark a server down, forces fastinter
11961 - mark-down: mark the server immediately down and force fastinter
11962
11963 See also the "check", "observe" and "error-limit".
11964
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011965on-marked-down <action>
11966 Modify what occurs when a server is marked down.
11967 Currently one action is available:
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011968 - shutdown-sessions: Shutdown peer sessions. When this setting is enabled,
11969 all connections to the server are immediately terminated when the server
11970 goes down. It might be used if the health check detects more complex cases
11971 than a simple connection status, and long timeouts would cause the service
11972 to remain unresponsive for too long a time. For instance, a health check
11973 might detect that a database is stuck and that there's no chance to reuse
11974 existing connections anymore. Connections killed this way are logged with
11975 a 'D' termination code (for "Down").
Simon Hormane0d1bfb2011-06-21 14:34:58 +090011976
11977 Actions are disabled by default
11978
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011979on-marked-up <action>
11980 Modify what occurs when a server is marked up.
11981 Currently one action is available:
11982 - shutdown-backup-sessions: Shutdown sessions on all backup servers. This is
11983 done only if the server is not in backup state and if it is not disabled
11984 (it must have an effective weight > 0). This can be used sometimes to force
11985 an active server to take all the traffic back after recovery when dealing
Davor Ocelice9ed2812017-12-25 17:49:28 +010011986 with long sessions (e.g. LDAP, SQL, ...). Doing this can cause more trouble
11987 than it tries to solve (e.g. incomplete transactions), so use this feature
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070011988 with extreme care. Sessions killed because a server comes up are logged
11989 with an 'U' termination code (for "Up").
11990
11991 Actions are disabled by default
11992
Olivier Houchard006e3102018-12-10 18:30:32 +010011993pool-max-conn <max>
11994 Set the maximum number of idling connections for a server. -1 means unlimited
11995 connections, 0 means no idle connections. The default is -1. When idle
11996 connections are enabled, orphaned idle connections which do not belong to any
11997 client session anymore are moved to a dedicated pool so that they remain
11998 usable by future clients. This only applies to connections that can be shared
11999 according to the same principles as those applying to "http-reuse".
12000
Olivier Houchardb7b3faa2018-12-14 18:15:36 +010012001pool-purge-delay <delay>
12002 Sets the delay to start purging idle connections. Each <delay> interval, half
12003 of the idle connections are closed. 0 means it's never purged. The default is
12004 1s.
12005
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012006port <port>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012007 Using the "port" parameter, it becomes possible to use a different port to
12008 send health-checks. On some servers, it may be desirable to dedicate a port
12009 to a specific component able to perform complex tests which are more suitable
12010 to health-checks than the application. It is common to run a simple script in
12011 inetd for instance. This parameter is ignored if the "check" parameter is not
12012 set. See also the "addr" parameter.
12013
Christopher Faulet8ed0a3e2018-04-10 14:45:45 +020012014proto <name>
12015
12016 Forces the multiplexer's protocol to use for the outgoing connections to this
12017 server. It must be compatible with the mode of the backend (TCP or HTTP). It
12018 must also be usable on the backend side. The list of available protocols is
12019 reported in haproxy -vv.
12020 Idea behind this optipon is to bypass the selection of the best multiplexer's
12021 protocol for all connections established to this server.
12022
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012023redir <prefix>
12024 The "redir" parameter enables the redirection mode for all GET and HEAD
12025 requests addressing this server. This means that instead of having HAProxy
12026 forward the request to the server, it will send an "HTTP 302" response with
12027 the "Location" header composed of this prefix immediately followed by the
12028 requested URI beginning at the leading '/' of the path component. That means
12029 that no trailing slash should be used after <prefix>. All invalid requests
12030 will be rejected, and all non-GET or HEAD requests will be normally served by
12031 the server. Note that since the response is completely forged, no header
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012032 mangling nor cookie insertion is possible in the response. However, cookies in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012033 requests are still analyzed, making this solution completely usable to direct
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012034 users to a remote location in case of local disaster. Main use consists in
12035 increasing bandwidth for static servers by having the clients directly
12036 connect to them. Note: never use a relative location here, it would cause a
12037 loop between the client and HAProxy!
12038
12039 Example : server srv1 192.168.1.1:80 redir http://image1.mydomain.com check
12040
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012041rise <count>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012042 The "rise" parameter states that a server will be considered as operational
12043 after <count> consecutive successful health checks. This value defaults to 2
12044 if unspecified. See also the "check", "inter" and "fall" parameters.
12045
Baptiste Assmann8e2d9432018-06-22 15:04:43 +020012046resolve-opts <option>,<option>,...
12047 Comma separated list of options to apply to DNS resolution linked to this
12048 server.
12049
12050 Available options:
12051
12052 * allow-dup-ip
12053 By default, HAProxy prevents IP address duplication in a backend when DNS
12054 resolution at runtime is in operation.
12055 That said, for some cases, it makes sense that two servers (in the same
12056 backend, being resolved by the same FQDN) have the same IP address.
12057 For such case, simply enable this option.
12058 This is the opposite of prevent-dup-ip.
12059
12060 * prevent-dup-ip
12061 Ensure HAProxy's default behavior is enforced on a server: prevent re-using
12062 an IP address already set to a server in the same backend and sharing the
12063 same fqdn.
12064 This is the opposite of allow-dup-ip.
12065
12066 Example:
12067 backend b_myapp
12068 default-server init-addr none resolvers dns
12069 server s1 myapp.example.com:80 check resolve-opts allow-dup-ip
12070 server s2 myapp.example.com:81 check resolve-opts allow-dup-ip
12071
12072 With the option allow-dup-ip set:
12073 * if the nameserver returns a single IP address, then both servers will use
12074 it
12075 * If the nameserver returns 2 IP addresses, then each server will pick up a
12076 different address
12077
12078 Default value: not set
12079
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012080resolve-prefer <family>
12081 When DNS resolution is enabled for a server and multiple IP addresses from
12082 different families are returned, HAProxy will prefer using an IP address
12083 from the family mentioned in the "resolve-prefer" parameter.
12084 Available families: "ipv4" and "ipv6"
12085
Baptiste Assmannc4aabae2015-08-04 22:43:06 +020012086 Default value: ipv6
12087
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012088 Example:
12089
12090 server s1 app1.domain.com:80 resolvers mydns resolve-prefer ipv6
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012091
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012092resolve-net <network>[,<network[,...]]
12093 This options prioritize th choice of an ip address matching a network. This is
12094 useful with clouds to prefer a local ip. In some cases, a cloud high
Tim Düsterhus4896c442016-11-29 02:15:19 +010012095 availability service can be announced with many ip addresses on many
Davor Ocelice9ed2812017-12-25 17:49:28 +010012096 different datacenters. The latency between datacenter is not negligible, so
12097 this patch permits to prefer a local datacenter. If no address matches the
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012098 configured network, another address is selected.
12099
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012100 Example:
12101
12102 server s1 app1.domain.com:80 resolvers mydns resolve-net 10.0.0.0/8
Thierry Fournierac88cfe2016-02-17 22:05:30 +010012103
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012104resolvers <id>
12105 Points to an existing "resolvers" section to resolve current server's
12106 hostname.
12107
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012108 Example:
12109
12110 server s1 app1.domain.com:80 check resolvers mydns
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012111
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012112 See also section 5.3
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012113
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012114send-proxy
12115 The "send-proxy" parameter enforces use of the PROXY protocol over any
12116 connection established to this server. The PROXY protocol informs the other
12117 end about the layer 3/4 addresses of the incoming connection, so that it can
12118 know the client's address or the public address it accessed to, whatever the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010012119 upper layer protocol. For connections accepted by an "accept-proxy" or
12120 "accept-netscaler-cip" listener, the advertised address will be used. Only
12121 TCPv4 and TCPv6 address families are supported. Other families such as
12122 Unix sockets, will report an UNKNOWN family. Servers using this option can
12123 fully be chained to another instance of haproxy listening with an
12124 "accept-proxy" setting. This setting must not be used if the server isn't
12125 aware of the protocol. When health checks are sent to the server, the PROXY
12126 protocol is automatically used when this option is set, unless there is an
12127 explicit "port" or "addr" directive, in which case an explicit
12128 "check-send-proxy" directive would also be needed to use the PROXY protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012129 See also the "no-send-proxy" option of this section and "accept-proxy" and
12130 "accept-netscaler-cip" option of the "bind" keyword.
Willy Tarreau5ab04ec2011-03-20 10:32:26 +010012131
David Safb76832014-05-08 23:42:08 -040012132send-proxy-v2
12133 The "send-proxy-v2" parameter enforces use of the PROXY protocol version 2
12134 over any connection established to this server. The PROXY protocol informs
12135 the other end about the layer 3/4 addresses of the incoming connection, so
12136 that it can know the client's address or the public address it accessed to,
Emmanuel Hocdet404d9782017-10-24 10:55:14 +020012137 whatever the upper layer protocol. It also send ALPN information if an alpn
12138 have been negotiated. This setting must not be used if the server isn't aware
12139 of this version of the protocol. See also the "no-send-proxy-v2" option of
12140 this section and send-proxy" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012141
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012142proxy-v2-options <option>[,<option>]*
12143 The "proxy-v2-options" parameter add option to send in PROXY protocol version
12144 2 when "send-proxy-v2" is used. Options available are "ssl" (see also
Emmanuel Hocdetfa8d0f12018-02-01 15:53:52 +010012145 send-proxy-v2-ssl), "cert-cn" (see also "send-proxy-v2-ssl-cn"), "ssl-cipher":
12146 name of the used cipher, "cert-sig": signature algorithm of the used
Emmanuel Hocdet253c3b72018-02-01 18:29:59 +010012147 certificate, "cert-key": key algorithm of the used certificate), "authority":
12148 host name value passed by the client (only sni from a tls connection is
Emmanuel Hocdet4399c752018-02-05 15:26:43 +010012149 supported), "crc32c": checksum of the proxy protocol v2 header.
Emmanuel Hocdetf643b802018-02-01 15:20:32 +010012150
David Safb76832014-05-08 23:42:08 -040012151send-proxy-v2-ssl
12152 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12153 2 over any connection established to this server. The PROXY protocol informs
12154 the other end about the layer 3/4 addresses of the incoming connection, so
12155 that it can know the client's address or the public address it accessed to,
12156 whatever the upper layer protocol. In addition, the SSL information extension
12157 of the PROXY protocol is added to the PROXY protocol header. This setting
12158 must not be used if the server isn't aware of this version of the protocol.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012159 See also the "no-send-proxy-v2-ssl" option of this section and the
12160 "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012161
12162send-proxy-v2-ssl-cn
12163 The "send-proxy-v2-ssl" parameter enforces use of the PROXY protocol version
12164 2 over any connection established to this server. The PROXY protocol informs
12165 the other end about the layer 3/4 addresses of the incoming connection, so
12166 that it can know the client's address or the public address it accessed to,
12167 whatever the upper layer protocol. In addition, the SSL information extension
12168 of the PROXY protocol, along along with the Common Name from the subject of
12169 the client certificate (if any), is added to the PROXY protocol header. This
12170 setting must not be used if the server isn't aware of this version of the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012171 protocol. See also the "no-send-proxy-v2-ssl-cn" option of this section and
12172 the "send-proxy-v2" option of the "bind" keyword.
David Safb76832014-05-08 23:42:08 -040012173
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012174slowstart <start_time_in_ms>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012175 The "slowstart" parameter for a server accepts a value in milliseconds which
12176 indicates after how long a server which has just come back up will run at
12177 full speed. Just as with every other time-based parameter, it can be entered
12178 in any other explicit unit among { us, ms, s, m, h, d }. The speed grows
12179 linearly from 0 to 100% during this time. The limitation applies to two
12180 parameters :
12181
12182 - maxconn: the number of connections accepted by the server will grow from 1
12183 to 100% of the usual dynamic limit defined by (minconn,maxconn,fullconn).
12184
12185 - weight: when the backend uses a dynamic weighted algorithm, the weight
12186 grows linearly from 1 to 100%. In this case, the weight is updated at every
12187 health-check. For this reason, it is important that the "inter" parameter
12188 is smaller than the "slowstart", in order to maximize the number of steps.
12189
12190 The slowstart never applies when haproxy starts, otherwise it would cause
12191 trouble to running servers. It only applies when a server has been previously
12192 seen as failed.
12193
Willy Tarreau732eac42015-07-09 11:40:25 +020012194sni <expression>
12195 The "sni" parameter evaluates the sample fetch expression, converts it to a
12196 string and uses the result as the host name sent in the SNI TLS extension to
12197 the server. A typical use case is to send the SNI received from the client in
12198 a bridged HTTPS scenario, using the "ssl_fc_sni" sample fetch for the
Willy Tarreau2ab88672017-07-05 18:23:03 +020012199 expression, though alternatives such as req.hdr(host) can also make sense. If
12200 "verify required" is set (which is the recommended setting), the resulting
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012201 name will also be matched against the server certificate's names. See the
Jérôme Magninb36a6d22018-12-09 16:03:40 +010012202 "verify" directive for more details. If you want to set a SNI for health
12203 checks, see the "check-sni" directive for more details.
Willy Tarreau732eac42015-07-09 11:40:25 +020012204
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012205source <addr>[:<pl>[-<ph>]] [usesrc { <addr2>[:<port2>] | client | clientip } ]
Willy Tarreaubce70882009-09-07 11:51:47 +020012206source <addr>[:<port>] [usesrc { <addr2>[:<port2>] | hdr_ip(<hdr>[,<occ>]) } ]
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012207source <addr>[:<pl>[-<ph>]] [interface <name>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012208 The "source" parameter sets the source address which will be used when
12209 connecting to the server. It follows the exact same parameters and principle
12210 as the backend "source" keyword, except that it only applies to the server
12211 referencing it. Please consult the "source" keyword for details.
12212
Willy Tarreauc6f4ce82009-06-10 11:09:37 +020012213 Additionally, the "source" statement on a server line allows one to specify a
12214 source port range by indicating the lower and higher bounds delimited by a
12215 dash ('-'). Some operating systems might require a valid IP address when a
12216 source port range is specified. It is permitted to have the same IP/range for
12217 several servers. Doing so makes it possible to bypass the maximum of 64k
12218 total concurrent connections. The limit will then reach 64k connections per
12219 server.
12220
Lukas Tribus7d56c6d2016-09-13 09:51:15 +000012221 Since Linux 4.2/libc 2.23 IP_BIND_ADDRESS_NO_PORT is set for connections
12222 specifying the source address without port(s).
12223
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012224ssl
Willy Tarreau44f65392013-06-25 07:56:20 +020012225 This option enables SSL ciphering on outgoing connections to the server. It
12226 is critical to verify server certificates using "verify" when using SSL to
12227 connect to servers, otherwise the communication is prone to trivial man in
12228 the-middle attacks rendering SSL useless. When this option is used, health
12229 checks are automatically sent in SSL too unless there is a "port" or an
12230 "addr" directive indicating the check should be sent to a different location.
Frédéric Lécailled2376272017-03-21 18:52:12 +010012231 See the "no-ssl" to disable "ssl" option and "check-ssl" option to force
12232 SSL health checks.
Willy Tarreau763a95b2012-10-04 23:15:39 +020012233
Emmanuel Hocdete1c722b2017-03-31 15:02:54 +020012234ssl-max-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12235 This option enforces use of <version> or lower when SSL is used to communicate
12236 with the server. This option is also available on global statement
12237 "ssl-default-server-options". See also "ssl-min-ver".
12238
12239ssl-min-ver [ SSLv3 | TLSv1.0 | TLSv1.1 | TLSv1.2 | TLSv1.3 ]
12240 This option enforces use of <version> or upper when SSL is used to communicate
12241 with the server. This option is also available on global statement
12242 "ssl-default-server-options". See also "ssl-max-ver".
12243
Frédéric Lécailled2376272017-03-21 18:52:12 +010012244ssl-reuse
12245 This option may be used as "server" setting to reset any "no-ssl-reuse"
12246 setting which would have been inherited from "default-server" directive as
12247 default value.
12248 It may also be used as "default-server" setting to reset any previous
12249 "default-server" "no-ssl-reuse" setting.
12250
12251stick
12252 This option may be used as "server" setting to reset any "non-stick"
12253 setting which would have been inherited from "default-server" directive as
12254 default value.
12255 It may also be used as "default-server" setting to reset any previous
12256 "default-server" "non-stick" setting.
Willy Tarreaua0ee1d02012-09-10 09:01:23 +020012257
Willy Tarreau163d4622015-10-13 16:16:41 +020012258tcp-ut <delay>
12259 Sets the TCP User Timeout for all outgoing connections to this server. This
12260 option is available on Linux since version 2.6.37. It allows haproxy to
12261 configure a timeout for sockets which contain data not receiving an
Davor Ocelice9ed2812017-12-25 17:49:28 +010012262 acknowledgment for the configured delay. This is especially useful on
Willy Tarreau163d4622015-10-13 16:16:41 +020012263 long-lived connections experiencing long idle periods such as remote
12264 terminals or database connection pools, where the client and server timeouts
12265 must remain high to allow a long period of idle, but where it is important to
12266 detect that the server has disappeared in order to release all resources
12267 associated with its connection (and the client's session). One typical use
12268 case is also to force dead server connections to die when health checks are
12269 too slow or during a soft reload since health checks are then disabled. The
12270 argument is a delay expressed in milliseconds by default. This only works for
12271 regular TCP connections, and is ignored for other protocols.
12272
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012273track [<proxy>/]<server>
Willy Tarreau32091232014-05-16 13:52:00 +020012274 This option enables ability to set the current state of the server by tracking
12275 another one. It is possible to track a server which itself tracks another
12276 server, provided that at the end of the chain, a server has health checks
12277 enabled. If <proxy> is omitted the current one is used. If disable-on-404 is
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012278 used, it has to be enabled on both proxies.
12279
Frédéric Lécailled2376272017-03-21 18:52:12 +010012280tls-tickets
12281 This option may be used as "server" setting to reset any "no-tls-tickets"
12282 setting which would have been inherited from "default-server" directive as
12283 default value.
12284 It may also be used as "default-server" setting to reset any previous
12285 "default-server" "no-tlsv-tickets" setting.
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012286
Emeric Brunef42d922012-10-11 16:11:36 +020012287verify [none|required]
12288 This setting is only available when support for OpenSSL was built in. If set
Emeric Brun850efd52014-01-29 12:24:34 +010012289 to 'none', server certificate is not verified. In the other case, The
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012290 certificate provided by the server is verified using CAs from 'ca-file' and
12291 optional CRLs from 'crl-file' after having checked that the names provided in
Davor Ocelice9ed2812017-12-25 17:49:28 +010012292 the certificate's subject and subjectAlternateNames attributes match either
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012293 the name passed using the "sni" directive, or if not provided, the static
12294 host name passed using the "verifyhost" directive. When no name is found, the
12295 certificate's names are ignored. For this reason, without SNI it's important
12296 to use "verifyhost". On verification failure the handshake is aborted. It is
12297 critically important to verify server certificates when using SSL to connect
12298 to servers, otherwise the communication is prone to trivial man-in-the-middle
12299 attacks rendering SSL totally useless. Unless "ssl_server_verify" appears in
12300 the global section, "verify" is set to "required" by default.
Emeric Brunef42d922012-10-11 16:11:36 +020012301
Evan Broderbe554312013-06-27 00:05:25 -070012302verifyhost <hostname>
12303 This setting is only available when support for OpenSSL was built in, and
Willy Tarreauad92a9a2017-07-28 11:38:41 +020012304 only takes effect if 'verify required' is also specified. This directive sets
12305 a default static hostname to check the server's certificate against when no
12306 SNI was used to connect to the server. If SNI is not used, this is the only
12307 way to enable hostname verification. This static hostname, when set, will
12308 also be used for health checks (which cannot provide an SNI value). If none
12309 of the hostnames in the certificate match the specified hostname, the
12310 handshake is aborted. The hostnames in the server-provided certificate may
12311 include wildcards. See also "verify", "sni" and "no-verifyhost" options.
Evan Broderbe554312013-06-27 00:05:25 -070012312
Krzysztof Piotr Oledzkic53601c2010-01-06 10:50:42 +010012313weight <weight>
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012314 The "weight" parameter is used to adjust the server's weight relative to
12315 other servers. All servers will receive a load proportional to their weight
12316 relative to the sum of all weights, so the higher the weight, the higher the
Willy Tarreau6704d672009-06-15 10:56:05 +020012317 load. The default weight is 1, and the maximal value is 256. A value of 0
12318 means the server will not participate in load-balancing but will still accept
12319 persistent connections. If this parameter is used to distribute the load
12320 according to server's capacity, it is recommended to start with values which
12321 can both grow and shrink, for instance between 10 and 100 to leave enough
12322 room above and below for later adjustments.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012323
12324
Cyril Bonté46175dd2015-07-02 22:45:32 +0200123255.3. Server IP address resolution using DNS
12326-------------------------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012327
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012328HAProxy allows using a host name on the server line to retrieve its IP address
12329using name servers. By default, HAProxy resolves the name when parsing the
12330configuration file, at startup and cache the result for the process' life.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012331This is not sufficient in some cases, such as in Amazon where a server's IP
12332can change after a reboot or an ELB Virtual IP can change based on current
12333workload.
12334This chapter describes how HAProxy can be configured to process server's name
12335resolution at run time.
12336Whether run time server name resolution has been enable or not, HAProxy will
12337carry on doing the first resolution when parsing the configuration.
12338
12339
Cyril Bonté46175dd2015-07-02 22:45:32 +0200123405.3.1. Global overview
12341----------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012342
12343As we've seen in introduction, name resolution in HAProxy occurs at two
12344different steps of the process life:
12345
12346 1. when starting up, HAProxy parses the server line definition and matches a
12347 host name. It uses libc functions to get the host name resolved. This
12348 resolution relies on /etc/resolv.conf file.
12349
Christopher Faulet67957bd2017-09-27 11:00:59 +020012350 2. at run time, HAProxy performs periodically name resolutions for servers
12351 requiring DNS resolutions.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012352
12353A few other events can trigger a name resolution at run time:
12354 - when a server's health check ends up in a connection timeout: this may be
12355 because the server has a new IP address. So we need to trigger a name
12356 resolution to know this new IP.
12357
Christopher Faulet67957bd2017-09-27 11:00:59 +020012358When using resolvers, the server name can either be a hostname, or a SRV label.
Davor Ocelice9ed2812017-12-25 17:49:28 +010012359HAProxy considers anything that starts with an underscore as a SRV label. If a
Christopher Faulet67957bd2017-09-27 11:00:59 +020012360SRV label is specified, then the corresponding SRV records will be retrieved
12361from the DNS server, and the provided hostnames will be used. The SRV label
12362will be checked periodically, and if any server are added or removed, haproxy
12363will automatically do the same.
Olivier Houchardecfa18d2017-08-07 17:30:03 +020012364
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012365A few things important to notice:
12366 - all the name servers are queried in the mean time. HAProxy will process the
12367 first valid response.
12368
12369 - a resolution is considered as invalid (NX, timeout, refused), when all the
12370 servers return an error.
12371
12372
Cyril Bonté46175dd2015-07-02 22:45:32 +0200123735.3.2. The resolvers section
12374----------------------------
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012375
12376This section is dedicated to host information related to name resolution in
Christopher Faulet67957bd2017-09-27 11:00:59 +020012377HAProxy. There can be as many as resolvers section as needed. Each section can
12378contain many name servers.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012379
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012380When multiple name servers are configured in a resolvers section, then HAProxy
12381uses the first valid response. In case of invalid responses, only the last one
12382is treated. Purpose is to give the chance to a slow server to deliver a valid
12383answer after a fast faulty or outdated server.
12384
12385When each server returns a different error type, then only the last error is
Christopher Faulet67957bd2017-09-27 11:00:59 +020012386used by HAProxy. The following processing is applied on this error:
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012387
Christopher Faulet67957bd2017-09-27 11:00:59 +020012388 1. HAProxy retries the same DNS query with a new query type. The A queries are
12389 switch to AAAA or the opposite. SRV queries are not concerned here. Timeout
12390 errors are also excluded.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012391
Christopher Faulet67957bd2017-09-27 11:00:59 +020012392 2. When the fallback on the query type was done (or not applicable), HAProxy
12393 retries the original DNS query, with the preferred query type.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012394
Christopher Faulet67957bd2017-09-27 11:00:59 +020012395 3. HAProxy retries previous steps <resolve_retires> times. If no valid
12396 response is received after that, it stops the DNS resolution and reports
12397 the error.
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012398
Christopher Faulet67957bd2017-09-27 11:00:59 +020012399For example, with 2 name servers configured in a resolvers section, the
12400following scenarios are possible:
12401
12402 - First response is valid and is applied directly, second response is
12403 ignored
12404
12405 - First response is invalid and second one is valid, then second response is
12406 applied
12407
12408 - First response is a NX domain and second one a truncated response, then
12409 HAProxy retries the query with a new type
12410
12411 - First response is a NX domain and second one is a timeout, then HAProxy
12412 retries the query with a new type
12413
12414 - Query timed out for both name servers, then HAProxy retries it with the
12415 same query type
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012416
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012417As a DNS server may not answer all the IPs in one DNS request, haproxy keeps
12418a cache of previous answers, an answer will be considered obsolete after
Christopher Faulet67957bd2017-09-27 11:00:59 +020012419<hold obsolete> seconds without the IP returned.
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012420
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012421
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012422resolvers <resolvers id>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012423 Creates a new name server list labeled <resolvers id>
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012424
12425A resolvers section accept the following parameters:
12426
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012427accepted_payload_size <nb>
Davor Ocelice9ed2812017-12-25 17:49:28 +010012428 Defines the maximum payload size accepted by HAProxy and announced to all the
Christopher Faulet67957bd2017-09-27 11:00:59 +020012429 name servers configured in this resolvers section.
Baptiste Assmann2af08fe2017-08-14 00:13:01 +020012430 <nb> is in bytes. If not set, HAProxy announces 512. (minimal value defined
12431 by RFC 6891)
12432
Baptiste Assmann9d8dbbc2017-08-18 23:35:08 +020012433 Note: the maximum allowed value is 8192.
12434
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012435nameserver <id> <ip>:<port>
12436 DNS server description:
12437 <id> : label of the server, should be unique
12438 <ip> : IP address of the server
12439 <port> : port where the DNS service actually runs
12440
Ben Draut44e609b2018-05-29 15:40:08 -060012441parse-resolv-conf
12442 Adds all nameservers found in /etc/resolv.conf to this resolvers nameservers
12443 list. Ordered as if each nameserver in /etc/resolv.conf was individually
12444 placed in the resolvers section in place of this directive.
12445
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012446hold <status> <period>
12447 Defines <period> during which the last name resolution should be kept based
12448 on last resolution <status>
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012449 <status> : last name resolution status. Acceptable values are "nx",
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012450 "other", "refused", "timeout", "valid", "obsolete".
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012451 <period> : interval between two successive name resolution when the last
12452 answer was in <status>. It follows the HAProxy time format.
12453 <period> is in milliseconds by default.
12454
Baptiste Assmann686408b2017-08-18 10:15:42 +020012455 Default value is 10s for "valid", 0s for "obsolete" and 30s for others.
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012456
Christopher Faulet67957bd2017-09-27 11:00:59 +020012457resolution_pool_size <nb> (deprecated)
Baptiste Assmann201c07f2017-05-22 15:17:15 +020012458 Defines the number of resolutions available in the pool for this resolvers.
12459 If not defines, it defaults to 64. If your configuration requires more than
12460 <nb>, then HAProxy will return an error when parsing the configuration.
12461
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012462resolve_retries <nb>
12463 Defines the number <nb> of queries to send to resolve a server name before
12464 giving up.
12465 Default value: 3
12466
Baptiste Assmann62b75b42015-09-09 01:11:36 +020012467 A retry occurs on name server timeout or when the full sequence of DNS query
12468 type failover is over and we need to start up from the default ANY query
12469 type.
12470
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012471timeout <event> <time>
12472 Defines timeouts related to name resolution
12473 <event> : the event on which the <time> timeout period applies to.
12474 events available are:
Frédéric Lécaille93d33162019-03-06 09:35:59 +010012475 - resolve : default time to trigger name resolutions when no
12476 other time applied.
Christopher Faulet67957bd2017-09-27 11:00:59 +020012477 Default value: 1s
12478 - retry : time between two DNS queries, when no valid response
Frédéric Lécaille93d33162019-03-06 09:35:59 +010012479 have been received.
Christopher Faulet67957bd2017-09-27 11:00:59 +020012480 Default value: 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012481 <time> : time related to the event. It follows the HAProxy time format.
12482 <time> is expressed in milliseconds.
12483
Olivier Doucetaa1ea8a2016-08-05 17:15:20 +020012484 Example:
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012485
12486 resolvers mydns
12487 nameserver dns1 10.0.0.1:53
12488 nameserver dns2 10.0.0.2:53
Ben Draut44e609b2018-05-29 15:40:08 -060012489 parse-resolv-conf
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012490 resolve_retries 3
Christopher Faulet67957bd2017-09-27 11:00:59 +020012491 timeout resolve 1s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012492 timeout retry 1s
Baptiste Assmann987e16d2016-11-02 22:23:31 +010012493 hold other 30s
12494 hold refused 30s
12495 hold nx 30s
12496 hold timeout 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012497 hold valid 10s
Olivier Houcharda8c6db82017-07-06 18:46:47 +020012498 hold obsolete 30s
Baptiste Assmann1fa66662015-04-14 00:28:47 +020012499
12500
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200125016. HTTP header manipulation
12502---------------------------
12503
12504In HTTP mode, it is possible to rewrite, add or delete some of the request and
12505response headers based on regular expressions. It is also possible to block a
12506request or a response if a particular header matches a regular expression,
12507which is enough to stop most elementary protocol attacks, and to protect
Willy Tarreau70dffda2014-01-30 03:07:23 +010012508against information leak from the internal network.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012509
Willy Tarreau70dffda2014-01-30 03:07:23 +010012510If HAProxy encounters an "Informational Response" (status code 1xx), it is able
12511to process all rsp* rules which can allow, deny, rewrite or delete a header,
12512but it will refuse to add a header to any such messages as this is not
12513HTTP-compliant. The reason for still processing headers in such responses is to
12514stop and/or fix any possible information leak which may happen, for instance
12515because another downstream equipment would unconditionally add a header, or if
12516a server name appears there. When such messages are seen, normal processing
12517still occurs on the next non-informational messages.
Willy Tarreau816b9792009-09-15 21:25:21 +020012518
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012519This section covers common usage of the following keywords, described in detail
12520in section 4.2 :
12521
12522 - reqadd <string>
12523 - reqallow <search>
12524 - reqiallow <search>
12525 - reqdel <search>
12526 - reqidel <search>
12527 - reqdeny <search>
12528 - reqideny <search>
12529 - reqpass <search>
12530 - reqipass <search>
12531 - reqrep <search> <replace>
12532 - reqirep <search> <replace>
12533 - reqtarpit <search>
12534 - reqitarpit <search>
12535 - rspadd <string>
12536 - rspdel <search>
12537 - rspidel <search>
12538 - rspdeny <search>
12539 - rspideny <search>
12540 - rsprep <search> <replace>
12541 - rspirep <search> <replace>
12542
12543With all these keywords, the same conventions are used. The <search> parameter
12544is a POSIX extended regular expression (regex) which supports grouping through
12545parenthesis (without the backslash). Spaces and other delimiters must be
12546prefixed with a backslash ('\') to avoid confusion with a field delimiter.
12547Other characters may be prefixed with a backslash to change their meaning :
12548
12549 \t for a tab
12550 \r for a carriage return (CR)
12551 \n for a new line (LF)
12552 \ to mark a space and differentiate it from a delimiter
12553 \# to mark a sharp and differentiate it from a comment
12554 \\ to use a backslash in a regex
12555 \\\\ to use a backslash in the text (*2 for regex, *2 for haproxy)
12556 \xXX to write the ASCII hex code XX as in the C language
12557
12558The <replace> parameter contains the string to be used to replace the largest
12559portion of text matching the regex. It can make use of the special characters
12560above, and can reference a substring which is delimited by parenthesis in the
12561regex, by writing a backslash ('\') immediately followed by one digit from 0 to
125629 indicating the group position (0 designating the entire line). This practice
12563is very common to users of the "sed" program.
12564
12565The <string> parameter represents the string which will systematically be added
12566after the last header line. It can also use special character sequences above.
12567
12568Notes related to these keywords :
12569---------------------------------
12570 - these keywords are not always convenient to allow/deny based on header
12571 contents. It is strongly recommended to use ACLs with the "block" keyword
12572 instead, resulting in far more flexible and manageable rules.
12573
12574 - lines are always considered as a whole. It is not possible to reference
12575 a header name only or a value only. This is important because of the way
12576 headers are written (notably the number of spaces after the colon).
12577
12578 - the first line is always considered as a header, which makes it possible to
12579 rewrite or filter HTTP requests URIs or response codes, but in turn makes
12580 it harder to distinguish between headers and request line. The regex prefix
12581 ^[^\ \t]*[\ \t] matches any HTTP method followed by a space, and the prefix
12582 ^[^ \t:]*: matches any header name followed by a colon.
12583
12584 - for performances reasons, the number of characters added to a request or to
12585 a response is limited at build time to values between 1 and 4 kB. This
12586 should normally be far more than enough for most usages. If it is too short
12587 on occasional usages, it is possible to gain some space by removing some
12588 useless headers before adding new ones.
12589
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010012590 - keywords beginning with "reqi" and "rspi" are the same as their counterpart
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012591 without the 'i' letter except that they ignore case when matching patterns.
12592
12593 - when a request passes through a frontend then a backend, all req* rules
12594 from the frontend will be evaluated, then all req* rules from the backend
12595 will be evaluated. The reverse path is applied to responses.
12596
12597 - req* statements are applied after "block" statements, so that "block" is
12598 always the first one, but before "use_backend" in order to permit rewriting
Willy Tarreaud72758d2010-01-12 10:42:19 +010012599 before switching.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012600
12601
Willy Tarreau74ca5042013-06-11 23:12:07 +0200126027. Using ACLs and fetching samples
12603----------------------------------
12604
Davor Ocelice9ed2812017-12-25 17:49:28 +010012605HAProxy is capable of extracting data from request or response streams, from
Willy Tarreau74ca5042013-06-11 23:12:07 +020012606client or server information, from tables, environmental information etc...
12607The action of extracting such data is called fetching a sample. Once retrieved,
12608these samples may be used for various purposes such as a key to a stick-table,
12609but most common usages consist in matching them against predefined constant
12610data called patterns.
12611
12612
126137.1. ACL basics
12614---------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012615
12616The use of Access Control Lists (ACL) provides a flexible solution to perform
12617content switching and generally to take decisions based on content extracted
12618from the request, the response or any environmental status. The principle is
12619simple :
12620
Willy Tarreau74ca5042013-06-11 23:12:07 +020012621 - extract a data sample from a stream, table or the environment
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012622 - optionally apply some format conversion to the extracted sample
Willy Tarreau74ca5042013-06-11 23:12:07 +020012623 - apply one or multiple pattern matching methods on this sample
12624 - perform actions only when a pattern matches the sample
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012625
Willy Tarreau74ca5042013-06-11 23:12:07 +020012626The actions generally consist in blocking a request, selecting a backend, or
12627adding a header.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012628
12629In order to define a test, the "acl" keyword is used. The syntax is :
12630
Willy Tarreau74ca5042013-06-11 23:12:07 +020012631 acl <aclname> <criterion> [flags] [operator] [<value>] ...
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012632
12633This creates a new ACL <aclname> or completes an existing one with new tests.
12634Those tests apply to the portion of request/response specified in <criterion>
12635and may be adjusted with optional flags [flags]. Some criteria also support
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012636an operator which may be specified before the set of values. Optionally some
12637conversion operators may be applied to the sample, and they will be specified
12638as a comma-delimited list of keywords just after the first keyword. The values
12639are of the type supported by the criterion, and are separated by spaces.
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012640
12641ACL names must be formed from upper and lower case letters, digits, '-' (dash),
12642'_' (underscore) , '.' (dot) and ':' (colon). ACL names are case-sensitive,
12643which means that "my_acl" and "My_Acl" are two different ACLs.
12644
12645There is no enforced limit to the number of ACLs. The unused ones do not affect
12646performance, they just consume a small amount of memory.
12647
Willy Tarreau74ca5042013-06-11 23:12:07 +020012648The criterion generally is the name of a sample fetch method, or one of its ACL
12649specific declinations. The default test method is implied by the output type of
12650this sample fetch method. The ACL declinations can describe alternate matching
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012651methods of a same sample fetch method. The sample fetch methods are the only
12652ones supporting a conversion.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012653
12654Sample fetch methods return data which can be of the following types :
12655 - boolean
12656 - integer (signed or unsigned)
12657 - IPv4 or IPv6 address
12658 - string
12659 - data block
12660
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012661Converters transform any of these data into any of these. For example, some
12662converters might convert a string to a lower-case string while other ones
12663would turn a string to an IPv4 address, or apply a netmask to an IP address.
12664The resulting sample is of the type of the last converter applied to the list,
12665which defaults to the type of the sample fetch method.
12666
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012667Each sample or converter returns data of a specific type, specified with its
12668keyword in this documentation. When an ACL is declared using a standard sample
12669fetch method, certain types automatically involved a default matching method
12670which are summarized in the table below :
12671
12672 +---------------------+-----------------+
12673 | Sample or converter | Default |
12674 | output type | matching method |
12675 +---------------------+-----------------+
12676 | boolean | bool |
12677 +---------------------+-----------------+
12678 | integer | int |
12679 +---------------------+-----------------+
12680 | ip | ip |
12681 +---------------------+-----------------+
12682 | string | str |
12683 +---------------------+-----------------+
12684 | binary | none, use "-m" |
12685 +---------------------+-----------------+
12686
12687Note that in order to match a binary samples, it is mandatory to specify a
12688matching method, see below.
12689
Willy Tarreau74ca5042013-06-11 23:12:07 +020012690The ACL engine can match these types against patterns of the following types :
12691 - boolean
12692 - integer or integer range
12693 - IP address / network
12694 - string (exact, substring, suffix, prefix, subdir, domain)
12695 - regular expression
12696 - hex block
12697
Willy Tarreauc57f0e22009-05-10 13:12:33 +020012698The following ACL flags are currently supported :
12699
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012700 -i : ignore case during matching of all subsequent patterns.
12701 -f : load patterns from a file.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012702 -m : use a specific pattern matching method
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012703 -n : forbid the DNS resolutions
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012704 -M : load the file pointed by -f like a map file.
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012705 -u : force the unique id of the ACL
Willy Tarreau6a06a402007-07-15 20:15:28 +020012706 -- : force end of flags. Useful when a string looks like one of the flags.
12707
Willy Tarreau74ca5042013-06-11 23:12:07 +020012708The "-f" flag is followed by the name of a file from which all lines will be
12709read as individual values. It is even possible to pass multiple "-f" arguments
12710if the patterns are to be loaded from multiple files. Empty lines as well as
12711lines beginning with a sharp ('#') will be ignored. All leading spaces and tabs
12712will be stripped. If it is absolutely necessary to insert a valid pattern
12713beginning with a sharp, just prefix it with a space so that it is not taken for
12714a comment. Depending on the data type and match method, haproxy may load the
12715lines into a binary tree, allowing very fast lookups. This is true for IPv4 and
12716exact string matching. In this case, duplicates will automatically be removed.
12717
Thierry FOURNIER9860c412014-01-29 14:23:29 +010012718The "-M" flag allows an ACL to use a map file. If this flag is set, the file is
12719parsed as two column file. The first column contains the patterns used by the
12720ACL, and the second column contain the samples. The sample can be used later by
12721a map. This can be useful in some rare cases where an ACL would just be used to
12722check for the existence of a pattern in a map before a mapping is applied.
12723
Thierry FOURNIER3534d882014-01-20 17:01:44 +010012724The "-u" flag forces the unique id of the ACL. This unique id is used with the
12725socket interface to identify ACL and dynamically change its values. Note that a
12726file is always identified by its name even if an id is set.
12727
Willy Tarreau74ca5042013-06-11 23:12:07 +020012728Also, note that the "-i" flag applies to subsequent entries and not to entries
12729loaded from files preceding it. For instance :
12730
12731 acl valid-ua hdr(user-agent) -f exact-ua.lst -i -f generic-ua.lst test
12732
12733In this example, each line of "exact-ua.lst" will be exactly matched against
12734the "user-agent" header of the request. Then each line of "generic-ua" will be
12735case-insensitively matched. Then the word "test" will be insensitively matched
12736as well.
12737
12738The "-m" flag is used to select a specific pattern matching method on the input
12739sample. All ACL-specific criteria imply a pattern matching method and generally
12740do not need this flag. However, this flag is useful with generic sample fetch
12741methods to describe how they're going to be matched against the patterns. This
12742is required for sample fetches which return data type for which there is no
Davor Ocelice9ed2812017-12-25 17:49:28 +010012743obvious matching method (e.g. string or binary). When "-m" is specified and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012744followed by a pattern matching method name, this method is used instead of the
12745default one for the criterion. This makes it possible to match contents in ways
12746that were not initially planned, or with sample fetch methods which return a
12747string. The matching method also affects the way the patterns are parsed.
12748
Thierry FOURNIERb7729c92014-02-11 16:24:41 +010012749The "-n" flag forbids the dns resolutions. It is used with the load of ip files.
12750By default, if the parser cannot parse ip address it considers that the parsed
12751string is maybe a domain name and try dns resolution. The flag "-n" disable this
12752resolution. It is useful for detecting malformed ip lists. Note that if the DNS
12753server is not reachable, the haproxy configuration parsing may last many minutes
12754waiting fir the timeout. During this time no error messages are displayed. The
12755flag "-n" disable this behavior. Note also that during the runtime, this
12756function is disabled for the dynamic acl modifications.
12757
Willy Tarreau74ca5042013-06-11 23:12:07 +020012758There are some restrictions however. Not all methods can be used with all
12759sample fetch methods. Also, if "-m" is used in conjunction with "-f", it must
12760be placed first. The pattern matching method must be one of the following :
Willy Tarreau5adeda12013-03-31 22:13:34 +020012761
12762 - "found" : only check if the requested sample could be found in the stream,
12763 but do not compare it against any pattern. It is recommended not
Willy Tarreau74ca5042013-06-11 23:12:07 +020012764 to pass any pattern to avoid confusion. This matching method is
12765 particularly useful to detect presence of certain contents such
12766 as headers, cookies, etc... even if they are empty and without
12767 comparing them to anything nor counting them.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012768
12769 - "bool" : check the value as a boolean. It can only be applied to fetches
12770 which return a boolean or integer value, and takes no pattern.
Willy Tarreau74ca5042013-06-11 23:12:07 +020012771 Value zero or false does not match, all other values do match.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012772
12773 - "int" : match the value as an integer. It can be used with integer and
Willy Tarreau74ca5042013-06-11 23:12:07 +020012774 boolean samples. Boolean false is integer 0, true is integer 1.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012775
12776 - "ip" : match the value as an IPv4 or IPv6 address. It is compatible
Willy Tarreau74ca5042013-06-11 23:12:07 +020012777 with IP address samples only, so it is implied and never needed.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012778
Davor Ocelice9ed2812017-12-25 17:49:28 +010012779 - "bin" : match the contents against a hexadecimal string representing a
Willy Tarreau5adeda12013-03-31 22:13:34 +020012780 binary sequence. This may be used with binary or string samples.
12781
12782 - "len" : match the sample's length as an integer. This may be used with
12783 binary or string samples.
12784
Willy Tarreau74ca5042013-06-11 23:12:07 +020012785 - "str" : exact match : match the contents against a string. This may be
12786 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012787
Willy Tarreau74ca5042013-06-11 23:12:07 +020012788 - "sub" : substring match : check that the contents contain at least one of
12789 the provided string patterns. This may be used with binary or
12790 string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012791
Willy Tarreau74ca5042013-06-11 23:12:07 +020012792 - "reg" : regex match : match the contents against a list of regular
12793 expressions. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012794
Willy Tarreau74ca5042013-06-11 23:12:07 +020012795 - "beg" : prefix match : check that the contents begin like the provided
12796 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012797
Willy Tarreau74ca5042013-06-11 23:12:07 +020012798 - "end" : suffix match : check that the contents end like the provided
12799 string patterns. This may be used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012800
Willy Tarreau74ca5042013-06-11 23:12:07 +020012801 - "dir" : subdir match : check that a slash-delimited portion of the
12802 contents exactly matches one of the provided string patterns.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012803 This may be used with binary or string samples.
12804
Willy Tarreau74ca5042013-06-11 23:12:07 +020012805 - "dom" : domain match : check that a dot-delimited portion of the contents
12806 exactly match one of the provided string patterns. This may be
12807 used with binary or string samples.
Willy Tarreau5adeda12013-03-31 22:13:34 +020012808
12809For example, to quickly detect the presence of cookie "JSESSIONID" in an HTTP
12810request, it is possible to do :
12811
12812 acl jsess_present cook(JSESSIONID) -m found
12813
12814In order to apply a regular expression on the 500 first bytes of data in the
12815buffer, one would use the following acl :
12816
12817 acl script_tag payload(0,500) -m reg -i <script>
12818
Willy Tarreaue6b11e42013-11-26 19:02:32 +010012819On systems where the regex library is much slower when using "-i", it is
12820possible to convert the sample to lowercase before matching, like this :
12821
12822 acl script_tag payload(0,500),lower -m reg <script>
12823
Willy Tarreau74ca5042013-06-11 23:12:07 +020012824All ACL-specific criteria imply a default matching method. Most often, these
12825criteria are composed by concatenating the name of the original sample fetch
12826method and the matching method. For example, "hdr_beg" applies the "beg" match
12827to samples retrieved using the "hdr" fetch method. Since all ACL-specific
12828criteria rely on a sample fetch method, it is always possible instead to use
12829the original sample fetch method and the explicit matching method using "-m".
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012830
Willy Tarreau74ca5042013-06-11 23:12:07 +020012831If an alternate match is specified using "-m" on an ACL-specific criterion,
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030012832the matching method is simply applied to the underlying sample fetch method.
12833For example, all ACLs below are exact equivalent :
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012834
Willy Tarreau74ca5042013-06-11 23:12:07 +020012835 acl short_form hdr_beg(host) www.
12836 acl alternate1 hdr_beg(host) -m beg www.
12837 acl alternate2 hdr_dom(host) -m beg www.
12838 acl alternate3 hdr(host) -m beg www.
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012839
Willy Tarreau2b5285d2010-05-09 23:45:24 +020012840
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012841The table below summarizes the compatibility matrix between sample or converter
12842types and the pattern types to fetch against. It indicates for each compatible
12843combination the name of the matching method to be used, surrounded with angle
12844brackets ">" and "<" when the method is the default one and will work by
12845default without "-m".
Willy Tarreau0ba27502007-12-24 16:55:16 +010012846
Willy Tarreau74ca5042013-06-11 23:12:07 +020012847 +-------------------------------------------------+
12848 | Input sample type |
12849 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012850 | pattern type | boolean | integer | ip | string | binary |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012851 +----------------------+---------+---------+---------+---------+---------+
12852 | none (presence only) | found | found | found | found | found |
12853 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012854 | none (boolean value) |> bool <| bool | | bool | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012855 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012856 | integer (value) | int |> int <| int | int | |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012857 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012858 | integer (length) | len | len | len | len | len |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012859 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012860 | IP address | | |> ip <| ip | ip |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012861 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIER2a06e392014-05-11 15:49:55 +020012862 | exact string | str | str | str |> str <| str |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012863 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012864 | prefix | beg | beg | beg | beg | beg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012865 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012866 | suffix | end | end | end | end | end |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012867 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012868 | substring | sub | sub | sub | sub | sub |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012869 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012870 | subdir | dir | dir | dir | dir | dir |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012871 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012872 | domain | dom | dom | dom | dom | dom |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012873 +----------------------+---------+---------+---------+---------+---------+
Thierry FOURNIERe3ded592013-12-06 15:36:54 +010012874 | regex | reg | reg | reg | reg | reg |
Willy Tarreau74ca5042013-06-11 23:12:07 +020012875 +----------------------+---------+---------+---------+---------+---------+
12876 | hex block | | | | bin | bin |
12877 +----------------------+---------+---------+---------+---------+---------+
Willy Tarreau6a06a402007-07-15 20:15:28 +020012878
12879
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128807.1.1. Matching booleans
12881------------------------
12882
12883In order to match a boolean, no value is needed and all values are ignored.
12884Boolean matching is used by default for all fetch methods of type "boolean".
12885When boolean matching is used, the fetched value is returned as-is, which means
12886that a boolean "true" will always match and a boolean "false" will never match.
12887
12888Boolean matching may also be enforced using "-m bool" on fetch methods which
12889return an integer value. Then, integer value 0 is converted to the boolean
12890"false" and all other values are converted to "true".
12891
Willy Tarreau6a06a402007-07-15 20:15:28 +020012892
Willy Tarreau74ca5042013-06-11 23:12:07 +0200128937.1.2. Matching integers
12894------------------------
12895
12896Integer matching applies by default to integer fetch methods. It can also be
12897enforced on boolean fetches using "-m int". In this case, "false" is converted
12898to the integer 0, and "true" is converted to the integer 1.
12899
12900Integer matching also supports integer ranges and operators. Note that integer
12901matching only applies to positive values. A range is a value expressed with a
12902lower and an upper bound separated with a colon, both of which may be omitted.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012903
12904For instance, "1024:65535" is a valid range to represent a range of
12905unprivileged ports, and "1024:" would also work. "0:1023" is a valid
12906representation of privileged ports, and ":1023" would also work.
12907
Willy Tarreau62644772008-07-16 18:36:06 +020012908As a special case, some ACL functions support decimal numbers which are in fact
12909two integers separated by a dot. This is used with some version checks for
12910instance. All integer properties apply to those decimal numbers, including
12911ranges and operators.
12912
Willy Tarreau6a06a402007-07-15 20:15:28 +020012913For an easier usage, comparison operators are also supported. Note that using
Willy Tarreau0ba27502007-12-24 16:55:16 +010012914operators with ranges does not make much sense and is strongly discouraged.
12915Similarly, it does not make much sense to perform order comparisons with a set
12916of values.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012917
Willy Tarreau0ba27502007-12-24 16:55:16 +010012918Available operators for integer matching are :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012919
12920 eq : true if the tested value equals at least one value
12921 ge : true if the tested value is greater than or equal to at least one value
12922 gt : true if the tested value is greater than at least one value
12923 le : true if the tested value is less than or equal to at least one value
12924 lt : true if the tested value is less than at least one value
12925
Willy Tarreau0ba27502007-12-24 16:55:16 +010012926For instance, the following ACL matches any negative Content-Length header :
Willy Tarreau6a06a402007-07-15 20:15:28 +020012927
12928 acl negative-length hdr_val(content-length) lt 0
12929
Willy Tarreau62644772008-07-16 18:36:06 +020012930This one matches SSL versions between 3.0 and 3.1 (inclusive) :
12931
12932 acl sslv3 req_ssl_ver 3:3.1
12933
Willy Tarreau6a06a402007-07-15 20:15:28 +020012934
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129357.1.3. Matching strings
12936-----------------------
12937
12938String matching applies to string or binary fetch methods, and exists in 6
12939different forms :
12940
12941 - exact match (-m str) : the extracted string must exactly match the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012942 patterns;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012943
12944 - substring match (-m sub) : the patterns are looked up inside the
Davor Ocelice9ed2812017-12-25 17:49:28 +010012945 extracted string, and the ACL matches if any of them is found inside;
Willy Tarreau74ca5042013-06-11 23:12:07 +020012946
12947 - prefix match (-m beg) : the patterns are compared with the beginning of
12948 the extracted string, and the ACL matches if any of them matches.
12949
12950 - suffix match (-m end) : the patterns are compared with the end of the
12951 extracted string, and the ACL matches if any of them matches.
12952
Baptiste Assmann33db6002016-03-06 23:32:10 +010012953 - subdir match (-m dir) : the patterns are looked up inside the extracted
Willy Tarreau74ca5042013-06-11 23:12:07 +020012954 string, delimited with slashes ("/"), and the ACL matches if any of them
12955 matches.
12956
12957 - domain match (-m dom) : the patterns are looked up inside the extracted
12958 string, delimited with dots ("."), and the ACL matches if any of them
12959 matches.
Willy Tarreau6a06a402007-07-15 20:15:28 +020012960
12961String matching applies to verbatim strings as they are passed, with the
12962exception of the backslash ("\") which makes it possible to escape some
12963characters such as the space. If the "-i" flag is passed before the first
12964string, then the matching will be performed ignoring the case. In order
12965to match the string "-i", either set it second, or pass the "--" flag
Willy Tarreau0ba27502007-12-24 16:55:16 +010012966before the first string. Same applies of course to match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012967
12968
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129697.1.4. Matching regular expressions (regexes)
12970---------------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012971
12972Just like with string matching, regex matching applies to verbatim strings as
12973they are passed, with the exception of the backslash ("\") which makes it
12974possible to escape some characters such as the space. If the "-i" flag is
12975passed before the first regex, then the matching will be performed ignoring
12976the case. In order to match the string "-i", either set it second, or pass
Willy Tarreau0ba27502007-12-24 16:55:16 +010012977the "--" flag before the first string. Same principle applies of course to
12978match the string "--".
Willy Tarreau6a06a402007-07-15 20:15:28 +020012979
12980
Willy Tarreau74ca5042013-06-11 23:12:07 +0200129817.1.5. Matching arbitrary data blocks
12982-------------------------------------
12983
12984It is possible to match some extracted samples against a binary block which may
12985not safely be represented as a string. For this, the patterns must be passed as
12986a series of hexadecimal digits in an even number, when the match method is set
12987to binary. Each sequence of two digits will represent a byte. The hexadecimal
12988digits may be used upper or lower case.
12989
12990Example :
12991 # match "Hello\n" in the input stream (\x48 \x65 \x6c \x6c \x6f \x0a)
12992 acl hello payload(0,6) -m bin 48656c6c6f0a
12993
12994
129957.1.6. Matching IPv4 and IPv6 addresses
12996---------------------------------------
Willy Tarreau6a06a402007-07-15 20:15:28 +020012997
12998IPv4 addresses values can be specified either as plain addresses or with a
12999netmask appended, in which case the IPv4 address matches whenever it is
13000within the network. Plain addresses may also be replaced with a resolvable
Willy Tarreaud2a4aa22008-01-31 15:28:22 +010013001host name, but this practice is generally discouraged as it makes it more
Willy Tarreau0ba27502007-12-24 16:55:16 +010013002difficult to read and debug configurations. If hostnames are used, you should
13003at least ensure that they are present in /etc/hosts so that the configuration
13004does not depend on any random DNS match at the moment the configuration is
13005parsed.
Willy Tarreau6a06a402007-07-15 20:15:28 +020013006
Daniel Schnellereba56342016-04-13 00:26:52 +020013007The dotted IPv4 address notation is supported in both regular as well as the
13008abbreviated form with all-0-octets omitted:
13009
13010 +------------------+------------------+------------------+
13011 | Example 1 | Example 2 | Example 3 |
13012 +------------------+------------------+------------------+
13013 | 192.168.0.1 | 10.0.0.12 | 127.0.0.1 |
13014 | 192.168.1 | 10.12 | 127.1 |
13015 | 192.168.0.1/22 | 10.0.0.12/8 | 127.0.0.1/8 |
13016 | 192.168.1/22 | 10.12/8 | 127.1/8 |
13017 +------------------+------------------+------------------+
13018
13019Notice that this is different from RFC 4632 CIDR address notation in which
13020192.168.42/24 would be equivalent to 192.168.42.0/24.
13021
Willy Tarreauceb4ac92012-04-28 00:41:46 +020013022IPv6 may be entered in their usual form, with or without a netmask appended.
13023Only bit counts are accepted for IPv6 netmasks. In order to avoid any risk of
13024trouble with randomly resolved IP addresses, host names are never allowed in
13025IPv6 patterns.
13026
13027HAProxy is also able to match IPv4 addresses with IPv6 addresses in the
13028following situations :
13029 - tested address is IPv4, pattern address is IPv4, the match applies
13030 in IPv4 using the supplied mask if any.
13031 - tested address is IPv6, pattern address is IPv6, the match applies
13032 in IPv6 using the supplied mask if any.
13033 - tested address is IPv6, pattern address is IPv4, the match applies in IPv4
13034 using the pattern's mask if the IPv6 address matches with 2002:IPV4::,
13035 ::IPV4 or ::ffff:IPV4, otherwise it fails.
13036 - tested address is IPv4, pattern address is IPv6, the IPv4 address is first
13037 converted to IPv6 by prefixing ::ffff: in front of it, then the match is
13038 applied in IPv6 using the supplied IPv6 mask.
13039
Willy Tarreau74ca5042013-06-11 23:12:07 +020013040
130417.2. Using ACLs to form conditions
13042----------------------------------
13043
13044Some actions are only performed upon a valid condition. A condition is a
13045combination of ACLs with operators. 3 operators are supported :
13046
13047 - AND (implicit)
13048 - OR (explicit with the "or" keyword or the "||" operator)
13049 - Negation with the exclamation mark ("!")
Willy Tarreau6a06a402007-07-15 20:15:28 +020013050
Willy Tarreau74ca5042013-06-11 23:12:07 +020013051A condition is formed as a disjunctive form:
Willy Tarreau6a06a402007-07-15 20:15:28 +020013052
Willy Tarreau74ca5042013-06-11 23:12:07 +020013053 [!]acl1 [!]acl2 ... [!]acln { or [!]acl1 [!]acl2 ... [!]acln } ...
Willy Tarreaubef91e72013-03-31 23:14:46 +020013054
Willy Tarreau74ca5042013-06-11 23:12:07 +020013055Such conditions are generally used after an "if" or "unless" statement,
13056indicating when the condition will trigger the action.
Willy Tarreaubef91e72013-03-31 23:14:46 +020013057
Willy Tarreau74ca5042013-06-11 23:12:07 +020013058For instance, to block HTTP requests to the "*" URL with methods other than
13059"OPTIONS", as well as POST requests without content-length, and GET or HEAD
13060requests with a content-length greater than 0, and finally every request which
13061is not either GET/HEAD/POST/OPTIONS !
13062
13063 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013064 http-request deny if HTTP_URL_STAR !METH_OPTIONS || METH_POST missing_cl
13065 http-request deny if METH_GET HTTP_CONTENT
13066 http-request deny unless METH_GET or METH_POST or METH_OPTIONS
Willy Tarreau74ca5042013-06-11 23:12:07 +020013067
13068To select a different backend for requests to static contents on the "www" site
13069and to every request on the "img", "video", "download" and "ftp" hosts :
13070
13071 acl url_static path_beg /static /images /img /css
13072 acl url_static path_end .gif .png .jpg .css .js
13073 acl host_www hdr_beg(host) -i www
13074 acl host_static hdr_beg(host) -i img. video. download. ftp.
13075
Davor Ocelice9ed2812017-12-25 17:49:28 +010013076 # now use backend "static" for all static-only hosts, and for static URLs
Willy Tarreau74ca5042013-06-11 23:12:07 +020013077 # of host "www". Use backend "www" for the rest.
13078 use_backend static if host_static or host_www url_static
13079 use_backend www if host_www
13080
13081It is also possible to form rules using "anonymous ACLs". Those are unnamed ACL
13082expressions that are built on the fly without needing to be declared. They must
13083be enclosed between braces, with a space before and after each brace (because
13084the braces must be seen as independent words). Example :
13085
13086 The following rule :
13087
13088 acl missing_cl hdr_cnt(Content-length) eq 0
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013089 http-request deny if METH_POST missing_cl
Willy Tarreau74ca5042013-06-11 23:12:07 +020013090
13091 Can also be written that way :
13092
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013093 http-request deny if METH_POST { hdr_cnt(Content-length) eq 0 }
Willy Tarreau74ca5042013-06-11 23:12:07 +020013094
13095It is generally not recommended to use this construct because it's a lot easier
13096to leave errors in the configuration when written that way. However, for very
13097simple rules matching only one source IP address for instance, it can make more
13098sense to use them than to declare ACLs with random names. Another example of
13099good use is the following :
13100
13101 With named ACLs :
13102
13103 acl site_dead nbsrv(dynamic) lt 2
13104 acl site_dead nbsrv(static) lt 2
13105 monitor fail if site_dead
13106
13107 With anonymous ACLs :
13108
13109 monitor fail if { nbsrv(dynamic) lt 2 } || { nbsrv(static) lt 2 }
13110
Jarno Huuskonen84c51ec2017-04-03 14:20:34 +030013111See section 4.2 for detailed help on the "http-request deny" and "use_backend"
13112keywords.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013113
13114
131157.3. Fetching samples
13116---------------------
13117
13118Historically, sample fetch methods were only used to retrieve data to match
13119against patterns using ACLs. With the arrival of stick-tables, a new class of
13120sample fetch methods was created, most often sharing the same syntax as their
13121ACL counterpart. These sample fetch methods are also known as "fetches". As
13122of now, ACLs and fetches have converged. All ACL fetch methods have been made
13123available as fetch methods, and ACLs may use any sample fetch method as well.
13124
13125This section details all available sample fetch methods and their output type.
13126Some sample fetch methods have deprecated aliases that are used to maintain
13127compatibility with existing configurations. They are then explicitly marked as
13128deprecated and should not be used in new setups.
13129
13130The ACL derivatives are also indicated when available, with their respective
13131matching methods. These ones all have a well defined default pattern matching
13132method, so it is never necessary (though allowed) to pass the "-m" option to
13133indicate how the sample will be matched using ACLs.
13134
13135As indicated in the sample type versus matching compatibility matrix above,
13136when using a generic sample fetch method in an ACL, the "-m" option is
13137mandatory unless the sample type is one of boolean, integer, IPv4 or IPv6. When
13138the same keyword exists as an ACL keyword and as a standard fetch method, the
13139ACL engine will automatically pick the ACL-only one by default.
13140
13141Some of these keywords support one or multiple mandatory arguments, and one or
13142multiple optional arguments. These arguments are strongly typed and are checked
13143when the configuration is parsed so that there is no risk of running with an
Davor Ocelice9ed2812017-12-25 17:49:28 +010013144incorrect argument (e.g. an unresolved backend name). Fetch function arguments
13145are passed between parenthesis and are delimited by commas. When an argument
Willy Tarreau74ca5042013-06-11 23:12:07 +020013146is optional, it will be indicated below between square brackets ('[ ]'). When
13147all arguments are optional, the parenthesis may be omitted.
13148
13149Thus, the syntax of a standard sample fetch method is one of the following :
13150 - name
13151 - name(arg1)
13152 - name(arg1,arg2)
13153
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013154
131557.3.1. Converters
13156-----------------
13157
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013158Sample fetch methods may be combined with transformations to be applied on top
13159of the fetched sample (also called "converters"). These combinations form what
13160is called "sample expressions" and the result is a "sample". Initially this
13161was only supported by "stick on" and "stick store-request" directives but this
Davor Ocelice9ed2812017-12-25 17:49:28 +010013162has now be extended to all places where samples may be used (ACLs, log-format,
Willy Tarreaue6b11e42013-11-26 19:02:32 +010013163unique-id-format, add-header, ...).
13164
13165These transformations are enumerated as a series of specific keywords after the
13166sample fetch method. These keywords may equally be appended immediately after
13167the fetch keyword's argument, delimited by a comma. These keywords can also
Davor Ocelice9ed2812017-12-25 17:49:28 +010013168support some arguments (e.g. a netmask) which must be passed in parenthesis.
Willy Tarreau0ba27502007-12-24 16:55:16 +010013169
Willy Tarreau97707872015-01-27 15:12:13 +010013170A certain category of converters are bitwise and arithmetic operators which
13171support performing basic operations on integers. Some bitwise operations are
13172supported (and, or, xor, cpl) and some arithmetic operations are supported
13173(add, sub, mul, div, mod, neg). Some comparators are provided (odd, even, not,
13174bool) which make it possible to report a match without having to write an ACL.
13175
Willy Tarreau74ca5042013-06-11 23:12:07 +020013176The currently available list of transformation keywords include :
Willy Tarreau0ba27502007-12-24 16:55:16 +010013177
Ben Shillitof25e8e52016-12-02 14:25:37 +00001317851d.single(<prop>[,<prop>*])
13179 Returns values for the properties requested as a string, where values are
13180 separated by the delimiter specified with "51degrees-property-separator".
13181 The device is identified using the User-Agent header passed to the
13182 converter. The function can be passed up to five property names, and if a
13183 property name can't be found, the value "NoData" is returned.
13184
13185 Example :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013186 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request,
13187 # containing values for the three properties requested by using the
Ben Shillitof25e8e52016-12-02 14:25:37 +000013188 # User-Agent passed to the converter.
13189 frontend http-in
13190 bind *:8081
13191 default_backend servers
13192 http-request set-header X-51D-DeviceTypeMobileTablet \
13193 %[req.fhdr(User-Agent),51d.single(DeviceType,IsMobile,IsTablet)]
13194
Willy Tarreau97707872015-01-27 15:12:13 +010013195add(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013196 Adds <value> to the input value of type signed integer, and returns the
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013197 result as a signed integer. <value> can be a numeric value or a variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013198 name. The name of the variable starts with an indication about its scope. The
13199 scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013200 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013201 "sess" : the variable is shared with the whole session
13202 "txn" : the variable is shared with the transaction (request and response)
13203 "req" : the variable is shared only during request processing
13204 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013205 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013206 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013207
13208and(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013209 Performs a bitwise "AND" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013210 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013211 numeric value or a variable name. The name of the variable starts with an
13212 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013213 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013214 "sess" : the variable is shared with the whole session
13215 "txn" : the variable is shared with the transaction (request and response)
13216 "req" : the variable is shared only during request processing
13217 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013218 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013219 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013220
Holger Just1bfc24b2017-05-06 00:56:53 +020013221b64dec
13222 Converts (decodes) a base64 encoded input string to its binary
13223 representation. It performs the inverse operation of base64().
13224
Emeric Brun53d1a982014-04-30 18:21:37 +020013225base64
13226 Converts a binary input sample to a base64 string. It is used to log or
Davor Ocelice9ed2812017-12-25 17:49:28 +010013227 transfer binary content in a way that can be reliably transferred (e.g.
Emeric Brun53d1a982014-04-30 18:21:37 +020013228 an SSL ID can be copied in a header).
13229
Willy Tarreau97707872015-01-27 15:12:13 +010013230bool
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013231 Returns a boolean TRUE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013232 non-null, otherwise returns FALSE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013233 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013234 presence of a flag).
13235
Emeric Brun54c4ac82014-11-03 15:32:43 +010013236bytes(<offset>[,<length>])
13237 Extracts some bytes from an input binary sample. The result is a binary
13238 sample starting at an offset (in bytes) of the original sample and
Tim Düsterhus4896c442016-11-29 02:15:19 +010013239 optionally truncated at the given length.
Emeric Brun54c4ac82014-11-03 15:32:43 +010013240
Willy Tarreau280f42b2018-02-19 15:34:12 +010013241concat([<start>],[<var>],[<end>])
13242 Concatenates up to 3 fields after the current sample which is then turned to
13243 a string. The first one, <start>, is a constant string, that will be appended
13244 immediately after the existing sample. It may be omitted if not used. The
13245 second one, <var>, is a variable name. The variable will be looked up, its
13246 contents converted to a string, and it will be appended immediately after the
13247 <first> part. If the variable is not found, nothing is appended. It may be
13248 omitted as well. The third field, <end> is a constant string that will be
13249 appended after the variable. It may also be omitted. Together, these elements
13250 allow to concatenate variables with delimiters to an existing set of
13251 variables. This can be used to build new variables made of a succession of
13252 other variables, such as colon-delimited varlues. Note that due to the config
13253 parser, it is not possible to use a comma nor a closing parenthesis as
13254 delimitors.
13255
13256 Example:
13257 tcp-request session set-var(sess.src) src
13258 tcp-request session set-var(sess.dn) ssl_c_s_dn
13259 tcp-request session set-var(txn.sig) str(),concat(<ip=,sess.ip,>),concat(<dn=,sess.dn,>)
13260 http-request set-header x-hap-sig %[var(txn.sig)]
13261
Willy Tarreau97707872015-01-27 15:12:13 +010013262cpl
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013263 Takes the input value of type signed integer, applies a ones-complement
13264 (flips all bits) and returns the result as an signed integer.
Willy Tarreau97707872015-01-27 15:12:13 +010013265
Willy Tarreau80599772015-01-20 19:35:24 +010013266crc32([<avalanche>])
13267 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32
13268 hash function. Optionally, it is possible to apply a full avalanche hash
13269 function to the output if the optional <avalanche> argument equals 1. This
13270 converter uses the same functions as used by the various hash-based load
13271 balancing algorithms, so it will provide exactly the same results. It is
13272 provided for compatibility with other software which want a CRC32 to be
13273 computed on some input keys, so it follows the most common implementation as
13274 found in Ethernet, Gzip, PNG, etc... It is slower than the other algorithms
13275 but may provide a better or at least less predictable distribution. It must
13276 not be used for security purposes as a 32-bit hash is trivial to break. See
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013277 also "djb2", "sdbm", "wt6", "crc32c" and the "hash-type" directive.
13278
13279crc32c([<avalanche>])
13280 Hashes a binary input sample into an unsigned 32-bit quantity using the CRC32C
13281 hash function. Optionally, it is possible to apply a full avalanche hash
13282 function to the output if the optional <avalanche> argument equals 1. This
13283 converter uses the same functions as described in RFC4960, Appendix B [8].
13284 It is provided for compatibility with other software which want a CRC32C to be
13285 computed on some input keys. It is slower than the other algorithms and it must
13286 not be used for security purposes as a 32-bit hash is trivial to break. See
13287 also "djb2", "sdbm", "wt6", "crc32" and the "hash-type" directive.
Willy Tarreau80599772015-01-20 19:35:24 +010013288
David Carlier29b3ca32015-09-25 14:09:21 +010013289da-csv-conv(<prop>[,<prop>*])
David Carlier4542b102015-06-01 13:54:29 +020013290 Asks the DeviceAtlas converter to identify the User Agent string passed on
13291 input, and to emit a string made of the concatenation of the properties
13292 enumerated in argument, delimited by the separator defined by the global
13293 keyword "deviceatlas-property-separator", or by default the pipe character
David Carlier840b0242016-03-16 10:09:55 +000013294 ('|'). There's a limit of 12 different properties imposed by the haproxy
David Carlier4542b102015-06-01 13:54:29 +020013295 configuration language.
13296
13297 Example:
13298 frontend www
Cyril Bonté307ee1e2015-09-28 23:16:06 +020013299 bind *:8881
13300 default_backend servers
David Carlier840b0242016-03-16 10:09:55 +000013301 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 +020013302
Thierry FOURNIER9687c772015-05-07 15:46:29 +020013303debug
13304 This converter is used as debug tool. It dumps on screen the content and the
13305 type of the input sample. The sample is returned as is on its output. This
13306 converter only exists when haproxy was built with debugging enabled.
13307
Willy Tarreau97707872015-01-27 15:12:13 +010013308div(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013309 Divides the input value of type signed integer by <value>, and returns the
13310 result as an signed integer. If <value> is null, the largest unsigned
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013311 integer is returned (typically 2^63-1). <value> can be a numeric value or a
Daniel Schneller0b547052016-03-21 20:46:57 +010013312 variable name. The name of the variable starts with an indication about its
13313 scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013314 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013315 "sess" : the variable is shared with the whole session
13316 "txn" : the variable is shared with the transaction (request and response)
13317 "req" : the variable is shared only during request processing
13318 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013319 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013320 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013321
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013322djb2([<avalanche>])
13323 Hashes a binary input sample into an unsigned 32-bit quantity using the DJB2
13324 hash function. Optionally, it is possible to apply a full avalanche hash
13325 function to the output if the optional <avalanche> argument equals 1. This
13326 converter uses the same functions as used by the various hash-based load
13327 balancing algorithms, so it will provide exactly the same results. It is
13328 mostly intended for debugging, but can be used as a stick-table entry to
13329 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013330 32-bit hash is trivial to break. See also "crc32", "sdbm", "wt6", "crc32c",
13331 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013332
Willy Tarreau97707872015-01-27 15:12:13 +010013333even
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013334 Returns a boolean TRUE if the input value of type signed integer is even
Willy Tarreau97707872015-01-27 15:12:13 +010013335 otherwise returns FALSE. It is functionally equivalent to "not,and(1),bool".
13336
Marcin Deranek9631a282018-04-16 14:30:46 +020013337field(<index>,<delimiters>[,<count>])
13338 Extracts the substring at the given index counting from the beginning
13339 (positive index) or from the end (negative index) considering given delimiters
13340 from an input string. Indexes start at 1 or -1 and delimiters are a string
13341 formatted list of chars. Optionally you can specify <count> of fields to
13342 extract (default: 1). Value of 0 indicates extraction of all remaining
13343 fields.
13344
13345 Example :
13346 str(f1_f2_f3__f5),field(5,_) # f5
13347 str(f1_f2_f3__f5),field(2,_,0) # f2_f3__f5
13348 str(f1_f2_f3__f5),field(2,_,2) # f2_f3
13349 str(f1_f2_f3__f5),field(-2,_,3) # f2_f3_
13350 str(f1_f2_f3__f5),field(-3,_,0) # f1_f2_f3
Emeric Brunf399b0d2014-11-03 17:07:03 +010013351
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013352hex
Davor Ocelice9ed2812017-12-25 17:49:28 +010013353 Converts a binary input sample to a hex string containing two hex digits per
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013354 input byte. It is used to log or transfer hex dumps of some binary input data
Davor Ocelice9ed2812017-12-25 17:49:28 +010013355 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 +020013356 header).
Thierry FOURNIER2f49d6d2014-03-12 15:01:52 +010013357
Dragan Dosen3f957b22017-10-24 09:27:34 +020013358hex2i
13359 Converts a hex string containing two hex digits per input byte to an
13360 integer. If the input value can not be converted, then zero is returned.
13361
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013362http_date([<offset>])
13363 Converts an integer supposed to contain a date since epoch to a string
13364 representing this date in a format suitable for use in HTTP header fields. If
13365 an offset value is specified, then it is a number of seconds that is added to
13366 the date before the conversion is operated. This is particularly useful to
13367 emit Date header fields, Expires values in responses when combined with a
13368 positive offset, or Last-Modified values when the offset is negative.
Willy Tarreau74ca5042013-06-11 23:12:07 +020013369
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013370in_table(<table>)
13371 Uses the string representation of the input sample to perform a look up in
13372 the specified table. If the key is not found in the table, a boolean false
13373 is returned. Otherwise a boolean true is returned. This can be used to verify
Davor Ocelice9ed2812017-12-25 17:49:28 +010013374 the presence of a certain key in a table tracking some elements (e.g. whether
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013375 or not a source IP address or an Authorization header was already seen).
13376
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013377ipmask(<mask4>, [<mask6>])
13378 Apply a mask to an IP address, and use the result for lookups and storage.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013379 This can be used to make all hosts within a certain mask to share the same
Tim Duesterhus1478aa72018-01-25 16:24:51 +010013380 table entries and as such use the same server. The mask4 can be passed in
13381 dotted form (e.g. 255.255.255.0) or in CIDR form (e.g. 24). The mask6 can
13382 be passed in quadruplet form (e.g. ffff:ffff::) or in CIDR form (e.g. 64).
13383 If no mask6 is given IPv6 addresses will fail to convert for backwards
13384 compatibility reasons.
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013385
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013386json([<input-code>])
Davor Ocelice9ed2812017-12-25 17:49:28 +010013387 Escapes the input string and produces an ASCII output string ready to use as a
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013388 JSON string. The converter tries to decode the input string according to the
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013389 <input-code> parameter. It can be "ascii", "utf8", "utf8s", "utf8p" or
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013390 "utf8ps". The "ascii" decoder never fails. The "utf8" decoder detects 3 types
13391 of errors:
13392 - bad UTF-8 sequence (lone continuation byte, bad number of continuation
13393 bytes, ...)
13394 - invalid range (the decoded value is within a UTF-8 prohibited range),
13395 - code overlong (the value is encoded with more bytes than necessary).
13396
13397 The UTF-8 JSON encoding can produce a "too long value" error when the UTF-8
13398 character is greater than 0xffff because the JSON string escape specification
13399 only authorizes 4 hex digits for the value encoding. The UTF-8 decoder exists
13400 in 4 variants designated by a combination of two suffix letters : "p" for
13401 "permissive" and "s" for "silently ignore". The behaviors of the decoders
13402 are :
Davor Ocelice9ed2812017-12-25 17:49:28 +010013403 - "ascii" : never fails;
13404 - "utf8" : fails on any detected errors;
13405 - "utf8s" : never fails, but removes characters corresponding to errors;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013406 - "utf8p" : accepts and fixes the overlong errors, but fails on any other
Davor Ocelice9ed2812017-12-25 17:49:28 +010013407 error;
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013408 - "utf8ps" : never fails, accepts and fixes the overlong errors, but removes
13409 characters corresponding to the other errors.
13410
13411 This converter is particularly useful for building properly escaped JSON for
Davor Ocelice9ed2812017-12-25 17:49:28 +010013412 logging to servers which consume JSON-formatted traffic logs.
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013413
13414 Example:
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013415 capture request header Host len 15
Herve COMMOWICK8dfe8632016-08-05 12:01:20 +020013416 capture request header user-agent len 150
13417 log-format '{"ip":"%[src]","user-agent":"%[capture.req.hdr(1),json(utf8s)]"}'
Thierry FOURNIER317e1c42014-08-12 10:20:47 +020013418
13419 Input request from client 127.0.0.1:
13420 GET / HTTP/1.0
13421 User-Agent: Very "Ugly" UA 1/2
13422
13423 Output log:
13424 {"ip":"127.0.0.1","user-agent":"Very \"Ugly\" UA 1\/2"}
13425
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013426language(<value>[,<default>])
13427 Returns the value with the highest q-factor from a list as extracted from the
13428 "accept-language" header using "req.fhdr". Values with no q-factor have a
13429 q-factor of 1. Values with a q-factor of 0 are dropped. Only values which
13430 belong to the list of semi-colon delimited <values> will be considered. The
13431 argument <value> syntax is "lang[;lang[;lang[;...]]]". If no value matches the
13432 given list and a default value is provided, it is returned. Note that language
13433 names may have a variant after a dash ('-'). If this variant is present in the
13434 list, it will be matched, but if it is not, only the base language is checked.
13435 The match is case-sensitive, and the output string is always one of those
Davor Ocelice9ed2812017-12-25 17:49:28 +010013436 provided in arguments. The ordering of arguments is meaningless, only the
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013437 ordering of the values in the request counts, as the first value among
13438 multiple sharing the same q-factor is used.
Thierry FOURNIERad903512014-04-11 17:51:01 +020013439
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013440 Example :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013441
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013442 # this configuration switches to the backend matching a
13443 # given language based on the request :
Thierry FOURNIERad903512014-04-11 17:51:01 +020013444
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013445 acl es req.fhdr(accept-language),language(es;fr;en) -m str es
13446 acl fr req.fhdr(accept-language),language(es;fr;en) -m str fr
13447 acl en req.fhdr(accept-language),language(es;fr;en) -m str en
13448 use_backend spanish if es
13449 use_backend french if fr
13450 use_backend english if en
13451 default_backend choose_your_language
Thierry FOURNIERad903512014-04-11 17:51:01 +020013452
Willy Tarreau60a2ee72017-12-15 07:13:48 +010013453length
Etienne Carriereed0d24e2017-12-13 13:41:34 +010013454 Get the length of the string. This can only be placed after a string
13455 sample fetch function or after a transformation keyword returning a string
13456 type. The result is of type integer.
13457
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013458lower
13459 Convert a string sample to lower case. This can only be placed after a string
13460 sample fetch function or after a transformation keyword returning a string
13461 type. The result is of type string.
13462
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013463ltime(<format>[,<offset>])
13464 Converts an integer supposed to contain a date since epoch to a string
13465 representing this date in local time using a format defined by the <format>
13466 string using strftime(3). The purpose is to allow any date format to be used
13467 in logs. An optional <offset> in seconds may be applied to the input date
13468 (positive or negative). See the strftime() man page for the format supported
13469 by your operating system. See also the utime converter.
13470
13471 Example :
13472
13473 # Emit two colons, one with the local time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013474 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013475 log-format %[date,ltime(%Y%m%d%H%M%S)]\ %ci:%cp
13476
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013477map(<map_file>[,<default_value>])
13478map_<match_type>(<map_file>[,<default_value>])
13479map_<match_type>_<output_type>(<map_file>[,<default_value>])
13480 Search the input value from <map_file> using the <match_type> matching method,
13481 and return the associated value converted to the type <output_type>. If the
13482 input value cannot be found in the <map_file>, the converter returns the
13483 <default_value>. If the <default_value> is not set, the converter fails and
13484 acts as if no input value could be fetched. If the <match_type> is not set, it
13485 defaults to "str". Likewise, if the <output_type> is not set, it defaults to
13486 "str". For convenience, the "map" keyword is an alias for "map_str" and maps a
13487 string to another string.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013488
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013489 It is important to avoid overlapping between the keys : IP addresses and
13490 strings are stored in trees, so the first of the finest match will be used.
13491 Other keys are stored in lists, so the first matching occurrence will be used.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013492
Tim Düsterhus4896c442016-11-29 02:15:19 +010013493 The following array contains the list of all map functions available sorted by
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013494 input type, match type and output type.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013495
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013496 input type | match method | output type str | output type int | output type ip
13497 -----------+--------------+-----------------+-----------------+---------------
13498 str | str | map_str | map_str_int | map_str_ip
13499 -----------+--------------+-----------------+-----------------+---------------
Willy Tarreau787a4c02014-05-10 07:55:30 +020013500 str | beg | map_beg | map_beg_int | map_end_ip
13501 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013502 str | sub | map_sub | map_sub_int | map_sub_ip
13503 -----------+--------------+-----------------+-----------------+---------------
13504 str | dir | map_dir | map_dir_int | map_dir_ip
13505 -----------+--------------+-----------------+-----------------+---------------
13506 str | dom | map_dom | map_dom_int | map_dom_ip
13507 -----------+--------------+-----------------+-----------------+---------------
13508 str | end | map_end | map_end_int | map_end_ip
13509 -----------+--------------+-----------------+-----------------+---------------
Ruoshan Huang3c5e3742016-12-02 16:25:31 +080013510 str | reg | map_reg | map_reg_int | map_reg_ip
13511 -----------+--------------+-----------------+-----------------+---------------
13512 str | reg | map_regm | map_reg_int | map_reg_ip
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013513 -----------+--------------+-----------------+-----------------+---------------
13514 int | int | map_int | map_int_int | map_int_ip
13515 -----------+--------------+-----------------+-----------------+---------------
13516 ip | ip | map_ip | map_ip_int | map_ip_ip
13517 -----------+--------------+-----------------+-----------------+---------------
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013518
Thierry Fournier8feaa662016-02-10 22:55:20 +010013519 The special map called "map_regm" expect matching zone in the regular
13520 expression and modify the output replacing back reference (like "\1") by
13521 the corresponding match text.
13522
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013523 The file contains one key + value per line. Lines which start with '#' are
13524 ignored, just like empty lines. Leading tabs and spaces are stripped. The key
13525 is then the first "word" (series of non-space/tabs characters), and the value
13526 is what follows this series of space/tab till the end of the line excluding
13527 trailing spaces/tabs.
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010013528
Thierry FOURNIER060762e2014-04-23 13:29:15 +020013529 Example :
13530
13531 # this is a comment and is ignored
13532 2.22.246.0/23 United Kingdom \n
13533 <-><-----------><--><------------><---->
13534 | | | | `- trailing spaces ignored
13535 | | | `---------- value
13536 | | `-------------------- middle spaces ignored
13537 | `---------------------------- key
13538 `------------------------------------ leading spaces ignored
13539
Willy Tarreau97707872015-01-27 15:12:13 +010013540mod(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013541 Divides the input value of type signed integer by <value>, and returns the
13542 remainder as an signed integer. If <value> is null, then zero is returned.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013543 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013544 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013545 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013546 "sess" : the variable is shared with the whole session
13547 "txn" : the variable is shared with the transaction (request and response)
13548 "req" : the variable is shared only during request processing
13549 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013550 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013551 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013552
13553mul(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013554 Multiplies the input value of type signed integer by <value>, and returns
Thierry FOURNIER00c005c2015-07-08 01:10:21 +020013555 the product as an signed integer. In case of overflow, the largest possible
13556 value for the sign is returned so that the operation doesn't wrap around.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013557 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013558 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013559 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013560 "sess" : the variable is shared with the whole session
13561 "txn" : the variable is shared with the transaction (request and response)
13562 "req" : the variable is shared only during request processing
13563 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013564 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013565 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013566
Nenad Merdanovicb7e7c472017-03-12 21:56:55 +010013567nbsrv
13568 Takes an input value of type string, interprets it as a backend name and
13569 returns the number of usable servers in that backend. Can be used in places
13570 where we want to look up a backend from a dynamic name, like a result of a
13571 map lookup.
13572
Willy Tarreau97707872015-01-27 15:12:13 +010013573neg
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013574 Takes the input value of type signed integer, computes the opposite value,
13575 and returns the remainder as an signed integer. 0 is identity. This operator
13576 is provided for reversed subtracts : in order to subtract the input from a
13577 constant, simply perform a "neg,add(value)".
Willy Tarreau97707872015-01-27 15:12:13 +010013578
13579not
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013580 Returns a boolean FALSE if the input value of type signed integer is
Willy Tarreau97707872015-01-27 15:12:13 +010013581 non-null, otherwise returns TRUE. Used in conjunction with and(), it can be
Davor Ocelice9ed2812017-12-25 17:49:28 +010013582 used to report true/false for bit testing on input values (e.g. verify the
Willy Tarreau97707872015-01-27 15:12:13 +010013583 absence of a flag).
13584
13585odd
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013586 Returns a boolean TRUE if the input value of type signed integer is odd
Willy Tarreau97707872015-01-27 15:12:13 +010013587 otherwise returns FALSE. It is functionally equivalent to "and(1),bool".
13588
13589or(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013590 Performs a bitwise "OR" between <value> and the input value of type signed
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013591 integer, and returns the result as an signed integer. <value> can be a
Daniel Schneller0b547052016-03-21 20:46:57 +010013592 numeric value or a variable name. The name of the variable starts with an
13593 indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013594 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013595 "sess" : the variable is shared with the whole session
13596 "txn" : the variable is shared with the transaction (request and response)
13597 "req" : the variable is shared only during request processing
13598 "res" : the variable is shared only during response processing
Davor Ocelice9ed2812017-12-25 17:49:28 +010013599 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013600 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013601
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013602protobuf(<field_number>,[<field_type>])
13603 This extracts the protocol buffers message field in raw mode of an input binary
13604 sample representation of a protocol buffer message with <field_number> as field
13605 number (dotted notation) if <field_type> is not present, or as an integer sample
13606 if this field is present (see also "ungrpc" below).
13607 The list of the authorized types is the following one: "int32", "int64", "uint32",
13608 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
13609 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
13610 "float" for the wire type 5. Note that "string" is considered as a length-delimited
13611 type, so it does not require any <field_type> argument to be extracted.
13612 More information may be found here about the protocol buffers message field types:
13613 https://developers.google.com/protocol-buffers/docs/encoding
13614
Willy Tarreauc4dc3502015-01-23 20:39:28 +010013615regsub(<regex>,<subst>[,<flags>])
Willy Tarreau7eda8492015-01-20 19:47:06 +010013616 Applies a regex-based substitution to the input string. It does the same
13617 operation as the well-known "sed" utility with "s/<regex>/<subst>/". By
13618 default it will replace in the input string the first occurrence of the
13619 largest part matching the regular expression <regex> with the substitution
13620 string <subst>. It is possible to replace all occurrences instead by adding
13621 the flag "g" in the third argument <flags>. It is also possible to make the
13622 regex case insensitive by adding the flag "i" in <flags>. Since <flags> is a
13623 string, it is made up from the concatenation of all desired flags. Thus if
13624 both "i" and "g" are desired, using "gi" or "ig" will have the same effect.
13625 It is important to note that due to the current limitations of the
Baptiste Assmann66025d82016-03-06 23:36:48 +010013626 configuration parser, some characters such as closing parenthesis, closing
13627 square brackets or comma are not possible to use in the arguments. The first
13628 use of this converter is to replace certain characters or sequence of
13629 characters with other ones.
Willy Tarreau7eda8492015-01-20 19:47:06 +010013630
13631 Example :
13632
13633 # de-duplicate "/" in header "x-path".
13634 # input: x-path: /////a///b/c/xzxyz/
13635 # output: x-path: /a/b/c/xzxyz/
13636 http-request set-header x-path %[hdr(x-path),regsub(/+,/,g)]
13637
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013638capture-req(<id>)
13639 Capture the string entry in the request slot <id> and returns the entry as
13640 is. If the slot doesn't exist, the capture fails silently.
13641
13642 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013643 "http-response capture", "capture.req.hdr" and
13644 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013645
13646capture-res(<id>)
13647 Capture the string entry in the response slot <id> and returns the entry as
13648 is. If the slot doesn't exist, the capture fails silently.
13649
13650 See also: "declare capture", "http-request capture",
Baptiste Assmann5ac425c2015-10-21 23:13:46 +020013651 "http-response capture", "capture.req.hdr" and
13652 "capture.res.hdr" (sample fetches).
Thierry FOURNIER35ab2752015-05-28 13:22:03 +020013653
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013654sdbm([<avalanche>])
13655 Hashes a binary input sample into an unsigned 32-bit quantity using the SDBM
13656 hash function. Optionally, it is possible to apply a full avalanche hash
13657 function to the output if the optional <avalanche> argument equals 1. This
13658 converter uses the same functions as used by the various hash-based load
13659 balancing algorithms, so it will provide exactly the same results. It is
13660 mostly intended for debugging, but can be used as a stick-table entry to
13661 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013662 32-bit hash is trivial to break. See also "crc32", "djb2", "wt6", "crc32c",
13663 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013664
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013665set-var(<var name>)
Davor Ocelice9ed2812017-12-25 17:49:28 +010013666 Sets a variable with the input content and returns the content on the output
13667 as-is. The variable keeps the value and the associated input type. The name of
13668 the variable starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013669 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013670 "sess" : the variable is shared with the whole session
13671 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013672 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013673 "req" : the variable is shared only during request processing,
13674 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013675 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013676 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020013677
Dragan Dosen6e5a9ca2017-10-24 09:18:23 +020013678sha1
13679 Converts a binary input sample to a SHA1 digest. The result is a binary
13680 sample with length of 20 bytes.
13681
Tim Duesterhusca097c12018-04-27 21:18:45 +020013682strcmp(<var>)
13683 Compares the contents of <var> with the input value of type string. Returns
13684 the result as a signed integer compatible with strcmp(3): 0 if both strings
13685 are identical. A value less than 0 if the left string is lexicographically
13686 smaller than the right string or if the left string is shorter. A value greater
13687 than 0 otherwise (right string greater than left string or the right string is
13688 shorter).
13689
13690 Example :
13691
13692 http-request set-var(txn.host) hdr(host)
13693 # Check whether the client is attempting domain fronting.
13694 acl ssl_sni_http_host_match ssl_fc_sni,strcmp(txn.host) eq 0
13695
13696
Willy Tarreau97707872015-01-27 15:12:13 +010013697sub(<value>)
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013698 Subtracts <value> from the input value of type signed integer, and returns
13699 the result as an signed integer. Note: in order to subtract the input from
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013700 a constant, simply perform a "neg,add(value)". <value> can be a numeric value
Daniel Schneller0b547052016-03-21 20:46:57 +010013701 or a variable name. The name of the variable starts with an indication about
13702 its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013703 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013704 "sess" : the variable is shared with the whole session
13705 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013706 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010013707 "req" : the variable is shared only during request processing,
13708 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013709 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010013710 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010013711
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013712table_bytes_in_rate(<table>)
13713 Uses the string representation of the input sample to perform a look up in
13714 the specified table. If the key is not found in the table, integer value zero
13715 is returned. Otherwise the converter returns the average client-to-server
13716 bytes rate associated with the input sample in the designated table, measured
13717 in amount of bytes over the period configured in the table. See also the
13718 sc_bytes_in_rate sample fetch keyword.
13719
13720
13721table_bytes_out_rate(<table>)
13722 Uses the string representation of the input sample to perform a look up in
13723 the specified table. If the key is not found in the table, integer value zero
13724 is returned. Otherwise the converter returns the average server-to-client
13725 bytes rate associated with the input sample in the designated table, measured
13726 in amount of bytes over the period configured in the table. See also the
13727 sc_bytes_out_rate sample fetch keyword.
13728
13729table_conn_cnt(<table>)
13730 Uses the string representation of the input sample to perform a look up in
13731 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013732 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013733 connections associated with the input sample in the designated table. See
13734 also the sc_conn_cnt sample fetch keyword.
13735
13736table_conn_cur(<table>)
13737 Uses the string representation of the input sample to perform a look up in
13738 the specified table. If the key is not found in the table, integer value zero
13739 is returned. Otherwise the converter returns the current amount of concurrent
13740 tracked connections associated with the input sample in the designated table.
13741 See also the sc_conn_cur sample fetch keyword.
13742
13743table_conn_rate(<table>)
13744 Uses the string representation of the input sample to perform a look up in
13745 the specified table. If the key is not found in the table, integer value zero
13746 is returned. Otherwise the converter returns the average incoming connection
13747 rate associated with the input sample in the designated table. See also the
13748 sc_conn_rate sample fetch keyword.
13749
Thierry FOURNIER236657b2015-08-19 08:25:14 +020013750table_gpt0(<table>)
13751 Uses the string representation of the input sample to perform a look up in
13752 the specified table. If the key is not found in the table, boolean value zero
13753 is returned. Otherwise the converter returns the current value of the first
13754 general purpose tag associated with the input sample in the designated table.
13755 See also the sc_get_gpt0 sample fetch keyword.
13756
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013757table_gpc0(<table>)
13758 Uses the string representation of the input sample to perform a look up in
13759 the specified table. If the key is not found in the table, integer value zero
13760 is returned. Otherwise the converter returns the current value of the first
13761 general purpose counter associated with the input sample in the designated
13762 table. See also the sc_get_gpc0 sample fetch keyword.
13763
13764table_gpc0_rate(<table>)
13765 Uses the string representation of the input sample to perform a look up in
13766 the specified table. If the key is not found in the table, integer value zero
13767 is returned. Otherwise the converter returns the frequency which the gpc0
13768 counter was incremented over the configured period in the table, associated
13769 with the input sample in the designated table. See also the sc_get_gpc0_rate
13770 sample fetch keyword.
13771
Frédéric Lécaille6778b272018-01-29 15:22:53 +010013772table_gpc1(<table>)
13773 Uses the string representation of the input sample to perform a look up in
13774 the specified table. If the key is not found in the table, integer value zero
13775 is returned. Otherwise the converter returns the current value of the second
13776 general purpose counter associated with the input sample in the designated
13777 table. See also the sc_get_gpc1 sample fetch keyword.
13778
13779table_gpc1_rate(<table>)
13780 Uses the string representation of the input sample to perform a look up in
13781 the specified table. If the key is not found in the table, integer value zero
13782 is returned. Otherwise the converter returns the frequency which the gpc1
13783 counter was incremented over the configured period in the table, associated
13784 with the input sample in the designated table. See also the sc_get_gpc1_rate
13785 sample fetch keyword.
13786
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013787table_http_err_cnt(<table>)
13788 Uses the string representation of the input sample to perform a look up in
13789 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013790 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013791 errors associated with the input sample in the designated table. See also the
13792 sc_http_err_cnt sample fetch keyword.
13793
13794table_http_err_rate(<table>)
13795 Uses the string representation of the input sample to perform a look up in
13796 the specified table. If the key is not found in the table, integer value zero
13797 is returned. Otherwise the average rate of HTTP errors associated with the
13798 input sample in the designated table, measured in amount of errors over the
13799 period configured in the table. See also the sc_http_err_rate sample fetch
13800 keyword.
13801
13802table_http_req_cnt(<table>)
13803 Uses the string representation of the input sample to perform a look up in
13804 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013805 is returned. Otherwise the converter returns the cumulative number of HTTP
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013806 requests associated with the input sample in the designated table. See also
13807 the sc_http_req_cnt sample fetch keyword.
13808
13809table_http_req_rate(<table>)
13810 Uses the string representation of the input sample to perform a look up in
13811 the specified table. If the key is not found in the table, integer value zero
13812 is returned. Otherwise the average rate of HTTP requests associated with the
13813 input sample in the designated table, measured in amount of requests over the
13814 period configured in the table. See also the sc_http_req_rate sample fetch
13815 keyword.
13816
13817table_kbytes_in(<table>)
13818 Uses the string representation of the input sample to perform a look up in
13819 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013820 is returned. Otherwise the converter returns the cumulative number of client-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013821 to-server data associated with the input sample in the designated table,
13822 measured in kilobytes. The test is currently performed on 32-bit integers,
13823 which limits values to 4 terabytes. See also the sc_kbytes_in sample fetch
13824 keyword.
13825
13826table_kbytes_out(<table>)
13827 Uses the string representation of the input sample to perform a look up in
13828 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013829 is returned. Otherwise the converter returns the cumulative number of server-
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013830 to-client data associated with the input sample in the designated table,
13831 measured in kilobytes. The test is currently performed on 32-bit integers,
13832 which limits values to 4 terabytes. See also the sc_kbytes_out sample fetch
13833 keyword.
13834
13835table_server_id(<table>)
13836 Uses the string representation of the input sample to perform a look up in
13837 the specified table. If the key is not found in the table, integer value zero
13838 is returned. Otherwise the converter returns the server ID associated with
13839 the input sample in the designated table. A server ID is associated to a
13840 sample by a "stick" rule when a connection to a server succeeds. A server ID
13841 zero means that no server is associated with this key.
13842
13843table_sess_cnt(<table>)
13844 Uses the string representation of the input sample to perform a look up in
13845 the specified table. If the key is not found in the table, integer value zero
Davor Ocelice9ed2812017-12-25 17:49:28 +010013846 is returned. Otherwise the converter returns the cumulative number of incoming
Willy Tarreaud9f316a2014-07-10 14:03:38 +020013847 sessions associated with the input sample in the designated table. Note that
13848 a session here refers to an incoming connection being accepted by the
13849 "tcp-request connection" rulesets. See also the sc_sess_cnt sample fetch
13850 keyword.
13851
13852table_sess_rate(<table>)
13853 Uses the string representation of the input sample to perform a look up in
13854 the specified table. If the key is not found in the table, integer value zero
13855 is returned. Otherwise the converter returns the average incoming session
13856 rate associated with the input sample in the designated table. Note that a
13857 session here refers to an incoming connection being accepted by the
13858 "tcp-request connection" rulesets. See also the sc_sess_rate sample fetch
13859 keyword.
13860
13861table_trackers(<table>)
13862 Uses the string representation of the input sample to perform a look up in
13863 the specified table. If the key is not found in the table, integer value zero
13864 is returned. Otherwise the converter returns the current amount of concurrent
13865 connections tracking the same key as the input sample in the designated
13866 table. It differs from table_conn_cur in that it does not rely on any stored
13867 information but on the table's reference count (the "use" value which is
13868 returned by "show table" on the CLI). This may sometimes be more suited for
13869 layer7 tracking. It can be used to tell a server how many concurrent
13870 connections there are from a given address for example. See also the
13871 sc_trackers sample fetch keyword.
13872
Willy Tarreauffcb2e42014-07-10 16:29:08 +020013873upper
13874 Convert a string sample to upper case. This can only be placed after a string
13875 sample fetch function or after a transformation keyword returning a string
13876 type. The result is of type string.
13877
Thierry FOURNIER82ff3c92015-05-07 15:46:20 +020013878url_dec
13879 Takes an url-encoded string provided as input and returns the decoded
13880 version as output. The input and the output are of type string.
13881
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013882ungrpc(<field_number>,[<field_type>])
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013883 This extracts the protocol buffers message field in raw mode of an input binary
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013884 sample representation of a gRPC message with <field_number> as field number
13885 (dotted notation) if <field_type> is not present, or as an integer sample if this
13886 field is present.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013887 The list of the authorized types is the following one: "int32", "int64", "uint32",
13888 "uint64", "sint32", "sint64", "bool", "enum" for the "varint" wire type 0
13889 "fixed64", "sfixed64", "double" for the 64bit wire type 1, "fixed32", "sfixed32",
13890 "float" for the wire type 5. Note that "string" is considered as a length-delimited
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013891 type, so it does not require any <field_type> argument to be extracted.
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013892 More information may be found here about the protocol buffers message field types:
13893 https://developers.google.com/protocol-buffers/docs/encoding
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013894
13895 Example:
13896 // with such a protocol buffer .proto file content adapted from
13897 // https://github.com/grpc/grpc/blob/master/examples/protos/route_guide.proto
13898
13899 message Point {
13900 int32 latitude = 1;
13901 int32 longitude = 2;
13902 }
13903
13904 message PPoint {
13905 Point point = 59;
13906 }
13907
13908 message Rectangle {
13909 // One corner of the rectangle.
13910 PPoint lo = 48;
13911 // The other corner of the rectangle.
13912 PPoint hi = 49;
13913 }
13914
13915 let's say a body request is made of a "Rectangle" object value (two PPoint
13916 protocol buffers messages), the four protocol buffers fields could be
13917 extracted with these "ungrpc" directives:
13918
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013919 req.body,ungrpc(48.59.1,int32) # "latitude" of "lo" first PPoint
13920 req.body,ungrpc(48.59.2,int32) # "longitude" of "lo" first PPoint
13921 req.body,ungrpc(49.59.1,int32) # "latidude" of "hi" second PPoint
13922 req.body,ungrpc(49.59.2,int32) # "longitude" of "hi" second PPoint
13923
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013924 We could also extract the intermediary 48.59 field as a binary sample as follows:
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013925
Frédéric Lécaille93d33162019-03-06 09:35:59 +010013926 req.body,ungrpc(48.59)
Frédéric Lécaille756d97f2019-03-04 19:03:48 +010013927
Frédéric Lécaillebfe61382019-03-06 14:34:36 +010013928 As a gRPC message is alway made of a gRPC header followed by protocol buffers
13929 messages, in the previous example the "latitude" of "lo" first PPoint
13930 could be extracted with these equivalent directives:
13931
13932 req.body,ungrpc(48.59),protobuf(1,int32)
13933 req.body,ungrpc(48),protobuf(59.1,int32)
13934 req.body,ungrpc(48),protobuf(59),protobuf(1,int32)
13935
13936 Note that the first convert must be "ungrpc", the remaining ones must be
13937 "protobuf" and only the last one may have or not a second argument to
13938 interpret the previous binary sample.
13939
Frédéric Lécaille50290fb2019-02-27 14:34:51 +010013940
Christopher Faulet85d79c92016-11-09 16:54:56 +010013941unset-var(<var name>)
13942 Unsets a variable if the input content is defined. The name of the variable
13943 starts with an indication about its scope. The scopes allowed are:
13944 "proc" : the variable is shared with the whole process
13945 "sess" : the variable is shared with the whole session
13946 "txn" : the variable is shared with the transaction (request and
13947 response),
13948 "req" : the variable is shared only during request processing,
13949 "res" : the variable is shared only during response processing.
13950 This prefix is followed by a name. The separator is a '.'. The name may only
13951 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
13952
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013953utime(<format>[,<offset>])
13954 Converts an integer supposed to contain a date since epoch to a string
13955 representing this date in UTC time using a format defined by the <format>
13956 string using strftime(3). The purpose is to allow any date format to be used
13957 in logs. An optional <offset> in seconds may be applied to the input date
13958 (positive or negative). See the strftime() man page for the format supported
13959 by your operating system. See also the ltime converter.
13960
13961 Example :
13962
13963 # Emit two colons, one with the UTC time and another with ip:port
Davor Ocelice9ed2812017-12-25 17:49:28 +010013964 # e.g. 20140710162350 127.0.0.1:57325
Willy Tarreau0dbfdba2014-07-10 16:37:47 +020013965 log-format %[date,utime(%Y%m%d%H%M%S)]\ %ci:%cp
13966
Marcin Deranek9631a282018-04-16 14:30:46 +020013967word(<index>,<delimiters>[,<count>])
13968 Extracts the nth word counting from the beginning (positive index) or from
13969 the end (negative index) considering given delimiters from an input string.
13970 Indexes start at 1 or -1 and delimiters are a string formatted list of chars.
13971 Optionally you can specify <count> of words to extract (default: 1).
13972 Value of 0 indicates extraction of all remaining words.
13973
13974 Example :
13975 str(f1_f2_f3__f5),word(4,_) # f5
13976 str(f1_f2_f3__f5),word(2,_,0) # f2_f3__f5
13977 str(f1_f2_f3__f5),word(3,_,2) # f3__f5
13978 str(f1_f2_f3__f5),word(-2,_,3) # f1_f2_f3
13979 str(f1_f2_f3__f5),word(-3,_,0) # f1_f2
Emeric Brunc9a0f6d2014-11-25 14:09:01 +010013980
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013981wt6([<avalanche>])
13982 Hashes a binary input sample into an unsigned 32-bit quantity using the WT6
13983 hash function. Optionally, it is possible to apply a full avalanche hash
13984 function to the output if the optional <avalanche> argument equals 1. This
13985 converter uses the same functions as used by the various hash-based load
13986 balancing algorithms, so it will provide exactly the same results. It is
13987 mostly intended for debugging, but can be used as a stick-table entry to
13988 collect rough statistics. It must not be used for security purposes as a
Emmanuel Hocdet50791a72018-03-21 11:19:01 +010013989 32-bit hash is trivial to break. See also "crc32", "djb2", "sdbm", "crc32c",
13990 and the "hash-type" directive.
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020013991
Willy Tarreau97707872015-01-27 15:12:13 +010013992xor(<value>)
13993 Performs a bitwise "XOR" (exclusive OR) between <value> and the input value
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020013994 of type signed integer, and returns the result as an signed integer.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020013995 <value> can be a numeric value or a variable name. The name of the variable
Daniel Schneller0b547052016-03-21 20:46:57 +010013996 starts with an indication about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010013997 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010013998 "sess" : the variable is shared with the whole session
13999 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014000 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014001 "req" : the variable is shared only during request processing,
14002 "res" : the variable is shared only during response processing.
Thierry FOURNIER5d86fae2015-07-07 21:10:16 +020014003 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014004 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Willy Tarreau97707872015-01-27 15:12:13 +010014005
Thierry FOURNIER01e09742016-12-26 11:46:11 +010014006xxh32([<seed>])
14007 Hashes a binary input sample into an unsigned 32-bit quantity using the 32-bit
14008 variant of the XXHash hash function. This hash supports a seed which defaults
14009 to zero but a different value maybe passed as the <seed> argument. This hash
14010 is known to be very good and very fast so it can be used to hash URLs and/or
14011 URL parameters for use as stick-table keys to collect statistics with a low
14012 collision rate, though care must be taken as the algorithm is not considered
14013 as cryptographically secure.
14014
14015xxh64([<seed>])
14016 Hashes a binary input sample into a signed 64-bit quantity using the 64-bit
14017 variant of the XXHash hash function. This hash supports a seed which defaults
14018 to zero but a different value maybe passed as the <seed> argument. This hash
14019 is known to be very good and very fast so it can be used to hash URLs and/or
14020 URL parameters for use as stick-table keys to collect statistics with a low
14021 collision rate, though care must be taken as the algorithm is not considered
14022 as cryptographically secure.
14023
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014024
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200140257.3.2. Fetching samples from internal states
Willy Tarreau74ca5042013-06-11 23:12:07 +020014026--------------------------------------------
14027
14028A first set of sample fetch methods applies to internal information which does
14029not even relate to any client information. These ones are sometimes used with
14030"monitor-fail" directives to report an internal status to external watchers.
14031The sample fetch methods described in this section are usable anywhere.
14032
14033always_false : boolean
14034 Always returns the boolean "false" value. It may be used with ACLs as a
14035 temporary replacement for another one when adjusting configurations.
14036
14037always_true : boolean
14038 Always returns the boolean "true" value. It may be used with ACLs as a
14039 temporary replacement for another one when adjusting configurations.
14040
14041avg_queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014042 Returns the total number of queued connections of the designated backend
Willy Tarreau74ca5042013-06-11 23:12:07 +020014043 divided by the number of active servers. The current backend is used if no
14044 backend is specified. This is very similar to "queue" except that the size of
14045 the farm is considered, in order to give a more accurate measurement of the
14046 time it may take for a new connection to be processed. The main usage is with
14047 ACL to return a sorry page to new users when it becomes certain they will get
14048 a degraded service, or to pass to the backend servers in a header so that
14049 they decide to work in degraded mode or to disable some functions to speed up
14050 the processing a bit. Note that in the event there would not be any active
14051 server anymore, twice the number of queued connections would be considered as
14052 the measured value. This is a fair estimate, as we expect one server to get
14053 back soon anyway, but we still prefer to send new traffic to another backend
14054 if in better shape. See also the "queue", "be_conn", and "be_sess_rate"
14055 sample fetches.
Krzysztof Piotr Oledzki346f76d2010-01-12 21:59:30 +010014056
Willy Tarreau74ca5042013-06-11 23:12:07 +020014057be_conn([<backend>]) : integer
Willy Tarreaua36af912009-10-10 12:02:45 +020014058 Applies to the number of currently established connections on the backend,
14059 possibly including the connection being evaluated. If no backend name is
14060 specified, the current one is used. But it is also possible to check another
14061 backend. It can be used to use a specific farm when the nominal one is full.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014062 See also the "fe_conn", "queue", "be_conn_free", and "be_sess_rate" criteria.
14063
14064be_conn_free([<backend>]) : integer
14065 Returns an integer value corresponding to the number of available connections
14066 across available servers in the backend. Queue slots are not included. Backup
14067 servers are also not included, unless all other servers are down. If no
14068 backend name is specified, the current one is used. But it is also possible
14069 to check another backend. It can be used to use a specific farm when the
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014070 nominal one is full. See also the "be_conn", "connslots", and "srv_conn_free"
14071 criteria.
Patrick Hemmer4cdf3ab2018-06-14 17:10:27 -040014072
14073 OTHER CAVEATS AND NOTES: if any of the server maxconn, or maxqueue is 0
14074 (meaning unlimited), then this fetch clearly does not make sense, in which
14075 case the value returned will be -1.
Willy Tarreau6a06a402007-07-15 20:15:28 +020014076
Willy Tarreau74ca5042013-06-11 23:12:07 +020014077be_sess_rate([<backend>]) : integer
14078 Returns an integer value corresponding to the sessions creation rate on the
14079 backend, in number of new sessions per second. This is used with ACLs to
14080 switch to an alternate backend when an expensive or fragile one reaches too
Davor Ocelice9ed2812017-12-25 17:49:28 +010014081 high a session rate, or to limit abuse of service (e.g. prevent sucking of an
Willy Tarreau74ca5042013-06-11 23:12:07 +020014082 online dictionary). It can also be useful to add this element to logs using a
14083 log-format directive.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014084
14085 Example :
14086 # Redirect to an error page if the dictionary is requested too often
14087 backend dynamic
14088 mode http
14089 acl being_scanned be_sess_rate gt 100
14090 redirect location /denied.html if being_scanned
Willy Tarreau0ba27502007-12-24 16:55:16 +010014091
Davor Ocelice9ed2812017-12-25 17:49:28 +010014092bin(<hex>) : bin
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014093 Returns a binary chain. The input is the hexadecimal representation
14094 of the string.
14095
14096bool(<bool>) : bool
14097 Returns a boolean value. <bool> can be 'true', 'false', '1' or '0'.
14098 'false' and '0' are the same. 'true' and '1' are the same.
14099
Willy Tarreau74ca5042013-06-11 23:12:07 +020014100connslots([<backend>]) : integer
14101 Returns an integer value corresponding to the number of connection slots
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014102 still available in the backend, by totaling the maximum amount of
Willy Tarreau74ca5042013-06-11 23:12:07 +020014103 connections on all servers and the maximum queue size. This is probably only
14104 used with ACLs.
Tait Clarridge7896d522012-12-05 21:39:31 -050014105
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014106 The basic idea here is to be able to measure the number of connection "slots"
Willy Tarreau55165fe2009-05-10 12:02:55 +020014107 still available (connection + queue), so that anything beyond that (intended
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014108 usage; see "use_backend" keyword) can be redirected to a different backend.
14109
Willy Tarreau55165fe2009-05-10 12:02:55 +020014110 'connslots' = number of available server connection slots, + number of
14111 available server queue slots.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014112
Willy Tarreaua36af912009-10-10 12:02:45 +020014113 Note that while "fe_conn" may be used, "connslots" comes in especially
Willy Tarreau55165fe2009-05-10 12:02:55 +020014114 useful when you have a case of traffic going to one single ip, splitting into
Willy Tarreau74ca5042013-06-11 23:12:07 +020014115 multiple backends (perhaps using ACLs to do name-based load balancing) and
Willy Tarreau55165fe2009-05-10 12:02:55 +020014116 you want to be able to differentiate between different backends, and their
Davor Ocelice9ed2812017-12-25 17:49:28 +010014117 available "connslots". Also, whereas "nbsrv" only measures servers that are
Willy Tarreau74ca5042013-06-11 23:12:07 +020014118 actually *down*, this fetch is more fine-grained and looks into the number of
Willy Tarreaua36af912009-10-10 12:02:45 +020014119 available connection slots as well. See also "queue" and "avg_queue".
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014120
Willy Tarreau55165fe2009-05-10 12:02:55 +020014121 OTHER CAVEATS AND NOTES: at this point in time, the code does not take care
14122 of dynamic connections. Also, if any of the server maxconn, or maxqueue is 0,
Willy Tarreau74ca5042013-06-11 23:12:07 +020014123 then this fetch clearly does not make sense, in which case the value returned
Willy Tarreau55165fe2009-05-10 12:02:55 +020014124 will be -1.
Jeffrey 'jf' Lim5051d7b2008-09-04 01:03:03 +080014125
Willy Tarreau70fe9442018-11-22 16:07:39 +010014126cpu_calls : integer
14127 Returns the number of calls to the task processing the stream or current
14128 request since it was allocated. This number is reset for each new request on
14129 the same connections in case of HTTP keep-alive. This value should usually be
14130 low and stable (around 2 calls for a typically simple request) but may become
14131 high if some processing (compression, caching or analysis) is performed. This
14132 is purely for performance monitoring purposes.
14133
14134cpu_ns_avg : integer
14135 Returns the average number of nanoseconds spent in each call to the task
14136 processing the stream or current request. This number is reset for each new
14137 request on the same connections in case of HTTP keep-alive. This value
14138 indicates the overall cost of processing the request or the connection for
14139 each call. There is no good nor bad value but the time spent in a call
14140 automatically causes latency for other processing (see lat_ns_avg below),
14141 and may affect other connection's apparent response time. Certain operations
14142 like compression, complex regex matching or heavy Lua operations may directly
14143 affect this value, and having it in the logs will make it easier to spot the
14144 faulty processing that needs to be fixed to recover decent performance.
14145 Note: this value is exactly cpu_ns_tot divided by cpu_calls.
14146
14147cpu_ns_tot : integer
14148 Returns the total number of nanoseconds spent in each call to the task
14149 processing the stream or current request. This number is reset for each new
14150 request on the same connections in case of HTTP keep-alive. This value
14151 indicates the overall cost of processing the request or the connection for
14152 each call. There is no good nor bad value but the time spent in a call
14153 automatically causes latency for other processing (see lat_ns_avg below),
14154 induces CPU costs on the machine, and may affect other connection's apparent
14155 response time. Certain operations like compression, complex regex matching or
14156 heavy Lua operations may directly affect this value, and having it in the
14157 logs will make it easier to spot the faulty processing that needs to be fixed
14158 to recover decent performance. The value may be artificially high due to a
14159 high cpu_calls count, for example when processing many HTTP chunks, and for
14160 this reason it is often preferred to log cpu_ns_avg instead.
14161
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014162date([<offset>]) : integer
14163 Returns the current date as the epoch (number of seconds since 01/01/1970).
14164 If an offset value is specified, then it is a number of seconds that is added
14165 to the current date before returning the value. This is particularly useful
14166 to compute relative dates, as both positive and negative offsets are allowed.
Willy Tarreau276fae92013-07-25 14:36:01 +020014167 It is useful combined with the http_date converter.
14168
14169 Example :
14170
14171 # set an expires header to now+1 hour in every response
14172 http-response set-header Expires %[date(3600),http_date]
Willy Tarreau6236d3a2013-07-25 14:28:25 +020014173
Etienne Carrierea792a0a2018-01-17 13:43:24 +010014174date_us : integer
14175 Return the microseconds part of the date (the "second" part is returned by
14176 date sample). This sample is coherent with the date sample as it is comes
14177 from the same timeval structure.
14178
Willy Tarreaud716f9b2017-10-13 11:03:15 +020014179distcc_body(<token>[,<occ>]) : binary
14180 Parses a distcc message and returns the body associated to occurrence #<occ>
14181 of the token <token>. Occurrences start at 1, and when unspecified, any may
14182 match though in practice only the first one is checked for now. This can be
14183 used to extract file names or arguments in files built using distcc through
14184 haproxy. Please refer to distcc's protocol documentation for the complete
14185 list of supported tokens.
14186
14187distcc_param(<token>[,<occ>]) : integer
14188 Parses a distcc message and returns the parameter associated to occurrence
14189 #<occ> of the token <token>. Occurrences start at 1, and when unspecified,
14190 any may match though in practice only the first one is checked for now. This
14191 can be used to extract certain information such as the protocol version, the
14192 file size or the argument in files built using distcc through haproxy.
14193 Another use case consists in waiting for the start of the preprocessed file
14194 contents before connecting to the server to avoid keeping idle connections.
14195 Please refer to distcc's protocol documentation for the complete list of
14196 supported tokens.
14197
14198 Example :
14199 # wait up to 20s for the pre-processed file to be uploaded
14200 tcp-request inspect-delay 20s
14201 tcp-request content accept if { distcc_param(DOTI) -m found }
14202 # send large files to the big farm
14203 use_backend big_farm if { distcc_param(DOTI) gt 1000000 }
14204
Willy Tarreau595ec542013-06-12 21:34:28 +020014205env(<name>) : string
14206 Returns a string containing the value of environment variable <name>. As a
14207 reminder, environment variables are per-process and are sampled when the
14208 process starts. This can be useful to pass some information to a next hop
14209 server, or with ACLs to take specific action when the process is started a
14210 certain way.
14211
14212 Examples :
14213 # Pass the Via header to next hop with the local hostname in it
14214 http-request add-header Via 1.1\ %[env(HOSTNAME)]
14215
14216 # reject cookie-less requests when the STOP environment variable is set
14217 http-request deny if !{ cook(SESSIONID) -m found } { env(STOP) -m found }
14218
Willy Tarreau74ca5042013-06-11 23:12:07 +020014219fe_conn([<frontend>]) : integer
14220 Returns the number of currently established connections on the frontend,
Willy Tarreaud63335a2010-02-26 12:56:52 +010014221 possibly including the connection being evaluated. If no frontend name is
14222 specified, the current one is used. But it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014223 frontend. It can be used to return a sorry page before hard-blocking, or to
14224 use a specific backend to drain new requests when the farm is considered
Davor Ocelice9ed2812017-12-25 17:49:28 +010014225 full. This is mostly used with ACLs but can also be used to pass some
Willy Tarreau74ca5042013-06-11 23:12:07 +020014226 statistics to servers in HTTP headers. See also the "dst_conn", "be_conn",
14227 "fe_sess_rate" fetches.
Willy Tarreaua36af912009-10-10 12:02:45 +020014228
Nenad Merdanovicad9a7e92016-10-03 04:57:37 +020014229fe_req_rate([<frontend>]) : integer
14230 Returns an integer value corresponding to the number of HTTP requests per
14231 second sent to a frontend. This number can differ from "fe_sess_rate" in
14232 situations where client-side keep-alive is enabled.
14233
Willy Tarreau74ca5042013-06-11 23:12:07 +020014234fe_sess_rate([<frontend>]) : integer
14235 Returns an integer value corresponding to the sessions creation rate on the
14236 frontend, in number of new sessions per second. This is used with ACLs to
14237 limit the incoming session rate to an acceptable range in order to prevent
14238 abuse of service at the earliest moment, for example when combined with other
14239 layer 4 ACLs in order to force the clients to wait a bit for the rate to go
14240 down below the limit. It can also be useful to add this element to logs using
14241 a log-format directive. See also the "rate-limit sessions" directive for use
14242 in frontends.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014243
14244 Example :
14245 # This frontend limits incoming mails to 10/s with a max of 100
14246 # concurrent connections. We accept any connection below 10/s, and
14247 # force excess clients to wait for 100 ms. Since clients are limited to
14248 # 100 max, there cannot be more than 10 incoming mails per second.
14249 frontend mail
14250 bind :25
14251 mode tcp
14252 maxconn 100
14253 acl too_fast fe_sess_rate ge 10
14254 tcp-request inspect-delay 100ms
14255 tcp-request content accept if ! too_fast
14256 tcp-request content accept if WAIT_END
Willy Tarreaud72758d2010-01-12 10:42:19 +010014257
Nenad Merdanovic807a6e72017-03-12 22:00:00 +010014258hostname : string
14259 Returns the system hostname.
14260
Thierry FOURNIER07ee64e2015-07-06 23:43:03 +020014261int(<integer>) : signed integer
14262 Returns a signed integer.
14263
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014264ipv4(<ipv4>) : ipv4
14265 Returns an ipv4.
14266
14267ipv6(<ipv6>) : ipv6
14268 Returns an ipv6.
14269
Willy Tarreau70fe9442018-11-22 16:07:39 +010014270lat_ns_avg : integer
14271 Returns the average number of nanoseconds spent between the moment the task
14272 handling the stream is woken up and the moment it is effectively called. This
14273 number is reset for each new request on the same connections in case of HTTP
14274 keep-alive. This value indicates the overall latency inflicted to the current
14275 request by all other requests being processed in parallel, and is a direct
14276 indicator of perceived performance due to noisy neighbours. In order to keep
14277 the value low, it is possible to reduce the scheduler's run queue depth using
14278 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14279 once using "tune.maxpollevents", to decrease the stream's nice value using
14280 the "nice" option on the "bind" lines or in the frontend, or to look for
14281 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14282 whose processing needs to be adjusted or fixed. Compression of large buffers
14283 could be a culprit, like heavy regex or long lists of regex.
14284 Note: this value is exactly lat_ns_tot divided by cpu_calls.
14285
14286lat_ns_tot : integer
14287 Returns the total number of nanoseconds spent between the moment the task
14288 handling the stream is woken up and the moment it is effectively called. This
14289 number is reset for each new request on the same connections in case of HTTP
14290 keep-alive. This value indicates the overall latency inflicted to the current
14291 request by all other requests being processed in parallel, and is a direct
14292 indicator of perceived performance due to noisy neighbours. In order to keep
14293 the value low, it is possible to reduce the scheduler's run queue depth using
14294 "tune.runqueue-depth", to reduce the number of concurrent events processed at
14295 once using "tune.maxpollevents", to decrease the stream's nice value using
14296 the "nice" option on the "bind" lines or in the frontend, or to look for
14297 other heavy requests in logs (those exhibiting large values of "cpu_ns_avg"),
14298 whose processing needs to be adjusted or fixed. Compression of large buffers
14299 could be a culprit, like heavy regex or long lists of regex. Note: while it
14300 may intuitively seem that the total latency adds to a transfer time, it is
14301 almost never true because while a task waits for the CPU, network buffers
14302 continue to fill up and the next call will process more at once. The value
14303 may be artificially high due to a high cpu_calls count, for example when
14304 processing many HTTP chunks, and for this reason it is often preferred to log
14305 lat_ns_avg instead, which is a more relevant performance indicator.
14306
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014307meth(<method>) : method
14308 Returns a method.
14309
Willy Tarreau0f30d262014-11-24 16:02:05 +010014310nbproc : integer
14311 Returns an integer value corresponding to the number of processes that were
14312 started (it equals the global "nbproc" setting). This is useful for logging
14313 and debugging purposes.
14314
Willy Tarreau74ca5042013-06-11 23:12:07 +020014315nbsrv([<backend>]) : integer
14316 Returns an integer value corresponding to the number of usable servers of
14317 either the current backend or the named backend. This is mostly used with
14318 ACLs but can also be useful when added to logs. This is normally used to
Willy Tarreaud63335a2010-02-26 12:56:52 +010014319 switch to an alternate backend when the number of servers is too low to
14320 to handle some load. It is useful to report a failure when combined with
14321 "monitor fail".
Willy Tarreau079ff0a2009-03-05 21:34:28 +010014322
Patrick Hemmerfabb24f2018-08-13 14:07:57 -040014323prio_class : integer
14324 Returns the priority class of the current session for http mode or connection
14325 for tcp mode. The value will be that set by the last call to "http-request
14326 set-priority-class" or "tcp-request content set-priority-class".
14327
14328prio_offset : integer
14329 Returns the priority offset of the current session for http mode or
14330 connection for tcp mode. The value will be that set by the last call to
14331 "http-request set-priority-offset" or "tcp-request content
14332 set-priority-offset".
14333
Willy Tarreau0f30d262014-11-24 16:02:05 +010014334proc : integer
14335 Returns an integer value corresponding to the position of the process calling
14336 the function, between 1 and global.nbproc. This is useful for logging and
14337 debugging purposes.
14338
Willy Tarreau74ca5042013-06-11 23:12:07 +020014339queue([<backend>]) : integer
Willy Tarreaud63335a2010-02-26 12:56:52 +010014340 Returns the total number of queued connections of the designated backend,
14341 including all the connections in server queues. If no backend name is
14342 specified, the current one is used, but it is also possible to check another
Willy Tarreau74ca5042013-06-11 23:12:07 +020014343 one. This is useful with ACLs or to pass statistics to backend servers. This
14344 can be used to take actions when queuing goes above a known level, generally
14345 indicating a surge of traffic or a massive slowdown on the servers. One
14346 possible action could be to reject new users but still accept old ones. See
14347 also the "avg_queue", "be_conn", and "be_sess_rate" fetches.
14348
Willy Tarreau84310e22014-02-14 11:59:04 +010014349rand([<range>]) : integer
14350 Returns a random integer value within a range of <range> possible values,
14351 starting at zero. If the range is not specified, it defaults to 2^32, which
14352 gives numbers between 0 and 4294967295. It can be useful to pass some values
14353 needed to take some routing decisions for example, or just for debugging
14354 purposes. This random must not be used for security purposes.
14355
Willy Tarreau74ca5042013-06-11 23:12:07 +020014356srv_conn([<backend>/]<server>) : integer
14357 Returns an integer value corresponding to the number of currently established
14358 connections on the designated server, possibly including the connection being
14359 evaluated. If <backend> is omitted, then the server is looked up in the
14360 current backend. It can be used to use a specific farm when one server is
14361 full, or to inform the server about our view of the number of active
Patrick Hemmer155e93e2018-06-14 18:01:35 -040014362 connections with it. See also the "fe_conn", "be_conn", "queue", and
14363 "srv_conn_free" fetch methods.
14364
14365srv_conn_free([<backend>/]<server>) : integer
14366 Returns an integer value corresponding to the number of available connections
14367 on the designated server, possibly including the connection being evaluated.
14368 The value does not include queue slots. If <backend> is omitted, then the
14369 server is looked up in the current backend. It can be used to use a specific
14370 farm when one server is full, or to inform the server about our view of the
14371 number of active connections with it. See also the "be_conn_free" and
14372 "srv_conn" fetch methods.
14373
14374 OTHER CAVEATS AND NOTES: If the server maxconn is 0, then this fetch clearly
14375 does not make sense, in which case the value returned will be -1.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014376
14377srv_is_up([<backend>/]<server>) : boolean
14378 Returns true when the designated server is UP, and false when it is either
14379 DOWN or in maintenance mode. If <backend> is omitted, then the server is
14380 looked up in the current backend. It is mainly used to take action based on
Davor Ocelice9ed2812017-12-25 17:49:28 +010014381 an external status reported via a health check (e.g. a geographical site's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014382 availability). Another possible use which is more of a hack consists in
14383 using dummy servers as boolean variables that can be enabled or disabled from
14384 the CLI, so that rules depending on those ACLs can be tweaked in realtime.
14385
Willy Tarreauff2b7af2017-10-13 11:46:26 +020014386srv_queue([<backend>/]<server>) : integer
14387 Returns an integer value corresponding to the number of connections currently
14388 pending in the designated server's queue. If <backend> is omitted, then the
14389 server is looked up in the current backend. It can sometimes be used together
14390 with the "use-server" directive to force to use a known faster server when it
14391 is not much loaded. See also the "srv_conn", "avg_queue" and "queue" sample
14392 fetch methods.
14393
Willy Tarreau74ca5042013-06-11 23:12:07 +020014394srv_sess_rate([<backend>/]<server>) : integer
14395 Returns an integer corresponding to the sessions creation rate on the
14396 designated server, in number of new sessions per second. If <backend> is
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030014397 omitted, then the server is looked up in the current backend. This is mostly
Willy Tarreau74ca5042013-06-11 23:12:07 +020014398 used with ACLs but can make sense with logs too. This is used to switch to an
14399 alternate backend when an expensive or fragile one reaches too high a session
Davor Ocelice9ed2812017-12-25 17:49:28 +010014400 rate, or to limit abuse of service (e.g. prevent latent requests from
Willy Tarreau74ca5042013-06-11 23:12:07 +020014401 overloading servers).
14402
14403 Example :
14404 # Redirect to a separate back
14405 acl srv1_full srv_sess_rate(be1/srv1) gt 50
14406 acl srv2_full srv_sess_rate(be1/srv2) gt 50
14407 use_backend be2 if srv1_full or srv2_full
14408
Willy Tarreau0f30d262014-11-24 16:02:05 +010014409stopping : boolean
14410 Returns TRUE if the process calling the function is currently stopping. This
14411 can be useful for logging, or for relaxing certain checks or helping close
14412 certain connections upon graceful shutdown.
14413
Thierry FOURNIERcc103292015-06-06 19:30:17 +020014414str(<string>) : string
14415 Returns a string.
14416
Willy Tarreau74ca5042013-06-11 23:12:07 +020014417table_avl([<table>]) : integer
14418 Returns the total number of available entries in the current proxy's
14419 stick-table or in the designated stick-table. See also table_cnt.
14420
14421table_cnt([<table>]) : integer
14422 Returns the total number of entries currently in use in the current proxy's
14423 stick-table or in the designated stick-table. See also src_conn_cnt and
14424 table_avl for other entry counting methods.
14425
Christopher Faulet34adb2a2017-11-21 21:45:38 +010014426thread : integer
14427 Returns an integer value corresponding to the position of the thread calling
14428 the function, between 0 and (global.nbthread-1). This is useful for logging
14429 and debugging purposes.
14430
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014431var(<var-name>) : undefined
14432 Returns a variable with the stored type. If the variable is not set, the
Daniel Schneller0b547052016-03-21 20:46:57 +010014433 sample fetch fails. The name of the variable starts with an indication
14434 about its scope. The scopes allowed are:
Christopher Fauletff2613e2016-11-09 11:36:17 +010014435 "proc" : the variable is shared with the whole process
Daniel Schneller0b547052016-03-21 20:46:57 +010014436 "sess" : the variable is shared with the whole session
14437 "txn" : the variable is shared with the transaction (request and
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014438 response),
Daniel Schneller0b547052016-03-21 20:46:57 +010014439 "req" : the variable is shared only during request processing,
14440 "res" : the variable is shared only during response processing.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014441 This prefix is followed by a name. The separator is a '.'. The name may only
Christopher Fauletb71557a2016-10-31 10:49:03 +010014442 contain characters 'a-z', 'A-Z', '0-9', '.' and '_'.
Thierry FOURNIER4834bc72015-06-06 19:29:07 +020014443
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200144447.3.3. Fetching samples at Layer 4
Willy Tarreau74ca5042013-06-11 23:12:07 +020014445----------------------------------
14446
14447The layer 4 usually describes just the transport layer which in haproxy is
14448closest to the connection, where no content is yet made available. The fetch
14449methods described here are usable as low as the "tcp-request connection" rule
14450sets unless they require some future information. Those generally include
14451TCP/IP addresses and ports, as well as elements from stick-tables related to
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014452the incoming connection. For retrieving a value from a sticky counters, the
14453counter number can be explicitly set as 0, 1, or 2 using the pre-defined
Moemen MHEDHBI9cf46342018-09-25 17:50:53 +020014454"sc0_", "sc1_", or "sc2_" prefix. These three pre-defined prefixes can only be
14455used if MAX_SESS_STKCTR value does not exceed 3, otherwise the counter number
14456can be specified as the first integer argument when using the "sc_" prefix.
14457Starting from "sc_0" to "sc_N" where N is (MAX_SESS_STKCTR-1). An optional
14458table may be specified with the "sc*" form, in which case the currently
14459tracked key will be looked up into this alternate table instead of the table
14460currently being tracked.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014461
Jérôme Magnin35e53a62019-01-16 14:38:37 +010014462bc_http_major : integer
Jérôme Magnin86577422018-12-07 09:03:11 +010014463 Returns the backend connection's HTTP major version encoding, which may be 1
14464 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14465 encoding and not the version present in the request header.
14466
Willy Tarreau74ca5042013-06-11 23:12:07 +020014467be_id : integer
14468 Returns an integer containing the current backend's id. It can be used in
14469 frontends with responses to check which backend processed the request.
14470
Marcin Deranekd2471c22016-12-12 14:08:05 +010014471be_name : string
14472 Returns a string containing the current backend's name. It can be used in
14473 frontends with responses to check which backend processed the request.
14474
Willy Tarreau74ca5042013-06-11 23:12:07 +020014475dst : ip
14476 This is the destination IPv4 address of the connection on the client side,
14477 which is the address the client connected to. It can be useful when running
14478 in transparent mode. It is of type IP and works on both IPv4 and IPv6 tables.
14479 On IPv6 tables, IPv4 address is mapped to its IPv6 equivalent, according to
Willy Tarreau64ded3d2019-01-23 10:02:15 +010014480 RFC 4291. When the incoming connection passed through address translation or
14481 redirection involving connection tracking, the original destination address
14482 before the redirection will be reported. On Linux systems, the source and
14483 destination may seldom appear reversed if the nf_conntrack_tcp_loose sysctl
14484 is set, because a late response may reopen a timed out connection and switch
14485 what is believed to be the source and the destination.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014486
14487dst_conn : integer
14488 Returns an integer value corresponding to the number of currently established
14489 connections on the same socket including the one being evaluated. It is
14490 normally used with ACLs but can as well be used to pass the information to
14491 servers in an HTTP header or in logs. It can be used to either return a sorry
14492 page before hard-blocking, or to use a specific backend to drain new requests
14493 when the socket is considered saturated. This offers the ability to assign
14494 different limits to different listening ports or addresses. See also the
14495 "fe_conn" and "be_conn" fetches.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014496
Willy Tarreau16e01562016-08-09 16:46:18 +020014497dst_is_local : boolean
14498 Returns true if the destination address of the incoming connection is local
14499 to the system, or false if the address doesn't exist on the system, meaning
14500 that it was intercepted in transparent mode. It can be useful to apply
14501 certain rules by default to forwarded traffic and other rules to the traffic
Davor Ocelice9ed2812017-12-25 17:49:28 +010014502 targeting the real address of the machine. For example the stats page could
Willy Tarreau16e01562016-08-09 16:46:18 +020014503 be delivered only on this address, or SSH access could be locally redirected.
14504 Please note that the check involves a few system calls, so it's better to do
14505 it only once per connection.
14506
Willy Tarreau74ca5042013-06-11 23:12:07 +020014507dst_port : integer
14508 Returns an integer value corresponding to the destination TCP port of the
14509 connection on the client side, which is the port the client connected to.
14510 This might be used when running in transparent mode, when assigning dynamic
14511 ports to some clients for a whole application session, to stick all users to
14512 a same server, or to pass the destination port information to a server using
14513 an HTTP header.
14514
Willy Tarreau60ca10a2017-08-18 15:26:54 +020014515fc_http_major : integer
14516 Reports the front connection's HTTP major version encoding, which may be 1
14517 for HTTP/0.9 to HTTP/1.1 or 2 for HTTP/2. Note, this is based on the on-wire
14518 encoding and not on the version present in the request header.
14519
Emeric Brun4f603012017-01-05 15:11:44 +010014520fc_rcvd_proxy : boolean
14521 Returns true if the client initiated the connection with a PROXY protocol
14522 header.
14523
Thierry Fournier / OZON.IO6310bef2016-07-24 20:16:50 +020014524fc_rtt(<unit>) : integer
14525 Returns the Round Trip Time (RTT) measured by the kernel for the client
14526 connection. <unit> is facultative, by default the unit is milliseconds. <unit>
14527 can be set to "ms" for milliseconds or "us" for microseconds. If the server
14528 connection is not established, if the connection is not TCP or if the
14529 operating system does not support TCP_INFO, for example Linux kernels before
14530 2.4, the sample fetch fails.
14531
14532fc_rttvar(<unit>) : integer
14533 Returns the Round Trip Time (RTT) variance measured by the kernel for the
14534 client connection. <unit> is facultative, by default the unit is milliseconds.
14535 <unit> can be set to "ms" for milliseconds or "us" for microseconds. If the
14536 server connection is not established, if the connection is not TCP or if the
14537 operating system does not support TCP_INFO, for example Linux kernels before
14538 2.4, the sample fetch fails.
14539
Joe Williams30fcd392016-08-10 07:06:44 -070014540fc_unacked(<unit>) : integer
14541 Returns the unacked counter measured by the kernel for the client connection.
14542 If the server connection is not established, if the connection is not TCP or
14543 if the operating system does not support TCP_INFO, for example Linux kernels
14544 before 2.4, the sample fetch fails.
14545
14546fc_sacked(<unit>) : integer
14547 Returns the sacked counter measured by the kernel for the client connection.
14548 If the server connection is not established, if the connection is not TCP or
14549 if the operating system does not support TCP_INFO, for example Linux kernels
14550 before 2.4, the sample fetch fails.
14551
14552fc_retrans(<unit>) : integer
14553 Returns the retransmits counter measured by the kernel for the client
14554 connection. If the server connection is not established, if the connection is
14555 not TCP or if the operating system does not support TCP_INFO, for example
14556 Linux kernels before 2.4, the sample fetch fails.
14557
14558fc_fackets(<unit>) : integer
14559 Returns the fack counter measured by the kernel for the client
14560 connection. If the server connection is not established, if the connection is
14561 not TCP or if the operating system does not support TCP_INFO, for example
14562 Linux kernels before 2.4, the sample fetch fails.
14563
14564fc_lost(<unit>) : integer
14565 Returns the lost counter measured by the kernel for the client
14566 connection. If the server connection is not established, if the connection is
14567 not TCP or if the operating system does not support TCP_INFO, for example
14568 Linux kernels before 2.4, the sample fetch fails.
14569
14570fc_reordering(<unit>) : integer
14571 Returns the reordering counter measured by the kernel for the client
14572 connection. If the server connection is not established, if the connection is
14573 not TCP or if the operating system does not support TCP_INFO, for example
14574 Linux kernels before 2.4, the sample fetch fails.
14575
Marcin Deranek9a66dfb2018-04-13 14:37:50 +020014576fe_defbe : string
14577 Returns a string containing the frontend's default backend name. It can be
14578 used in frontends to check which backend will handle requests by default.
14579
Willy Tarreau74ca5042013-06-11 23:12:07 +020014580fe_id : integer
14581 Returns an integer containing the current frontend's id. It can be used in
Marcin Deranek6e413ed2016-12-13 12:40:01 +010014582 backends to check from which frontend it was called, or to stick all users
Willy Tarreau74ca5042013-06-11 23:12:07 +020014583 coming via a same frontend to the same server.
14584
Marcin Deranekd2471c22016-12-12 14:08:05 +010014585fe_name : string
14586 Returns a string containing the current frontend's name. It can be used in
14587 backends to check from which frontend it was called, or to stick all users
14588 coming via a same frontend to the same server.
14589
Cyril Bonté62ba8702014-04-22 23:52:25 +020014590sc_bytes_in_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014591sc0_bytes_in_rate([<table>]) : integer
14592sc1_bytes_in_rate([<table>]) : integer
14593sc2_bytes_in_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014594 Returns the average client-to-server bytes rate from the currently tracked
14595 counters, measured in amount of bytes over the period configured in the
14596 table. See also src_bytes_in_rate.
14597
Cyril Bonté62ba8702014-04-22 23:52:25 +020014598sc_bytes_out_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014599sc0_bytes_out_rate([<table>]) : integer
14600sc1_bytes_out_rate([<table>]) : integer
14601sc2_bytes_out_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014602 Returns the average server-to-client bytes rate from the currently tracked
14603 counters, measured in amount of bytes over the period configured in the
14604 table. See also src_bytes_out_rate.
14605
Cyril Bonté62ba8702014-04-22 23:52:25 +020014606sc_clr_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014607sc0_clr_gpc0([<table>]) : integer
14608sc1_clr_gpc0([<table>]) : integer
14609sc2_clr_gpc0([<table>]) : integer
Willy Tarreauf73cd112011-08-13 01:45:16 +020014610 Clears the first General Purpose Counter associated to the currently tracked
14611 counters, and returns its previous value. Before the first invocation, the
Willy Tarreau869948b2013-01-04 14:14:57 +010014612 stored value is zero, so first invocation will always return zero. This is
14613 typically used as a second ACL in an expression in order to mark a connection
14614 when a first ACL was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014615
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014616 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014617 # block if 5 consecutive requests continue to come faster than 10 sess
14618 # per second, and reset the counter as soon as the traffic slows down.
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014619 acl abuse sc0_http_req_rate gt 10
14620 acl kill sc0_inc_gpc0 gt 5
14621 acl save sc0_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014622 tcp-request connection accept if !abuse save
14623 tcp-request connection reject if abuse kill
14624
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014625sc_clr_gpc1(<ctr>[,<table>]) : integer
14626sc0_clr_gpc1([<table>]) : integer
14627sc1_clr_gpc1([<table>]) : integer
14628sc2_clr_gpc1([<table>]) : integer
14629 Clears the second General Purpose Counter associated to the currently tracked
14630 counters, and returns its previous value. Before the first invocation, the
14631 stored value is zero, so first invocation will always return zero. This is
14632 typically used as a second ACL in an expression in order to mark a connection
14633 when a first ACL was verified.
14634
Cyril Bonté62ba8702014-04-22 23:52:25 +020014635sc_conn_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014636sc0_conn_cnt([<table>]) : integer
14637sc1_conn_cnt([<table>]) : integer
14638sc2_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014639 Returns the cumulative number of incoming connections from currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014640 counters. See also src_conn_cnt.
14641
Cyril Bonté62ba8702014-04-22 23:52:25 +020014642sc_conn_cur(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014643sc0_conn_cur([<table>]) : integer
14644sc1_conn_cur([<table>]) : integer
14645sc2_conn_cur([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014646 Returns the current amount of concurrent connections tracking the same
14647 tracked counters. This number is automatically incremented when tracking
14648 begins and decremented when tracking stops. See also src_conn_cur.
14649
Cyril Bonté62ba8702014-04-22 23:52:25 +020014650sc_conn_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014651sc0_conn_rate([<table>]) : integer
14652sc1_conn_rate([<table>]) : integer
14653sc2_conn_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014654 Returns the average connection rate from the currently tracked counters,
14655 measured in amount of connections over the period configured in the table.
14656 See also src_conn_rate.
14657
Cyril Bonté62ba8702014-04-22 23:52:25 +020014658sc_get_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014659sc0_get_gpc0([<table>]) : integer
14660sc1_get_gpc0([<table>]) : integer
14661sc2_get_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014662 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014663 currently tracked counters. See also src_get_gpc0 and sc/sc0/sc1/sc2_inc_gpc0.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014664
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014665sc_get_gpc1(<ctr>[,<table>]) : integer
14666sc0_get_gpc1([<table>]) : integer
14667sc1_get_gpc1([<table>]) : integer
14668sc2_get_gpc1([<table>]) : integer
14669 Returns the value of the second General Purpose Counter associated to the
14670 currently tracked counters. See also src_get_gpc1 and sc/sc0/sc1/sc2_inc_gpc1.
14671
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014672sc_get_gpt0(<ctr>[,<table>]) : integer
14673sc0_get_gpt0([<table>]) : integer
14674sc1_get_gpt0([<table>]) : integer
14675sc2_get_gpt0([<table>]) : integer
14676 Returns the value of the first General Purpose Tag associated to the
14677 currently tracked counters. See also src_get_gpt0.
14678
Cyril Bonté62ba8702014-04-22 23:52:25 +020014679sc_gpc0_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014680sc0_gpc0_rate([<table>]) : integer
14681sc1_gpc0_rate([<table>]) : integer
14682sc2_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014683 Returns the average increment rate of the first General Purpose Counter
14684 associated to the currently tracked counters. It reports the frequency
14685 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014686 src_gpc0_rate, sc/sc0/sc1/sc2_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14687 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14688 be returned, as "gpc0" only holds the event count.
Willy Tarreaue9656522010-08-17 15:40:09 +020014689
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014690sc_gpc1_rate(<ctr>[,<table>]) : integer
14691sc0_gpc1_rate([<table>]) : integer
14692sc1_gpc1_rate([<table>]) : integer
14693sc2_gpc1_rate([<table>]) : integer
14694 Returns the average increment rate of the second General Purpose Counter
14695 associated to the currently tracked counters. It reports the frequency
14696 which the gpc1 counter was incremented over the configured period. See also
14697 src_gpcA_rate, sc/sc0/sc1/sc2_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14698 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14699 be returned, as "gpc1" only holds the event count.
14700
Cyril Bonté62ba8702014-04-22 23:52:25 +020014701sc_http_err_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014702sc0_http_err_cnt([<table>]) : integer
14703sc1_http_err_cnt([<table>]) : integer
14704sc2_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014705 Returns the cumulative number of HTTP errors from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014706 counters. This includes the both request errors and 4xx error responses.
14707 See also src_http_err_cnt.
14708
Cyril Bonté62ba8702014-04-22 23:52:25 +020014709sc_http_err_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014710sc0_http_err_rate([<table>]) : integer
14711sc1_http_err_rate([<table>]) : integer
14712sc2_http_err_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014713 Returns the average rate of HTTP errors from the currently tracked counters,
14714 measured in amount of errors over the period configured in the table. This
14715 includes the both request errors and 4xx error responses. See also
14716 src_http_err_rate.
14717
Cyril Bonté62ba8702014-04-22 23:52:25 +020014718sc_http_req_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014719sc0_http_req_cnt([<table>]) : integer
14720sc1_http_req_cnt([<table>]) : integer
14721sc2_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014722 Returns the cumulative number of HTTP requests from the currently tracked
Willy Tarreaue9656522010-08-17 15:40:09 +020014723 counters. This includes every started request, valid or not. See also
14724 src_http_req_cnt.
14725
Cyril Bonté62ba8702014-04-22 23:52:25 +020014726sc_http_req_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014727sc0_http_req_rate([<table>]) : integer
14728sc1_http_req_rate([<table>]) : integer
14729sc2_http_req_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014730 Returns the average rate of HTTP requests from the currently tracked
14731 counters, measured in amount of requests over the period configured in
14732 the table. This includes every started request, valid or not. See also
14733 src_http_req_rate.
14734
Cyril Bonté62ba8702014-04-22 23:52:25 +020014735sc_inc_gpc0(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014736sc0_inc_gpc0([<table>]) : integer
14737sc1_inc_gpc0([<table>]) : integer
14738sc2_inc_gpc0([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014739 Increments the first General Purpose Counter associated to the currently
Willy Tarreau869948b2013-01-04 14:14:57 +010014740 tracked counters, and returns its new value. Before the first invocation,
14741 the stored value is zero, so first invocation will increase it to 1 and will
14742 return 1. This is typically used as a second ACL in an expression in order
14743 to mark a connection when a first ACL was verified :
Willy Tarreaue9656522010-08-17 15:40:09 +020014744
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014745 Example:
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014746 acl abuse sc0_http_req_rate gt 10
14747 acl kill sc0_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014748 tcp-request connection reject if abuse kill
14749
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014750sc_inc_gpc1(<ctr>[,<table>]) : integer
14751sc0_inc_gpc1([<table>]) : integer
14752sc1_inc_gpc1([<table>]) : integer
14753sc2_inc_gpc1([<table>]) : integer
14754 Increments the second General Purpose Counter associated to the currently
14755 tracked counters, and returns its new value. Before the first invocation,
14756 the stored value is zero, so first invocation will increase it to 1 and will
14757 return 1. This is typically used as a second ACL in an expression in order
14758 to mark a connection when a first ACL was verified.
14759
Cyril Bonté62ba8702014-04-22 23:52:25 +020014760sc_kbytes_in(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014761sc0_kbytes_in([<table>]) : integer
14762sc1_kbytes_in([<table>]) : integer
14763sc2_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014764 Returns the total amount of client-to-server data from the currently tracked
14765 counters, measured in kilobytes. The test is currently performed on 32-bit
14766 integers, which limits values to 4 terabytes. See also src_kbytes_in.
Willy Tarreaue9656522010-08-17 15:40:09 +020014767
Cyril Bonté62ba8702014-04-22 23:52:25 +020014768sc_kbytes_out(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014769sc0_kbytes_out([<table>]) : integer
14770sc1_kbytes_out([<table>]) : integer
14771sc2_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014772 Returns the total amount of server-to-client data from the currently tracked
14773 counters, measured in kilobytes. The test is currently performed on 32-bit
14774 integers, which limits values to 4 terabytes. See also src_kbytes_out.
Willy Tarreaue9656522010-08-17 15:40:09 +020014775
Cyril Bonté62ba8702014-04-22 23:52:25 +020014776sc_sess_cnt(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014777sc0_sess_cnt([<table>]) : integer
14778sc1_sess_cnt([<table>]) : integer
14779sc2_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014780 Returns the cumulative number of incoming connections that were transformed
Willy Tarreaue9656522010-08-17 15:40:09 +020014781 into sessions, which means that they were accepted by a "tcp-request
14782 connection" rule, from the currently tracked counters. A backend may count
14783 more sessions than connections because each connection could result in many
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014784 backend sessions if some HTTP keep-alive is performed over the connection
Willy Tarreaue9656522010-08-17 15:40:09 +020014785 with the client. See also src_sess_cnt.
14786
Cyril Bonté62ba8702014-04-22 23:52:25 +020014787sc_sess_rate(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014788sc0_sess_rate([<table>]) : integer
14789sc1_sess_rate([<table>]) : integer
14790sc2_sess_rate([<table>]) : integer
Willy Tarreaue9656522010-08-17 15:40:09 +020014791 Returns the average session rate from the currently tracked counters,
14792 measured in amount of sessions over the period configured in the table. A
14793 session is a connection that got past the early "tcp-request connection"
14794 rules. A backend may count more sessions than connections because each
14795 connection could result in many backend sessions if some HTTP keep-alive is
Jamie Gloudonaaa21002012-08-25 00:18:33 -040014796 performed over the connection with the client. See also src_sess_rate.
Willy Tarreaue9656522010-08-17 15:40:09 +020014797
Cyril Bonté62ba8702014-04-22 23:52:25 +020014798sc_tracked(<ctr>[,<table>]) : boolean
Willy Tarreau0f791d42013-07-23 19:56:43 +020014799sc0_tracked([<table>]) : boolean
14800sc1_tracked([<table>]) : boolean
14801sc2_tracked([<table>]) : boolean
Willy Tarreau6f1615f2013-06-03 15:15:22 +020014802 Returns true if the designated session counter is currently being tracked by
14803 the current session. This can be useful when deciding whether or not we want
14804 to set some values in a header passed to the server.
14805
Cyril Bonté62ba8702014-04-22 23:52:25 +020014806sc_trackers(<ctr>[,<table>]) : integer
Willy Tarreau0f791d42013-07-23 19:56:43 +020014807sc0_trackers([<table>]) : integer
14808sc1_trackers([<table>]) : integer
14809sc2_trackers([<table>]) : integer
Willy Tarreau2406db42012-12-09 12:16:43 +010014810 Returns the current amount of concurrent connections tracking the same
14811 tracked counters. This number is automatically incremented when tracking
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014812 begins and decremented when tracking stops. It differs from sc0_conn_cur in
Willy Tarreau2406db42012-12-09 12:16:43 +010014813 that it does not rely on any stored information but on the table's reference
14814 count (the "use" value which is returned by "show table" on the CLI). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020014815 may sometimes be more suited for layer7 tracking. It can be used to tell a
14816 server how many concurrent connections there are from a given address for
14817 example.
Willy Tarreau2406db42012-12-09 12:16:43 +010014818
Willy Tarreau74ca5042013-06-11 23:12:07 +020014819so_id : integer
14820 Returns an integer containing the current listening socket's id. It is useful
14821 in frontends involving many "bind" lines, or to stick all users coming via a
14822 same socket to the same server.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014823
Willy Tarreau74ca5042013-06-11 23:12:07 +020014824src : ip
Davor Ocelice9ed2812017-12-25 17:49:28 +010014825 This is the source IPv4 address of the client of the session. It is of type
Willy Tarreau74ca5042013-06-11 23:12:07 +020014826 IP and works on both IPv4 and IPv6 tables. On IPv6 tables, IPv4 addresses are
14827 mapped to their IPv6 equivalent, according to RFC 4291. Note that it is the
14828 TCP-level source address which is used, and not the address of a client
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010014829 behind a proxy. However if the "accept-proxy" or "accept-netscaler-cip" bind
14830 directive is used, it can be the address of a client behind another
14831 PROXY-protocol compatible component for all rule sets except
Willy Tarreau64ded3d2019-01-23 10:02:15 +010014832 "tcp-request connection" which sees the real address. When the incoming
14833 connection passed through address translation or redirection involving
14834 connection tracking, the original destination address before the redirection
14835 will be reported. On Linux systems, the source and destination may seldom
14836 appear reversed if the nf_conntrack_tcp_loose sysctl is set, because a late
14837 response may reopen a timed out connection and switch what is believed to be
14838 the source and the destination.
Willy Tarreaud63335a2010-02-26 12:56:52 +010014839
Thierry FOURNIERd5f624d2013-11-26 11:52:33 +010014840 Example:
14841 # add an HTTP header in requests with the originating address' country
14842 http-request set-header X-Country %[src,map_ip(geoip.lst)]
14843
Willy Tarreau74ca5042013-06-11 23:12:07 +020014844src_bytes_in_rate([<table>]) : integer
14845 Returns the average bytes rate from the incoming connection's source address
14846 in the current proxy's stick-table or in the designated stick-table, measured
14847 in amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014848 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_in_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014849
Willy Tarreau74ca5042013-06-11 23:12:07 +020014850src_bytes_out_rate([<table>]) : integer
14851 Returns the average bytes rate to the incoming connection's source address in
14852 the current proxy's stick-table or in the designated stick-table, measured in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014853 amount of bytes over the period configured in the table. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014854 not found, zero is returned. See also sc/sc0/sc1/sc2_bytes_out_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014855
Willy Tarreau74ca5042013-06-11 23:12:07 +020014856src_clr_gpc0([<table>]) : integer
14857 Clears the first General Purpose Counter associated to the incoming
14858 connection's source address in the current proxy's stick-table or in the
14859 designated stick-table, and returns its previous value. If the address is not
14860 found, an entry is created and 0 is returned. This is typically used as a
14861 second ACL in an expression in order to mark a connection when a first ACL
14862 was verified :
Willy Tarreauf73cd112011-08-13 01:45:16 +020014863
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014864 Example:
Willy Tarreauf73cd112011-08-13 01:45:16 +020014865 # block if 5 consecutive requests continue to come faster than 10 sess
14866 # per second, and reset the counter as soon as the traffic slows down.
14867 acl abuse src_http_req_rate gt 10
14868 acl kill src_inc_gpc0 gt 5
Willy Tarreau869948b2013-01-04 14:14:57 +010014869 acl save src_clr_gpc0 ge 0
Willy Tarreauf73cd112011-08-13 01:45:16 +020014870 tcp-request connection accept if !abuse save
14871 tcp-request connection reject if abuse kill
14872
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014873src_clr_gpc1([<table>]) : integer
14874 Clears the second General Purpose Counter associated to the incoming
14875 connection's source address in the current proxy's stick-table or in the
14876 designated stick-table, and returns its previous value. If the address is not
14877 found, an entry is created and 0 is returned. This is typically used as a
14878 second ACL in an expression in order to mark a connection when a first ACL
14879 was verified.
14880
Willy Tarreau74ca5042013-06-11 23:12:07 +020014881src_conn_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014882 Returns the cumulative number of connections initiated from the current
Willy Tarreau74ca5042013-06-11 23:12:07 +020014883 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014884 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014885 See also sc/sc0/sc1/sc2_conn_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014886
Willy Tarreau74ca5042013-06-11 23:12:07 +020014887src_conn_cur([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014888 Returns the current amount of concurrent connections initiated from the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014889 current incoming connection's source address in the current proxy's
14890 stick-table or in the designated stick-table. If the address is not found,
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014891 zero is returned. See also sc/sc0/sc1/sc2_conn_cur.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014892
Willy Tarreau74ca5042013-06-11 23:12:07 +020014893src_conn_rate([<table>]) : integer
14894 Returns the average connection rate from the incoming connection's source
14895 address in the current proxy's stick-table or in the designated stick-table,
14896 measured in amount of connections over the period configured in the table. If
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014897 the address is not found, zero is returned. See also sc/sc0/sc1/sc2_conn_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014898
Willy Tarreau74ca5042013-06-11 23:12:07 +020014899src_get_gpc0([<table>]) : integer
Willy Tarreauc9705a12010-07-27 20:05:50 +020014900 Returns the value of the first General Purpose Counter associated to the
Willy Tarreau74ca5042013-06-11 23:12:07 +020014901 incoming connection's source address in the current proxy's stick-table or in
Willy Tarreauc9705a12010-07-27 20:05:50 +020014902 the designated stick-table. If the address is not found, zero is returned.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014903 See also sc/sc0/sc1/sc2_get_gpc0 and src_inc_gpc0.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014904
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014905src_get_gpc1([<table>]) : integer
14906 Returns the value of the second General Purpose Counter associated to the
14907 incoming connection's source address in the current proxy's stick-table or in
14908 the designated stick-table. If the address is not found, zero is returned.
14909 See also sc/sc0/sc1/sc2_get_gpc1 and src_inc_gpc1.
14910
Thierry FOURNIER236657b2015-08-19 08:25:14 +020014911src_get_gpt0([<table>]) : integer
14912 Returns the value of the first General Purpose Tag associated to the
14913 incoming connection's source address in the current proxy's stick-table or in
14914 the designated stick-table. If the address is not found, zero is returned.
14915 See also sc/sc0/sc1/sc2_get_gpt0.
14916
Willy Tarreau74ca5042013-06-11 23:12:07 +020014917src_gpc0_rate([<table>]) : integer
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014918 Returns the average increment rate of the first General Purpose Counter
Willy Tarreau74ca5042013-06-11 23:12:07 +020014919 associated to the incoming connection's source address in the current proxy's
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014920 stick-table or in the designated stick-table. It reports the frequency
14921 which the gpc0 counter was incremented over the configured period. See also
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014922 sc/sc0/sc1/sc2_gpc0_rate, src_get_gpc0, and sc/sc0/sc1/sc2_inc_gpc0. Note
14923 that the "gpc0_rate" counter must be stored in the stick-table for a value to
14924 be returned, as "gpc0" only holds the event count.
Willy Tarreauba2ffd12013-05-29 15:54:14 +020014925
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014926src_gpc1_rate([<table>]) : integer
14927 Returns the average increment rate of the second General Purpose Counter
14928 associated to the incoming connection's source address in the current proxy's
14929 stick-table or in the designated stick-table. It reports the frequency
14930 which the gpc1 counter was incremented over the configured period. See also
14931 sc/sc0/sc1/sc2_gpc1_rate, src_get_gpc1, and sc/sc0/sc1/sc2_inc_gpc1. Note
14932 that the "gpc1_rate" counter must be stored in the stick-table for a value to
14933 be returned, as "gpc1" only holds the event count.
14934
Willy Tarreau74ca5042013-06-11 23:12:07 +020014935src_http_err_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014936 Returns the cumulative number of HTTP errors from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014937 source address in the current proxy's stick-table or in the designated
Willy Tarreauc9705a12010-07-27 20:05:50 +020014938 stick-table. This includes the both request errors and 4xx error responses.
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014939 See also sc/sc0/sc1/sc2_http_err_cnt. If the address is not found, zero is
Willy Tarreau74ca5042013-06-11 23:12:07 +020014940 returned.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014941
Willy Tarreau74ca5042013-06-11 23:12:07 +020014942src_http_err_rate([<table>]) : integer
14943 Returns the average rate of HTTP errors from the incoming connection's source
14944 address in the current proxy's stick-table or in the designated stick-table,
14945 measured in amount of errors over the period configured in the table. This
14946 includes the both request errors and 4xx error responses. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014947 not found, zero is returned. See also sc/sc0/sc1/sc2_http_err_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014948
Willy Tarreau74ca5042013-06-11 23:12:07 +020014949src_http_req_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010014950 Returns the cumulative number of HTTP requests from the incoming connection's
Willy Tarreau74ca5042013-06-11 23:12:07 +020014951 source address in the current proxy's stick-table or in the designated stick-
14952 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014953 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014954
Willy Tarreau74ca5042013-06-11 23:12:07 +020014955src_http_req_rate([<table>]) : integer
14956 Returns the average rate of HTTP requests from the incoming connection's
14957 source address in the current proxy's stick-table or in the designated stick-
14958 table, measured in amount of requests over the period configured in the
Willy Tarreauc9705a12010-07-27 20:05:50 +020014959 table. This includes every started request, valid or not. If the address is
Willy Tarreau4d4149c2013-07-23 19:33:46 +020014960 not found, zero is returned. See also sc/sc0/sc1/sc2_http_req_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014961
Willy Tarreau74ca5042013-06-11 23:12:07 +020014962src_inc_gpc0([<table>]) : integer
14963 Increments the first General Purpose Counter associated to the incoming
14964 connection's source address in the current proxy's stick-table or in the
14965 designated stick-table, and returns its new value. If the address is not
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020014966 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc0.
Willy Tarreau74ca5042013-06-11 23:12:07 +020014967 This is typically used as a second ACL in an expression in order to mark a
14968 connection when a first ACL was verified :
Willy Tarreauc9705a12010-07-27 20:05:50 +020014969
Jarno Huuskonen676f6222017-03-30 09:19:45 +030014970 Example:
Willy Tarreauc9705a12010-07-27 20:05:50 +020014971 acl abuse src_http_req_rate gt 10
Willy Tarreau869948b2013-01-04 14:14:57 +010014972 acl kill src_inc_gpc0 gt 0
Willy Tarreaue9656522010-08-17 15:40:09 +020014973 tcp-request connection reject if abuse kill
Willy Tarreauc9705a12010-07-27 20:05:50 +020014974
Frédéric Lécaille6778b272018-01-29 15:22:53 +010014975src_inc_gpc1([<table>]) : integer
14976 Increments the second General Purpose Counter associated to the incoming
14977 connection's source address in the current proxy's stick-table or in the
14978 designated stick-table, and returns its new value. If the address is not
14979 found, an entry is created and 1 is returned. See also sc0/sc2/sc2_inc_gpc1.
14980 This is typically used as a second ACL in an expression in order to mark a
14981 connection when a first ACL was verified.
14982
Willy Tarreau16e01562016-08-09 16:46:18 +020014983src_is_local : boolean
14984 Returns true if the source address of the incoming connection is local to the
14985 system, or false if the address doesn't exist on the system, meaning that it
14986 comes from a remote machine. Note that UNIX addresses are considered local.
14987 It can be useful to apply certain access restrictions based on where the
Davor Ocelice9ed2812017-12-25 17:49:28 +010014988 client comes from (e.g. require auth or https for remote machines). Please
Willy Tarreau16e01562016-08-09 16:46:18 +020014989 note that the check involves a few system calls, so it's better to do it only
14990 once per connection.
14991
Willy Tarreau74ca5042013-06-11 23:12:07 +020014992src_kbytes_in([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020014993 Returns the total amount of data received from the incoming connection's
14994 source address in the current proxy's stick-table or in the designated
14995 stick-table, measured in kilobytes. If the address is not found, zero is
14996 returned. The test is currently performed on 32-bit integers, which limits
14997 values to 4 terabytes. See also sc/sc0/sc1/sc2_kbytes_in.
Willy Tarreauc9705a12010-07-27 20:05:50 +020014998
Willy Tarreau74ca5042013-06-11 23:12:07 +020014999src_kbytes_out([<table>]) : integer
Willy Tarreaua01b9742014-07-10 15:29:24 +020015000 Returns the total amount of data sent to the incoming connection's source
15001 address in the current proxy's stick-table or in the designated stick-table,
15002 measured in kilobytes. If the address is not found, zero is returned. The
15003 test is currently performed on 32-bit integers, which limits values to 4
15004 terabytes. See also sc/sc0/sc1/sc2_kbytes_out.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015005
Willy Tarreau74ca5042013-06-11 23:12:07 +020015006src_port : integer
15007 Returns an integer value corresponding to the TCP source port of the
15008 connection on the client side, which is the port the client connected from.
15009 Usage of this function is very limited as modern protocols do not care much
15010 about source ports nowadays.
Willy Tarreau079ff0a2009-03-05 21:34:28 +010015011
Willy Tarreau74ca5042013-06-11 23:12:07 +020015012src_sess_cnt([<table>]) : integer
Davor Ocelice9ed2812017-12-25 17:49:28 +010015013 Returns the cumulative number of connections initiated from the incoming
Willy Tarreauc9705a12010-07-27 20:05:50 +020015014 connection's source IPv4 address in the current proxy's stick-table or in the
15015 designated stick-table, that were transformed into sessions, which means that
15016 they were accepted by "tcp-request" rules. If the address is not found, zero
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015017 is returned. See also sc/sc0/sc1/sc2_sess_cnt.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015018
Willy Tarreau74ca5042013-06-11 23:12:07 +020015019src_sess_rate([<table>]) : integer
15020 Returns the average session rate from the incoming connection's source
15021 address in the current proxy's stick-table or in the designated stick-table,
15022 measured in amount of sessions over the period configured in the table. A
15023 session is a connection that went past the early "tcp-request" rules. If the
Willy Tarreau4d4149c2013-07-23 19:33:46 +020015024 address is not found, zero is returned. See also sc/sc0/sc1/sc2_sess_rate.
Willy Tarreauc9705a12010-07-27 20:05:50 +020015025
Willy Tarreau74ca5042013-06-11 23:12:07 +020015026src_updt_conn_cnt([<table>]) : integer
15027 Creates or updates the entry associated to the incoming connection's source
15028 address in the current proxy's stick-table or in the designated stick-table.
15029 This table must be configured to store the "conn_cnt" data type, otherwise
15030 the match will be ignored. The current count is incremented by one, and the
15031 expiration timer refreshed. The updated count is returned, so this match
15032 can't return zero. This was used to reject service abusers based on their
15033 source address. Note: it is recommended to use the more complete "track-sc*"
15034 actions in "tcp-request" rules instead.
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015035
15036 Example :
15037 # This frontend limits incoming SSH connections to 3 per 10 second for
15038 # each source address, and rejects excess connections until a 10 second
15039 # silence is observed. At most 20 addresses are tracked.
15040 listen ssh
15041 bind :22
15042 mode tcp
15043 maxconn 100
Willy Tarreauc9705a12010-07-27 20:05:50 +020015044 stick-table type ip size 20 expire 10s store conn_cnt
Willy Tarreau74ca5042013-06-11 23:12:07 +020015045 tcp-request content reject if { src_updt_conn_cnt gt 3 }
Willy Tarreaua975b8f2010-06-05 19:13:27 +020015046 server local 127.0.0.1:22
15047
Willy Tarreau74ca5042013-06-11 23:12:07 +020015048srv_id : integer
15049 Returns an integer containing the server's id when processing the response.
15050 While it's almost only used with ACLs, it may be used for logging or
15051 debugging.
Hervé COMMOWICKdaa824e2011-08-05 12:09:44 +020015052
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200150537.3.4. Fetching samples at Layer 5
Willy Tarreau74ca5042013-06-11 23:12:07 +020015054----------------------------------
Willy Tarreau0b1cd942010-05-16 22:18:27 +020015055
Willy Tarreau74ca5042013-06-11 23:12:07 +020015056The layer 5 usually describes just the session layer which in haproxy is
15057closest to the session once all the connection handshakes are finished, but
15058when no content is yet made available. The fetch methods described here are
15059usable as low as the "tcp-request content" rule sets unless they require some
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015060future information. Those generally include the results of SSL negotiations.
Willy Tarreauc735a072011-03-29 00:57:02 +020015061
Ben Shillitof25e8e52016-12-02 14:25:37 +00001506251d.all(<prop>[,<prop>*]) : string
15063 Returns values for the properties requested as a string, where values are
15064 separated by the delimiter specified with "51degrees-property-separator".
15065 The device is identified using all the important HTTP headers from the
15066 request. The function can be passed up to five property names, and if a
15067 property name can't be found, the value "NoData" is returned.
15068
15069 Example :
15070 # Here the header "X-51D-DeviceTypeMobileTablet" is added to the request
15071 # containing the three properties requested using all relevant headers from
15072 # the request.
15073 frontend http-in
15074 bind *:8081
15075 default_backend servers
15076 http-request set-header X-51D-DeviceTypeMobileTablet \
15077 %[51d.all(DeviceType,IsMobile,IsTablet)]
15078
Emeric Brun645ae792014-04-30 14:21:06 +020015079ssl_bc : boolean
15080 Returns true when the back connection was made via an SSL/TLS transport
15081 layer and is locally deciphered. This means the outgoing connection was made
15082 other a server with the "ssl" option.
15083
15084ssl_bc_alg_keysize : integer
15085 Returns the symmetric cipher key size supported in bits when the outgoing
15086 connection was made over an SSL/TLS transport layer.
15087
Olivier Houchard6b77f492018-11-22 18:18:29 +010015088ssl_bc_alpn : string
15089 This extracts the Application Layer Protocol Negotiation field from an
15090 outgoing connection made via a TLS transport layer.
15091 The result is a string containing the protocol name negociated with the
15092 server. The SSL library must have been built with support for TLS
15093 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15094 not advertised unless the "alpn" keyword on the "server" line specifies a
15095 protocol list. Also, nothing forces the server to pick a protocol from this
15096 list, any other one may be requested. The TLS ALPN extension is meant to
15097 replace the TLS NPN extension. See also "ssl_bc_npn".
15098
Emeric Brun645ae792014-04-30 14:21:06 +020015099ssl_bc_cipher : string
15100 Returns the name of the used cipher when the outgoing connection was made
15101 over an SSL/TLS transport layer.
15102
Emeric Brun74f7ffa2018-02-19 16:14:12 +010015103ssl_bc_is_resumed : boolean
15104 Returns true when the back connection was made over an SSL/TLS transport
15105 layer and the newly created SSL session was resumed using a cached
15106 session or a TLS ticket.
15107
Olivier Houchard6b77f492018-11-22 18:18:29 +010015108ssl_bc_npn : string
15109 This extracts the Next Protocol Negotiation field from an outgoing connection
15110 made via a TLS transport layer. The result is a string containing the
15111 protocol name negociated with the server . The SSL library must have been
15112 built with support for TLS extensions enabled (check haproxy -vv). Note that
15113 the TLS NPN extension is not advertised unless the "npn" keyword on the
15114 "server" line specifies a protocol list. Also, nothing forces the server to
15115 pick a protocol from this list, any other one may be used. Please note that
15116 the TLS NPN extension was replaced with ALPN.
15117
Emeric Brun645ae792014-04-30 14:21:06 +020015118ssl_bc_protocol : string
15119 Returns the name of the used protocol when the outgoing connection was made
15120 over an SSL/TLS transport layer.
15121
Emeric Brunb73a9b02014-04-30 18:49:19 +020015122ssl_bc_unique_id : binary
Emeric Brun645ae792014-04-30 14:21:06 +020015123 When the outgoing connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015124 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15125 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
Emeric Brun645ae792014-04-30 14:21:06 +020015126
15127ssl_bc_session_id : binary
15128 Returns the SSL ID of the back connection when the outgoing connection was
15129 made over an SSL/TLS transport layer. It is useful to log if we want to know
15130 if session was reused or not.
15131
Patrick Hemmere0275472018-04-28 19:15:51 -040015132ssl_bc_session_key : binary
15133 Returns the SSL session master key of the back connection when the outgoing
15134 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15135 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15136 BoringSSL.
15137
Emeric Brun645ae792014-04-30 14:21:06 +020015138ssl_bc_use_keysize : integer
15139 Returns the symmetric cipher key size used in bits when the outgoing
15140 connection was made over an SSL/TLS transport layer.
15141
Willy Tarreau74ca5042013-06-11 23:12:07 +020015142ssl_c_ca_err : integer
15143 When the incoming connection was made over an SSL/TLS transport layer,
15144 returns the ID of the first error detected during verification of the client
15145 certificate at depth > 0, or 0 if no error was encountered during this
15146 verification process. Please refer to your SSL library's documentation to
15147 find the exhaustive list of error codes.
Willy Tarreauc735a072011-03-29 00:57:02 +020015148
Willy Tarreau74ca5042013-06-11 23:12:07 +020015149ssl_c_ca_err_depth : integer
15150 When the incoming connection was made over an SSL/TLS transport layer,
15151 returns the depth in the CA chain of the first error detected during the
15152 verification of the client certificate. If no error is encountered, 0 is
15153 returned.
Willy Tarreau0ba27502007-12-24 16:55:16 +010015154
Emeric Brun43e79582014-10-29 19:03:26 +010015155ssl_c_der : binary
15156 Returns the DER formatted certificate presented by the client when the
15157 incoming connection was made over an SSL/TLS transport layer. When used for
15158 an ACL, the value(s) to match against can be passed in hexadecimal form.
15159
Willy Tarreau74ca5042013-06-11 23:12:07 +020015160ssl_c_err : integer
15161 When the incoming connection was made over an SSL/TLS transport layer,
15162 returns the ID of the first error detected during verification at depth 0, or
15163 0 if no error was encountered during this verification process. Please refer
15164 to your SSL library's documentation to find the exhaustive list of error
15165 codes.
Willy Tarreau62644772008-07-16 18:36:06 +020015166
Willy Tarreau74ca5042013-06-11 23:12:07 +020015167ssl_c_i_dn([<entry>[,<occ>]]) : string
15168 When the incoming connection was made over an SSL/TLS transport layer,
15169 returns the full distinguished name of the issuer of the certificate
15170 presented by the client when no <entry> is specified, or the value of the
15171 first given entry found from the beginning of the DN. If a positive/negative
15172 occurrence number is specified as the optional second argument, it returns
15173 the value of the nth given entry value from the beginning/end of the DN.
15174 For instance, "ssl_c_i_dn(OU,2)" the second organization unit, and
15175 "ssl_c_i_dn(CN)" retrieves the common name.
Willy Tarreau62644772008-07-16 18:36:06 +020015176
Willy Tarreau74ca5042013-06-11 23:12:07 +020015177ssl_c_key_alg : string
15178 Returns the name of the algorithm used to generate the key of the certificate
15179 presented by the client when the incoming connection was made over an SSL/TLS
15180 transport layer.
Willy Tarreau62644772008-07-16 18:36:06 +020015181
Willy Tarreau74ca5042013-06-11 23:12:07 +020015182ssl_c_notafter : string
15183 Returns the end date presented by the client as a formatted string
15184 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15185 transport layer.
Emeric Brunbede3d02009-06-30 17:54:00 +020015186
Willy Tarreau74ca5042013-06-11 23:12:07 +020015187ssl_c_notbefore : string
15188 Returns the start date presented by the client as a formatted string
15189 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15190 transport layer.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015191
Willy Tarreau74ca5042013-06-11 23:12:07 +020015192ssl_c_s_dn([<entry>[,<occ>]]) : string
15193 When the incoming connection was made over an SSL/TLS transport layer,
15194 returns the full distinguished name of the subject of the certificate
15195 presented by the client when no <entry> is specified, or the value of the
15196 first given entry found from the beginning of the DN. If a positive/negative
15197 occurrence number is specified as the optional second argument, it returns
15198 the value of the nth given entry value from the beginning/end of the DN.
15199 For instance, "ssl_c_s_dn(OU,2)" the second organization unit, and
15200 "ssl_c_s_dn(CN)" retrieves the common name.
Willy Tarreaub6672b52011-12-12 17:23:41 +010015201
Willy Tarreau74ca5042013-06-11 23:12:07 +020015202ssl_c_serial : binary
15203 Returns the serial of the certificate presented by the client when the
15204 incoming connection was made over an SSL/TLS transport layer. When used for
15205 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015206
Willy Tarreau74ca5042013-06-11 23:12:07 +020015207ssl_c_sha1 : binary
15208 Returns the SHA-1 fingerprint of the certificate presented by the client when
15209 the incoming connection was made over an SSL/TLS transport layer. This can be
15210 used to stick a client to a server, or to pass this information to a server.
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015211 Note that the output is binary, so if you want to pass that signature to the
15212 server, you need to encode it in hex or base64, such as in the example below:
15213
Jarno Huuskonen676f6222017-03-30 09:19:45 +030015214 Example:
Willy Tarreau2d0caa32014-07-02 19:01:22 +020015215 http-request set-header X-SSL-Client-SHA1 %[ssl_c_sha1,hex]
Emeric Brun2525b6b2012-10-18 15:59:43 +020015216
Willy Tarreau74ca5042013-06-11 23:12:07 +020015217ssl_c_sig_alg : string
15218 Returns the name of the algorithm used to sign the certificate presented by
15219 the client when the incoming connection was made over an SSL/TLS transport
15220 layer.
Emeric Brun87855892012-10-17 17:39:35 +020015221
Willy Tarreau74ca5042013-06-11 23:12:07 +020015222ssl_c_used : boolean
15223 Returns true if current SSL session uses a client certificate even if current
15224 connection uses SSL session resumption. See also "ssl_fc_has_crt".
Emeric Brun7f56e742012-10-19 18:15:40 +020015225
Willy Tarreau74ca5042013-06-11 23:12:07 +020015226ssl_c_verify : integer
15227 Returns the verify result error ID when the incoming connection was made over
15228 an SSL/TLS transport layer, otherwise zero if no error is encountered. Please
15229 refer to your SSL library's documentation for an exhaustive list of error
15230 codes.
Emeric Brunce5ad802012-10-22 14:11:22 +020015231
Willy Tarreau74ca5042013-06-11 23:12:07 +020015232ssl_c_version : integer
15233 Returns the version of the certificate presented by the client when the
15234 incoming connection was made over an SSL/TLS transport layer.
Emeric Brunce5ad802012-10-22 14:11:22 +020015235
Emeric Brun43e79582014-10-29 19:03:26 +010015236ssl_f_der : binary
15237 Returns the DER formatted certificate presented by the frontend when the
15238 incoming connection was made over an SSL/TLS transport layer. When used for
15239 an ACL, the value(s) to match against can be passed in hexadecimal form.
15240
Willy Tarreau74ca5042013-06-11 23:12:07 +020015241ssl_f_i_dn([<entry>[,<occ>]]) : string
15242 When the incoming connection was made over an SSL/TLS transport layer,
15243 returns the full distinguished name of the issuer of the certificate
15244 presented by the frontend when no <entry> is specified, or the value of the
15245 first given entry found from the beginning of the DN. If a positive/negative
Emeric Brun87855892012-10-17 17:39:35 +020015246 occurrence number is specified as the optional second argument, it returns
Willy Tarreau74ca5042013-06-11 23:12:07 +020015247 the value of the nth given entry value from the beginning/end of the DN.
15248 For instance, "ssl_f_i_dn(OU,2)" the second organization unit, and
15249 "ssl_f_i_dn(CN)" retrieves the common name.
Emeric Brun87855892012-10-17 17:39:35 +020015250
Willy Tarreau74ca5042013-06-11 23:12:07 +020015251ssl_f_key_alg : string
15252 Returns the name of the algorithm used to generate the key of the certificate
15253 presented by the frontend when the incoming connection was made over an
15254 SSL/TLS transport layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015255
Willy Tarreau74ca5042013-06-11 23:12:07 +020015256ssl_f_notafter : string
15257 Returns the end date presented by the frontend as a formatted string
15258 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15259 transport layer.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015260
Willy Tarreau74ca5042013-06-11 23:12:07 +020015261ssl_f_notbefore : string
15262 Returns the start date presented by the frontend as a formatted string
15263 YYMMDDhhmmss[Z] when the incoming connection was made over an SSL/TLS
15264 transport layer.
Emeric Brun87855892012-10-17 17:39:35 +020015265
Willy Tarreau74ca5042013-06-11 23:12:07 +020015266ssl_f_s_dn([<entry>[,<occ>]]) : string
15267 When the incoming connection was made over an SSL/TLS transport layer,
15268 returns the full distinguished name of the subject of the certificate
15269 presented by the frontend when no <entry> is specified, or the value of the
15270 first given entry found from the beginning of the DN. If a positive/negative
15271 occurrence number is specified as the optional second argument, it returns
15272 the value of the nth given entry value from the beginning/end of the DN.
15273 For instance, "ssl_f_s_dn(OU,2)" the second organization unit, and
15274 "ssl_f_s_dn(CN)" retrieves the common name.
Emeric Brunce5ad802012-10-22 14:11:22 +020015275
Willy Tarreau74ca5042013-06-11 23:12:07 +020015276ssl_f_serial : binary
15277 Returns the serial of the certificate presented by the frontend when the
15278 incoming connection was made over an SSL/TLS transport layer. When used for
15279 an ACL, the value(s) to match against can be passed in hexadecimal form.
Emeric Brun87855892012-10-17 17:39:35 +020015280
Emeric Brun55f4fa82014-04-30 17:11:25 +020015281ssl_f_sha1 : binary
15282 Returns the SHA-1 fingerprint of the certificate presented by the frontend
15283 when the incoming connection was made over an SSL/TLS transport layer. This
15284 can be used to know which certificate was chosen using SNI.
15285
Willy Tarreau74ca5042013-06-11 23:12:07 +020015286ssl_f_sig_alg : string
15287 Returns the name of the algorithm used to sign the certificate presented by
15288 the frontend when the incoming connection was made over an SSL/TLS transport
15289 layer.
Emeric Brun7f56e742012-10-19 18:15:40 +020015290
Willy Tarreau74ca5042013-06-11 23:12:07 +020015291ssl_f_version : integer
15292 Returns the version of the certificate presented by the frontend when the
15293 incoming connection was made over an SSL/TLS transport layer.
15294
15295ssl_fc : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015296 Returns true when the front connection was made via an SSL/TLS transport
15297 layer and is locally deciphered. This means it has matched a socket declared
15298 with a "bind" line having the "ssl" option.
15299
Willy Tarreau74ca5042013-06-11 23:12:07 +020015300 Example :
15301 # This passes "X-Proto: https" to servers when client connects over SSL
15302 listen http-https
15303 bind :80
15304 bind :443 ssl crt /etc/haproxy.pem
15305 http-request add-header X-Proto https if { ssl_fc }
15306
15307ssl_fc_alg_keysize : integer
15308 Returns the symmetric cipher key size supported in bits when the incoming
15309 connection was made over an SSL/TLS transport layer.
15310
15311ssl_fc_alpn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015312 This extracts the Application Layer Protocol Negotiation field from an
Willy Tarreau74ca5042013-06-11 23:12:07 +020015313 incoming connection made via a TLS transport layer and locally deciphered by
15314 haproxy. The result is a string containing the protocol name advertised by
15315 the client. The SSL library must have been built with support for TLS
15316 extensions enabled (check haproxy -vv). Note that the TLS ALPN extension is
15317 not advertised unless the "alpn" keyword on the "bind" line specifies a
15318 protocol list. Also, nothing forces the client to pick a protocol from this
15319 list, any other one may be requested. The TLS ALPN extension is meant to
15320 replace the TLS NPN extension. See also "ssl_fc_npn".
15321
Willy Tarreau74ca5042013-06-11 23:12:07 +020015322ssl_fc_cipher : string
15323 Returns the name of the used cipher when the incoming connection was made
15324 over an SSL/TLS transport layer.
Willy Tarreauab861d32013-04-02 02:30:41 +020015325
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015326ssl_fc_cipherlist_bin : binary
15327 Returns the binary form of the client hello cipher list. The maximum returned
15328 value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015329 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015330
15331ssl_fc_cipherlist_hex : string
15332 Returns the binary form of the client hello cipher list encoded as
15333 hexadecimal. The maximum returned value length is according with the value of
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015334 "tune.ssl.capture-cipherlist-size".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015335
15336ssl_fc_cipherlist_str : string
15337 Returns the decoded text form of the client hello cipher list. The maximum
15338 number of ciphers returned is according with the value of
15339 "tune.ssl.capture-cipherlist-size". Note that this sample-fetch is only
Davor Ocelice9ed2812017-12-25 17:49:28 +010015340 available with OpenSSL >= 1.0.2. If the function is not enabled, this
Emmanuel Hocdetddcde192017-09-01 17:32:08 +020015341 sample-fetch returns the hash like "ssl_fc_cipherlist_xxh".
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015342
15343ssl_fc_cipherlist_xxh : integer
15344 Returns a xxh64 of the cipher list. This hash can be return only is the value
15345 "tune.ssl.capture-cipherlist-size" is set greater than 0, however the hash
Emmanuel Hocdetaaee7502017-03-07 18:34:58 +010015346 take in account all the data of the cipher list.
Thierry FOURNIER5bf77322017-02-25 12:45:22 +010015347
Willy Tarreau74ca5042013-06-11 23:12:07 +020015348ssl_fc_has_crt : boolean
Emeric Brun2525b6b2012-10-18 15:59:43 +020015349 Returns true if a client certificate is present in an incoming connection over
15350 SSL/TLS transport layer. Useful if 'verify' statement is set to 'optional'.
Emeric Brun9143d372012-12-20 15:44:16 +010015351 Note: on SSL session resumption with Session ID or TLS ticket, client
15352 certificate is not present in the current connection but may be retrieved
15353 from the cache or the ticket. So prefer "ssl_c_used" if you want to check if
15354 current SSL session uses a client certificate.
Emeric Brun2525b6b2012-10-18 15:59:43 +020015355
Olivier Houchardccaa7de2017-10-02 11:51:03 +020015356ssl_fc_has_early : boolean
15357 Returns true if early data were sent, and the handshake didn't happen yet. As
15358 it has security implications, it is useful to be able to refuse those, or
15359 wait until the handshake happened.
15360
Willy Tarreau74ca5042013-06-11 23:12:07 +020015361ssl_fc_has_sni : boolean
15362 This checks for the presence of a Server Name Indication TLS extension (SNI)
Willy Tarreauf7bc57c2012-10-03 00:19:48 +020015363 in an incoming connection was made over an SSL/TLS transport layer. Returns
15364 true when the incoming connection presents a TLS SNI field. This requires
15365 that the SSL library is build with support for TLS extensions enabled (check
15366 haproxy -vv).
Willy Tarreau7875d092012-09-10 08:20:03 +020015367
Nenad Merdanovic1516fe32016-05-17 03:31:21 +020015368ssl_fc_is_resumed : boolean
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015369 Returns true if the SSL/TLS session has been resumed through the use of
Jérôme Magnin4a326cb2018-01-15 14:01:17 +010015370 SSL session cache or TLS tickets on an incoming connection over an SSL/TLS
15371 transport layer.
Nenad Merdanovic26ea8222015-05-18 02:28:57 +020015372
Willy Tarreau74ca5042013-06-11 23:12:07 +020015373ssl_fc_npn : string
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015374 This extracts the Next Protocol Negotiation field from an incoming connection
Willy Tarreau74ca5042013-06-11 23:12:07 +020015375 made via a TLS transport layer and locally deciphered by haproxy. The result
15376 is a string containing the protocol name advertised by the client. The SSL
15377 library must have been built with support for TLS extensions enabled (check
15378 haproxy -vv). Note that the TLS NPN extension is not advertised unless the
15379 "npn" keyword on the "bind" line specifies a protocol list. Also, nothing
15380 forces the client to pick a protocol from this list, any other one may be
15381 requested. Please note that the TLS NPN extension was replaced with ALPN.
Willy Tarreaua33c6542012-10-15 13:19:06 +020015382
Willy Tarreau74ca5042013-06-11 23:12:07 +020015383ssl_fc_protocol : string
15384 Returns the name of the used protocol when the incoming connection was made
15385 over an SSL/TLS transport layer.
Willy Tarreau7875d092012-09-10 08:20:03 +020015386
Emeric Brunb73a9b02014-04-30 18:49:19 +020015387ssl_fc_unique_id : binary
David Sc1ad52e2014-04-08 18:48:47 -040015388 When the incoming connection was made over an SSL/TLS transport layer,
Emeric Brunb73a9b02014-04-30 18:49:19 +020015389 returns the TLS unique ID as defined in RFC5929 section 3. The unique id
15390 can be encoded to base64 using the converter: "ssl_bc_unique_id,base64".
David Sc1ad52e2014-04-08 18:48:47 -040015391
Willy Tarreau74ca5042013-06-11 23:12:07 +020015392ssl_fc_session_id : binary
15393 Returns the SSL ID of the front connection when the incoming connection was
15394 made over an SSL/TLS transport layer. It is useful to stick a given client to
15395 a server. It is important to note that some browsers refresh their session ID
15396 every few minutes.
Willy Tarreau7875d092012-09-10 08:20:03 +020015397
Patrick Hemmere0275472018-04-28 19:15:51 -040015398ssl_fc_session_key : binary
15399 Returns the SSL session master key of the front connection when the incoming
15400 connection was made over an SSL/TLS transport layer. It is useful to decrypt
15401 traffic sent using ephemeral ciphers. This requires OpenSSL >= 1.1.0, or
15402 BoringSSL.
15403
15404
Willy Tarreau74ca5042013-06-11 23:12:07 +020015405ssl_fc_sni : string
15406 This extracts the Server Name Indication TLS extension (SNI) field from an
15407 incoming connection made via an SSL/TLS transport layer and locally
15408 deciphered by haproxy. The result (when present) typically is a string
15409 matching the HTTPS host name (253 chars or less). The SSL library must have
15410 been built with support for TLS extensions enabled (check haproxy -vv).
15411
15412 This fetch is different from "req_ssl_sni" above in that it applies to the
15413 connection being deciphered by haproxy and not to SSL contents being blindly
15414 forwarded. See also "ssl_fc_sni_end" and "ssl_fc_sni_reg" below. This
Cyril Bonté9c1eb1e2012-10-09 22:45:34 +020015415 requires that the SSL library is build with support for TLS extensions
15416 enabled (check haproxy -vv).
Willy Tarreau62644772008-07-16 18:36:06 +020015417
Willy Tarreau74ca5042013-06-11 23:12:07 +020015418 ACL derivatives :
Willy Tarreau74ca5042013-06-11 23:12:07 +020015419 ssl_fc_sni_end : suffix match
15420 ssl_fc_sni_reg : regex match
Emeric Brun589fcad2012-10-16 14:13:26 +020015421
Willy Tarreau74ca5042013-06-11 23:12:07 +020015422ssl_fc_use_keysize : integer
15423 Returns the symmetric cipher key size used in bits when the incoming
15424 connection was made over an SSL/TLS transport layer.
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015425
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015426
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200154277.3.5. Fetching samples from buffer contents (Layer 6)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015428------------------------------------------------------
Willy Tarreaub6fb4202008-07-20 11:18:28 +020015429
Willy Tarreau74ca5042013-06-11 23:12:07 +020015430Fetching samples from buffer contents is a bit different from the previous
15431sample fetches above because the sampled data are ephemeral. These data can
15432only be used when they're available and will be lost when they're forwarded.
15433For this reason, samples fetched from buffer contents during a request cannot
15434be used in a response for example. Even while the data are being fetched, they
15435can change. Sometimes it is necessary to set some delays or combine multiple
15436sample fetch methods to ensure that the expected data are complete and usable,
15437for example through TCP request content inspection. Please see the "tcp-request
15438content" keyword for more detailed information on the subject.
Willy Tarreau62644772008-07-16 18:36:06 +020015439
Willy Tarreau74ca5042013-06-11 23:12:07 +020015440payload(<offset>,<length>) : binary (deprecated)
Davor Ocelice9ed2812017-12-25 17:49:28 +010015441 This is an alias for "req.payload" when used in the context of a request (e.g.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015442 "stick on", "stick match"), and for "res.payload" when used in the context of
15443 a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015444
Willy Tarreau74ca5042013-06-11 23:12:07 +020015445payload_lv(<offset1>,<length>[,<offset2>]) : binary (deprecated)
15446 This is an alias for "req.payload_lv" when used in the context of a request
Davor Ocelice9ed2812017-12-25 17:49:28 +010015447 (e.g. "stick on", "stick match"), and for "res.payload_lv" when used in the
Willy Tarreau74ca5042013-06-11 23:12:07 +020015448 context of a response such as in "stick store response".
Willy Tarreau0ba27502007-12-24 16:55:16 +010015449
Thierry FOURNIERd7d88812017-04-19 15:15:14 +020015450req.hdrs : string
15451 Returns the current request headers as string including the last empty line
15452 separating headers from the request body. The last empty line can be used to
15453 detect a truncated header block. This sample fetch is useful for some SPOE
15454 headers analyzers and for advanced logging.
15455
Thierry FOURNIER5617dce2017-04-09 05:38:19 +020015456req.hdrs_bin : binary
15457 Returns the current request headers contained in preparsed binary form. This
15458 is useful for offloading some processing with SPOE. Each string is described
15459 by a length followed by the number of bytes indicated in the length. The
15460 length is represented using the variable integer encoding detailed in the
15461 SPOE documentation. The end of the list is marked by a couple of empty header
15462 names and values (length of 0 for both).
15463
15464 *(<str:header-name><str:header-value>)<empty string><empty string>
15465
15466 int: refer to the SPOE documentation for the encoding
15467 str: <int:length><bytes>
15468
Willy Tarreau74ca5042013-06-11 23:12:07 +020015469req.len : integer
15470req_len : integer (deprecated)
15471 Returns an integer value corresponding to the number of bytes present in the
15472 request buffer. This is mostly used in ACL. It is important to understand
15473 that this test does not return false as long as the buffer is changing. This
15474 means that a check with equality to zero will almost always immediately match
15475 at the beginning of the session, while a test for more data will wait for
15476 that data to come in and return false only when haproxy is certain that no
15477 more data will come in. This test was designed to be used with TCP request
15478 content inspection.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015479
Willy Tarreau74ca5042013-06-11 23:12:07 +020015480req.payload(<offset>,<length>) : binary
15481 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015482 in the request buffer. As a special case, if the <length> argument is zero,
15483 the the whole buffer from <offset> to the end is extracted. This can be used
15484 with ACLs in order to check for the presence of some content in a buffer at
15485 any location.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015486
Willy Tarreau74ca5042013-06-11 23:12:07 +020015487 ACL alternatives :
15488 payload(<offset>,<length>) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015489
Willy Tarreau74ca5042013-06-11 23:12:07 +020015490req.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15491 This extracts a binary block whose size is specified at <offset1> for <length>
15492 bytes, and which starts at <offset2> if specified or just after the length in
15493 the request buffer. The <offset2> parameter also supports relative offsets if
15494 prepended with a '+' or '-' sign.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015495
Willy Tarreau74ca5042013-06-11 23:12:07 +020015496 ACL alternatives :
15497 payload_lv(<offset1>,<length>[,<offset2>]) : hex binary match
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015498
Willy Tarreau74ca5042013-06-11 23:12:07 +020015499 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015500
Willy Tarreau74ca5042013-06-11 23:12:07 +020015501req.proto_http : boolean
15502req_proto_http : boolean (deprecated)
15503 Returns true when data in the request buffer look like HTTP and correctly
15504 parses as such. It is the same parser as the common HTTP request parser which
15505 is used so there should be no surprises. The test does not match until the
15506 request is complete, failed or timed out. This test may be used to report the
15507 protocol in TCP logs, but the biggest use is to block TCP request analysis
15508 until a complete HTTP request is present in the buffer, for example to track
15509 a header.
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015510
Willy Tarreau74ca5042013-06-11 23:12:07 +020015511 Example:
15512 # track request counts per "base" (concatenation of Host+URL)
15513 tcp-request inspect-delay 10s
15514 tcp-request content reject if !HTTP
Willy Tarreaube4a3ef2013-06-17 15:04:07 +020015515 tcp-request content track-sc0 base table req-rate
Willy Tarreaua7ad50c2012-04-29 15:39:40 +020015516
Willy Tarreau74ca5042013-06-11 23:12:07 +020015517req.rdp_cookie([<name>]) : string
15518rdp_cookie([<name>]) : string (deprecated)
15519 When the request buffer looks like the RDP protocol, extracts the RDP cookie
15520 <name>, or any cookie if unspecified. The parser only checks for the first
15521 cookie, as illustrated in the RDP protocol specification. The cookie name is
15522 case insensitive. Generally the "MSTS" cookie name will be used, as it can
15523 contain the user name of the client connecting to the server if properly
15524 configured on the client. The "MSTSHASH" cookie is often used as well for
15525 session stickiness to servers.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015526
Willy Tarreau74ca5042013-06-11 23:12:07 +020015527 This differs from "balance rdp-cookie" in that any balancing algorithm may be
15528 used and thus the distribution of clients to backend servers is not linked to
15529 a hash of the RDP cookie. It is envisaged that using a balancing algorithm
15530 such as "balance roundrobin" or "balance leastconn" will lead to a more even
15531 distribution of clients to backend servers than the hash used by "balance
15532 rdp-cookie".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015533
Willy Tarreau74ca5042013-06-11 23:12:07 +020015534 ACL derivatives :
15535 req_rdp_cookie([<name>]) : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015536
Willy Tarreau74ca5042013-06-11 23:12:07 +020015537 Example :
15538 listen tse-farm
15539 bind 0.0.0.0:3389
15540 # wait up to 5s for an RDP cookie in the request
15541 tcp-request inspect-delay 5s
15542 tcp-request content accept if RDP_COOKIE
15543 # apply RDP cookie persistence
15544 persist rdp-cookie
15545 # Persist based on the mstshash cookie
15546 # This is only useful makes sense if
15547 # balance rdp-cookie is not used
15548 stick-table type string size 204800
15549 stick on req.rdp_cookie(mstshash)
15550 server srv1 1.1.1.1:3389
15551 server srv1 1.1.1.2:3389
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015552
Willy Tarreau74ca5042013-06-11 23:12:07 +020015553 See also : "balance rdp-cookie", "persist rdp-cookie", "tcp-request" and the
15554 "req_rdp_cookie" ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015555
Willy Tarreau74ca5042013-06-11 23:12:07 +020015556req.rdp_cookie_cnt([name]) : integer
15557rdp_cookie_cnt([name]) : integer (deprecated)
15558 Tries to parse the request buffer as RDP protocol, then returns an integer
15559 corresponding to the number of RDP cookies found. If an optional cookie name
15560 is passed, only cookies matching this name are considered. This is mostly
15561 used in ACL.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015562
Willy Tarreau74ca5042013-06-11 23:12:07 +020015563 ACL derivatives :
15564 req_rdp_cookie_cnt([<name>]) : integer match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015565
Alex Zorin4afdd132018-12-30 13:56:28 +110015566req.ssl_alpn : string
15567 Returns a string containing the values of the Application-Layer Protocol
15568 Negotiation (ALPN) TLS extension (RFC7301), sent by the client within the SSL
15569 ClientHello message. Note that this only applies to raw contents found in the
15570 request buffer and not to the contents deciphered via an SSL data layer, so
15571 this will not work with "bind" lines having the "ssl" option. This is useful
15572 in ACL to make a routing decision based upon the ALPN preferences of a TLS
Jarno Huuskonene504f812019-01-03 07:56:49 +020015573 client, like in the example below. See also "ssl_fc_alpn".
Alex Zorin4afdd132018-12-30 13:56:28 +110015574
15575 Examples :
15576 # Wait for a client hello for at most 5 seconds
15577 tcp-request inspect-delay 5s
15578 tcp-request content accept if { req_ssl_hello_type 1 }
Jarno Huuskonene504f812019-01-03 07:56:49 +020015579 use_backend bk_acme if { req.ssl_alpn acme-tls/1 }
Alex Zorin4afdd132018-12-30 13:56:28 +110015580 default_backend bk_default
15581
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015582req.ssl_ec_ext : boolean
15583 Returns a boolean identifying if client sent the Supported Elliptic Curves
15584 Extension as defined in RFC4492, section 5.1. within the SSL ClientHello
Cyril Bonté307ee1e2015-09-28 23:16:06 +020015585 message. This can be used to present ECC compatible clients with EC
15586 certificate and to use RSA for all others, on the same IP address. Note that
15587 this only applies to raw contents found in the request buffer and not to
15588 contents deciphered via an SSL data layer, so this will not work with "bind"
15589 lines having the "ssl" option.
Nenad Merdanovic5fc7d7e2015-07-07 22:00:17 +020015590
Willy Tarreau74ca5042013-06-11 23:12:07 +020015591req.ssl_hello_type : integer
15592req_ssl_hello_type : integer (deprecated)
15593 Returns an integer value containing the type of the SSL hello message found
15594 in the request buffer if the buffer contains data that parse as a complete
15595 SSL (v3 or superior) client hello message. Note that this only applies to raw
15596 contents found in the request buffer and not to contents deciphered via an
15597 SSL data layer, so this will not work with "bind" lines having the "ssl"
15598 option. This is mostly used in ACL to detect presence of an SSL hello message
15599 that is supposed to contain an SSL session ID usable for stickiness.
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015600
Willy Tarreau74ca5042013-06-11 23:12:07 +020015601req.ssl_sni : string
15602req_ssl_sni : string (deprecated)
15603 Returns a string containing the value of the Server Name TLS extension sent
15604 by a client in a TLS stream passing through the request buffer if the buffer
15605 contains data that parse as a complete SSL (v3 or superior) client hello
15606 message. Note that this only applies to raw contents found in the request
15607 buffer and not to contents deciphered via an SSL data layer, so this will not
15608 work with "bind" lines having the "ssl" option. SNI normally contains the
15609 name of the host the client tries to connect to (for recent browsers). SNI is
15610 useful for allowing or denying access to certain hosts when SSL/TLS is used
15611 by the client. This test was designed to be used with TCP request content
15612 inspection. If content switching is needed, it is recommended to first wait
15613 for a complete client hello (type 1), like in the example below. See also
15614 "ssl_fc_sni".
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015615
Willy Tarreau74ca5042013-06-11 23:12:07 +020015616 ACL derivatives :
15617 req_ssl_sni : exact string match
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015618
Willy Tarreau74ca5042013-06-11 23:12:07 +020015619 Examples :
15620 # Wait for a client hello for at most 5 seconds
15621 tcp-request inspect-delay 5s
15622 tcp-request content accept if { req_ssl_hello_type 1 }
15623 use_backend bk_allow if { req_ssl_sni -f allowed_sites }
15624 default_backend bk_sorry_page
Willy Tarreau04aa6a92012-04-06 18:57:55 +020015625
Pradeep Jindalbb2acf52015-09-29 10:12:57 +053015626req.ssl_st_ext : integer
15627 Returns 0 if the client didn't send a SessionTicket TLS Extension (RFC5077)
15628 Returns 1 if the client sent SessionTicket TLS Extension
15629 Returns 2 if the client also sent non-zero length TLS SessionTicket
15630 Note that this only applies to raw contents found in the request buffer and
15631 not to contents deciphered via an SSL data layer, so this will not work with
15632 "bind" lines having the "ssl" option. This can for example be used to detect
15633 whether the client sent a SessionTicket or not and stick it accordingly, if
15634 no SessionTicket then stick on SessionID or don't stick as there's no server
15635 side state is there when SessionTickets are in use.
15636
Willy Tarreau74ca5042013-06-11 23:12:07 +020015637req.ssl_ver : integer
15638req_ssl_ver : integer (deprecated)
15639 Returns an integer value containing the version of the SSL/TLS protocol of a
15640 stream present in the request buffer. Both SSLv2 hello messages and SSLv3
15641 messages are supported. TLSv1 is announced as SSL version 3.1. The value is
15642 composed of the major version multiplied by 65536, added to the minor
15643 version. Note that this only applies to raw contents found in the request
15644 buffer and not to contents deciphered via an SSL data layer, so this will not
15645 work with "bind" lines having the "ssl" option. The ACL version of the test
Davor Ocelice9ed2812017-12-25 17:49:28 +010015646 matches against a decimal notation in the form MAJOR.MINOR (e.g. 3.1). This
Willy Tarreau74ca5042013-06-11 23:12:07 +020015647 fetch is mostly used in ACL.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015648
Willy Tarreau74ca5042013-06-11 23:12:07 +020015649 ACL derivatives :
15650 req_ssl_ver : decimal match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015651
Willy Tarreau47e8eba2013-09-11 23:28:46 +020015652res.len : integer
15653 Returns an integer value corresponding to the number of bytes present in the
15654 response buffer. This is mostly used in ACL. It is important to understand
15655 that this test does not return false as long as the buffer is changing. This
15656 means that a check with equality to zero will almost always immediately match
15657 at the beginning of the session, while a test for more data will wait for
15658 that data to come in and return false only when haproxy is certain that no
15659 more data will come in. This test was designed to be used with TCP response
15660 content inspection.
15661
Willy Tarreau74ca5042013-06-11 23:12:07 +020015662res.payload(<offset>,<length>) : binary
15663 This extracts a binary block of <length> bytes and starting at byte <offset>
Willy Tarreau00f00842013-08-02 11:07:32 +020015664 in the response buffer. As a special case, if the <length> argument is zero,
15665 the the whole buffer from <offset> to the end is extracted. This can be used
15666 with ACLs in order to check for the presence of some content in a buffer at
15667 any location.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015668
Willy Tarreau74ca5042013-06-11 23:12:07 +020015669res.payload_lv(<offset1>,<length>[,<offset2>]) : binary
15670 This extracts a binary block whose size is specified at <offset1> for <length>
15671 bytes, and which starts at <offset2> if specified or just after the length in
15672 the response buffer. The <offset2> parameter also supports relative offsets
15673 if prepended with a '+' or '-' sign.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015674
Willy Tarreau74ca5042013-06-11 23:12:07 +020015675 Example : please consult the example from the "stick store-response" keyword.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015676
Willy Tarreau971f7b62015-09-29 14:06:59 +020015677res.ssl_hello_type : integer
15678rep_ssl_hello_type : integer (deprecated)
15679 Returns an integer value containing the type of the SSL hello message found
15680 in the response buffer if the buffer contains data that parses as a complete
15681 SSL (v3 or superior) hello message. Note that this only applies to raw
15682 contents found in the response buffer and not to contents deciphered via an
15683 SSL data layer, so this will not work with "server" lines having the "ssl"
15684 option. This is mostly used in ACL to detect presence of an SSL hello message
15685 that is supposed to contain an SSL session ID usable for stickiness.
15686
Willy Tarreau74ca5042013-06-11 23:12:07 +020015687wait_end : boolean
15688 This fetch either returns true when the inspection period is over, or does
15689 not fetch. It is only used in ACLs, in conjunction with content analysis to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015690 avoid returning a wrong verdict early. It may also be used to delay some
Willy Tarreau74ca5042013-06-11 23:12:07 +020015691 actions, such as a delayed reject for some special addresses. Since it either
15692 stops the rules evaluation or immediately returns true, it is recommended to
Davor Ocelice9ed2812017-12-25 17:49:28 +010015693 use this acl as the last one in a rule. Please note that the default ACL
Willy Tarreau74ca5042013-06-11 23:12:07 +020015694 "WAIT_END" is always usable without prior declaration. This test was designed
15695 to be used with TCP request content inspection.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015696
Willy Tarreau74ca5042013-06-11 23:12:07 +020015697 Examples :
15698 # delay every incoming request by 2 seconds
15699 tcp-request inspect-delay 2s
15700 tcp-request content accept if WAIT_END
Willy Tarreaud63335a2010-02-26 12:56:52 +010015701
Willy Tarreau74ca5042013-06-11 23:12:07 +020015702 # don't immediately tell bad guys they are rejected
15703 tcp-request inspect-delay 10s
15704 acl goodguys src 10.0.0.0/24
15705 acl badguys src 10.0.1.0/24
15706 tcp-request content accept if goodguys
15707 tcp-request content reject if badguys WAIT_END
15708 tcp-request content reject
15709
15710
Thierry FOURNIER060762e2014-04-23 13:29:15 +0200157117.3.6. Fetching HTTP samples (Layer 7)
Willy Tarreau74ca5042013-06-11 23:12:07 +020015712--------------------------------------
15713
15714It is possible to fetch samples from HTTP contents, requests and responses.
15715This application layer is also called layer 7. It is only possible to fetch the
15716data in this section when a full HTTP request or response has been parsed from
15717its respective request or response buffer. This is always the case with all
15718HTTP specific rules and for sections running with "mode http". When using TCP
15719content inspection, it may be necessary to support an inspection delay in order
15720to let the request or response come in first. These fetches may require a bit
15721more CPU resources than the layer 4 ones, but not much since the request and
15722response are indexed.
15723
15724base : string
15725 This returns the concatenation of the first Host header and the path part of
15726 the request, which starts at the first slash and ends before the question
15727 mark. It can be useful in virtual hosted environments to detect URL abuses as
15728 well as to improve shared caches efficiency. Using this with a limited size
15729 stick table also allows one to collect statistics about most commonly
15730 requested objects by host/path. With ACLs it can allow simple content
15731 switching rules involving the host and the path at the same time, such as
15732 "www.example.com/favicon.ico". See also "path" and "uri".
15733
15734 ACL derivatives :
15735 base : exact string match
15736 base_beg : prefix match
15737 base_dir : subdir match
15738 base_dom : domain match
15739 base_end : suffix match
15740 base_len : length match
15741 base_reg : regex match
15742 base_sub : substring match
15743
15744base32 : integer
15745 This returns a 32-bit hash of the value returned by the "base" fetch method
15746 above. This is useful to track per-URL activity on high traffic sites without
15747 having to store all URLs. Instead a shorter hash is stored, saving a lot of
Willy Tarreau23ec4ca2014-07-15 20:15:37 +020015748 memory. The output type is an unsigned integer. The hash function used is
15749 SDBM with full avalanche on the output. Technically, base32 is exactly equal
15750 to "base,sdbm(1)".
Willy Tarreau74ca5042013-06-11 23:12:07 +020015751
15752base32+src : binary
15753 This returns the concatenation of the base32 fetch above and the src fetch
15754 below. The resulting type is of type binary, with a size of 8 or 20 bytes
15755 depending on the source address family. This can be used to track per-IP,
15756 per-URL counters.
15757
William Lallemand65ad6e12014-01-31 15:08:02 +010015758capture.req.hdr(<idx>) : string
15759 This extracts the content of the header captured by the "capture request
15760 header", idx is the position of the capture keyword in the configuration.
15761 The first entry is an index of 0. See also: "capture request header".
15762
15763capture.req.method : string
15764 This extracts the METHOD of an HTTP request. It can be used in both request
15765 and response. Unlike "method", it can be used in both request and response
15766 because it's allocated.
15767
15768capture.req.uri : string
15769 This extracts the request's URI, which starts at the first slash and ends
15770 before the first space in the request (without the host part). Unlike "path"
15771 and "url", it can be used in both request and response because it's
15772 allocated.
15773
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015774capture.req.ver : string
15775 This extracts the request's HTTP version and returns either "HTTP/1.0" or
15776 "HTTP/1.1". Unlike "req.ver", it can be used in both request, response, and
15777 logs because it relies on a persistent flag.
15778
William Lallemand65ad6e12014-01-31 15:08:02 +010015779capture.res.hdr(<idx>) : string
15780 This extracts the content of the header captured by the "capture response
15781 header", idx is the position of the capture keyword in the configuration.
15782 The first entry is an index of 0.
15783 See also: "capture response header"
15784
Willy Tarreau3c1b5ec2014-04-24 23:41:57 +020015785capture.res.ver : string
15786 This extracts the response's HTTP version and returns either "HTTP/1.0" or
15787 "HTTP/1.1". Unlike "res.ver", it can be used in logs because it relies on a
15788 persistent flag.
15789
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015790req.body : binary
15791 This returns the HTTP request's available body as a block of data. It
15792 requires that the request body has been buffered made available using
15793 "option http-buffer-request". In case of chunked-encoded body, currently only
15794 the first chunk is analyzed.
15795
Thierry FOURNIER9826c772015-05-20 15:50:54 +020015796req.body_param([<name>) : string
15797 This fetch assumes that the body of the POST request is url-encoded. The user
15798 can check if the "content-type" contains the value
15799 "application/x-www-form-urlencoded". This extracts the first occurrence of the
15800 parameter <name> in the body, which ends before '&'. The parameter name is
15801 case-sensitive. If no name is given, any parameter will match, and the first
15802 one will be returned. The result is a string corresponding to the value of the
15803 parameter <name> as presented in the request body (no URL decoding is
15804 performed). Note that the ACL version of this fetch iterates over multiple
15805 parameters and will iteratively report all parameters values if no name is
15806 given.
15807
Willy Tarreaua5910cc2015-05-02 00:46:08 +020015808req.body_len : integer
15809 This returns the length of the HTTP request's available body in bytes. It may
15810 be lower than the advertised length if the body is larger than the buffer. It
15811 requires that the request body has been buffered made available using
15812 "option http-buffer-request".
15813
15814req.body_size : integer
15815 This returns the advertised length of the HTTP request's body in bytes. It
15816 will represent the advertised Content-Length header, or the size of the first
15817 chunk in case of chunked encoding. In order to parse the chunks, it requires
15818 that the request body has been buffered made available using
15819 "option http-buffer-request".
15820
Willy Tarreau74ca5042013-06-11 23:12:07 +020015821req.cook([<name>]) : string
15822cook([<name>]) : string (deprecated)
15823 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15824 header line from the request, and returns its value as string. If no name is
15825 specified, the first cookie value is returned. When used with ACLs, all
15826 matching cookies are evaluated. Spaces around the name and the value are
15827 ignored as requested by the Cookie header specification (RFC6265). The cookie
15828 name is case-sensitive. Empty cookies are valid, so an empty cookie may very
15829 well return an empty value if it is present. Use the "found" match to detect
15830 presence. Use the res.cook() variant for response cookies sent by the server.
15831
15832 ACL derivatives :
15833 cook([<name>]) : exact string match
15834 cook_beg([<name>]) : prefix match
15835 cook_dir([<name>]) : subdir match
15836 cook_dom([<name>]) : domain match
15837 cook_end([<name>]) : suffix match
15838 cook_len([<name>]) : length match
15839 cook_reg([<name>]) : regex match
15840 cook_sub([<name>]) : substring match
Willy Tarreaud63335a2010-02-26 12:56:52 +010015841
Willy Tarreau74ca5042013-06-11 23:12:07 +020015842req.cook_cnt([<name>]) : integer
15843cook_cnt([<name>]) : integer (deprecated)
15844 Returns an integer value representing the number of occurrences of the cookie
15845 <name> in the request, or all cookies if <name> is not specified.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015846
Willy Tarreau74ca5042013-06-11 23:12:07 +020015847req.cook_val([<name>]) : integer
15848cook_val([<name>]) : integer (deprecated)
15849 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15850 header line from the request, and converts its value to an integer which is
15851 returned. If no name is specified, the first cookie value is returned. When
15852 used in ACLs, all matching names are iterated over until a value matches.
Willy Tarreau0e698542011-09-16 08:32:32 +020015853
Willy Tarreau74ca5042013-06-11 23:12:07 +020015854cookie([<name>]) : string (deprecated)
15855 This extracts the last occurrence of the cookie name <name> on a "Cookie"
15856 header line from the request, or a "Set-Cookie" header from the response, and
15857 returns its value as a string. A typical use is to get multiple clients
15858 sharing a same profile use the same server. This can be similar to what
Willy Tarreau294d0f02015-08-10 19:40:12 +020015859 "appsession" did with the "request-learn" statement, but with support for
Willy Tarreau74ca5042013-06-11 23:12:07 +020015860 multi-peer synchronization and state keeping across restarts. If no name is
15861 specified, the first cookie value is returned. This fetch should not be used
15862 anymore and should be replaced by req.cook() or res.cook() instead as it
15863 ambiguously uses the direction based on the context where it is used.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015864
Willy Tarreau74ca5042013-06-11 23:12:07 +020015865hdr([<name>[,<occ>]]) : string
15866 This is equivalent to req.hdr() when used on requests, and to res.hdr() when
15867 used on responses. Please refer to these respective fetches for more details.
15868 In case of doubt about the fetch direction, please use the explicit ones.
15869 Note that contrary to the hdr() sample fetch method, the hdr_* ACL keywords
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030015870 unambiguously apply to the request headers.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015871
Willy Tarreau74ca5042013-06-11 23:12:07 +020015872req.fhdr(<name>[,<occ>]) : string
15873 This extracts the last occurrence of header <name> in an HTTP request. When
15874 used from an ACL, all occurrences are iterated over until a match is found.
15875 Optionally, a specific occurrence might be specified as a position number.
15876 Positive values indicate a position from the first occurrence, with 1 being
15877 the first one. Negative values indicate positions relative to the last one,
15878 with -1 being the last one. It differs from req.hdr() in that any commas
15879 present in the value are returned and are not used as delimiters. This is
15880 sometimes useful with headers such as User-Agent.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015881
Willy Tarreau74ca5042013-06-11 23:12:07 +020015882req.fhdr_cnt([<name>]) : integer
15883 Returns an integer value representing the number of occurrences of request
15884 header field name <name>, or the total number of header fields if <name> is
15885 not specified. Contrary to its req.hdr_cnt() cousin, this function returns
15886 the number of full line headers and does not stop on commas.
Willy Tarreaud63335a2010-02-26 12:56:52 +010015887
Willy Tarreau74ca5042013-06-11 23:12:07 +020015888req.hdr([<name>[,<occ>]]) : string
15889 This extracts the last occurrence of header <name> in an HTTP request. When
15890 used from an ACL, all occurrences are iterated over until a match is found.
15891 Optionally, a specific occurrence might be specified as a position number.
15892 Positive values indicate a position from the first occurrence, with 1 being
15893 the first one. Negative values indicate positions relative to the last one,
15894 with -1 being the last one. A typical use is with the X-Forwarded-For header
15895 once converted to IP, associated with an IP stick-table. The function
15896 considers any comma as a delimiter for distinct values. If full-line headers
Lukas Tribus23953682017-04-28 13:24:30 +000015897 are desired instead, use req.fhdr(). Please carefully check RFC7231 to know
Willy Tarreau74ca5042013-06-11 23:12:07 +020015898 how certain headers are supposed to be parsed. Also, some of them are case
Davor Ocelice9ed2812017-12-25 17:49:28 +010015899 insensitive (e.g. Connection).
Willy Tarreaud63335a2010-02-26 12:56:52 +010015900
Willy Tarreau74ca5042013-06-11 23:12:07 +020015901 ACL derivatives :
15902 hdr([<name>[,<occ>]]) : exact string match
15903 hdr_beg([<name>[,<occ>]]) : prefix match
15904 hdr_dir([<name>[,<occ>]]) : subdir match
15905 hdr_dom([<name>[,<occ>]]) : domain match
15906 hdr_end([<name>[,<occ>]]) : suffix match
15907 hdr_len([<name>[,<occ>]]) : length match
15908 hdr_reg([<name>[,<occ>]]) : regex match
15909 hdr_sub([<name>[,<occ>]]) : substring match
15910
15911req.hdr_cnt([<name>]) : integer
15912hdr_cnt([<header>]) : integer (deprecated)
15913 Returns an integer value representing the number of occurrences of request
15914 header field name <name>, or the total number of header field values if
15915 <name> is not specified. It is important to remember that one header line may
15916 count as several headers if it has several values. The function considers any
15917 comma as a delimiter for distinct values. If full-line headers are desired
15918 instead, req.fhdr_cnt() should be used instead. With ACLs, it can be used to
15919 detect presence, absence or abuse of a specific header, as well as to block
15920 request smuggling attacks by rejecting requests which contain more than one
15921 of certain headers. See "req.hdr" for more information on header matching.
15922
15923req.hdr_ip([<name>[,<occ>]]) : ip
15924hdr_ip([<name>[,<occ>]]) : ip (deprecated)
15925 This extracts the last occurrence of header <name> in an HTTP request,
15926 converts it to an IPv4 or IPv6 address and returns this address. When used
15927 with ACLs, all occurrences are checked, and if <name> is omitted, every value
15928 of every header is checked. Optionally, a specific occurrence might be
15929 specified as a position number. Positive values indicate a position from the
Davor Ocelice9ed2812017-12-25 17:49:28 +010015930 first occurrence, with 1 being the first one. Negative values indicate
Willy Tarreau74ca5042013-06-11 23:12:07 +020015931 positions relative to the last one, with -1 being the last one. A typical use
15932 is with the X-Forwarded-For and X-Client-IP headers.
15933
15934req.hdr_val([<name>[,<occ>]]) : integer
15935hdr_val([<name>[,<occ>]]) : integer (deprecated)
15936 This extracts the last occurrence of header <name> in an HTTP request, and
15937 converts it to an integer value. When used with ACLs, all occurrences are
15938 checked, and if <name> is omitted, every value of every header is checked.
15939 Optionally, a specific occurrence might be specified as a position number.
15940 Positive values indicate a position from the first occurrence, with 1 being
15941 the first one. Negative values indicate positions relative to the last one,
15942 with -1 being the last one. A typical use is with the X-Forwarded-For header.
15943
Frédéric Lécailleec891192019-02-26 15:02:35 +010015944
15945
Willy Tarreau74ca5042013-06-11 23:12:07 +020015946http_auth(<userlist>) : boolean
15947 Returns a boolean indicating whether the authentication data received from
15948 the client match a username & password stored in the specified userlist. This
15949 fetch function is not really useful outside of ACLs. Currently only http
15950 basic auth is supported.
15951
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015952http_auth_group(<userlist>) : string
15953 Returns a string corresponding to the user name found in the authentication
15954 data received from the client if both the user name and password are valid
15955 according to the specified userlist. The main purpose is to use it in ACLs
15956 where it is then checked whether the user belongs to any group within a list.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015957 This fetch function is not really useful outside of ACLs. Currently only http
15958 basic auth is supported.
15959
15960 ACL derivatives :
Thierry FOURNIER9eec0a62014-01-22 18:38:02 +010015961 http_auth_group(<userlist>) : group ...
15962 Returns true when the user extracted from the request and whose password is
15963 valid according to the specified userlist belongs to at least one of the
15964 groups.
Willy Tarreau74ca5042013-06-11 23:12:07 +020015965
15966http_first_req : boolean
Willy Tarreau7f18e522010-10-22 20:04:13 +020015967 Returns true when the request being processed is the first one of the
15968 connection. This can be used to add or remove headers that may be missing
Willy Tarreau74ca5042013-06-11 23:12:07 +020015969 from some requests when a request is not the first one, or to help grouping
15970 requests in the logs.
Willy Tarreau7f18e522010-10-22 20:04:13 +020015971
Willy Tarreau74ca5042013-06-11 23:12:07 +020015972method : integer + string
15973 Returns an integer value corresponding to the method in the HTTP request. For
15974 example, "GET" equals 1 (check sources to establish the matching). Value 9
15975 means "other method" and may be converted to a string extracted from the
15976 stream. This should not be used directly as a sample, this is only meant to
15977 be used from ACLs, which transparently convert methods from patterns to these
15978 integer + string values. Some predefined ACL already check for most common
15979 methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015980
Willy Tarreau74ca5042013-06-11 23:12:07 +020015981 ACL derivatives :
15982 method : case insensitive method match
Willy Tarreau6a06a402007-07-15 20:15:28 +020015983
Willy Tarreau74ca5042013-06-11 23:12:07 +020015984 Example :
15985 # only accept GET and HEAD requests
15986 acl valid_method method GET HEAD
15987 http-request deny if ! valid_method
Willy Tarreau6a06a402007-07-15 20:15:28 +020015988
Willy Tarreau74ca5042013-06-11 23:12:07 +020015989path : string
15990 This extracts the request's URL path, which starts at the first slash and
15991 ends before the question mark (without the host part). A typical use is with
15992 prefetch-capable caches, and with portals which need to aggregate multiple
15993 information from databases and keep them in caches. Note that with outgoing
15994 caches, it would be wiser to use "url" instead. With ACLs, it's typically
Davor Ocelice9ed2812017-12-25 17:49:28 +010015995 used to match exact file names (e.g. "/login.php"), or directory parts using
Willy Tarreau74ca5042013-06-11 23:12:07 +020015996 the derivative forms. See also the "url" and "base" fetch methods.
Willy Tarreau6a06a402007-07-15 20:15:28 +020015997
Willy Tarreau74ca5042013-06-11 23:12:07 +020015998 ACL derivatives :
15999 path : exact string match
16000 path_beg : prefix match
16001 path_dir : subdir match
16002 path_dom : domain match
16003 path_end : suffix match
16004 path_len : length match
16005 path_reg : regex match
16006 path_sub : substring match
Willy Tarreau6a06a402007-07-15 20:15:28 +020016007
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016008query : string
16009 This extracts the request's query string, which starts after the first
16010 question mark. If no question mark is present, this fetch returns nothing. If
16011 a question mark is present but nothing follows, it returns an empty string.
16012 This means it's possible to easily know whether a query string is present
Tim Düsterhus4896c442016-11-29 02:15:19 +010016013 using the "found" matching method. This fetch is the complement of "path"
Willy Tarreau49ad95c2015-01-19 15:06:26 +010016014 which stops before the question mark.
16015
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016016req.hdr_names([<delim>]) : string
16017 This builds a string made from the concatenation of all header names as they
16018 appear in the request when the rule is evaluated. The default delimiter is
16019 the comma (',') but it may be overridden as an optional argument <delim>. In
16020 this case, only the first character of <delim> is considered.
16021
Willy Tarreau74ca5042013-06-11 23:12:07 +020016022req.ver : string
16023req_ver : string (deprecated)
16024 Returns the version string from the HTTP request, for example "1.1". This can
16025 be useful for logs, but is mostly there for ACL. Some predefined ACL already
16026 check for versions 1.0 and 1.1.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016027
Willy Tarreau74ca5042013-06-11 23:12:07 +020016028 ACL derivatives :
16029 req_ver : exact string match
Willy Tarreau0e698542011-09-16 08:32:32 +020016030
Willy Tarreau74ca5042013-06-11 23:12:07 +020016031res.comp : boolean
16032 Returns the boolean "true" value if the response has been compressed by
16033 HAProxy, otherwise returns boolean "false". This may be used to add
16034 information in the logs.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016035
Willy Tarreau74ca5042013-06-11 23:12:07 +020016036res.comp_algo : string
16037 Returns a string containing the name of the algorithm used if the response
16038 was compressed by HAProxy, for example : "deflate". This may be used to add
16039 some information in the logs.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016040
Willy Tarreau74ca5042013-06-11 23:12:07 +020016041res.cook([<name>]) : string
16042scook([<name>]) : string (deprecated)
16043 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16044 header line from the response, and returns its value as string. If no name is
16045 specified, the first cookie value is returned.
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016046
Willy Tarreau74ca5042013-06-11 23:12:07 +020016047 ACL derivatives :
16048 scook([<name>] : exact string match
Willy Tarreau0ce3aa02012-04-25 18:46:33 +020016049
Willy Tarreau74ca5042013-06-11 23:12:07 +020016050res.cook_cnt([<name>]) : integer
16051scook_cnt([<name>]) : integer (deprecated)
16052 Returns an integer value representing the number of occurrences of the cookie
16053 <name> in the response, or all cookies if <name> is not specified. This is
16054 mostly useful when combined with ACLs to detect suspicious responses.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016055
Willy Tarreau74ca5042013-06-11 23:12:07 +020016056res.cook_val([<name>]) : integer
16057scook_val([<name>]) : integer (deprecated)
16058 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16059 header line from the response, and converts its value to an integer which is
16060 returned. If no name is specified, the first cookie value is returned.
Willy Tarreaud63335a2010-02-26 12:56:52 +010016061
Willy Tarreau74ca5042013-06-11 23:12:07 +020016062res.fhdr([<name>[,<occ>]]) : string
16063 This extracts the last occurrence of header <name> in an HTTP response, or of
16064 the last header if no <name> is specified. Optionally, a specific occurrence
16065 might be specified as a position number. Positive values indicate a position
16066 from the first occurrence, with 1 being the first one. Negative values
16067 indicate positions relative to the last one, with -1 being the last one. It
16068 differs from res.hdr() in that any commas present in the value are returned
16069 and are not used as delimiters. If this is not desired, the res.hdr() fetch
16070 should be used instead. This is sometimes useful with headers such as Date or
16071 Expires.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016072
Willy Tarreau74ca5042013-06-11 23:12:07 +020016073res.fhdr_cnt([<name>]) : integer
16074 Returns an integer value representing the number of occurrences of response
16075 header field name <name>, or the total number of header fields if <name> is
16076 not specified. Contrary to its res.hdr_cnt() cousin, this function returns
16077 the number of full line headers and does not stop on commas. If this is not
16078 desired, the res.hdr_cnt() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016079
Willy Tarreau74ca5042013-06-11 23:12:07 +020016080res.hdr([<name>[,<occ>]]) : string
16081shdr([<name>[,<occ>]]) : string (deprecated)
16082 This extracts the last occurrence of header <name> in an HTTP response, or of
16083 the last header if no <name> is specified. Optionally, a specific occurrence
16084 might be specified as a position number. Positive values indicate a position
16085 from the first occurrence, with 1 being the first one. Negative values
16086 indicate positions relative to the last one, with -1 being the last one. This
16087 can be useful to learn some data into a stick-table. The function considers
16088 any comma as a delimiter for distinct values. If this is not desired, the
16089 res.fhdr() fetch should be used instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016090
Willy Tarreau74ca5042013-06-11 23:12:07 +020016091 ACL derivatives :
16092 shdr([<name>[,<occ>]]) : exact string match
16093 shdr_beg([<name>[,<occ>]]) : prefix match
16094 shdr_dir([<name>[,<occ>]]) : subdir match
16095 shdr_dom([<name>[,<occ>]]) : domain match
16096 shdr_end([<name>[,<occ>]]) : suffix match
16097 shdr_len([<name>[,<occ>]]) : length match
16098 shdr_reg([<name>[,<occ>]]) : regex match
16099 shdr_sub([<name>[,<occ>]]) : substring match
16100
16101res.hdr_cnt([<name>]) : integer
16102shdr_cnt([<name>]) : integer (deprecated)
16103 Returns an integer value representing the number of occurrences of response
16104 header field name <name>, or the total number of header fields if <name> is
16105 not specified. The function considers any comma as a delimiter for distinct
16106 values. If this is not desired, the res.fhdr_cnt() fetch should be used
16107 instead.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016108
Willy Tarreau74ca5042013-06-11 23:12:07 +020016109res.hdr_ip([<name>[,<occ>]]) : ip
16110shdr_ip([<name>[,<occ>]]) : ip (deprecated)
16111 This extracts the last occurrence of header <name> in an HTTP response,
16112 convert it to an IPv4 or IPv6 address and returns this address. Optionally, a
16113 specific occurrence might be specified as a position number. Positive values
16114 indicate a position from the first occurrence, with 1 being the first one.
16115 Negative values indicate positions relative to the last one, with -1 being
16116 the last one. This can be useful to learn some data into a stick table.
Willy Tarreau6a06a402007-07-15 20:15:28 +020016117
Willy Tarreaueb27ec72015-02-20 13:55:29 +010016118res.hdr_names([<delim>]) : string
16119 This builds a string made from the concatenation of all header names as they
16120 appear in the response when the rule is evaluated. The default delimiter is
16121 the comma (',') but it may be overridden as an optional argument <delim>. In
16122 this case, only the first character of <delim> is considered.
16123
Willy Tarreau74ca5042013-06-11 23:12:07 +020016124res.hdr_val([<name>[,<occ>]]) : integer
16125shdr_val([<name>[,<occ>]]) : integer (deprecated)
16126 This extracts the last occurrence of header <name> in an HTTP response, and
16127 converts it to an integer value. Optionally, a specific occurrence might be
16128 specified as a position number. Positive values indicate a position from the
16129 first occurrence, with 1 being the first one. Negative values indicate
16130 positions relative to the last one, with -1 being the last one. This can be
16131 useful to learn some data into a stick table.
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016132
Willy Tarreau74ca5042013-06-11 23:12:07 +020016133res.ver : string
16134resp_ver : string (deprecated)
16135 Returns the version string from the HTTP response, for example "1.1". This
16136 can be useful for logs, but is mostly there for ACL.
Willy Tarreau0e698542011-09-16 08:32:32 +020016137
Willy Tarreau74ca5042013-06-11 23:12:07 +020016138 ACL derivatives :
16139 resp_ver : exact string match
Alexandre Cassen5eb1a902007-11-29 15:43:32 +010016140
Willy Tarreau74ca5042013-06-11 23:12:07 +020016141set-cookie([<name>]) : string (deprecated)
16142 This extracts the last occurrence of the cookie name <name> on a "Set-Cookie"
16143 header line from the response and uses the corresponding value to match. This
Willy Tarreau294d0f02015-08-10 19:40:12 +020016144 can be comparable to what "appsession" did with default options, but with
Willy Tarreau74ca5042013-06-11 23:12:07 +020016145 support for multi-peer synchronization and state keeping across restarts.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016146
Willy Tarreau74ca5042013-06-11 23:12:07 +020016147 This fetch function is deprecated and has been superseded by the "res.cook"
16148 fetch. This keyword will disappear soon.
Krzysztof Piotr Oledzki6b35ce12010-02-01 23:35:44 +010016149
Willy Tarreau74ca5042013-06-11 23:12:07 +020016150status : integer
16151 Returns an integer containing the HTTP status code in the HTTP response, for
16152 example, 302. It is mostly used within ACLs and integer ranges, for example,
16153 to remove any Location header if the response is not a 3xx.
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016154
Thierry Fournier0e00dca2016-04-07 15:47:40 +020016155unique-id : string
16156 Returns the unique-id attached to the request. The directive
16157 "unique-id-format" must be set. If it is not set, the unique-id sample fetch
16158 fails. Note that the unique-id is usually used with HTTP requests, however this
16159 sample fetch can be used with other protocols. Obviously, if it is used with
16160 other protocols than HTTP, the unique-id-format directive must not contain
16161 HTTP parts. See: unique-id-format and unique-id-header
16162
Willy Tarreau74ca5042013-06-11 23:12:07 +020016163url : string
16164 This extracts the request's URL as presented in the request. A typical use is
16165 with prefetch-capable caches, and with portals which need to aggregate
16166 multiple information from databases and keep them in caches. With ACLs, using
16167 "path" is preferred over using "url", because clients may send a full URL as
16168 is normally done with proxies. The only real use is to match "*" which does
16169 not match in "path", and for which there is already a predefined ACL. See
16170 also "path" and "base".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016171
Willy Tarreau74ca5042013-06-11 23:12:07 +020016172 ACL derivatives :
16173 url : exact string match
16174 url_beg : prefix match
16175 url_dir : subdir match
16176 url_dom : domain match
16177 url_end : suffix match
16178 url_len : length match
16179 url_reg : regex match
16180 url_sub : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016181
Willy Tarreau74ca5042013-06-11 23:12:07 +020016182url_ip : ip
16183 This extracts the IP address from the request's URL when the host part is
16184 presented as an IP address. Its use is very limited. For instance, a
16185 monitoring system might use this field as an alternative for the source IP in
16186 order to test what path a given source address would follow, or to force an
16187 entry in a table for a given source address. With ACLs it can be used to
16188 restrict access to certain systems through a proxy, for example when combined
16189 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016190
Willy Tarreau74ca5042013-06-11 23:12:07 +020016191url_port : integer
16192 This extracts the port part from the request's URL. Note that if the port is
16193 not specified in the request, port 80 is assumed. With ACLs it can be used to
16194 restrict access to certain systems through a proxy, for example when combined
16195 with option "http_proxy".
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016196
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016197urlp([<name>[,<delim>]]) : string
16198url_param([<name>[,<delim>]]) : string
Willy Tarreau74ca5042013-06-11 23:12:07 +020016199 This extracts the first occurrence of the parameter <name> in the query
16200 string, which begins after either '?' or <delim>, and which ends before '&',
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016201 ';' or <delim>. The parameter name is case-sensitive. If no name is given,
16202 any parameter will match, and the first one will be returned. The result is
16203 a string corresponding to the value of the parameter <name> as presented in
16204 the request (no URL decoding is performed). This can be used for session
Willy Tarreau74ca5042013-06-11 23:12:07 +020016205 stickiness based on a client ID, to extract an application cookie passed as a
16206 URL parameter, or in ACLs to apply some checks. Note that the ACL version of
Willy Tarreau1ede1da2015-05-07 16:06:18 +020016207 this fetch iterates over multiple parameters and will iteratively report all
16208 parameters values if no name is given
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016209
Willy Tarreau74ca5042013-06-11 23:12:07 +020016210 ACL derivatives :
16211 urlp(<name>[,<delim>]) : exact string match
16212 urlp_beg(<name>[,<delim>]) : prefix match
16213 urlp_dir(<name>[,<delim>]) : subdir match
16214 urlp_dom(<name>[,<delim>]) : domain match
16215 urlp_end(<name>[,<delim>]) : suffix match
16216 urlp_len(<name>[,<delim>]) : length match
16217 urlp_reg(<name>[,<delim>]) : regex match
16218 urlp_sub(<name>[,<delim>]) : substring match
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016219
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016220
Willy Tarreau74ca5042013-06-11 23:12:07 +020016221 Example :
16222 # match http://example.com/foo?PHPSESSIONID=some_id
16223 stick on urlp(PHPSESSIONID)
16224 # match http://example.com/foo;JSESSIONID=some_id
16225 stick on urlp(JSESSIONID,;)
Willy Tarreau25c1ebc2012-04-25 16:21:44 +020016226
Jarno Huuskonen676f6222017-03-30 09:19:45 +030016227urlp_val([<name>[,<delim>]]) : integer
Willy Tarreau74ca5042013-06-11 23:12:07 +020016228 See "urlp" above. This one extracts the URL parameter <name> in the request
16229 and converts it to an integer value. This can be used for session stickiness
16230 based on a user ID for example, or with ACLs to match a page number or price.
Willy Tarreaua9fddca2012-07-31 07:51:48 +020016231
Dragan Dosen0070cd52016-06-16 12:19:49 +020016232url32 : integer
16233 This returns a 32-bit hash of the value obtained by concatenating the first
16234 Host header and the whole URL including parameters (not only the path part of
16235 the request, as in the "base32" fetch above). This is useful to track per-URL
16236 activity. A shorter hash is stored, saving a lot of memory. The output type
16237 is an unsigned integer.
16238
16239url32+src : binary
16240 This returns the concatenation of the "url32" fetch and the "src" fetch. The
16241 resulting type is of type binary, with a size of 8 or 20 bytes depending on
16242 the source address family. This can be used to track per-IP, per-URL counters.
16243
Willy Tarreau198a7442008-01-17 12:05:32 +010016244
Willy Tarreau74ca5042013-06-11 23:12:07 +0200162457.4. Pre-defined ACLs
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016246---------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016247
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016248Some predefined ACLs are hard-coded so that they do not have to be declared in
16249every frontend which needs them. They all have their names in upper case in
Patrick Mézard2382ad62010-05-09 10:43:32 +020016250order to avoid confusion. Their equivalence is provided below.
Willy Tarreauced27012008-01-17 20:35:34 +010016251
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016252ACL name Equivalent to Usage
16253---------------+-----------------------------+---------------------------------
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016254FALSE always_false never match
Willy Tarreau2492d5b2009-07-11 00:06:00 +020016255HTTP req_proto_http match if protocol is valid HTTP
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016256HTTP_1.0 req_ver 1.0 match HTTP version 1.0
16257HTTP_1.1 req_ver 1.1 match HTTP version 1.1
Willy Tarreaud63335a2010-02-26 12:56:52 +010016258HTTP_CONTENT hdr_val(content-length) gt 0 match an existing content-length
16259HTTP_URL_ABS url_reg ^[^/:]*:// match absolute URL with scheme
16260HTTP_URL_SLASH url_beg / match URL beginning with "/"
16261HTTP_URL_STAR url * match URL equal to "*"
16262LOCALHOST src 127.0.0.1/8 match connection from local host
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016263METH_CONNECT method CONNECT match HTTP CONNECT method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016264METH_DELETE method DELETE match HTTP DELETE method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016265METH_GET method GET HEAD match HTTP GET or HEAD method
16266METH_HEAD method HEAD match HTTP HEAD method
16267METH_OPTIONS method OPTIONS match HTTP OPTIONS method
16268METH_POST method POST match HTTP POST method
Daniel Schneller9ff96c72016-04-11 17:45:29 +020016269METH_PUT method PUT match HTTP PUT method
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016270METH_TRACE method TRACE match HTTP TRACE method
Emeric Brunbede3d02009-06-30 17:54:00 +020016271RDP_COOKIE req_rdp_cookie_cnt gt 0 match presence of an RDP cookie
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016272REQ_CONTENT req_len gt 0 match data in the request buffer
Willy Tarreaud63335a2010-02-26 12:56:52 +010016273TRUE always_true always match
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016274WAIT_END wait_end wait for end of content analysis
16275---------------+-----------------------------+---------------------------------
Willy Tarreauced27012008-01-17 20:35:34 +010016276
Willy Tarreaub937b7e2010-01-12 15:27:54 +010016277
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200162788. Logging
16279----------
Willy Tarreau844e3c52008-01-11 16:28:18 +010016280
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016281One of HAProxy's strong points certainly lies is its precise logs. It probably
16282provides the finest level of information available for such a product, which is
16283very important for troubleshooting complex environments. Standard information
16284provided in logs include client ports, TCP/HTTP state timers, precise session
16285state at termination and precise termination cause, information about decisions
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010016286to direct traffic to a server, and of course the ability to capture arbitrary
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016287headers.
16288
16289In order to improve administrators reactivity, it offers a great transparency
16290about encountered problems, both internal and external, and it is possible to
16291send logs to different sources at the same time with different level filters :
16292
16293 - global process-level logs (system errors, start/stop, etc..)
16294 - per-instance system and internal errors (lack of resource, bugs, ...)
16295 - per-instance external troubles (servers up/down, max connections)
16296 - per-instance activity (client connections), either at the establishment or
16297 at the termination.
Davor Ocelice9ed2812017-12-25 17:49:28 +010016298 - per-request control of log-level, e.g.
Jim Freeman9e8714b2015-05-26 09:16:34 -060016299 http-request set-log-level silent if sensitive_request
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016300
16301The ability to distribute different levels of logs to different log servers
16302allow several production teams to interact and to fix their problems as soon
16303as possible. For example, the system team might monitor system-wide errors,
16304while the application team might be monitoring the up/down for their servers in
16305real time, and the security team might analyze the activity logs with one hour
16306delay.
16307
16308
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163098.1. Log levels
16310---------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016311
Simon Hormandf791f52011-05-29 15:01:10 +090016312TCP and HTTP connections can be logged with information such as the date, time,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016313source IP address, destination address, connection duration, response times,
Simon Hormandf791f52011-05-29 15:01:10 +090016314HTTP request, HTTP return code, number of bytes transmitted, conditions
16315in which the session ended, and even exchanged cookies values. For example
16316track a particular user's problems. All messages may be sent to up to two
16317syslog servers. Check the "log" keyword in section 4.2 for more information
16318about log facilities.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016319
16320
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163218.2. Log formats
16322----------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016323
William Lallemand48940402012-01-30 16:47:22 +010016324HAProxy supports 5 log formats. Several fields are common between these formats
Simon Hormandf791f52011-05-29 15:01:10 +090016325and will be detailed in the following sections. A few of them may vary
16326slightly with the configuration, due to indicators specific to certain
16327options. The supported formats are as follows :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016328
16329 - the default format, which is very basic and very rarely used. It only
16330 provides very basic information about the incoming connection at the moment
16331 it is accepted : source IP:port, destination IP:port, and frontend-name.
16332 This mode will eventually disappear so it will not be described to great
16333 extents.
16334
16335 - the TCP format, which is more advanced. This format is enabled when "option
16336 tcplog" is set on the frontend. HAProxy will then usually wait for the
16337 connection to terminate before logging. This format provides much richer
16338 information, such as timers, connection counts, queue size, etc... This
16339 format is recommended for pure TCP proxies.
16340
16341 - the HTTP format, which is the most advanced for HTTP proxying. This format
16342 is enabled when "option httplog" is set on the frontend. It provides the
16343 same information as the TCP format with some HTTP-specific fields such as
16344 the request, the status code, and captures of headers and cookies. This
16345 format is recommended for HTTP proxies.
16346
Emeric Brun3a058f32009-06-30 18:26:00 +020016347 - the CLF HTTP format, which is equivalent to the HTTP format, but with the
16348 fields arranged in the same order as the CLF format. In this mode, all
16349 timers, captures, flags, etc... appear one per field after the end of the
16350 common fields, in the same order they appear in the standard HTTP format.
16351
William Lallemand48940402012-01-30 16:47:22 +010016352 - the custom log format, allows you to make your own log line.
16353
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016354Next sections will go deeper into details for each of these formats. Format
16355specification will be performed on a "field" basis. Unless stated otherwise, a
16356field is a portion of text delimited by any number of spaces. Since syslog
16357servers are susceptible of inserting fields at the beginning of a line, it is
16358always assumed that the first field is the one containing the process name and
16359identifier.
16360
16361Note : Since log lines may be quite long, the log examples in sections below
16362 might be broken into multiple lines. The example log lines will be
16363 prefixed with 3 closing angle brackets ('>>>') and each time a log is
16364 broken into multiple lines, each non-final line will end with a
16365 backslash ('\') and the next line will start indented by two characters.
16366
16367
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200163688.2.1. Default log format
16369-------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016370
16371This format is used when no specific option is set. The log is emitted as soon
16372as the connection is accepted. One should note that this currently is the only
16373format which logs the request's destination IP and ports.
16374
16375 Example :
16376 listen www
16377 mode http
16378 log global
16379 server srv1 127.0.0.1:8000
16380
16381 >>> Feb 6 12:12:09 localhost \
16382 haproxy[14385]: Connect from 10.0.1.2:33312 to 10.0.3.31:8012 \
16383 (www/HTTP)
16384
16385 Field Format Extract from the example above
16386 1 process_name '[' pid ']:' haproxy[14385]:
16387 2 'Connect from' Connect from
16388 3 source_ip ':' source_port 10.0.1.2:33312
16389 4 'to' to
16390 5 destination_ip ':' destination_port 10.0.3.31:8012
16391 6 '(' frontend_name '/' mode ')' (www/HTTP)
16392
16393Detailed fields description :
16394 - "source_ip" is the IP address of the client which initiated the connection.
16395 - "source_port" is the TCP port of the client which initiated the connection.
16396 - "destination_ip" is the IP address the client connected to.
16397 - "destination_port" is the TCP port the client connected to.
16398 - "frontend_name" is the name of the frontend (or listener) which received
16399 and processed the connection.
16400 - "mode is the mode the frontend is operating (TCP or HTTP).
16401
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016402In case of a UNIX socket, the source and destination addresses are marked as
16403"unix:" and the ports reflect the internal ID of the socket which accepted the
16404connection (the same ID as reported in the stats).
16405
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016406It is advised not to use this deprecated format for newer installations as it
16407will eventually disappear.
16408
16409
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200164108.2.2. TCP log format
16411---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016412
16413The TCP format is used when "option tcplog" is specified in the frontend, and
16414is the recommended format for pure TCP proxies. It provides a lot of precious
16415information for troubleshooting. Since this format includes timers and byte
16416counts, the log is normally emitted at the end of the session. It can be
16417emitted earlier if "option logasap" is specified, which makes sense in most
16418environments with long sessions such as remote terminals. Sessions which match
16419the "monitor" rules are never logged. It is also possible not to emit logs for
16420sessions for which no data were exchanged between the client and the server, by
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016421specifying "option dontlognull" in the frontend. Successful connections will
16422not be logged if "option dontlog-normal" is specified in the frontend. A few
16423fields may slightly vary depending on some configuration options, those are
16424marked with a star ('*') after the field name below.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016425
16426 Example :
16427 frontend fnt
16428 mode tcp
16429 option tcplog
16430 log global
16431 default_backend bck
16432
16433 backend bck
16434 server srv1 127.0.0.1:8000
16435
16436 >>> Feb 6 12:12:56 localhost \
16437 haproxy[14387]: 10.0.1.2:33313 [06/Feb/2009:12:12:51.443] fnt \
16438 bck/srv1 0/0/5007 212 -- 0/0/0/0/3 0/0
16439
16440 Field Format Extract from the example above
16441 1 process_name '[' pid ']:' haproxy[14387]:
16442 2 client_ip ':' client_port 10.0.1.2:33313
16443 3 '[' accept_date ']' [06/Feb/2009:12:12:51.443]
16444 4 frontend_name fnt
16445 5 backend_name '/' server_name bck/srv1
16446 6 Tw '/' Tc '/' Tt* 0/0/5007
16447 7 bytes_read* 212
16448 8 termination_state --
16449 9 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 0/0/0/0/3
16450 10 srv_queue '/' backend_queue 0/0
16451
16452Detailed fields description :
16453 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016454 connection to haproxy. If the connection was accepted on a UNIX socket
16455 instead, the IP address would be replaced with the word "unix". Note that
16456 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016457 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016458 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016459 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016460
16461 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016462 If the connection was accepted on a UNIX socket instead, the port would be
16463 replaced with the ID of the accepting socket, which is also reported in the
16464 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016465
16466 - "accept_date" is the exact date when the connection was received by haproxy
16467 (which might be very slightly different from the date observed on the
16468 network if there was some queuing in the system's backlog). This is usually
Willy Tarreau590a0512018-09-05 11:56:48 +020016469 the same date which may appear in any upstream firewall's log. When used in
16470 HTTP mode, the accept_date field will be reset to the first moment the
16471 connection is ready to receive a new request (end of previous response for
16472 HTTP/1, immediately after previous request for HTTP/2).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016473
16474 - "frontend_name" is the name of the frontend (or listener) which received
16475 and processed the connection.
16476
16477 - "backend_name" is the name of the backend (or listener) which was selected
16478 to manage the connection to the server. This will be the same as the
16479 frontend if no switching rule has been applied, which is common for TCP
16480 applications.
16481
16482 - "server_name" is the name of the last server to which the connection was
16483 sent, which might differ from the first one if there were connection errors
16484 and a redispatch occurred. Note that this server belongs to the backend
16485 which processed the request. If the connection was aborted before reaching
16486 a server, "<NOSRV>" is indicated instead of a server name.
16487
16488 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16489 It can be "-1" if the connection was aborted before reaching the queue.
16490 See "Timers" below for more details.
16491
16492 - "Tc" is the total time in milliseconds spent waiting for the connection to
16493 establish to the final server, including retries. It can be "-1" if the
16494 connection was aborted before a connection could be established. See
16495 "Timers" below for more details.
16496
16497 - "Tt" is the total time in milliseconds elapsed between the accept and the
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016498 last close. It covers all possible processing. There is one exception, if
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016499 "option logasap" was specified, then the time counting stops at the moment
16500 the log is emitted. In this case, a '+' sign is prepended before the value,
16501 indicating that the final one will be larger. See "Timers" below for more
16502 details.
16503
16504 - "bytes_read" is the total number of bytes transmitted from the server to
16505 the client when the log is emitted. If "option logasap" is specified, the
16506 this value will be prefixed with a '+' sign indicating that the final one
16507 may be larger. Please note that this value is a 64-bit counter, so log
16508 analysis tools must be able to handle it without overflowing.
16509
16510 - "termination_state" is the condition the session was in when the session
16511 ended. This indicates the session state, which side caused the end of
16512 session to happen, and for what reason (timeout, error, ...). The normal
16513 flags should be "--", indicating the session was closed by either end with
16514 no data remaining in buffers. See below "Session state at disconnection"
16515 for more details.
16516
16517 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016518 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016519 limits have been reached. For instance, if actconn is close to 512 when
16520 multiple connection errors occur, chances are high that the system limits
16521 the process to use a maximum of 1024 file descriptors and that all of them
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016522 are used. See section 3 "Global parameters" to find how to tune the system.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016523
16524 - "feconn" is the total number of concurrent connections on the frontend when
16525 the session was logged. It is useful to estimate the amount of resource
16526 required to sustain high loads, and to detect when the frontend's "maxconn"
16527 has been reached. Most often when this value increases by huge jumps, it is
16528 because there is congestion on the backend servers, but sometimes it can be
16529 caused by a denial of service attack.
16530
16531 - "beconn" is the total number of concurrent connections handled by the
16532 backend when the session was logged. It includes the total number of
16533 concurrent connections active on servers as well as the number of
16534 connections pending in queues. It is useful to estimate the amount of
16535 additional servers needed to support high loads for a given application.
16536 Most often when this value increases by huge jumps, it is because there is
16537 congestion on the backend servers, but sometimes it can be caused by a
16538 denial of service attack.
16539
16540 - "srv_conn" is the total number of concurrent connections still active on
16541 the server when the session was logged. It can never exceed the server's
16542 configured "maxconn" parameter. If this value is very often close or equal
16543 to the server's "maxconn", it means that traffic regulation is involved a
16544 lot, meaning that either the server's maxconn value is too low, or that
16545 there aren't enough servers to process the load with an optimal response
16546 time. When only one of the server's "srv_conn" is high, it usually means
16547 that this server has some trouble causing the connections to take longer to
16548 be processed than on other servers.
16549
16550 - "retries" is the number of connection retries experienced by this session
16551 when trying to connect to the server. It must normally be zero, unless a
16552 server is being stopped at the same moment the connection was attempted.
16553 Frequent retries generally indicate either a network problem between
16554 haproxy and the server, or a misconfigured system backlog on the server
16555 preventing new connections from being queued. This field may optionally be
16556 prefixed with a '+' sign, indicating that the session has experienced a
16557 redispatch after the maximal retry count has been reached on the initial
16558 server. In this case, the server name appearing in the log is the one the
16559 connection was redispatched to, and not the first one, though both may
16560 sometimes be the same in case of hashing for instance. So as a general rule
16561 of thumb, when a '+' is present in front of the retry count, this count
16562 should not be attributed to the logged server.
16563
16564 - "srv_queue" is the total number of requests which were processed before
16565 this one in the server queue. It is zero when the request has not gone
16566 through the server queue. It makes it possible to estimate the approximate
16567 server's response time by dividing the time spent in queue by the number of
16568 requests in the queue. It is worth noting that if a session experiences a
16569 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016570 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016571 backend queue unless a redispatch occurs.
16572
16573 - "backend_queue" is the total number of requests which were processed before
16574 this one in the backend's global queue. It is zero when the request has not
16575 gone through the global queue. It makes it possible to estimate the average
16576 queue length, which easily translates into a number of missing servers when
16577 divided by a server's "maxconn" parameter. It is worth noting that if a
16578 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016579 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016580 through both the server queue and the backend queue unless a redispatch
16581 occurs.
16582
16583
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200165848.2.3. HTTP log format
16585----------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016586
16587The HTTP format is the most complete and the best suited for HTTP proxies. It
16588is enabled by when "option httplog" is specified in the frontend. It provides
16589the same level of information as the TCP format with additional features which
16590are specific to the HTTP protocol. Just like the TCP format, the log is usually
16591emitted at the end of the session, unless "option logasap" is specified, which
16592generally only makes sense for download sites. A session which matches the
16593"monitor" rules will never logged. It is also possible not to log sessions for
16594which no data were sent by the client by specifying "option dontlognull" in the
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020016595frontend. Successful connections will not be logged if "option dontlog-normal"
16596is specified in the frontend.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016597
16598Most fields are shared with the TCP log, some being different. A few fields may
16599slightly vary depending on some configuration options. Those ones are marked
16600with a star ('*') after the field name below.
16601
16602 Example :
16603 frontend http-in
16604 mode http
16605 option httplog
16606 log global
16607 default_backend bck
16608
16609 backend static
16610 server srv1 127.0.0.1:8000
16611
16612 >>> Feb 6 12:14:14 localhost \
16613 haproxy[14389]: 10.0.1.2:33317 [06/Feb/2009:12:14:14.655] http-in \
16614 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 +010016615 {} "GET /index.html HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016616
16617 Field Format Extract from the example above
16618 1 process_name '[' pid ']:' haproxy[14389]:
16619 2 client_ip ':' client_port 10.0.1.2:33317
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016620 3 '[' request_date ']' [06/Feb/2009:12:14:14.655]
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016621 4 frontend_name http-in
16622 5 backend_name '/' server_name static/srv1
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016623 6 TR '/' Tw '/' Tc '/' Tr '/' Ta* 10/0/30/69/109
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016624 7 status_code 200
16625 8 bytes_read* 2750
16626 9 captured_request_cookie -
16627 10 captured_response_cookie -
16628 11 termination_state ----
16629 12 actconn '/' feconn '/' beconn '/' srv_conn '/' retries* 1/1/1/1/0
16630 13 srv_queue '/' backend_queue 0/0
16631 14 '{' captured_request_headers* '}' {haproxy.1wt.eu}
16632 15 '{' captured_response_headers* '}' {}
16633 16 '"' http_request '"' "GET /index.html HTTP/1.1"
Willy Tarreaud72758d2010-01-12 10:42:19 +010016634
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016635Detailed fields description :
16636 - "client_ip" is the IP address of the client which initiated the TCP
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016637 connection to haproxy. If the connection was accepted on a UNIX socket
16638 instead, the IP address would be replaced with the word "unix". Note that
16639 when the connection is accepted on a socket configured with "accept-proxy"
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016640 and the PROXY protocol is correctly used, or with a "accept-netscaler-cip"
Davor Ocelice9ed2812017-12-25 17:49:28 +010016641 and the NetScaler Client IP insertion protocol is correctly used, then the
Bertrand Jacquin93b227d2016-06-04 15:11:10 +010016642 logs will reflect the forwarded connection's information.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016643
16644 - "client_port" is the TCP port of the client which initiated the connection.
Willy Tarreauceb24bc2010-11-09 12:46:41 +010016645 If the connection was accepted on a UNIX socket instead, the port would be
16646 replaced with the ID of the accepting socket, which is also reported in the
16647 stats interface.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016648
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016649 - "request_date" is the exact date when the first byte of the HTTP request
16650 was received by haproxy (log field %tr).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016651
16652 - "frontend_name" is the name of the frontend (or listener) which received
16653 and processed the connection.
16654
16655 - "backend_name" is the name of the backend (or listener) which was selected
16656 to manage the connection to the server. This will be the same as the
16657 frontend if no switching rule has been applied.
16658
16659 - "server_name" is the name of the last server to which the connection was
16660 sent, which might differ from the first one if there were connection errors
16661 and a redispatch occurred. Note that this server belongs to the backend
16662 which processed the request. If the request was aborted before reaching a
16663 server, "<NOSRV>" is indicated instead of a server name. If the request was
16664 intercepted by the stats subsystem, "<STATS>" is indicated instead.
16665
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016666 - "TR" is the total time in milliseconds spent waiting for a full HTTP
16667 request from the client (not counting body) after the first byte was
16668 received. It can be "-1" if the connection was aborted before a complete
16669 request could be received or the a bad request was received. It should
16670 always be very small because a request generally fits in one single packet.
16671 Large times here generally indicate network issues between the client and
Willy Tarreau590a0512018-09-05 11:56:48 +020016672 haproxy or requests being typed by hand. See section 8.4 "Timing Events"
16673 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016674
16675 - "Tw" is the total time in milliseconds spent waiting in the various queues.
16676 It can be "-1" if the connection was aborted before reaching the queue.
Willy Tarreau590a0512018-09-05 11:56:48 +020016677 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016678
16679 - "Tc" is the total time in milliseconds spent waiting for the connection to
16680 establish to the final server, including retries. It can be "-1" if the
Willy Tarreau590a0512018-09-05 11:56:48 +020016681 request was aborted before a connection could be established. See section
16682 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016683
16684 - "Tr" is the total time in milliseconds spent waiting for the server to send
16685 a full HTTP response, not counting data. It can be "-1" if the request was
16686 aborted before a complete response could be received. It generally matches
16687 the server's processing time for the request, though it may be altered by
16688 the amount of data sent by the client to the server. Large times here on
Willy Tarreau590a0512018-09-05 11:56:48 +020016689 "GET" requests generally indicate an overloaded server. See section 8.4
16690 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016691
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016692 - "Ta" is the time the request remained active in haproxy, which is the total
16693 time in milliseconds elapsed between the first byte of the request was
16694 received and the last byte of response was sent. It covers all possible
16695 processing except the handshake (see Th) and idle time (see Ti). There is
16696 one exception, if "option logasap" was specified, then the time counting
16697 stops at the moment the log is emitted. In this case, a '+' sign is
16698 prepended before the value, indicating that the final one will be larger.
Willy Tarreau590a0512018-09-05 11:56:48 +020016699 See section 8.4 "Timing Events" for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016700
16701 - "status_code" is the HTTP status code returned to the client. This status
16702 is generally set by the server, but it might also be set by haproxy when
16703 the server cannot be reached or when its response is blocked by haproxy.
16704
16705 - "bytes_read" is the total number of bytes transmitted to the client when
16706 the log is emitted. This does include HTTP headers. If "option logasap" is
16707 specified, the this value will be prefixed with a '+' sign indicating that
16708 the final one may be larger. Please note that this value is a 64-bit
16709 counter, so log analysis tools must be able to handle it without
16710 overflowing.
16711
16712 - "captured_request_cookie" is an optional "name=value" entry indicating that
16713 the client had this cookie in the request. The cookie name and its maximum
16714 length are defined by the "capture cookie" statement in the frontend
16715 configuration. The field is a single dash ('-') when the option is not
16716 set. Only one cookie may be captured, it is generally used to track session
16717 ID exchanges between a client and a server to detect session crossing
16718 between clients due to application bugs. For more details, please consult
16719 the section "Capturing HTTP headers and cookies" below.
16720
16721 - "captured_response_cookie" is an optional "name=value" entry indicating
16722 that the server has returned a cookie with its response. The cookie name
16723 and its maximum length are defined by the "capture cookie" statement in the
16724 frontend configuration. The field is a single dash ('-') when the option is
16725 not set. Only one cookie may be captured, it is generally used to track
16726 session ID exchanges between a client and a server to detect session
16727 crossing between clients due to application bugs. For more details, please
16728 consult the section "Capturing HTTP headers and cookies" below.
16729
16730 - "termination_state" is the condition the session was in when the session
16731 ended. This indicates the session state, which side caused the end of
16732 session to happen, for what reason (timeout, error, ...), just like in TCP
16733 logs, and information about persistence operations on cookies in the last
16734 two characters. The normal flags should begin with "--", indicating the
16735 session was closed by either end with no data remaining in buffers. See
16736 below "Session state at disconnection" for more details.
16737
16738 - "actconn" is the total number of concurrent connections on the process when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016739 the session was logged. It is useful to detect when some per-process system
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016740 limits have been reached. For instance, if actconn is close to 512 or 1024
16741 when multiple connection errors occur, chances are high that the system
16742 limits the process to use a maximum of 1024 file descriptors and that all
Willy Tarreauc57f0e22009-05-10 13:12:33 +020016743 of them are used. See section 3 "Global parameters" to find how to tune the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016744 system.
16745
16746 - "feconn" is the total number of concurrent connections on the frontend when
16747 the session was logged. It is useful to estimate the amount of resource
16748 required to sustain high loads, and to detect when the frontend's "maxconn"
16749 has been reached. Most often when this value increases by huge jumps, it is
16750 because there is congestion on the backend servers, but sometimes it can be
16751 caused by a denial of service attack.
16752
16753 - "beconn" is the total number of concurrent connections handled by the
16754 backend when the session was logged. It includes the total number of
16755 concurrent connections active on servers as well as the number of
16756 connections pending in queues. It is useful to estimate the amount of
16757 additional servers needed to support high loads for a given application.
16758 Most often when this value increases by huge jumps, it is because there is
16759 congestion on the backend servers, but sometimes it can be caused by a
16760 denial of service attack.
16761
16762 - "srv_conn" is the total number of concurrent connections still active on
16763 the server when the session was logged. It can never exceed the server's
16764 configured "maxconn" parameter. If this value is very often close or equal
16765 to the server's "maxconn", it means that traffic regulation is involved a
16766 lot, meaning that either the server's maxconn value is too low, or that
16767 there aren't enough servers to process the load with an optimal response
16768 time. When only one of the server's "srv_conn" is high, it usually means
16769 that this server has some trouble causing the requests to take longer to be
16770 processed than on other servers.
16771
16772 - "retries" is the number of connection retries experienced by this session
16773 when trying to connect to the server. It must normally be zero, unless a
16774 server is being stopped at the same moment the connection was attempted.
16775 Frequent retries generally indicate either a network problem between
16776 haproxy and the server, or a misconfigured system backlog on the server
16777 preventing new connections from being queued. This field may optionally be
16778 prefixed with a '+' sign, indicating that the session has experienced a
16779 redispatch after the maximal retry count has been reached on the initial
16780 server. In this case, the server name appearing in the log is the one the
16781 connection was redispatched to, and not the first one, though both may
16782 sometimes be the same in case of hashing for instance. So as a general rule
16783 of thumb, when a '+' is present in front of the retry count, this count
16784 should not be attributed to the logged server.
16785
16786 - "srv_queue" is the total number of requests which were processed before
16787 this one in the server queue. It is zero when the request has not gone
16788 through the server queue. It makes it possible to estimate the approximate
16789 server's response time by dividing the time spent in queue by the number of
16790 requests in the queue. It is worth noting that if a session experiences a
16791 redispatch and passes through two server queues, their positions will be
Davor Ocelice9ed2812017-12-25 17:49:28 +010016792 cumulative. A request should not pass through both the server queue and the
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016793 backend queue unless a redispatch occurs.
16794
16795 - "backend_queue" is the total number of requests which were processed before
16796 this one in the backend's global queue. It is zero when the request has not
16797 gone through the global queue. It makes it possible to estimate the average
16798 queue length, which easily translates into a number of missing servers when
16799 divided by a server's "maxconn" parameter. It is worth noting that if a
16800 session experiences a redispatch, it may pass twice in the backend's queue,
Davor Ocelice9ed2812017-12-25 17:49:28 +010016801 and then both positions will be cumulative. A request should not pass
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016802 through both the server queue and the backend queue unless a redispatch
16803 occurs.
16804
16805 - "captured_request_headers" is a list of headers captured in the request due
16806 to the presence of the "capture request header" statement in the frontend.
16807 Multiple headers can be captured, they will be delimited by a vertical bar
16808 ('|'). When no capture is enabled, the braces do not appear, causing a
16809 shift of remaining fields. It is important to note that this field may
16810 contain spaces, and that using it requires a smarter log parser than when
16811 it's not used. Please consult the section "Capturing HTTP headers and
16812 cookies" below for more details.
16813
16814 - "captured_response_headers" is a list of headers captured in the response
16815 due to the presence of the "capture response header" statement in the
16816 frontend. Multiple headers can be captured, they will be delimited by a
16817 vertical bar ('|'). When no capture is enabled, the braces do not appear,
16818 causing a shift of remaining fields. It is important to note that this
16819 field may contain spaces, and that using it requires a smarter log parser
16820 than when it's not used. Please consult the section "Capturing HTTP headers
16821 and cookies" below for more details.
16822
16823 - "http_request" is the complete HTTP request line, including the method,
16824 request and HTTP version string. Non-printable characters are encoded (see
16825 below the section "Non-printable characters"). This is always the last
16826 field, and it is always delimited by quotes and is the only one which can
16827 contain quotes. If new fields are added to the log format, they will be
16828 added before this field. This field might be truncated if the request is
16829 huge and does not fit in the standard syslog buffer (1024 characters). This
16830 is the reason why this field must always remain the last one.
16831
16832
Cyril Bontédc4d9032012-04-08 21:57:39 +0200168338.2.4. Custom log format
16834------------------------
William Lallemand48940402012-01-30 16:47:22 +010016835
Willy Tarreau2beef582012-12-20 17:22:52 +010016836The directive log-format allows you to customize the logs in http mode and tcp
William Lallemandbddd4fd2012-02-27 11:23:10 +010016837mode. It takes a string as argument.
William Lallemand48940402012-01-30 16:47:22 +010016838
Davor Ocelice9ed2812017-12-25 17:49:28 +010016839HAProxy understands some log format variables. % precedes log format variables.
William Lallemand48940402012-01-30 16:47:22 +010016840Variables can take arguments using braces ('{}'), and multiple arguments are
16841separated by commas within the braces. Flags may be added or removed by
16842prefixing them with a '+' or '-' sign.
16843
16844Special variable "%o" may be used to propagate its flags to all other
16845variables on the same format string. This is particularly handy with quoted
Dragan Dosen835b9212016-02-12 13:23:03 +010016846("Q") and escaped ("E") string formats.
William Lallemand48940402012-01-30 16:47:22 +010016847
Willy Tarreauc8368452012-12-21 00:09:23 +010016848If a variable is named between square brackets ('[' .. ']') then it is used
Willy Tarreaube722a22014-06-13 16:31:59 +020016849as a sample expression rule (see section 7.3). This it useful to add some
Willy Tarreauc8368452012-12-21 00:09:23 +010016850less common information such as the client's SSL certificate's DN, or to log
16851the key that would be used to store an entry into a stick table.
16852
William Lallemand48940402012-01-30 16:47:22 +010016853Note: spaces must be escaped. A space character is considered as a separator.
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030016854In order to emit a verbatim '%', it must be preceded by another '%' resulting
Willy Tarreau06d97f92013-12-02 17:45:48 +010016855in '%%'. HAProxy will automatically merge consecutive separators.
William Lallemand48940402012-01-30 16:47:22 +010016856
Dragan Dosen835b9212016-02-12 13:23:03 +010016857Note: when using the RFC5424 syslog message format, the characters '"',
16858'\' and ']' inside PARAM-VALUE should be escaped with '\' as prefix (see
16859https://tools.ietf.org/html/rfc5424#section-6.3.3 for more details). In
16860such cases, the use of the flag "E" should be considered.
16861
William Lallemand48940402012-01-30 16:47:22 +010016862Flags are :
16863 * Q: quote a string
Jamie Gloudonaaa21002012-08-25 00:18:33 -040016864 * X: hexadecimal representation (IPs, Ports, %Ts, %rt, %pid)
Dragan Dosen835b9212016-02-12 13:23:03 +010016865 * E: escape characters '"', '\' and ']' in a string with '\' as prefix
16866 (intended purpose is for the RFC5424 structured-data log formats)
William Lallemand48940402012-01-30 16:47:22 +010016867
16868 Example:
16869
16870 log-format %T\ %t\ Some\ Text
16871 log-format %{+Q}o\ %t\ %s\ %{-Q}r
16872
Dragan Dosen835b9212016-02-12 13:23:03 +010016873 log-format-sd %{+Q,+E}o\ [exampleSDID@1234\ header=%[capture.req.hdr(0)]]
16874
William Lallemand48940402012-01-30 16:47:22 +010016875At the moment, the default HTTP format is defined this way :
16876
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016877 log-format "%ci:%cp [%tr] %ft %b/%s %TR/%Tw/%Tc/%Tr/%Ta %ST %B %CC \
16878 %CS %tsc %ac/%fc/%bc/%sc/%rc %sq/%bq %hr %hs %{+Q}r"
William Lallemand48940402012-01-30 16:47:22 +010016879
William Lallemandbddd4fd2012-02-27 11:23:10 +010016880the default CLF format is defined this way :
William Lallemand48940402012-01-30 16:47:22 +010016881
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016882 log-format "%{+Q}o %{-Q}ci - - [%trg] %r %ST %B \"\" \"\" %cp \
16883 %ms %ft %b %s %TR %Tw %Tc %Tr %Ta %tsc %ac %fc \
16884 %bc %sc %rc %sq %bq %CC %CS %hrl %hsl"
William Lallemand48940402012-01-30 16:47:22 +010016885
William Lallemandbddd4fd2012-02-27 11:23:10 +010016886and the default TCP format is defined this way :
16887
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016888 log-format "%ci:%cp [%t] %ft %b/%s %Tw/%Tc/%Tt %B %ts \
16889 %ac/%fc/%bc/%sc/%rc %sq/%bq"
William Lallemandbddd4fd2012-02-27 11:23:10 +010016890
William Lallemand48940402012-01-30 16:47:22 +010016891Please refer to the table below for currently defined variables :
16892
William Lallemandbddd4fd2012-02-27 11:23:10 +010016893 +---+------+-----------------------------------------------+-------------+
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016894 | R | var | field name (8.2.2 and 8.2.3 for description) | type |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016895 +---+------+-----------------------------------------------+-------------+
16896 | | %o | special variable, apply flags on all next var | |
16897 +---+------+-----------------------------------------------+-------------+
Willy Tarreau2beef582012-12-20 17:22:52 +010016898 | | %B | bytes_read (from server to client) | numeric |
16899 | H | %CC | captured_request_cookie | string |
16900 | H | %CS | captured_response_cookie | string |
William Lallemand5f232402012-04-05 18:02:55 +020016901 | | %H | hostname | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016902 | H | %HM | HTTP method (ex: POST) | string |
16903 | H | %HP | HTTP request URI without query string (path) | string |
Andrew Hayworthe63ac872015-07-31 16:14:16 +000016904 | H | %HQ | HTTP request URI query string (ex: ?bar=baz) | string |
Andrew Hayworth0ebc55f2015-04-27 21:37:03 +000016905 | H | %HU | HTTP request URI (ex: /foo?bar=baz) | string |
16906 | H | %HV | HTTP version (ex: HTTP/1.0) | string |
William Lallemanda73203e2012-03-12 12:48:57 +010016907 | | %ID | unique-id | string |
Willy Tarreau4bf99632014-06-13 12:21:40 +020016908 | | %ST | status_code | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016909 | | %T | gmt_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016910 | | %Ta | Active time of the request (from TR to end) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016911 | | %Tc | Tc | numeric |
Willy Tarreau27b639d2016-05-17 17:55:27 +020016912 | | %Td | Td = Tt - (Tq + Tw + Tc + Tr) | numeric |
Yuxans Yao4e25b012012-10-19 10:36:09 +080016913 | | %Tl | local_date_time | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016914 | | %Th | connection handshake time (SSL, PROXY proto) | numeric |
16915 | H | %Ti | idle time before the HTTP request | numeric |
16916 | H | %Tq | Th + Ti + TR | numeric |
16917 | H | %TR | time to receive the full request from 1st byte| numeric |
16918 | H | %Tr | Tr (response time) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016919 | | %Ts | timestamp | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016920 | | %Tt | Tt | numeric |
16921 | | %Tw | Tw | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016922 | | %U | bytes_uploaded (from client to server) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016923 | | %ac | actconn | numeric |
16924 | | %b | backend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016925 | | %bc | beconn (backend concurrent connections) | numeric |
16926 | | %bi | backend_source_ip (connecting address) | IP |
16927 | | %bp | backend_source_port (connecting address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016928 | | %bq | backend_queue | numeric |
Willy Tarreau2beef582012-12-20 17:22:52 +010016929 | | %ci | client_ip (accepted address) | IP |
16930 | | %cp | client_port (accepted address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016931 | | %f | frontend_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016932 | | %fc | feconn (frontend concurrent connections) | numeric |
16933 | | %fi | frontend_ip (accepting address) | IP |
16934 | | %fp | frontend_port (accepting address) | numeric |
Willy Tarreau773d65f2012-10-12 14:56:11 +020016935 | | %ft | frontend_name_transport ('~' suffix for SSL) | string |
Willy Tarreau7346acb2014-08-28 15:03:15 +020016936 | | %lc | frontend_log_counter | numeric |
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020016937 | | %hr | captured_request_headers default style | string |
16938 | | %hrl | captured_request_headers CLF style | string list |
16939 | | %hs | captured_response_headers default style | string |
16940 | | %hsl | captured_response_headers CLF style | string list |
Willy Tarreau812c88e2015-08-09 10:56:35 +020016941 | | %ms | accept date milliseconds (left-padded with 0) | numeric |
William Lallemand5f232402012-04-05 18:02:55 +020016942 | | %pid | PID | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016943 | H | %r | http_request | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016944 | | %rc | retries | numeric |
Willy Tarreau1f0da242014-01-25 11:01:50 +010016945 | | %rt | request_counter (HTTP req or TCP session) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016946 | | %s | server_name | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016947 | | %sc | srv_conn (server concurrent connections) | numeric |
16948 | | %si | server_IP (target address) | IP |
16949 | | %sp | server_port (target address) | numeric |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016950 | | %sq | srv_queue | numeric |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016951 | S | %sslc| ssl_ciphers (ex: AES-SHA) | string |
16952 | S | %sslv| ssl_version (ex: TLSv1) | string |
Willy Tarreau2beef582012-12-20 17:22:52 +010016953 | | %t | date_time (with millisecond resolution) | date |
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020016954 | H | %tr | date_time of HTTP request | date |
16955 | H | %trg | gmt_date_time of start of HTTP request | date |
Jens Bissinger15c64ff2018-08-23 14:11:27 +020016956 | H | %trl | local_date_time of start of HTTP request | date |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016957 | | %ts | termination_state | string |
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016958 | H | %tsc | termination_state with cookie status | string |
William Lallemandbddd4fd2012-02-27 11:23:10 +010016959 +---+------+-----------------------------------------------+-------------+
William Lallemand48940402012-01-30 16:47:22 +010016960
Willy Tarreauffc3fcd2012-10-12 20:17:54 +020016961 R = Restrictions : H = mode http only ; S = SSL only
William Lallemand48940402012-01-30 16:47:22 +010016962
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016963
169648.2.5. Error log format
16965-----------------------
16966
16967When an incoming connection fails due to an SSL handshake or an invalid PROXY
16968protocol header, haproxy will log the event using a shorter, fixed line format.
16969By default, logs are emitted at the LOG_INFO level, unless the option
16970"log-separate-errors" is set in the backend, in which case the LOG_ERR level
Davor Ocelice9ed2812017-12-25 17:49:28 +010016971will be used. Connections on which no data are exchanged (e.g. probes) are not
Willy Tarreau5f51e1a2012-12-03 18:40:10 +010016972logged if the "dontlognull" option is set.
16973
16974The format looks like this :
16975
16976 >>> Dec 3 18:27:14 localhost \
16977 haproxy[6103]: 127.0.0.1:56059 [03/Dec/2012:17:35:10.380] frt/f1: \
16978 Connection error during SSL handshake
16979
16980 Field Format Extract from the example above
16981 1 process_name '[' pid ']:' haproxy[6103]:
16982 2 client_ip ':' client_port 127.0.0.1:56059
16983 3 '[' accept_date ']' [03/Dec/2012:17:35:10.380]
16984 4 frontend_name "/" bind_name ":" frt/f1:
16985 5 message Connection error during SSL handshake
16986
16987These fields just provide minimal information to help debugging connection
16988failures.
16989
16990
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200169918.3. Advanced logging options
16992-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010016993
16994Some advanced logging options are often looked for but are not easy to find out
16995just by looking at the various options. Here is an entry point for the few
16996options which can enable better logging. Please refer to the keywords reference
16997for more information about their usage.
16998
16999
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170008.3.1. Disabling logging of external tests
17001------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017002
17003It is quite common to have some monitoring tools perform health checks on
17004haproxy. Sometimes it will be a layer 3 load-balancer such as LVS or any
17005commercial load-balancer, and sometimes it will simply be a more complete
17006monitoring system such as Nagios. When the tests are very frequent, users often
17007ask how to disable logging for those checks. There are three possibilities :
17008
17009 - if connections come from everywhere and are just TCP probes, it is often
17010 desired to simply disable logging of connections without data exchange, by
17011 setting "option dontlognull" in the frontend. It also disables logging of
17012 port scans, which may or may not be desired.
17013
17014 - if the connection come from a known source network, use "monitor-net" to
17015 declare this network as monitoring only. Any host in this network will then
17016 only be able to perform health checks, and their requests will not be
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017017 logged. This is generally appropriate to designate a list of equipment
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017018 such as other load-balancers.
17019
17020 - if the tests are performed on a known URI, use "monitor-uri" to declare
17021 this URI as dedicated to monitoring. Any host sending this request will
17022 only get the result of a health-check, and the request will not be logged.
17023
17024
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170258.3.2. Logging before waiting for the session to terminate
17026----------------------------------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017027
17028The problem with logging at end of connection is that you have no clue about
17029what is happening during very long sessions, such as remote terminal sessions
17030or large file downloads. This problem can be worked around by specifying
Davor Ocelice9ed2812017-12-25 17:49:28 +010017031"option logasap" in the frontend. HAProxy will then log as soon as possible,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017032just before data transfer begins. This means that in case of TCP, it will still
17033log the connection status to the server, and in case of HTTP, it will log just
17034after processing the server headers. In this case, the number of bytes reported
17035is the number of header bytes sent to the client. In order to avoid confusion
17036with normal logs, the total time field and the number of bytes are prefixed
17037with a '+' sign which means that real numbers are certainly larger.
17038
17039
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170408.3.3. Raising log level upon errors
17041------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017042
17043Sometimes it is more convenient to separate normal traffic from errors logs,
17044for instance in order to ease error monitoring from log files. When the option
17045"log-separate-errors" is used, connections which experience errors, timeouts,
17046retries, redispatches or HTTP status codes 5xx will see their syslog level
17047raised from "info" to "err". This will help a syslog daemon store the log in
17048a separate file. It is very important to keep the errors in the normal traffic
17049file too, so that log ordering is not altered. You should also be careful if
17050you already have configured your syslog daemon to store all logs higher than
17051"notice" in an "admin" file, because the "err" level is higher than "notice".
17052
17053
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170548.3.4. Disabling logging of successful connections
17055--------------------------------------------------
Willy Tarreauc9bd0cc2009-05-10 11:57:02 +020017056
17057Although this may sound strange at first, some large sites have to deal with
17058multiple thousands of logs per second and are experiencing difficulties keeping
17059them intact for a long time or detecting errors within them. If the option
17060"dontlog-normal" is set on the frontend, all normal connections will not be
17061logged. In this regard, a normal connection is defined as one without any
17062error, timeout, retry nor redispatch. In HTTP, the status code is checked too,
17063and a response with a status 5xx is not considered normal and will be logged
17064too. Of course, doing is is really discouraged as it will remove most of the
17065useful information from the logs. Do this only if you have no other
17066alternative.
17067
17068
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200170698.4. Timing events
17070------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017071
17072Timers provide a great help in troubleshooting network problems. All values are
17073reported in milliseconds (ms). These timers should be used in conjunction with
17074the session termination flags. In TCP mode with "option tcplog" set on the
17075frontend, 3 control points are reported under the form "Tw/Tc/Tt", and in HTTP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017076mode, 5 control points are reported under the form "TR/Tw/Tc/Tr/Ta". In
17077addition, three other measures are provided, "Th", "Ti", and "Tq".
17078
Guillaume de Lafondf27cddc2016-12-23 17:32:43 +010017079Timings events in HTTP mode:
17080
17081 first request 2nd request
17082 |<-------------------------------->|<-------------- ...
17083 t tr t tr ...
17084 ---|----|----|----|----|----|----|----|----|--
17085 : Th Ti TR Tw Tc Tr Td : Ti ...
17086 :<---- Tq ---->: :
17087 :<-------------- Tt -------------->:
17088 :<--------- Ta --------->:
17089
17090Timings events in TCP mode:
17091
17092 TCP session
17093 |<----------------->|
17094 t t
17095 ---|----|----|----|----|---
17096 | Th Tw Tc Td |
17097 |<------ Tt ------->|
17098
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017099 - Th: total time to accept tcp connection and execute handshakes for low level
Davor Ocelice9ed2812017-12-25 17:49:28 +010017100 protocols. Currently, these protocols are proxy-protocol and SSL. This may
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017101 only happen once during the whole connection's lifetime. A large time here
17102 may indicate that the client only pre-established the connection without
17103 speaking, that it is experiencing network issues preventing it from
Davor Ocelice9ed2812017-12-25 17:49:28 +010017104 completing a handshake in a reasonable time (e.g. MTU issues), or that an
Willy Tarreau590a0512018-09-05 11:56:48 +020017105 SSL handshake was very expensive to compute. Please note that this time is
17106 reported only before the first request, so it is safe to average it over
17107 all request to calculate the amortized value. The second and subsequent
17108 request will always report zero here.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017109
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017110 - Ti: is the idle time before the HTTP request (HTTP mode only). This timer
17111 counts between the end of the handshakes and the first byte of the HTTP
17112 request. When dealing with a second request in keep-alive mode, it starts
Willy Tarreau590a0512018-09-05 11:56:48 +020017113 to count after the end of the transmission the previous response. When a
17114 multiplexed protocol such as HTTP/2 is used, it starts to count immediately
17115 after the previous request. Some browsers pre-establish connections to a
17116 server in order to reduce the latency of a future request, and keep them
17117 pending until they need it. This delay will be reported as the idle time. A
17118 value of -1 indicates that nothing was received on the connection.
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017119
17120 - TR: total time to get the client request (HTTP mode only). It's the time
17121 elapsed between the first bytes received and the moment the proxy received
17122 the empty line marking the end of the HTTP headers. The value "-1"
17123 indicates that the end of headers has never been seen. This happens when
17124 the client closes prematurely or times out. This time is usually very short
17125 since most requests fit in a single packet. A large time may indicate a
17126 request typed by hand during a test.
17127
17128 - Tq: total time to get the client request from the accept date or since the
17129 emission of the last byte of the previous response (HTTP mode only). It's
Davor Ocelice9ed2812017-12-25 17:49:28 +010017130 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 +020017131 returns -1 as well. This timer used to be very useful before the arrival of
17132 HTTP keep-alive and browsers' pre-connect feature. It's recommended to drop
17133 it in favor of TR nowadays, as the idle time adds a lot of noise to the
17134 reports.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017135
17136 - Tw: total time spent in the queues waiting for a connection slot. It
17137 accounts for backend queue as well as the server queues, and depends on the
17138 queue size, and the time needed for the server to complete previous
17139 requests. The value "-1" means that the request was killed before reaching
17140 the queue, which is generally what happens with invalid or denied requests.
17141
17142 - Tc: total time to establish the TCP connection to the server. It's the time
17143 elapsed between the moment the proxy sent the connection request, and the
17144 moment it was acknowledged by the server, or between the TCP SYN packet and
17145 the matching SYN/ACK packet in return. The value "-1" means that the
17146 connection never established.
17147
17148 - Tr: server response time (HTTP mode only). It's the time elapsed between
17149 the moment the TCP connection was established to the server and the moment
17150 the server sent its complete response headers. It purely shows its request
17151 processing time, without the network overhead due to the data transmission.
17152 It is worth noting that when the client has data to send to the server, for
17153 instance during a POST request, the time already runs, and this can distort
17154 apparent response time. For this reason, it's generally wise not to trust
17155 too much this field for POST requests initiated from clients behind an
17156 untrusted network. A value of "-1" here means that the last the response
17157 header (empty line) was never seen, most likely because the server timeout
17158 stroke before the server managed to process the request.
17159
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017160 - Ta: total active time for the HTTP request, between the moment the proxy
17161 received the first byte of the request header and the emission of the last
17162 byte of the response body. The exception is when the "logasap" option is
17163 specified. In this case, it only equals (TR+Tw+Tc+Tr), and is prefixed with
17164 a '+' sign. From this field, we can deduce "Td", the data transmission time,
17165 by subtracting other timers when valid :
17166
17167 Td = Ta - (TR + Tw + Tc + Tr)
17168
17169 Timers with "-1" values have to be excluded from this equation. Note that
17170 "Ta" can never be negative.
17171
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017172 - Tt: total session duration time, between the moment the proxy accepted it
17173 and the moment both ends were closed. The exception is when the "logasap"
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017174 option is specified. In this case, it only equals (Th+Ti+TR+Tw+Tc+Tr), and
17175 is prefixed with a '+' sign. From this field, we can deduce "Td", the data
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017176 transmission time, by subtracting other timers when valid :
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017177
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017178 Td = Tt - (Th + Ti + TR + Tw + Tc + Tr)
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017179
17180 Timers with "-1" values have to be excluded from this equation. In TCP
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017181 mode, "Ti", "Tq" and "Tr" have to be excluded too. Note that "Tt" can never
17182 be negative and that for HTTP, Tt is simply equal to (Th+Ti+Ta).
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017183
17184These timers provide precious indications on trouble causes. Since the TCP
17185protocol defines retransmit delays of 3, 6, 12... seconds, we know for sure
17186that timers close to multiples of 3s are nearly always related to lost packets
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017187due to network problems (wires, negotiation, congestion). Moreover, if "Ta" or
17188"Tt" is close to a timeout value specified in the configuration, it often means
17189that a session has been aborted on timeout.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017190
17191Most common cases :
17192
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017193 - If "Th" or "Ti" are close to 3000, a packet has probably been lost between
17194 the client and the proxy. This is very rare on local networks but might
17195 happen when clients are on far remote networks and send large requests. It
17196 may happen that values larger than usual appear here without any network
17197 cause. Sometimes, during an attack or just after a resource starvation has
17198 ended, haproxy may accept thousands of connections in a few milliseconds.
17199 The time spent accepting these connections will inevitably slightly delay
17200 processing of other connections, and it can happen that request times in the
17201 order of a few tens of milliseconds are measured after a few thousands of
17202 new connections have been accepted at once. Using one of the keep-alive
17203 modes may display larger idle times since "Ti" measures the time spent
Patrick Mezard105faca2010-06-12 17:02:46 +020017204 waiting for additional requests.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017205
17206 - If "Tc" is close to 3000, a packet has probably been lost between the
17207 server and the proxy during the server connection phase. This value should
17208 always be very low, such as 1 ms on local networks and less than a few tens
17209 of ms on remote networks.
17210
Willy Tarreau55165fe2009-05-10 12:02:55 +020017211 - If "Tr" is nearly always lower than 3000 except some rare values which seem
17212 to be the average majored by 3000, there are probably some packets lost
17213 between the proxy and the server.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017214
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017215 - If "Ta" is large even for small byte counts, it generally is because
17216 neither the client nor the server decides to close the connection while
17217 haproxy is running in tunnel mode and both have agreed on a keep-alive
17218 connection mode. In order to solve this issue, it will be needed to specify
17219 one of the HTTP options to manipulate keep-alive or close options on either
17220 the frontend or the backend. Having the smallest possible 'Ta' or 'Tt' is
17221 important when connection regulation is used with the "maxconn" option on
17222 the servers, since no new connection will be sent to the server until
17223 another one is released.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017224
17225Other noticeable HTTP log cases ('xx' means any value to be ignored) :
17226
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017227 TR/Tw/Tc/Tr/+Ta The "option logasap" is present on the frontend and the log
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017228 was emitted before the data phase. All the timers are valid
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017229 except "Ta" which is shorter than reality.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017230
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017231 -1/xx/xx/xx/Ta The client was not able to send a complete request in time
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017232 or it aborted too early. Check the session termination flags
17233 then "timeout http-request" and "timeout client" settings.
17234
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017235 TR/-1/xx/xx/Ta It was not possible to process the request, maybe because
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017236 servers were out of order, because the request was invalid
17237 or forbidden by ACL rules. Check the session termination
17238 flags.
17239
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017240 TR/Tw/-1/xx/Ta The connection could not establish on the server. Either it
17241 actively refused it or it timed out after Ta-(TR+Tw) ms.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017242 Check the session termination flags, then check the
17243 "timeout connect" setting. Note that the tarpit action might
17244 return similar-looking patterns, with "Tw" equal to the time
17245 the client connection was maintained open.
17246
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017247 TR/Tw/Tc/-1/Ta The server has accepted the connection but did not return
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017248 a complete response in time, or it closed its connection
Thierry FOURNIER / OZON.IO4cac3592016-07-28 17:19:45 +020017249 unexpectedly after Ta-(TR+Tw+Tc) ms. Check the session
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017250 termination flags, then check the "timeout server" setting.
17251
17252
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200172538.5. Session state at disconnection
17254-----------------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017255
17256TCP and HTTP logs provide a session termination indicator in the
17257"termination_state" field, just before the number of active connections. It is
172582-characters long in TCP mode, and is extended to 4 characters in HTTP mode,
17259each of which has a special meaning :
17260
17261 - On the first character, a code reporting the first event which caused the
17262 session to terminate :
17263
17264 C : the TCP session was unexpectedly aborted by the client.
17265
17266 S : the TCP session was unexpectedly aborted by the server, or the
17267 server explicitly refused it.
17268
17269 P : the session was prematurely aborted by the proxy, because of a
17270 connection limit enforcement, because a DENY filter was matched,
17271 because of a security check which detected and blocked a dangerous
17272 error in server response which might have caused information leak
Davor Ocelice9ed2812017-12-25 17:49:28 +010017273 (e.g. cacheable cookie).
Willy Tarreau570f2212013-06-10 16:42:09 +020017274
17275 L : the session was locally processed by haproxy and was not passed to
17276 a server. This is what happens for stats and redirects.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017277
17278 R : a resource on the proxy has been exhausted (memory, sockets, source
17279 ports, ...). Usually, this appears during the connection phase, and
17280 system logs should contain a copy of the precise error. If this
17281 happens, it must be considered as a very serious anomaly which
17282 should be fixed as soon as possible by any means.
17283
17284 I : an internal error was identified by the proxy during a self-check.
17285 This should NEVER happen, and you are encouraged to report any log
17286 containing this, because this would almost certainly be a bug. It
17287 would be wise to preventively restart the process after such an
17288 event too, in case it would be caused by memory corruption.
17289
Simon Horman752dc4a2011-06-21 14:34:59 +090017290 D : the session was killed by haproxy because the server was detected
17291 as down and was configured to kill all connections when going down.
17292
Justin Karnegeseb2c24a2012-05-24 15:28:52 -070017293 U : the session was killed by haproxy on this backup server because an
17294 active server was detected as up and was configured to kill all
17295 backup connections when going up.
17296
Willy Tarreaua2a64e92011-09-07 23:01:56 +020017297 K : the session was actively killed by an admin operating on haproxy.
17298
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017299 c : the client-side timeout expired while waiting for the client to
17300 send or receive data.
17301
17302 s : the server-side timeout expired while waiting for the server to
17303 send or receive data.
17304
17305 - : normal session completion, both the client and the server closed
17306 with nothing left in the buffers.
17307
17308 - on the second character, the TCP or HTTP session state when it was closed :
17309
Willy Tarreauf7b30a92010-12-06 22:59:17 +010017310 R : the proxy was waiting for a complete, valid REQUEST from the client
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017311 (HTTP mode only). Nothing was sent to any server.
17312
17313 Q : the proxy was waiting in the QUEUE for a connection slot. This can
17314 only happen when servers have a 'maxconn' parameter set. It can
17315 also happen in the global queue after a redispatch consecutive to
17316 a failed attempt to connect to a dying server. If no redispatch is
17317 reported, then no connection attempt was made to any server.
17318
17319 C : the proxy was waiting for the CONNECTION to establish on the
17320 server. The server might at most have noticed a connection attempt.
17321
17322 H : the proxy was waiting for complete, valid response HEADERS from the
17323 server (HTTP only).
17324
17325 D : the session was in the DATA phase.
17326
17327 L : the proxy was still transmitting LAST data to the client while the
17328 server had already finished. This one is very rare as it can only
17329 happen when the client dies while receiving the last packets.
17330
17331 T : the request was tarpitted. It has been held open with the client
17332 during the whole "timeout tarpit" duration or until the client
17333 closed, both of which will be reported in the "Tw" timer.
17334
17335 - : normal session completion after end of data transfer.
17336
17337 - the third character tells whether the persistence cookie was provided by
17338 the client (only in HTTP mode) :
17339
17340 N : the client provided NO cookie. This is usually the case for new
17341 visitors, so counting the number of occurrences of this flag in the
17342 logs generally indicate a valid trend for the site frequentation.
17343
17344 I : the client provided an INVALID cookie matching no known server.
17345 This might be caused by a recent configuration change, mixed
Cyril Bontéa8e7bbc2010-04-25 22:29:29 +020017346 cookies between HTTP/HTTPS sites, persistence conditionally
17347 ignored, or an attack.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017348
17349 D : the client provided a cookie designating a server which was DOWN,
17350 so either "option persist" was used and the client was sent to
17351 this server, or it was not set and the client was redispatched to
17352 another server.
17353
Willy Tarreau996a92c2010-10-13 19:30:47 +020017354 V : the client provided a VALID cookie, and was sent to the associated
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017355 server.
17356
Willy Tarreau996a92c2010-10-13 19:30:47 +020017357 E : the client provided a valid cookie, but with a last date which was
17358 older than what is allowed by the "maxidle" cookie parameter, so
17359 the cookie is consider EXPIRED and is ignored. The request will be
17360 redispatched just as if there was no cookie.
17361
17362 O : the client provided a valid cookie, but with a first date which was
17363 older than what is allowed by the "maxlife" cookie parameter, so
17364 the cookie is consider too OLD and is ignored. The request will be
17365 redispatched just as if there was no cookie.
17366
Willy Tarreauc89ccb62012-04-05 21:18:22 +020017367 U : a cookie was present but was not used to select the server because
17368 some other server selection mechanism was used instead (typically a
17369 "use-server" rule).
17370
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017371 - : does not apply (no cookie set in configuration).
17372
17373 - the last character reports what operations were performed on the persistence
17374 cookie returned by the server (only in HTTP mode) :
17375
17376 N : NO cookie was provided by the server, and none was inserted either.
17377
17378 I : no cookie was provided by the server, and the proxy INSERTED one.
17379 Note that in "cookie insert" mode, if the server provides a cookie,
17380 it will still be overwritten and reported as "I" here.
17381
Willy Tarreau996a92c2010-10-13 19:30:47 +020017382 U : the proxy UPDATED the last date in the cookie that was presented by
17383 the client. This can only happen in insert mode with "maxidle". It
Jarno Huuskonen0e82b922014-04-12 18:22:19 +030017384 happens every time there is activity at a different date than the
Willy Tarreau996a92c2010-10-13 19:30:47 +020017385 date indicated in the cookie. If any other change happens, such as
17386 a redispatch, then the cookie will be marked as inserted instead.
17387
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017388 P : a cookie was PROVIDED by the server and transmitted as-is.
17389
17390 R : the cookie provided by the server was REWRITTEN by the proxy, which
17391 happens in "cookie rewrite" or "cookie prefix" modes.
17392
17393 D : the cookie provided by the server was DELETED by the proxy.
17394
17395 - : does not apply (no cookie set in configuration).
17396
Willy Tarreau996a92c2010-10-13 19:30:47 +020017397The combination of the two first flags gives a lot of information about what
17398was happening when the session terminated, and why it did terminate. It can be
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017399helpful to detect server saturation, network troubles, local system resource
17400starvation, attacks, etc...
17401
17402The most common termination flags combinations are indicated below. They are
17403alphabetically sorted, with the lowercase set just after the upper case for
17404easier finding and understanding.
17405
17406 Flags Reason
17407
17408 -- Normal termination.
17409
17410 CC The client aborted before the connection could be established to the
17411 server. This can happen when haproxy tries to connect to a recently
17412 dead (or unchecked) server, and the client aborts while haproxy is
17413 waiting for the server to respond or for "timeout connect" to expire.
17414
17415 CD The client unexpectedly aborted during data transfer. This can be
17416 caused by a browser crash, by an intermediate equipment between the
17417 client and haproxy which decided to actively break the connection,
17418 by network routing issues between the client and haproxy, or by a
17419 keep-alive session between the server and the client terminated first
17420 by the client.
Willy Tarreaud72758d2010-01-12 10:42:19 +010017421
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017422 cD The client did not send nor acknowledge any data for as long as the
17423 "timeout client" delay. This is often caused by network failures on
Cyril Bontédc4d9032012-04-08 21:57:39 +020017424 the client side, or the client simply leaving the net uncleanly.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017425
17426 CH The client aborted while waiting for the server to start responding.
17427 It might be the server taking too long to respond or the client
17428 clicking the 'Stop' button too fast.
17429
17430 cH The "timeout client" stroke while waiting for client data during a
17431 POST request. This is sometimes caused by too large TCP MSS values
17432 for PPPoE networks which cannot transport full-sized packets. It can
17433 also happen when client timeout is smaller than server timeout and
17434 the server takes too long to respond.
17435
17436 CQ The client aborted while its session was queued, waiting for a server
17437 with enough empty slots to accept it. It might be that either all the
17438 servers were saturated or that the assigned server was taking too
17439 long a time to respond.
17440
17441 CR The client aborted before sending a full HTTP request. Most likely
17442 the request was typed by hand using a telnet client, and aborted
17443 too early. The HTTP status code is likely a 400 here. Sometimes this
17444 might also be caused by an IDS killing the connection between haproxy
Willy Tarreau0f228a02015-05-01 15:37:53 +020017445 and the client. "option http-ignore-probes" can be used to ignore
17446 connections without any data transfer.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017447
17448 cR The "timeout http-request" stroke before the client sent a full HTTP
17449 request. This is sometimes caused by too large TCP MSS values on the
17450 client side for PPPoE networks which cannot transport full-sized
17451 packets, or by clients sending requests by hand and not typing fast
17452 enough, or forgetting to enter the empty line at the end of the
Willy Tarreau2705a612014-05-23 17:38:34 +020017453 request. The HTTP status code is likely a 408 here. Note: recently,
Willy Tarreau0f228a02015-05-01 15:37:53 +020017454 some browsers started to implement a "pre-connect" feature consisting
17455 in speculatively connecting to some recently visited web sites just
17456 in case the user would like to visit them. This results in many
17457 connections being established to web sites, which end up in 408
17458 Request Timeout if the timeout strikes first, or 400 Bad Request when
17459 the browser decides to close them first. These ones pollute the log
17460 and feed the error counters. Some versions of some browsers have even
17461 been reported to display the error code. It is possible to work
Davor Ocelice9ed2812017-12-25 17:49:28 +010017462 around the undesirable effects of this behavior by adding "option
Willy Tarreau0f228a02015-05-01 15:37:53 +020017463 http-ignore-probes" in the frontend, resulting in connections with
17464 zero data transfer to be totally ignored. This will definitely hide
17465 the errors of people experiencing connectivity issues though.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017466
17467 CT The client aborted while its session was tarpitted. It is important to
17468 check if this happens on valid requests, in order to be sure that no
Willy Tarreau55165fe2009-05-10 12:02:55 +020017469 wrong tarpit rules have been written. If a lot of them happen, it
17470 might make sense to lower the "timeout tarpit" value to something
17471 closer to the average reported "Tw" timer, in order not to consume
17472 resources for just a few attackers.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017473
Willy Tarreau570f2212013-06-10 16:42:09 +020017474 LR The request was intercepted and locally handled by haproxy. Generally
17475 it means that this was a redirect or a stats request.
17476
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017477 SC The server or an equipment between it and haproxy explicitly refused
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017478 the TCP connection (the proxy received a TCP RST or an ICMP message
17479 in return). Under some circumstances, it can also be the network
Davor Ocelice9ed2812017-12-25 17:49:28 +010017480 stack telling the proxy that the server is unreachable (e.g. no route,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017481 or no ARP response on local network). When this happens in HTTP mode,
17482 the status code is likely a 502 or 503 here.
17483
17484 sC The "timeout connect" stroke before a connection to the server could
17485 complete. When this happens in HTTP mode, the status code is likely a
17486 503 or 504 here.
17487
17488 SD The connection to the server died with an error during the data
17489 transfer. This usually means that haproxy has received an RST from
17490 the server or an ICMP message from an intermediate equipment while
17491 exchanging data with the server. This can be caused by a server crash
17492 or by a network issue on an intermediate equipment.
17493
17494 sD The server did not send nor acknowledge any data for as long as the
17495 "timeout server" setting during the data phase. This is often caused
Davor Ocelice9ed2812017-12-25 17:49:28 +010017496 by too short timeouts on L4 equipment before the server (firewalls,
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017497 load-balancers, ...), as well as keep-alive sessions maintained
17498 between the client and the server expiring first on haproxy.
17499
17500 SH The server aborted before sending its full HTTP response headers, or
17501 it crashed while processing the request. Since a server aborting at
17502 this moment is very rare, it would be wise to inspect its logs to
17503 control whether it crashed and why. The logged request may indicate a
17504 small set of faulty requests, demonstrating bugs in the application.
17505 Sometimes this might also be caused by an IDS killing the connection
17506 between haproxy and the server.
17507
17508 sH The "timeout server" stroke before the server could return its
17509 response headers. This is the most common anomaly, indicating too
17510 long transactions, probably caused by server or database saturation.
17511 The immediate workaround consists in increasing the "timeout server"
17512 setting, but it is important to keep in mind that the user experience
17513 will suffer from these long response times. The only long term
17514 solution is to fix the application.
17515
17516 sQ The session spent too much time in queue and has been expired. See
17517 the "timeout queue" and "timeout connect" settings to find out how to
17518 fix this if it happens too often. If it often happens massively in
17519 short periods, it may indicate general problems on the affected
17520 servers due to I/O or database congestion, or saturation caused by
17521 external attacks.
17522
17523 PC The proxy refused to establish a connection to the server because the
17524 process' socket limit has been reached while attempting to connect.
Cyril Bontédc4d9032012-04-08 21:57:39 +020017525 The global "maxconn" parameter may be increased in the configuration
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017526 so that it does not happen anymore. This status is very rare and
17527 might happen when the global "ulimit-n" parameter is forced by hand.
17528
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017529 PD The proxy blocked an incorrectly formatted chunked encoded message in
17530 a request or a response, after the server has emitted its headers. In
17531 most cases, this will indicate an invalid message from the server to
Davor Ocelice9ed2812017-12-25 17:49:28 +010017532 the client. HAProxy supports chunk sizes of up to 2GB - 1 (2147483647
Willy Tarreauf3a3e132013-08-31 08:16:26 +020017533 bytes). Any larger size will be considered as an error.
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017534
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017535 PH The proxy blocked the server's response, because it was invalid,
17536 incomplete, dangerous (cache control), or matched a security filter.
17537 In any case, an HTTP 502 error is sent to the client. One possible
17538 cause for this error is an invalid syntax in an HTTP header name
Willy Tarreaued2fd2d2010-12-29 11:23:27 +010017539 containing unauthorized characters. It is also possible but quite
17540 rare, that the proxy blocked a chunked-encoding request from the
17541 client due to an invalid syntax, before the server responded. In this
17542 case, an HTTP 400 error is sent to the client and reported in the
17543 logs.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017544
17545 PR The proxy blocked the client's HTTP request, either because of an
17546 invalid HTTP syntax, in which case it returned an HTTP 400 error to
17547 the client, or because a deny filter matched, in which case it
17548 returned an HTTP 403 error.
17549
17550 PT The proxy blocked the client's request and has tarpitted its
17551 connection before returning it a 500 server error. Nothing was sent
17552 to the server. The connection was maintained open for as long as
17553 reported by the "Tw" timer field.
17554
17555 RC A local resource has been exhausted (memory, sockets, source ports)
17556 preventing the connection to the server from establishing. The error
17557 logs will tell precisely what was missing. This is very rare and can
17558 only be solved by proper system tuning.
17559
Willy Tarreau996a92c2010-10-13 19:30:47 +020017560The combination of the two last flags gives a lot of information about how
17561persistence was handled by the client, the server and by haproxy. This is very
17562important to troubleshoot disconnections, when users complain they have to
17563re-authenticate. The commonly encountered flags are :
17564
17565 -- Persistence cookie is not enabled.
17566
17567 NN No cookie was provided by the client, none was inserted in the
17568 response. For instance, this can be in insert mode with "postonly"
17569 set on a GET request.
17570
17571 II A cookie designating an invalid server was provided by the client,
17572 a valid one was inserted in the response. This typically happens when
Jamie Gloudonaaa21002012-08-25 00:18:33 -040017573 a "server" entry is removed from the configuration, since its cookie
Willy Tarreau996a92c2010-10-13 19:30:47 +020017574 value can be presented by a client when no other server knows it.
17575
17576 NI No cookie was provided by the client, one was inserted in the
17577 response. This typically happens for first requests from every user
17578 in "insert" mode, which makes it an easy way to count real users.
17579
17580 VN A cookie was provided by the client, none was inserted in the
17581 response. This happens for most responses for which the client has
17582 already got a cookie.
17583
17584 VU A cookie was provided by the client, with a last visit date which is
17585 not completely up-to-date, so an updated cookie was provided in
17586 response. This can also happen if there was no date at all, or if
17587 there was a date but the "maxidle" parameter was not set, so that the
17588 cookie can be switched to unlimited time.
17589
17590 EI A cookie was provided by the client, with a last visit date which is
17591 too old for the "maxidle" parameter, so the cookie was ignored and a
17592 new cookie was inserted in the response.
17593
17594 OI A cookie was provided by the client, with a first visit date which is
17595 too old for the "maxlife" parameter, so the cookie was ignored and a
17596 new cookie was inserted in the response.
17597
17598 DI The server designated by the cookie was down, a new server was
17599 selected and a new cookie was emitted in the response.
17600
17601 VI The server designated by the cookie was not marked dead but could not
17602 be reached. A redispatch happened and selected another one, which was
17603 then advertised in the response.
17604
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017605
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176068.6. Non-printable characters
17607-----------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017608
17609In order not to cause trouble to log analysis tools or terminals during log
17610consulting, non-printable characters are not sent as-is into log files, but are
17611converted to the two-digits hexadecimal representation of their ASCII code,
17612prefixed by the character '#'. The only characters that can be logged without
17613being escaped are comprised between 32 and 126 (inclusive). Obviously, the
17614escape character '#' itself is also encoded to avoid any ambiguity ("#23"). It
17615is the same for the character '"' which becomes "#22", as well as '{', '|' and
17616'}' when logging headers.
17617
17618Note that the space character (' ') is not encoded in headers, which can cause
17619issues for tools relying on space count to locate fields. A typical header
17620containing spaces is "User-Agent".
17621
17622Last, it has been observed that some syslog daemons such as syslog-ng escape
17623the quote ('"') with a backslash ('\'). The reverse operation can safely be
17624performed since no quote may appear anywhere else in the logs.
17625
17626
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176278.7. Capturing HTTP cookies
17628---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017629
17630Cookie capture simplifies the tracking a complete user session. This can be
17631achieved using the "capture cookie" statement in the frontend. Please refer to
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017632section 4.2 for more details. Only one cookie can be captured, and the same
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017633cookie will simultaneously be checked in the request ("Cookie:" header) and in
17634the response ("Set-Cookie:" header). The respective values will be reported in
17635the HTTP logs at the "captured_request_cookie" and "captured_response_cookie"
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017636locations (see section 8.2.3 about HTTP log format). When either cookie is
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017637not seen, a dash ('-') replaces the value. This way, it's easy to detect when a
17638user switches to a new session for example, because the server will reassign it
17639a new cookie. It is also possible to detect if a server unexpectedly sets a
17640wrong cookie to a client, leading to session crossing.
17641
17642 Examples :
17643 # capture the first cookie whose name starts with "ASPSESSION"
17644 capture cookie ASPSESSION len 32
17645
17646 # capture the first cookie whose name is exactly "vgnvisitor"
17647 capture cookie vgnvisitor= len 32
17648
17649
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200176508.8. Capturing HTTP headers
17651---------------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017652
17653Header captures are useful to track unique request identifiers set by an upper
17654proxy, virtual host names, user-agents, POST content-length, referrers, etc. In
17655the response, one can search for information about the response length, how the
17656server asked the cache to behave, or an object location during a redirection.
17657
17658Header captures are performed using the "capture request header" and "capture
17659response header" statements in the frontend. Please consult their definition in
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017660section 4.2 for more details.
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017661
17662It is possible to include both request headers and response headers at the same
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017663time. Non-existent headers are logged as empty strings, and if one header
17664appears more than once, only its last occurrence will be logged. Request headers
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017665are grouped within braces '{' and '}' in the same order as they were declared,
17666and delimited with a vertical bar '|' without any space. Response headers
17667follow the same representation, but are displayed after a space following the
17668request headers block. These blocks are displayed just before the HTTP request
17669in the logs.
17670
Willy Tarreaud9ed3d22014-06-13 12:23:06 +020017671As a special case, it is possible to specify an HTTP header capture in a TCP
17672frontend. The purpose is to enable logging of headers which will be parsed in
17673an HTTP backend if the request is then switched to this HTTP backend.
17674
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017675 Example :
17676 # This instance chains to the outgoing proxy
17677 listen proxy-out
17678 mode http
17679 option httplog
17680 option logasap
17681 log global
17682 server cache1 192.168.1.1:3128
17683
17684 # log the name of the virtual server
17685 capture request header Host len 20
17686
17687 # log the amount of data uploaded during a POST
17688 capture request header Content-Length len 10
17689
17690 # log the beginning of the referrer
17691 capture request header Referer len 20
17692
17693 # server name (useful for outgoing proxies only)
17694 capture response header Server len 20
17695
17696 # logging the content-length is useful with "option logasap"
17697 capture response header Content-Length len 10
17698
Davor Ocelice9ed2812017-12-25 17:49:28 +010017699 # log the expected cache behavior on the response
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017700 capture response header Cache-Control len 8
17701
17702 # the Via header will report the next proxy's name
17703 capture response header Via len 20
17704
17705 # log the URL location during a redirection
17706 capture response header Location len 20
17707
17708 >>> Aug 9 20:26:09 localhost \
17709 haproxy[2022]: 127.0.0.1:34014 [09/Aug/2004:20:26:09] proxy-out \
17710 proxy-out/cache1 0/0/0/162/+162 200 +350 - - ---- 0/0/0/0/0 0/0 \
17711 {fr.adserver.yahoo.co||http://fr.f416.mail.} {|864|private||} \
17712 "GET http://fr.adserver.yahoo.com/"
17713
17714 >>> Aug 9 20:30:46 localhost \
17715 haproxy[2022]: 127.0.0.1:34020 [09/Aug/2004:20:30:46] proxy-out \
17716 proxy-out/cache1 0/0/0/182/+182 200 +279 - - ---- 0/0/0/0/0 0/0 \
17717 {w.ods.org||} {Formilux/0.1.8|3495|||} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017718 "GET http://trafic.1wt.eu/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017719
17720 >>> Aug 9 20:30:46 localhost \
17721 haproxy[2022]: 127.0.0.1:34028 [09/Aug/2004:20:30:46] proxy-out \
17722 proxy-out/cache1 0/0/2/126/+128 301 +223 - - ---- 0/0/0/0/0 0/0 \
17723 {www.sytadin.equipement.gouv.fr||http://trafic.1wt.eu/} \
17724 {Apache|230|||http://www.sytadin.} \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017725 "GET http://www.sytadin.equipement.gouv.fr/ HTTP/1.1"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017726
17727
Willy Tarreauc57f0e22009-05-10 13:12:33 +0200177288.9. Examples of logs
17729---------------------
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017730
17731These are real-world examples of logs accompanied with an explanation. Some of
17732them have been made up by hand. The syslog part has been removed for better
17733reading. Their sole purpose is to explain how to decipher them.
17734
17735 >>> haproxy[674]: 127.0.0.1:33318 [15/Oct/2003:08:31:57.130] px-http \
17736 px-http/srv1 6559/0/7/147/6723 200 243 - - ---- 5/3/3/1/0 0/0 \
17737 "HEAD / HTTP/1.0"
17738
17739 => long request (6.5s) entered by hand through 'telnet'. The server replied
17740 in 147 ms, and the session ended normally ('----')
17741
17742 >>> haproxy[674]: 127.0.0.1:33319 [15/Oct/2003:08:31:57.149] px-http \
17743 px-http/srv1 6559/1230/7/147/6870 200 243 - - ---- 324/239/239/99/0 \
17744 0/9 "HEAD / HTTP/1.0"
17745
17746 => Idem, but the request was queued in the global queue behind 9 other
17747 requests, and waited there for 1230 ms.
17748
17749 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.654] px-http \
17750 px-http/srv1 9/0/7/14/+30 200 +243 - - ---- 3/3/3/1/0 0/0 \
17751 "GET /image.iso HTTP/1.0"
17752
17753 => request for a long data transfer. The "logasap" option was specified, so
Krzysztof Piotr Oledzkif8645332009-12-13 21:55:50 +010017754 the log was produced just before transferring data. The server replied in
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017755 14 ms, 243 bytes of headers were sent to the client, and total time from
17756 accept to first data byte is 30 ms.
17757
17758 >>> haproxy[674]: 127.0.0.1:33320 [15/Oct/2003:08:32:17.925] px-http \
17759 px-http/srv1 9/0/7/14/30 502 243 - - PH-- 3/2/2/0/0 0/0 \
17760 "GET /cgi-bin/bug.cgi? HTTP/1.0"
17761
17762 => the proxy blocked a server response either because of an "rspdeny" or
17763 "rspideny" filter, or because the response was improperly formatted and
Willy Tarreau3c92c5f2011-08-28 09:45:47 +020017764 not HTTP-compliant, or because it blocked sensitive information which
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017765 risked being cached. In this case, the response is replaced with a "502
17766 bad gateway". The flags ("PH--") tell us that it was haproxy who decided
17767 to return the 502 and not the server.
17768
17769 >>> haproxy[18113]: 127.0.0.1:34548 [15/Oct/2003:15:18:55.798] px-http \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017770 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 +010017771
17772 => the client never completed its request and aborted itself ("C---") after
17773 8.5s, while the proxy was waiting for the request headers ("-R--").
17774 Nothing was sent to any server.
17775
17776 >>> haproxy[18113]: 127.0.0.1:34549 [15/Oct/2003:15:19:06.103] px-http \
17777 px-http/<NOSRV> -1/-1/-1/-1/50001 408 0 - - cR-- 2/2/2/0/0 0/0 ""
17778
17779 => The client never completed its request, which was aborted by the
17780 time-out ("c---") after 50s, while the proxy was waiting for the request
Davor Ocelice9ed2812017-12-25 17:49:28 +010017781 headers ("-R--"). Nothing was sent to any server, but the proxy could
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017782 send a 408 return code to the client.
17783
17784 >>> haproxy[18989]: 127.0.0.1:34550 [15/Oct/2003:15:24:28.312] px-tcp \
17785 px-tcp/srv1 0/0/5007 0 cD 0/0/0/0/0 0/0
17786
17787 => This log was produced with "option tcplog". The client timed out after
17788 5 seconds ("c----").
17789
17790 >>> haproxy[18989]: 10.0.0.1:34552 [15/Oct/2003:15:26:31.462] px-http \
17791 px-http/srv1 3183/-1/-1/-1/11215 503 0 - - SC-- 205/202/202/115/3 \
Willy Tarreaud72758d2010-01-12 10:42:19 +010017792 0/0 "HEAD / HTTP/1.0"
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017793
17794 => The request took 3s to complete (probably a network problem), and the
Willy Tarreauc57f0e22009-05-10 13:12:33 +020017795 connection to the server failed ('SC--') after 4 attempts of 2 seconds
Willy Tarreaucc6c8912009-02-22 10:53:55 +010017796 (config says 'retries 3'), and no redispatch (otherwise we would have
17797 seen "/+3"). Status code 503 was returned to the client. There were 115
17798 connections on this server, 202 connections on this proxy, and 205 on
17799 the global process. It is possible that the server refused the
17800 connection because of too many already established.
Willy Tarreau844e3c52008-01-11 16:28:18 +010017801
Willy Tarreau52b2d222011-09-07 23:48:48 +020017802
Christopher Fauletc3fe5332016-04-07 15:30:10 +0200178039. Supported filters
17804--------------------
17805
17806Here are listed officially supported filters with the list of parameters they
17807accept. Depending on compile options, some of these filters might be
17808unavailable. The list of available filters is reported in haproxy -vv.
17809
17810See also : "filter"
17811
178129.1. Trace
17813----------
17814
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017815filter trace [name <name>] [random-parsing] [random-forwarding] [hexdump]
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017816
17817 Arguments:
17818 <name> is an arbitrary name that will be reported in
17819 messages. If no name is provided, "TRACE" is used.
17820
17821 <random-parsing> enables the random parsing of data exchanged between
17822 the client and the server. By default, this filter
17823 parses all available data. With this parameter, it
17824 only parses a random amount of the available data.
17825
Davor Ocelice9ed2812017-12-25 17:49:28 +010017826 <random-forwarding> enables the random forwarding of parsed data. By
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017827 default, this filter forwards all previously parsed
17828 data. With this parameter, it only forwards a random
17829 amount of the parsed data.
17830
Davor Ocelice9ed2812017-12-25 17:49:28 +010017831 <hexdump> dumps all forwarded data to the server and the client.
Christopher Faulet31bfe1f2016-12-09 17:42:38 +010017832
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017833This filter can be used as a base to develop new filters. It defines all
17834callbacks and print a message on the standard error stream (stderr) with useful
17835information for all of them. It may be useful to debug the activity of other
17836filters or, quite simply, HAProxy's activity.
17837
17838Using <random-parsing> and/or <random-forwarding> parameters is a good way to
17839tests the behavior of a filter that parses data exchanged between a client and
17840a server by adding some latencies in the processing.
17841
17842
178439.2. HTTP compression
17844---------------------
17845
17846filter compression
17847
17848The HTTP compression has been moved in a filter in HAProxy 1.7. "compression"
17849keyword must still be used to enable and configure the HTTP compression. And
Christopher Faulet27d93c32018-12-15 22:32:02 +010017850when no other filter is used, it is enough. When used with the cache enabled,
17851it is also enough. In this case, the compression is always done after the
17852response is stored in the cache. But it is mandatory to explicitly use a filter
17853line to enable the HTTP compression when at least one filter other than the
17854cache is used for the same listener/frontend/backend. This is important to know
17855the filters evaluation order.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017856
Christopher Faulet27d93c32018-12-15 22:32:02 +010017857See also : "compression" and section 9.4 about the cache filter.
Christopher Fauletc3fe5332016-04-07 15:30:10 +020017858
17859
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +0200178609.3. Stream Processing Offload Engine (SPOE)
17861--------------------------------------------
17862
17863filter spoe [engine <name>] config <file>
17864
17865 Arguments :
17866
17867 <name> is the engine name that will be used to find the right scope in
17868 the configuration file. If not provided, all the file will be
17869 parsed.
17870
17871 <file> is the path of the engine configuration file. This file can
17872 contain configuration of several engines. In this case, each
17873 part must be placed in its own scope.
17874
17875The Stream Processing Offload Engine (SPOE) is a filter communicating with
17876external components. It allows the offload of some specifics processing on the
Davor Ocelice9ed2812017-12-25 17:49:28 +010017877streams in tiered applications. These external components and information
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017878exchanged with them are configured in dedicated files, for the main part. It
17879also requires dedicated backends, defined in HAProxy configuration.
17880
17881SPOE communicates with external components using an in-house binary protocol,
17882the Stream Processing Offload Protocol (SPOP).
17883
Tim Düsterhus4896c442016-11-29 02:15:19 +010017884For all information about the SPOE configuration and the SPOP specification, see
Christopher Fauletf7e4e7e2016-10-27 22:29:49 +020017885"doc/SPOE.txt".
17886
17887Important note:
17888 The SPOE filter is highly experimental for now and was not heavily
17889 tested. It is really not production ready. So use it carefully.
17890
Christopher Faulet99a17a22018-12-11 09:18:27 +0100178919.4. Cache
17892----------
17893
17894filter cache <name>
17895
17896 Arguments :
17897
17898 <name> is name of the cache section this filter will use.
17899
17900The cache uses a filter to store cacheable responses. The HTTP rules
17901"cache-store" and "cache-use" must be used to define how and when to use a
17902cache. By default the correpsonding filter is implicitly defined. And when no
Christopher Faulet27d93c32018-12-15 22:32:02 +010017903other filters than cache or compression are used, it is enough. In such case,
17904the compression filter is always evaluated after the cache filter. But it is
17905mandatory to explicitly use a filter line to use a cache when at least one
17906filter other than the compression is used for the same
17907listener/frontend/backend. This is important to know the filters evaluation
17908order.
Christopher Faulet99a17a22018-12-11 09:18:27 +010017909
Christopher Faulet27d93c32018-12-15 22:32:02 +010017910See also : section 9.2 about the compression filter and section 10 about cache.
Christopher Faulet99a17a22018-12-11 09:18:27 +010017911
William Lallemand86d0df02017-11-24 21:36:45 +01001791210. Cache
17913---------
17914
17915HAProxy provides a cache, which was designed to perform cache on small objects
17916(favicon, css...). This is a minimalist low-maintenance cache which runs in
17917RAM.
17918
17919The cache is based on a memory which is shared between processes and threads,
Cyril Bonté7b888f12017-11-26 22:24:31 +010017920this memory is split in blocks of 1k.
William Lallemand86d0df02017-11-24 21:36:45 +010017921
17922If an object is not used anymore, it can be deleted to store a new object
17923independently of its expiration date. The oldest objects are deleted first
17924when we try to allocate a new one.
17925
Cyril Bonté7b888f12017-11-26 22:24:31 +010017926The cache uses a hash of the host header and the URI as the key.
William Lallemand86d0df02017-11-24 21:36:45 +010017927
17928It's possible to view the status of a cache using the Unix socket command
17929"show cache" consult section 9.3 "Unix Socket commands" of Management Guide
17930for more details.
17931
17932When an object is delivered from the cache, the server name in the log is
17933replaced by "<CACHE>".
17934
Cyril Bonté7b888f12017-11-26 22:24:31 +01001793510.1. Limitation
17936----------------
William Lallemand86d0df02017-11-24 21:36:45 +010017937
17938The cache won't store and won't deliver objects in these cases:
17939
17940- If the response is not a 200
17941- If the response contains a Vary header
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017942- If the Content-Length + the headers size is greater than "max-object-size"
William Lallemand86d0df02017-11-24 21:36:45 +010017943- If the response is not cacheable
17944
17945- If the request is not a GET
17946- If the HTTP version of the request is smaller than 1.1
William Lallemand8a16fe02018-05-22 11:04:33 +020017947- If the request contains an Authorization header
William Lallemand86d0df02017-11-24 21:36:45 +010017948
Christopher Faulet99a17a22018-12-11 09:18:27 +010017949Caution!: For HAProxy version prior to 1.9, due to the limitation of the
17950filters, it is not recommended to use the cache with other filters. Using them
17951can cause undefined behavior if they modify the response (compression for
17952example). For HAProxy 1.9 and greater, it is safe, for HTX proxies only (see
17953"option http-use-htx" for details).
William Lallemand86d0df02017-11-24 21:36:45 +010017954
Cyril Bonté7b888f12017-11-26 22:24:31 +01001795510.2. Setup
17956-----------
William Lallemand86d0df02017-11-24 21:36:45 +010017957
17958To setup a cache, you must define a cache section and use it in a proxy with
17959the corresponding http-request and response actions.
17960
Cyril Bonté7b888f12017-11-26 22:24:31 +01001796110.2.1. Cache section
17962---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017963
17964cache <name>
17965 Declare a cache section, allocate a shared cache memory named <name>, the
17966 size of cache is mandatory.
17967
17968total-max-size <megabytes>
Davor Ocelice9ed2812017-12-25 17:49:28 +010017969 Define the size in RAM of the cache in megabytes. This size is split in
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020017970 blocks of 1kB which are used by the cache entries. Its maximum value is 4095.
William Lallemand86d0df02017-11-24 21:36:45 +010017971
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017972max-object-size <bytes>
Frédéric Lécaillee3c83d82018-10-25 10:46:40 +020017973 Define the maximum size of the objects to be cached. Must not be greater than
17974 an half of "total-max-size". If not set, it equals to a 256th of the cache size.
17975 All objects with sizes larger than "max-object-size" will not be cached.
Frédéric Lécaille5f8bea62018-10-23 10:09:19 +020017976
William Lallemand86d0df02017-11-24 21:36:45 +010017977max-age <seconds>
17978 Define the maximum expiration duration. The expiration is set has the lowest
17979 value between the s-maxage or max-age (in this order) directive in the
17980 Cache-Control response header and this value. The default value is 60
17981 seconds, which means that you can't cache an object more than 60 seconds by
17982 default.
17983
Cyril Bonté7b888f12017-11-26 22:24:31 +01001798410.2.2. Proxy section
17985---------------------
William Lallemand86d0df02017-11-24 21:36:45 +010017986
Jarno Huuskonen251a6b72019-01-04 14:05:02 +020017987http-request cache-use <name> [ { if | unless } <condition> ]
William Lallemand86d0df02017-11-24 21:36:45 +010017988 Try to deliver a cached object from the cache <name>. This directive is also
17989 mandatory to store the cache as it calculates the cache hash. If you want to
17990 use a condition for both storage and delivering that's a good idea to put it
17991 after this one.
17992
Jarno Huuskonen251a6b72019-01-04 14:05:02 +020017993http-response cache-store <name> [ { if | unless } <condition> ]
William Lallemand86d0df02017-11-24 21:36:45 +010017994 Store an http-response within the cache. The storage of the response headers
17995 is done at this step, which means you can use others http-response actions
17996 to modify headers before or after the storage of the response. This action
17997 is responsible for the setup of the cache storage filter.
17998
17999
18000Example:
18001
18002 backend bck1
18003 mode http
18004
18005 http-request cache-use foobar
18006 http-response cache-store foobar
18007 server srv1 127.0.0.1:80
18008
18009 cache foobar
18010 total-max-size 4
18011 max-age 240
18012
Willy Tarreau0ba27502007-12-24 16:55:16 +010018013/*
18014 * Local variables:
18015 * fill-column: 79
18016 * End:
18017 */